{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.399936035819941,
  "eval_steps": 500,
  "global_step": 45024,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0005330348338263906,
      "grad_norm": 22.384035169953112,
      "learning_rate": 1.0660980810234541e-08,
      "loss": 0.8117,
      "step": 10
    },
    {
      "epoch": 0.0010660696676527812,
      "grad_norm": 12.56156999016183,
      "learning_rate": 2.1321961620469082e-08,
      "loss": 0.8163,
      "step": 20
    },
    {
      "epoch": 0.0015991045014791716,
      "grad_norm": 33.76307089269726,
      "learning_rate": 3.1982942430703625e-08,
      "loss": 0.7923,
      "step": 30
    },
    {
      "epoch": 0.0021321393353055623,
      "grad_norm": 16.83726744730983,
      "learning_rate": 4.2643923240938164e-08,
      "loss": 0.8066,
      "step": 40
    },
    {
      "epoch": 0.002665174169131953,
      "grad_norm": 49.634523742971666,
      "learning_rate": 5.330490405117271e-08,
      "loss": 0.7986,
      "step": 50
    },
    {
      "epoch": 0.0031982090029583432,
      "grad_norm": 13.026076869540546,
      "learning_rate": 6.396588486140725e-08,
      "loss": 0.8245,
      "step": 60
    },
    {
      "epoch": 0.003731243836784734,
      "grad_norm": 38.37983400304547,
      "learning_rate": 7.46268656716418e-08,
      "loss": 0.8151,
      "step": 70
    },
    {
      "epoch": 0.004264278670611125,
      "grad_norm": 24.292803186984795,
      "learning_rate": 8.528784648187633e-08,
      "loss": 0.8024,
      "step": 80
    },
    {
      "epoch": 0.004797313504437515,
      "grad_norm": 20.63481080073996,
      "learning_rate": 9.594882729211088e-08,
      "loss": 0.8165,
      "step": 90
    },
    {
      "epoch": 0.005330348338263906,
      "grad_norm": 8.059503536005575,
      "learning_rate": 1.0660980810234542e-07,
      "loss": 0.795,
      "step": 100
    },
    {
      "epoch": 0.005863383172090296,
      "grad_norm": 5.470117214566169,
      "learning_rate": 1.1727078891257995e-07,
      "loss": 0.7837,
      "step": 110
    },
    {
      "epoch": 0.0063964180059166865,
      "grad_norm": 67.1429137488128,
      "learning_rate": 1.279317697228145e-07,
      "loss": 0.7636,
      "step": 120
    },
    {
      "epoch": 0.006929452839743077,
      "grad_norm": 27.57553604148236,
      "learning_rate": 1.3859275053304905e-07,
      "loss": 0.7615,
      "step": 130
    },
    {
      "epoch": 0.007462487673569468,
      "grad_norm": 96.95190578421065,
      "learning_rate": 1.492537313432836e-07,
      "loss": 0.7654,
      "step": 140
    },
    {
      "epoch": 0.007995522507395858,
      "grad_norm": 6.513880444574477,
      "learning_rate": 1.5991471215351813e-07,
      "loss": 0.7314,
      "step": 150
    },
    {
      "epoch": 0.00852855734122225,
      "grad_norm": 17.34658807824645,
      "learning_rate": 1.7057569296375266e-07,
      "loss": 0.7413,
      "step": 160
    },
    {
      "epoch": 0.009061592175048639,
      "grad_norm": 4.68951448214306,
      "learning_rate": 1.812366737739872e-07,
      "loss": 0.7237,
      "step": 170
    },
    {
      "epoch": 0.00959462700887503,
      "grad_norm": 31.904806594531866,
      "learning_rate": 1.9189765458422176e-07,
      "loss": 0.6961,
      "step": 180
    },
    {
      "epoch": 0.01012766184270142,
      "grad_norm": 8.610528268372763,
      "learning_rate": 2.0255863539445632e-07,
      "loss": 0.7138,
      "step": 190
    },
    {
      "epoch": 0.010660696676527812,
      "grad_norm": 27.407492655186093,
      "learning_rate": 2.1321961620469084e-07,
      "loss": 0.7031,
      "step": 200
    },
    {
      "epoch": 0.011193731510354202,
      "grad_norm": 15.23960798334783,
      "learning_rate": 2.2388059701492537e-07,
      "loss": 0.6725,
      "step": 210
    },
    {
      "epoch": 0.011726766344180592,
      "grad_norm": 14.339840399958527,
      "learning_rate": 2.345415778251599e-07,
      "loss": 0.7094,
      "step": 220
    },
    {
      "epoch": 0.012259801178006983,
      "grad_norm": 40.257564402060794,
      "learning_rate": 2.4520255863539447e-07,
      "loss": 0.6648,
      "step": 230
    },
    {
      "epoch": 0.012792836011833373,
      "grad_norm": 38.12287684104614,
      "learning_rate": 2.55863539445629e-07,
      "loss": 0.6624,
      "step": 240
    },
    {
      "epoch": 0.013325870845659765,
      "grad_norm": 16.667179938633755,
      "learning_rate": 2.665245202558635e-07,
      "loss": 0.6578,
      "step": 250
    },
    {
      "epoch": 0.013858905679486154,
      "grad_norm": 30.69626401265156,
      "learning_rate": 2.771855010660981e-07,
      "loss": 0.6452,
      "step": 260
    },
    {
      "epoch": 0.014391940513312544,
      "grad_norm": 53.91940567739014,
      "learning_rate": 2.8784648187633263e-07,
      "loss": 0.679,
      "step": 270
    },
    {
      "epoch": 0.014924975347138936,
      "grad_norm": 6.14720521465269,
      "learning_rate": 2.985074626865672e-07,
      "loss": 0.6497,
      "step": 280
    },
    {
      "epoch": 0.015458010180965326,
      "grad_norm": 37.24542287618424,
      "learning_rate": 3.0916844349680174e-07,
      "loss": 0.6759,
      "step": 290
    },
    {
      "epoch": 0.015991045014791715,
      "grad_norm": 18.31377801368936,
      "learning_rate": 3.1982942430703626e-07,
      "loss": 0.6243,
      "step": 300
    },
    {
      "epoch": 0.016524079848618107,
      "grad_norm": 8.048409639692276,
      "learning_rate": 3.304904051172708e-07,
      "loss": 0.6416,
      "step": 310
    },
    {
      "epoch": 0.0170571146824445,
      "grad_norm": 18.265003051468547,
      "learning_rate": 3.411513859275053e-07,
      "loss": 0.6474,
      "step": 320
    },
    {
      "epoch": 0.01759014951627089,
      "grad_norm": 22.044928666366484,
      "learning_rate": 3.518123667377399e-07,
      "loss": 0.6612,
      "step": 330
    },
    {
      "epoch": 0.018123184350097278,
      "grad_norm": 6.29550754706564,
      "learning_rate": 3.624733475479744e-07,
      "loss": 0.6526,
      "step": 340
    },
    {
      "epoch": 0.01865621918392367,
      "grad_norm": 23.655194210354356,
      "learning_rate": 3.73134328358209e-07,
      "loss": 0.6335,
      "step": 350
    },
    {
      "epoch": 0.01918925401775006,
      "grad_norm": 18.457694770192315,
      "learning_rate": 3.837953091684435e-07,
      "loss": 0.6215,
      "step": 360
    },
    {
      "epoch": 0.01972228885157645,
      "grad_norm": 3.1958736725407744,
      "learning_rate": 3.9445628997867805e-07,
      "loss": 0.6317,
      "step": 370
    },
    {
      "epoch": 0.02025532368540284,
      "grad_norm": 19.973374225574325,
      "learning_rate": 4.0511727078891263e-07,
      "loss": 0.618,
      "step": 380
    },
    {
      "epoch": 0.020788358519229232,
      "grad_norm": 6.883229999182656,
      "learning_rate": 4.157782515991471e-07,
      "loss": 0.6513,
      "step": 390
    },
    {
      "epoch": 0.021321393353055624,
      "grad_norm": 34.58236951166425,
      "learning_rate": 4.264392324093817e-07,
      "loss": 0.6385,
      "step": 400
    },
    {
      "epoch": 0.021854428186882012,
      "grad_norm": 4.896875627926621,
      "learning_rate": 4.371002132196162e-07,
      "loss": 0.6121,
      "step": 410
    },
    {
      "epoch": 0.022387463020708404,
      "grad_norm": 9.571937291164867,
      "learning_rate": 4.4776119402985074e-07,
      "loss": 0.6372,
      "step": 420
    },
    {
      "epoch": 0.022920497854534795,
      "grad_norm": 7.7820482020663215,
      "learning_rate": 4.584221748400853e-07,
      "loss": 0.6205,
      "step": 430
    },
    {
      "epoch": 0.023453532688361183,
      "grad_norm": 17.09258095859951,
      "learning_rate": 4.690831556503198e-07,
      "loss": 0.6497,
      "step": 440
    },
    {
      "epoch": 0.023986567522187575,
      "grad_norm": 15.810404773206715,
      "learning_rate": 4.797441364605544e-07,
      "loss": 0.6007,
      "step": 450
    },
    {
      "epoch": 0.024519602356013966,
      "grad_norm": 2.4769181005915932,
      "learning_rate": 4.904051172707889e-07,
      "loss": 0.6212,
      "step": 460
    },
    {
      "epoch": 0.025052637189840354,
      "grad_norm": 2.7466335080586206,
      "learning_rate": 5.010660980810235e-07,
      "loss": 0.6024,
      "step": 470
    },
    {
      "epoch": 0.025585672023666746,
      "grad_norm": 7.819298025658019,
      "learning_rate": 5.11727078891258e-07,
      "loss": 0.6024,
      "step": 480
    },
    {
      "epoch": 0.026118706857493137,
      "grad_norm": 11.993606063767926,
      "learning_rate": 5.223880597014926e-07,
      "loss": 0.618,
      "step": 490
    },
    {
      "epoch": 0.02665174169131953,
      "grad_norm": 6.092448929136046,
      "learning_rate": 5.33049040511727e-07,
      "loss": 0.6181,
      "step": 500
    },
    {
      "epoch": 0.027184776525145917,
      "grad_norm": 8.52801558890943,
      "learning_rate": 5.437100213219616e-07,
      "loss": 0.608,
      "step": 510
    },
    {
      "epoch": 0.02771781135897231,
      "grad_norm": 3.4101231810925174,
      "learning_rate": 5.543710021321962e-07,
      "loss": 0.6033,
      "step": 520
    },
    {
      "epoch": 0.0282508461927987,
      "grad_norm": 17.453953167558357,
      "learning_rate": 5.650319829424307e-07,
      "loss": 0.586,
      "step": 530
    },
    {
      "epoch": 0.02878388102662509,
      "grad_norm": 3.7853039467463434,
      "learning_rate": 5.756929637526653e-07,
      "loss": 0.5761,
      "step": 540
    },
    {
      "epoch": 0.02931691586045148,
      "grad_norm": 19.26469192065809,
      "learning_rate": 5.863539445628998e-07,
      "loss": 0.6047,
      "step": 550
    },
    {
      "epoch": 0.02984995069427787,
      "grad_norm": 4.660799518343717,
      "learning_rate": 5.970149253731344e-07,
      "loss": 0.5951,
      "step": 560
    },
    {
      "epoch": 0.030382985528104263,
      "grad_norm": 11.724266047364479,
      "learning_rate": 6.076759061833688e-07,
      "loss": 0.6186,
      "step": 570
    },
    {
      "epoch": 0.03091602036193065,
      "grad_norm": 12.045147025901208,
      "learning_rate": 6.183368869936035e-07,
      "loss": 0.5995,
      "step": 580
    },
    {
      "epoch": 0.031449055195757046,
      "grad_norm": 6.611619546789754,
      "learning_rate": 6.28997867803838e-07,
      "loss": 0.5916,
      "step": 590
    },
    {
      "epoch": 0.03198209002958343,
      "grad_norm": 2.6668492445044536,
      "learning_rate": 6.396588486140725e-07,
      "loss": 0.5986,
      "step": 600
    },
    {
      "epoch": 0.03251512486340982,
      "grad_norm": 2.409130712574357,
      "learning_rate": 6.50319829424307e-07,
      "loss": 0.5943,
      "step": 610
    },
    {
      "epoch": 0.033048159697236214,
      "grad_norm": 7.3172633211536615,
      "learning_rate": 6.609808102345416e-07,
      "loss": 0.5985,
      "step": 620
    },
    {
      "epoch": 0.033581194531062605,
      "grad_norm": 2.5021776516635104,
      "learning_rate": 6.716417910447762e-07,
      "loss": 0.5688,
      "step": 630
    },
    {
      "epoch": 0.034114229364889,
      "grad_norm": 3.092066217506789,
      "learning_rate": 6.823027718550106e-07,
      "loss": 0.5968,
      "step": 640
    },
    {
      "epoch": 0.03464726419871539,
      "grad_norm": 2.5459200872949483,
      "learning_rate": 6.929637526652453e-07,
      "loss": 0.5908,
      "step": 650
    },
    {
      "epoch": 0.03518029903254178,
      "grad_norm": 6.494246936455836,
      "learning_rate": 7.036247334754798e-07,
      "loss": 0.5744,
      "step": 660
    },
    {
      "epoch": 0.035713333866368165,
      "grad_norm": 4.162093112846186,
      "learning_rate": 7.142857142857143e-07,
      "loss": 0.5924,
      "step": 670
    },
    {
      "epoch": 0.036246368700194556,
      "grad_norm": 6.283815107667488,
      "learning_rate": 7.249466950959488e-07,
      "loss": 0.5918,
      "step": 680
    },
    {
      "epoch": 0.03677940353402095,
      "grad_norm": 13.747666320600384,
      "learning_rate": 7.356076759061834e-07,
      "loss": 0.5875,
      "step": 690
    },
    {
      "epoch": 0.03731243836784734,
      "grad_norm": 6.495457820303132,
      "learning_rate": 7.46268656716418e-07,
      "loss": 0.5655,
      "step": 700
    },
    {
      "epoch": 0.03784547320167373,
      "grad_norm": 2.6123117632996795,
      "learning_rate": 7.569296375266524e-07,
      "loss": 0.5656,
      "step": 710
    },
    {
      "epoch": 0.03837850803550012,
      "grad_norm": 2.169732322924637,
      "learning_rate": 7.67590618336887e-07,
      "loss": 0.5554,
      "step": 720
    },
    {
      "epoch": 0.038911542869326514,
      "grad_norm": 2.3438244121416956,
      "learning_rate": 7.782515991471216e-07,
      "loss": 0.5791,
      "step": 730
    },
    {
      "epoch": 0.0394445777031529,
      "grad_norm": 2.6726563129303824,
      "learning_rate": 7.889125799573561e-07,
      "loss": 0.575,
      "step": 740
    },
    {
      "epoch": 0.03997761253697929,
      "grad_norm": 2.8732249315321723,
      "learning_rate": 7.995735607675906e-07,
      "loss": 0.5628,
      "step": 750
    },
    {
      "epoch": 0.04051064737080568,
      "grad_norm": 1.90982824669672,
      "learning_rate": 8.102345415778253e-07,
      "loss": 0.5526,
      "step": 760
    },
    {
      "epoch": 0.04104368220463207,
      "grad_norm": 6.766200067447573,
      "learning_rate": 8.208955223880597e-07,
      "loss": 0.572,
      "step": 770
    },
    {
      "epoch": 0.041576717038458465,
      "grad_norm": 9.412682209911177,
      "learning_rate": 8.315565031982942e-07,
      "loss": 0.5702,
      "step": 780
    },
    {
      "epoch": 0.042109751872284856,
      "grad_norm": 6.025336896305503,
      "learning_rate": 8.422174840085288e-07,
      "loss": 0.5465,
      "step": 790
    },
    {
      "epoch": 0.04264278670611125,
      "grad_norm": 1.9706615562451428,
      "learning_rate": 8.528784648187634e-07,
      "loss": 0.5568,
      "step": 800
    },
    {
      "epoch": 0.04317582153993763,
      "grad_norm": 2.939411355491427,
      "learning_rate": 8.635394456289978e-07,
      "loss": 0.5502,
      "step": 810
    },
    {
      "epoch": 0.043708856373764024,
      "grad_norm": 5.142970989416449,
      "learning_rate": 8.742004264392324e-07,
      "loss": 0.5517,
      "step": 820
    },
    {
      "epoch": 0.044241891207590416,
      "grad_norm": 3.053201788230296,
      "learning_rate": 8.84861407249467e-07,
      "loss": 0.5663,
      "step": 830
    },
    {
      "epoch": 0.04477492604141681,
      "grad_norm": 1.8449978083615863,
      "learning_rate": 8.955223880597015e-07,
      "loss": 0.5371,
      "step": 840
    },
    {
      "epoch": 0.0453079608752432,
      "grad_norm": 3.952962229474722,
      "learning_rate": 9.06183368869936e-07,
      "loss": 0.5518,
      "step": 850
    },
    {
      "epoch": 0.04584099570906959,
      "grad_norm": 2.3114170550600517,
      "learning_rate": 9.168443496801706e-07,
      "loss": 0.5339,
      "step": 860
    },
    {
      "epoch": 0.046374030542895975,
      "grad_norm": 2.2912760175353872,
      "learning_rate": 9.275053304904052e-07,
      "loss": 0.5326,
      "step": 870
    },
    {
      "epoch": 0.046907065376722366,
      "grad_norm": 4.298637664764211,
      "learning_rate": 9.381663113006396e-07,
      "loss": 0.5602,
      "step": 880
    },
    {
      "epoch": 0.04744010021054876,
      "grad_norm": 2.208073245907923,
      "learning_rate": 9.488272921108742e-07,
      "loss": 0.5555,
      "step": 890
    },
    {
      "epoch": 0.04797313504437515,
      "grad_norm": 2.0989454369450793,
      "learning_rate": 9.594882729211088e-07,
      "loss": 0.5431,
      "step": 900
    },
    {
      "epoch": 0.04850616987820154,
      "grad_norm": 4.868043085397458,
      "learning_rate": 9.701492537313434e-07,
      "loss": 0.5554,
      "step": 910
    },
    {
      "epoch": 0.04903920471202793,
      "grad_norm": 2.163101936478925,
      "learning_rate": 9.808102345415779e-07,
      "loss": 0.5575,
      "step": 920
    },
    {
      "epoch": 0.049572239545854324,
      "grad_norm": 2.5780084223910076,
      "learning_rate": 9.914712153518124e-07,
      "loss": 0.537,
      "step": 930
    },
    {
      "epoch": 0.05010527437968071,
      "grad_norm": 7.951366322245886,
      "learning_rate": 1.002132196162047e-06,
      "loss": 0.5213,
      "step": 940
    },
    {
      "epoch": 0.0506383092135071,
      "grad_norm": 4.603313864051097,
      "learning_rate": 1.0127931769722815e-06,
      "loss": 0.5477,
      "step": 950
    },
    {
      "epoch": 0.05117134404733349,
      "grad_norm": 2.8636243672532316,
      "learning_rate": 1.023454157782516e-06,
      "loss": 0.5342,
      "step": 960
    },
    {
      "epoch": 0.05170437888115988,
      "grad_norm": 3.0321492345169703,
      "learning_rate": 1.0341151385927505e-06,
      "loss": 0.55,
      "step": 970
    },
    {
      "epoch": 0.052237413714986275,
      "grad_norm": 9.474189998020558,
      "learning_rate": 1.0447761194029853e-06,
      "loss": 0.5481,
      "step": 980
    },
    {
      "epoch": 0.052770448548812667,
      "grad_norm": 2.8402495149841402,
      "learning_rate": 1.0554371002132196e-06,
      "loss": 0.5358,
      "step": 990
    },
    {
      "epoch": 0.05330348338263906,
      "grad_norm": 2.83582514386478,
      "learning_rate": 1.066098081023454e-06,
      "loss": 0.522,
      "step": 1000
    },
    {
      "epoch": 0.05383651821646544,
      "grad_norm": 2.8322815447404937,
      "learning_rate": 1.0767590618336888e-06,
      "loss": 0.5339,
      "step": 1010
    },
    {
      "epoch": 0.054369553050291834,
      "grad_norm": 2.673931322281884,
      "learning_rate": 1.0874200426439232e-06,
      "loss": 0.5337,
      "step": 1020
    },
    {
      "epoch": 0.054902587884118226,
      "grad_norm": 1.984491606409298,
      "learning_rate": 1.0980810234541577e-06,
      "loss": 0.5238,
      "step": 1030
    },
    {
      "epoch": 0.05543562271794462,
      "grad_norm": 2.0699566932848374,
      "learning_rate": 1.1087420042643924e-06,
      "loss": 0.5281,
      "step": 1040
    },
    {
      "epoch": 0.05596865755177101,
      "grad_norm": 2.152694343376385,
      "learning_rate": 1.119402985074627e-06,
      "loss": 0.5291,
      "step": 1050
    },
    {
      "epoch": 0.0565016923855974,
      "grad_norm": 2.3181371881447985,
      "learning_rate": 1.1300639658848615e-06,
      "loss": 0.5341,
      "step": 1060
    },
    {
      "epoch": 0.05703472721942379,
      "grad_norm": 2.9679916174369407,
      "learning_rate": 1.140724946695096e-06,
      "loss": 0.5279,
      "step": 1070
    },
    {
      "epoch": 0.05756776205325018,
      "grad_norm": 1.6467045554688482,
      "learning_rate": 1.1513859275053305e-06,
      "loss": 0.5259,
      "step": 1080
    },
    {
      "epoch": 0.05810079688707657,
      "grad_norm": 2.9173913321251193,
      "learning_rate": 1.162046908315565e-06,
      "loss": 0.5326,
      "step": 1090
    },
    {
      "epoch": 0.05863383172090296,
      "grad_norm": 1.9327154653150131,
      "learning_rate": 1.1727078891257996e-06,
      "loss": 0.5444,
      "step": 1100
    },
    {
      "epoch": 0.05916686655472935,
      "grad_norm": 2.053579315100549,
      "learning_rate": 1.183368869936034e-06,
      "loss": 0.5328,
      "step": 1110
    },
    {
      "epoch": 0.05969990138855574,
      "grad_norm": 1.8406327414695598,
      "learning_rate": 1.1940298507462688e-06,
      "loss": 0.5179,
      "step": 1120
    },
    {
      "epoch": 0.060232936222382134,
      "grad_norm": 2.3056097145291674,
      "learning_rate": 1.2046908315565032e-06,
      "loss": 0.5204,
      "step": 1130
    },
    {
      "epoch": 0.060765971056208526,
      "grad_norm": 2.896462098442556,
      "learning_rate": 1.2153518123667377e-06,
      "loss": 0.5165,
      "step": 1140
    },
    {
      "epoch": 0.06129900589003491,
      "grad_norm": 2.789749420612799,
      "learning_rate": 1.2260127931769724e-06,
      "loss": 0.5238,
      "step": 1150
    },
    {
      "epoch": 0.0618320407238613,
      "grad_norm": 1.8478165324546423,
      "learning_rate": 1.236673773987207e-06,
      "loss": 0.5179,
      "step": 1160
    },
    {
      "epoch": 0.062365075557687694,
      "grad_norm": 1.8375243417739309,
      "learning_rate": 1.2473347547974413e-06,
      "loss": 0.5284,
      "step": 1170
    },
    {
      "epoch": 0.06289811039151409,
      "grad_norm": 1.9052439579336735,
      "learning_rate": 1.257995735607676e-06,
      "loss": 0.5258,
      "step": 1180
    },
    {
      "epoch": 0.06343114522534048,
      "grad_norm": 5.901778710118485,
      "learning_rate": 1.2686567164179105e-06,
      "loss": 0.5316,
      "step": 1190
    },
    {
      "epoch": 0.06396418005916686,
      "grad_norm": 7.426400086233918,
      "learning_rate": 1.279317697228145e-06,
      "loss": 0.5041,
      "step": 1200
    },
    {
      "epoch": 0.06449721489299326,
      "grad_norm": 1.7416422626181,
      "learning_rate": 1.2899786780383796e-06,
      "loss": 0.5388,
      "step": 1210
    },
    {
      "epoch": 0.06503024972681964,
      "grad_norm": 5.260557965260364,
      "learning_rate": 1.300639658848614e-06,
      "loss": 0.5127,
      "step": 1220
    },
    {
      "epoch": 0.06556328456064604,
      "grad_norm": 2.038127736521098,
      "learning_rate": 1.3113006396588486e-06,
      "loss": 0.5139,
      "step": 1230
    },
    {
      "epoch": 0.06609631939447243,
      "grad_norm": 1.8823373915151784,
      "learning_rate": 1.3219616204690832e-06,
      "loss": 0.5215,
      "step": 1240
    },
    {
      "epoch": 0.06662935422829883,
      "grad_norm": 2.15503695451607,
      "learning_rate": 1.3326226012793177e-06,
      "loss": 0.5199,
      "step": 1250
    },
    {
      "epoch": 0.06716238906212521,
      "grad_norm": 2.1383222903150276,
      "learning_rate": 1.3432835820895524e-06,
      "loss": 0.5015,
      "step": 1260
    },
    {
      "epoch": 0.0676954238959516,
      "grad_norm": 2.2511503739891987,
      "learning_rate": 1.3539445628997867e-06,
      "loss": 0.519,
      "step": 1270
    },
    {
      "epoch": 0.068228458729778,
      "grad_norm": 1.917158848873185,
      "learning_rate": 1.3646055437100213e-06,
      "loss": 0.5136,
      "step": 1280
    },
    {
      "epoch": 0.06876149356360438,
      "grad_norm": 2.1325053191679006,
      "learning_rate": 1.375266524520256e-06,
      "loss": 0.5372,
      "step": 1290
    },
    {
      "epoch": 0.06929452839743078,
      "grad_norm": 2.4772885935630407,
      "learning_rate": 1.3859275053304905e-06,
      "loss": 0.5194,
      "step": 1300
    },
    {
      "epoch": 0.06982756323125716,
      "grad_norm": 1.8973258351709514,
      "learning_rate": 1.3965884861407248e-06,
      "loss": 0.531,
      "step": 1310
    },
    {
      "epoch": 0.07036059806508356,
      "grad_norm": 1.9395058463946968,
      "learning_rate": 1.4072494669509596e-06,
      "loss": 0.5183,
      "step": 1320
    },
    {
      "epoch": 0.07089363289890994,
      "grad_norm": 2.011140162870199,
      "learning_rate": 1.417910447761194e-06,
      "loss": 0.4906,
      "step": 1330
    },
    {
      "epoch": 0.07142666773273633,
      "grad_norm": 2.365711885922838,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.5066,
      "step": 1340
    },
    {
      "epoch": 0.07195970256656273,
      "grad_norm": 2.3177942824451674,
      "learning_rate": 1.4392324093816632e-06,
      "loss": 0.5173,
      "step": 1350
    },
    {
      "epoch": 0.07249273740038911,
      "grad_norm": 1.7555752174395958,
      "learning_rate": 1.4498933901918977e-06,
      "loss": 0.5141,
      "step": 1360
    },
    {
      "epoch": 0.07302577223421551,
      "grad_norm": 1.889980839900935,
      "learning_rate": 1.4605543710021322e-06,
      "loss": 0.5027,
      "step": 1370
    },
    {
      "epoch": 0.0735588070680419,
      "grad_norm": 1.7582289396362074,
      "learning_rate": 1.4712153518123667e-06,
      "loss": 0.5276,
      "step": 1380
    },
    {
      "epoch": 0.0740918419018683,
      "grad_norm": 1.7602655151027362,
      "learning_rate": 1.4818763326226013e-06,
      "loss": 0.5158,
      "step": 1390
    },
    {
      "epoch": 0.07462487673569468,
      "grad_norm": 1.8001183881260645,
      "learning_rate": 1.492537313432836e-06,
      "loss": 0.4986,
      "step": 1400
    },
    {
      "epoch": 0.07515791156952106,
      "grad_norm": 2.0725353366183694,
      "learning_rate": 1.5031982942430703e-06,
      "loss": 0.5217,
      "step": 1410
    },
    {
      "epoch": 0.07569094640334746,
      "grad_norm": 1.7278480431234928,
      "learning_rate": 1.5138592750533048e-06,
      "loss": 0.5215,
      "step": 1420
    },
    {
      "epoch": 0.07622398123717385,
      "grad_norm": 1.9474652366857994,
      "learning_rate": 1.5245202558635396e-06,
      "loss": 0.517,
      "step": 1430
    },
    {
      "epoch": 0.07675701607100024,
      "grad_norm": 2.1422897034566413,
      "learning_rate": 1.535181236673774e-06,
      "loss": 0.5026,
      "step": 1440
    },
    {
      "epoch": 0.07729005090482663,
      "grad_norm": 1.7905462109235504,
      "learning_rate": 1.5458422174840086e-06,
      "loss": 0.5119,
      "step": 1450
    },
    {
      "epoch": 0.07782308573865303,
      "grad_norm": 2.493897087442035,
      "learning_rate": 1.5565031982942432e-06,
      "loss": 0.5165,
      "step": 1460
    },
    {
      "epoch": 0.07835612057247941,
      "grad_norm": 1.8945720968238624,
      "learning_rate": 1.5671641791044775e-06,
      "loss": 0.5251,
      "step": 1470
    },
    {
      "epoch": 0.0788891554063058,
      "grad_norm": 1.9361905130988597,
      "learning_rate": 1.5778251599147122e-06,
      "loss": 0.5119,
      "step": 1480
    },
    {
      "epoch": 0.0794221902401322,
      "grad_norm": 1.6540771960472516,
      "learning_rate": 1.5884861407249467e-06,
      "loss": 0.5132,
      "step": 1490
    },
    {
      "epoch": 0.07995522507395858,
      "grad_norm": 2.0366997500257087,
      "learning_rate": 1.5991471215351813e-06,
      "loss": 0.4875,
      "step": 1500
    },
    {
      "epoch": 0.08048825990778498,
      "grad_norm": 2.481201500484451,
      "learning_rate": 1.6098081023454158e-06,
      "loss": 0.5168,
      "step": 1510
    },
    {
      "epoch": 0.08102129474161136,
      "grad_norm": 2.843400113923369,
      "learning_rate": 1.6204690831556505e-06,
      "loss": 0.4963,
      "step": 1520
    },
    {
      "epoch": 0.08155432957543776,
      "grad_norm": 1.7815166397160784,
      "learning_rate": 1.631130063965885e-06,
      "loss": 0.524,
      "step": 1530
    },
    {
      "epoch": 0.08208736440926415,
      "grad_norm": 2.1943964965228533,
      "learning_rate": 1.6417910447761194e-06,
      "loss": 0.5223,
      "step": 1540
    },
    {
      "epoch": 0.08262039924309053,
      "grad_norm": 1.7228556034029094,
      "learning_rate": 1.6524520255863539e-06,
      "loss": 0.5189,
      "step": 1550
    },
    {
      "epoch": 0.08315343407691693,
      "grad_norm": 3.2132263175111073,
      "learning_rate": 1.6631130063965884e-06,
      "loss": 0.4928,
      "step": 1560
    },
    {
      "epoch": 0.08368646891074331,
      "grad_norm": 3.2032275982706997,
      "learning_rate": 1.6737739872068232e-06,
      "loss": 0.5154,
      "step": 1570
    },
    {
      "epoch": 0.08421950374456971,
      "grad_norm": 1.783161362038307,
      "learning_rate": 1.6844349680170577e-06,
      "loss": 0.5008,
      "step": 1580
    },
    {
      "epoch": 0.0847525385783961,
      "grad_norm": 2.776910027467588,
      "learning_rate": 1.6950959488272922e-06,
      "loss": 0.5113,
      "step": 1590
    },
    {
      "epoch": 0.0852855734122225,
      "grad_norm": 4.011219424278999,
      "learning_rate": 1.7057569296375267e-06,
      "loss": 0.5123,
      "step": 1600
    },
    {
      "epoch": 0.08581860824604888,
      "grad_norm": 1.8879771864726966,
      "learning_rate": 1.716417910447761e-06,
      "loss": 0.5084,
      "step": 1610
    },
    {
      "epoch": 0.08635164307987526,
      "grad_norm": 1.604188895583308,
      "learning_rate": 1.7270788912579956e-06,
      "loss": 0.4971,
      "step": 1620
    },
    {
      "epoch": 0.08688467791370166,
      "grad_norm": 3.4755012148328883,
      "learning_rate": 1.7377398720682303e-06,
      "loss": 0.5014,
      "step": 1630
    },
    {
      "epoch": 0.08741771274752805,
      "grad_norm": 1.7670914630278758,
      "learning_rate": 1.7484008528784648e-06,
      "loss": 0.4961,
      "step": 1640
    },
    {
      "epoch": 0.08795074758135445,
      "grad_norm": 1.8279328976983698,
      "learning_rate": 1.7590618336886994e-06,
      "loss": 0.5114,
      "step": 1650
    },
    {
      "epoch": 0.08848378241518083,
      "grad_norm": 6.293161998539346,
      "learning_rate": 1.769722814498934e-06,
      "loss": 0.4984,
      "step": 1660
    },
    {
      "epoch": 0.08901681724900722,
      "grad_norm": 1.597910745986726,
      "learning_rate": 1.7803837953091686e-06,
      "loss": 0.5396,
      "step": 1670
    },
    {
      "epoch": 0.08954985208283361,
      "grad_norm": 1.6653635349148304,
      "learning_rate": 1.791044776119403e-06,
      "loss": 0.5303,
      "step": 1680
    },
    {
      "epoch": 0.09008288691666,
      "grad_norm": 1.4482133951906653,
      "learning_rate": 1.8017057569296375e-06,
      "loss": 0.5043,
      "step": 1690
    },
    {
      "epoch": 0.0906159217504864,
      "grad_norm": 1.7349298164399845,
      "learning_rate": 1.812366737739872e-06,
      "loss": 0.4874,
      "step": 1700
    },
    {
      "epoch": 0.09114895658431278,
      "grad_norm": 1.6070828987136645,
      "learning_rate": 1.8230277185501067e-06,
      "loss": 0.4832,
      "step": 1710
    },
    {
      "epoch": 0.09168199141813918,
      "grad_norm": 1.9743951285284194,
      "learning_rate": 1.8336886993603413e-06,
      "loss": 0.4964,
      "step": 1720
    },
    {
      "epoch": 0.09221502625196557,
      "grad_norm": 1.717799481918251,
      "learning_rate": 1.8443496801705758e-06,
      "loss": 0.5089,
      "step": 1730
    },
    {
      "epoch": 0.09274806108579195,
      "grad_norm": 1.9356727490694972,
      "learning_rate": 1.8550106609808103e-06,
      "loss": 0.5109,
      "step": 1740
    },
    {
      "epoch": 0.09328109591961835,
      "grad_norm": 1.9580272645326764,
      "learning_rate": 1.8656716417910446e-06,
      "loss": 0.4917,
      "step": 1750
    },
    {
      "epoch": 0.09381413075344473,
      "grad_norm": 2.727933294298909,
      "learning_rate": 1.8763326226012792e-06,
      "loss": 0.5025,
      "step": 1760
    },
    {
      "epoch": 0.09434716558727113,
      "grad_norm": 1.8659027091038973,
      "learning_rate": 1.8869936034115139e-06,
      "loss": 0.4967,
      "step": 1770
    },
    {
      "epoch": 0.09488020042109752,
      "grad_norm": 4.100730576838372,
      "learning_rate": 1.8976545842217484e-06,
      "loss": 0.4976,
      "step": 1780
    },
    {
      "epoch": 0.09541323525492391,
      "grad_norm": 1.7911168265329525,
      "learning_rate": 1.908315565031983e-06,
      "loss": 0.5013,
      "step": 1790
    },
    {
      "epoch": 0.0959462700887503,
      "grad_norm": 1.7463986183897582,
      "learning_rate": 1.9189765458422177e-06,
      "loss": 0.4761,
      "step": 1800
    },
    {
      "epoch": 0.09647930492257668,
      "grad_norm": 2.082040030487041,
      "learning_rate": 1.929637526652452e-06,
      "loss": 0.4984,
      "step": 1810
    },
    {
      "epoch": 0.09701233975640308,
      "grad_norm": 3.405347729207966,
      "learning_rate": 1.9402985074626867e-06,
      "loss": 0.4988,
      "step": 1820
    },
    {
      "epoch": 0.09754537459022947,
      "grad_norm": 2.0886731520776167,
      "learning_rate": 1.950959488272921e-06,
      "loss": 0.5064,
      "step": 1830
    },
    {
      "epoch": 0.09807840942405587,
      "grad_norm": 2.051895534884051,
      "learning_rate": 1.9616204690831558e-06,
      "loss": 0.4901,
      "step": 1840
    },
    {
      "epoch": 0.09861144425788225,
      "grad_norm": 1.709668302914341,
      "learning_rate": 1.9722814498933903e-06,
      "loss": 0.4918,
      "step": 1850
    },
    {
      "epoch": 0.09914447909170865,
      "grad_norm": 1.7128961914267717,
      "learning_rate": 1.982942430703625e-06,
      "loss": 0.5093,
      "step": 1860
    },
    {
      "epoch": 0.09967751392553503,
      "grad_norm": 1.6581351982370527,
      "learning_rate": 1.9936034115138594e-06,
      "loss": 0.513,
      "step": 1870
    },
    {
      "epoch": 0.10021054875936142,
      "grad_norm": 1.6911304413453174,
      "learning_rate": 2.004264392324094e-06,
      "loss": 0.4966,
      "step": 1880
    },
    {
      "epoch": 0.10074358359318782,
      "grad_norm": 1.9285274642420187,
      "learning_rate": 2.0149253731343284e-06,
      "loss": 0.4908,
      "step": 1890
    },
    {
      "epoch": 0.1012766184270142,
      "grad_norm": 1.6229726939701612,
      "learning_rate": 2.025586353944563e-06,
      "loss": 0.4919,
      "step": 1900
    },
    {
      "epoch": 0.1018096532608406,
      "grad_norm": 1.6722981844499845,
      "learning_rate": 2.0362473347547975e-06,
      "loss": 0.5038,
      "step": 1910
    },
    {
      "epoch": 0.10234268809466698,
      "grad_norm": 1.8525929467338784,
      "learning_rate": 2.046908315565032e-06,
      "loss": 0.4851,
      "step": 1920
    },
    {
      "epoch": 0.10287572292849338,
      "grad_norm": 1.7651203389199857,
      "learning_rate": 2.0575692963752665e-06,
      "loss": 0.4978,
      "step": 1930
    },
    {
      "epoch": 0.10340875776231977,
      "grad_norm": 3.0399113460745153,
      "learning_rate": 2.068230277185501e-06,
      "loss": 0.5093,
      "step": 1940
    },
    {
      "epoch": 0.10394179259614615,
      "grad_norm": 1.694188069109413,
      "learning_rate": 2.0788912579957356e-06,
      "loss": 0.4852,
      "step": 1950
    },
    {
      "epoch": 0.10447482742997255,
      "grad_norm": 1.4749248767326193,
      "learning_rate": 2.0895522388059705e-06,
      "loss": 0.4991,
      "step": 1960
    },
    {
      "epoch": 0.10500786226379893,
      "grad_norm": 1.6545205597942974,
      "learning_rate": 2.1002132196162046e-06,
      "loss": 0.5046,
      "step": 1970
    },
    {
      "epoch": 0.10554089709762533,
      "grad_norm": 1.789844021052239,
      "learning_rate": 2.110874200426439e-06,
      "loss": 0.4856,
      "step": 1980
    },
    {
      "epoch": 0.10607393193145172,
      "grad_norm": 2.389747097340645,
      "learning_rate": 2.1215351812366737e-06,
      "loss": 0.492,
      "step": 1990
    },
    {
      "epoch": 0.10660696676527812,
      "grad_norm": 1.9999702735537674,
      "learning_rate": 2.132196162046908e-06,
      "loss": 0.4776,
      "step": 2000
    },
    {
      "epoch": 0.1071400015991045,
      "grad_norm": 1.708572902721768,
      "learning_rate": 2.142857142857143e-06,
      "loss": 0.5082,
      "step": 2010
    },
    {
      "epoch": 0.10767303643293089,
      "grad_norm": 1.6616552169351602,
      "learning_rate": 2.1535181236673777e-06,
      "loss": 0.4991,
      "step": 2020
    },
    {
      "epoch": 0.10820607126675728,
      "grad_norm": 1.9961949316657603,
      "learning_rate": 2.164179104477612e-06,
      "loss": 0.4916,
      "step": 2030
    },
    {
      "epoch": 0.10873910610058367,
      "grad_norm": 1.947126393495355,
      "learning_rate": 2.1748400852878463e-06,
      "loss": 0.5075,
      "step": 2040
    },
    {
      "epoch": 0.10927214093441007,
      "grad_norm": 2.1974453403429144,
      "learning_rate": 2.185501066098081e-06,
      "loss": 0.477,
      "step": 2050
    },
    {
      "epoch": 0.10980517576823645,
      "grad_norm": 2.831922532150069,
      "learning_rate": 2.1961620469083154e-06,
      "loss": 0.4886,
      "step": 2060
    },
    {
      "epoch": 0.11033821060206285,
      "grad_norm": 1.684708827472965,
      "learning_rate": 2.2068230277185503e-06,
      "loss": 0.502,
      "step": 2070
    },
    {
      "epoch": 0.11087124543588923,
      "grad_norm": 2.0331477351606724,
      "learning_rate": 2.217484008528785e-06,
      "loss": 0.4804,
      "step": 2080
    },
    {
      "epoch": 0.11140428026971562,
      "grad_norm": 1.717450101243865,
      "learning_rate": 2.2281449893390194e-06,
      "loss": 0.4947,
      "step": 2090
    },
    {
      "epoch": 0.11193731510354202,
      "grad_norm": 1.8174981757295043,
      "learning_rate": 2.238805970149254e-06,
      "loss": 0.4929,
      "step": 2100
    },
    {
      "epoch": 0.1124703499373684,
      "grad_norm": 1.6889988447513715,
      "learning_rate": 2.249466950959488e-06,
      "loss": 0.5005,
      "step": 2110
    },
    {
      "epoch": 0.1130033847711948,
      "grad_norm": 1.6480565960509483,
      "learning_rate": 2.260127931769723e-06,
      "loss": 0.4948,
      "step": 2120
    },
    {
      "epoch": 0.11353641960502119,
      "grad_norm": 1.6952436618404707,
      "learning_rate": 2.2707889125799575e-06,
      "loss": 0.5001,
      "step": 2130
    },
    {
      "epoch": 0.11406945443884758,
      "grad_norm": 1.5994581672256405,
      "learning_rate": 2.281449893390192e-06,
      "loss": 0.4866,
      "step": 2140
    },
    {
      "epoch": 0.11460248927267397,
      "grad_norm": 1.714526187540524,
      "learning_rate": 2.2921108742004265e-06,
      "loss": 0.4797,
      "step": 2150
    },
    {
      "epoch": 0.11513552410650035,
      "grad_norm": 1.721179665627559,
      "learning_rate": 2.302771855010661e-06,
      "loss": 0.5066,
      "step": 2160
    },
    {
      "epoch": 0.11566855894032675,
      "grad_norm": 1.639833643560835,
      "learning_rate": 2.3134328358208956e-06,
      "loss": 0.4781,
      "step": 2170
    },
    {
      "epoch": 0.11620159377415314,
      "grad_norm": 1.690239965307617,
      "learning_rate": 2.32409381663113e-06,
      "loss": 0.4688,
      "step": 2180
    },
    {
      "epoch": 0.11673462860797953,
      "grad_norm": 1.8205864495338293,
      "learning_rate": 2.3347547974413646e-06,
      "loss": 0.4897,
      "step": 2190
    },
    {
      "epoch": 0.11726766344180592,
      "grad_norm": 2.0410830299330716,
      "learning_rate": 2.345415778251599e-06,
      "loss": 0.4926,
      "step": 2200
    },
    {
      "epoch": 0.11780069827563232,
      "grad_norm": 2.3481204247479006,
      "learning_rate": 2.3560767590618337e-06,
      "loss": 0.4907,
      "step": 2210
    },
    {
      "epoch": 0.1183337331094587,
      "grad_norm": 3.3565284556836095,
      "learning_rate": 2.366737739872068e-06,
      "loss": 0.4942,
      "step": 2220
    },
    {
      "epoch": 0.11886676794328509,
      "grad_norm": 1.765724500598554,
      "learning_rate": 2.3773987206823027e-06,
      "loss": 0.4559,
      "step": 2230
    },
    {
      "epoch": 0.11939980277711149,
      "grad_norm": 1.6685529628290774,
      "learning_rate": 2.3880597014925377e-06,
      "loss": 0.4796,
      "step": 2240
    },
    {
      "epoch": 0.11993283761093787,
      "grad_norm": 2.025184489439367,
      "learning_rate": 2.3987206823027718e-06,
      "loss": 0.4974,
      "step": 2250
    },
    {
      "epoch": 0.12046587244476427,
      "grad_norm": 1.5877609639194825,
      "learning_rate": 2.4093816631130063e-06,
      "loss": 0.4862,
      "step": 2260
    },
    {
      "epoch": 0.12099890727859065,
      "grad_norm": 1.75789703015658,
      "learning_rate": 2.420042643923241e-06,
      "loss": 0.4758,
      "step": 2270
    },
    {
      "epoch": 0.12153194211241705,
      "grad_norm": 1.6715829585184263,
      "learning_rate": 2.4307036247334754e-06,
      "loss": 0.4875,
      "step": 2280
    },
    {
      "epoch": 0.12206497694624344,
      "grad_norm": 1.7692896856499511,
      "learning_rate": 2.4413646055437103e-06,
      "loss": 0.4849,
      "step": 2290
    },
    {
      "epoch": 0.12259801178006982,
      "grad_norm": 1.5709069675016976,
      "learning_rate": 2.452025586353945e-06,
      "loss": 0.4749,
      "step": 2300
    },
    {
      "epoch": 0.12313104661389622,
      "grad_norm": 1.725433914578352,
      "learning_rate": 2.4626865671641794e-06,
      "loss": 0.4731,
      "step": 2310
    },
    {
      "epoch": 0.1236640814477226,
      "grad_norm": 1.6116971493392425,
      "learning_rate": 2.473347547974414e-06,
      "loss": 0.4803,
      "step": 2320
    },
    {
      "epoch": 0.124197116281549,
      "grad_norm": 1.6656362662222297,
      "learning_rate": 2.484008528784648e-06,
      "loss": 0.5,
      "step": 2330
    },
    {
      "epoch": 0.12473015111537539,
      "grad_norm": 1.7717514478220375,
      "learning_rate": 2.4946695095948825e-06,
      "loss": 0.4696,
      "step": 2340
    },
    {
      "epoch": 0.12526318594920177,
      "grad_norm": 1.7157129500673733,
      "learning_rate": 2.5053304904051175e-06,
      "loss": 0.4848,
      "step": 2350
    },
    {
      "epoch": 0.12579622078302818,
      "grad_norm": 1.6394167467054033,
      "learning_rate": 2.515991471215352e-06,
      "loss": 0.4882,
      "step": 2360
    },
    {
      "epoch": 0.12632925561685457,
      "grad_norm": 2.0113883742254712,
      "learning_rate": 2.5266524520255865e-06,
      "loss": 0.4791,
      "step": 2370
    },
    {
      "epoch": 0.12686229045068095,
      "grad_norm": 1.5991081210634723,
      "learning_rate": 2.537313432835821e-06,
      "loss": 0.4899,
      "step": 2380
    },
    {
      "epoch": 0.12739532528450734,
      "grad_norm": 1.7140376837663376,
      "learning_rate": 2.5479744136460556e-06,
      "loss": 0.4596,
      "step": 2390
    },
    {
      "epoch": 0.12792836011833372,
      "grad_norm": 1.7508722678866806,
      "learning_rate": 2.55863539445629e-06,
      "loss": 0.4698,
      "step": 2400
    },
    {
      "epoch": 0.12846139495216014,
      "grad_norm": 1.9199249496948412,
      "learning_rate": 2.5692963752665246e-06,
      "loss": 0.4598,
      "step": 2410
    },
    {
      "epoch": 0.12899442978598652,
      "grad_norm": 1.8211627501867966,
      "learning_rate": 2.579957356076759e-06,
      "loss": 0.4921,
      "step": 2420
    },
    {
      "epoch": 0.1295274646198129,
      "grad_norm": 1.7122266715459322,
      "learning_rate": 2.5906183368869937e-06,
      "loss": 0.4949,
      "step": 2430
    },
    {
      "epoch": 0.1300604994536393,
      "grad_norm": 1.843237819990565,
      "learning_rate": 2.601279317697228e-06,
      "loss": 0.4927,
      "step": 2440
    },
    {
      "epoch": 0.13059353428746567,
      "grad_norm": 1.5700771392593305,
      "learning_rate": 2.6119402985074627e-06,
      "loss": 0.4811,
      "step": 2450
    },
    {
      "epoch": 0.13112656912129209,
      "grad_norm": 1.7209228243885426,
      "learning_rate": 2.6226012793176973e-06,
      "loss": 0.4853,
      "step": 2460
    },
    {
      "epoch": 0.13165960395511847,
      "grad_norm": 1.7261663158334686,
      "learning_rate": 2.6332622601279318e-06,
      "loss": 0.4725,
      "step": 2470
    },
    {
      "epoch": 0.13219263878894486,
      "grad_norm": 1.5867794526316623,
      "learning_rate": 2.6439232409381663e-06,
      "loss": 0.4711,
      "step": 2480
    },
    {
      "epoch": 0.13272567362277124,
      "grad_norm": 1.651517768428253,
      "learning_rate": 2.654584221748401e-06,
      "loss": 0.4626,
      "step": 2490
    },
    {
      "epoch": 0.13325870845659765,
      "grad_norm": 1.7942624515241987,
      "learning_rate": 2.6652452025586354e-06,
      "loss": 0.4762,
      "step": 2500
    },
    {
      "epoch": 0.13379174329042404,
      "grad_norm": 1.838220915085115,
      "learning_rate": 2.67590618336887e-06,
      "loss": 0.4873,
      "step": 2510
    },
    {
      "epoch": 0.13432477812425042,
      "grad_norm": 1.6777720734017567,
      "learning_rate": 2.686567164179105e-06,
      "loss": 0.4752,
      "step": 2520
    },
    {
      "epoch": 0.1348578129580768,
      "grad_norm": 1.6952704077043272,
      "learning_rate": 2.6972281449893394e-06,
      "loss": 0.4733,
      "step": 2530
    },
    {
      "epoch": 0.1353908477919032,
      "grad_norm": 1.7370410349742849,
      "learning_rate": 2.7078891257995735e-06,
      "loss": 0.4893,
      "step": 2540
    },
    {
      "epoch": 0.1359238826257296,
      "grad_norm": 1.5855106890772668,
      "learning_rate": 2.718550106609808e-06,
      "loss": 0.4757,
      "step": 2550
    },
    {
      "epoch": 0.136456917459556,
      "grad_norm": 1.767876350225791,
      "learning_rate": 2.7292110874200425e-06,
      "loss": 0.4786,
      "step": 2560
    },
    {
      "epoch": 0.13698995229338237,
      "grad_norm": 1.921436725621705,
      "learning_rate": 2.739872068230277e-06,
      "loss": 0.4775,
      "step": 2570
    },
    {
      "epoch": 0.13752298712720876,
      "grad_norm": 1.9038990150261386,
      "learning_rate": 2.750533049040512e-06,
      "loss": 0.4832,
      "step": 2580
    },
    {
      "epoch": 0.13805602196103514,
      "grad_norm": 1.7501851035936127,
      "learning_rate": 2.7611940298507465e-06,
      "loss": 0.4762,
      "step": 2590
    },
    {
      "epoch": 0.13858905679486155,
      "grad_norm": 1.631228035358413,
      "learning_rate": 2.771855010660981e-06,
      "loss": 0.476,
      "step": 2600
    },
    {
      "epoch": 0.13912209162868794,
      "grad_norm": 1.5680229889762707,
      "learning_rate": 2.782515991471215e-06,
      "loss": 0.4724,
      "step": 2610
    },
    {
      "epoch": 0.13965512646251432,
      "grad_norm": 1.7053595348929826,
      "learning_rate": 2.7931769722814497e-06,
      "loss": 0.4652,
      "step": 2620
    },
    {
      "epoch": 0.1401881612963407,
      "grad_norm": 1.815725483080737,
      "learning_rate": 2.8038379530916846e-06,
      "loss": 0.499,
      "step": 2630
    },
    {
      "epoch": 0.14072119613016712,
      "grad_norm": 1.7622924765811276,
      "learning_rate": 2.814498933901919e-06,
      "loss": 0.4652,
      "step": 2640
    },
    {
      "epoch": 0.1412542309639935,
      "grad_norm": 1.8000297407262529,
      "learning_rate": 2.8251599147121537e-06,
      "loss": 0.4671,
      "step": 2650
    },
    {
      "epoch": 0.1417872657978199,
      "grad_norm": 1.7028419578488374,
      "learning_rate": 2.835820895522388e-06,
      "loss": 0.4724,
      "step": 2660
    },
    {
      "epoch": 0.14232030063164627,
      "grad_norm": 1.557502405149442,
      "learning_rate": 2.8464818763326227e-06,
      "loss": 0.4877,
      "step": 2670
    },
    {
      "epoch": 0.14285333546547266,
      "grad_norm": 1.682514725509012,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.4676,
      "step": 2680
    },
    {
      "epoch": 0.14338637029929907,
      "grad_norm": 1.6917198647698848,
      "learning_rate": 2.8678038379530918e-06,
      "loss": 0.4972,
      "step": 2690
    },
    {
      "epoch": 0.14391940513312546,
      "grad_norm": 1.647667906805554,
      "learning_rate": 2.8784648187633263e-06,
      "loss": 0.4809,
      "step": 2700
    },
    {
      "epoch": 0.14445243996695184,
      "grad_norm": 1.6304257308791026,
      "learning_rate": 2.889125799573561e-06,
      "loss": 0.4832,
      "step": 2710
    },
    {
      "epoch": 0.14498547480077822,
      "grad_norm": 1.776749710670109,
      "learning_rate": 2.8997867803837954e-06,
      "loss": 0.4712,
      "step": 2720
    },
    {
      "epoch": 0.1455185096346046,
      "grad_norm": 1.4680748829527526,
      "learning_rate": 2.91044776119403e-06,
      "loss": 0.4715,
      "step": 2730
    },
    {
      "epoch": 0.14605154446843102,
      "grad_norm": 1.6920857612601896,
      "learning_rate": 2.9211087420042644e-06,
      "loss": 0.4931,
      "step": 2740
    },
    {
      "epoch": 0.1465845793022574,
      "grad_norm": 1.7897945480485358,
      "learning_rate": 2.931769722814499e-06,
      "loss": 0.4874,
      "step": 2750
    },
    {
      "epoch": 0.1471176141360838,
      "grad_norm": 1.5377484124553449,
      "learning_rate": 2.9424307036247335e-06,
      "loss": 0.4714,
      "step": 2760
    },
    {
      "epoch": 0.14765064896991018,
      "grad_norm": 1.5803752311002037,
      "learning_rate": 2.953091684434968e-06,
      "loss": 0.4738,
      "step": 2770
    },
    {
      "epoch": 0.1481836838037366,
      "grad_norm": 1.6654580252453022,
      "learning_rate": 2.9637526652452025e-06,
      "loss": 0.4576,
      "step": 2780
    },
    {
      "epoch": 0.14871671863756297,
      "grad_norm": 1.5853688122421294,
      "learning_rate": 2.974413646055437e-06,
      "loss": 0.4845,
      "step": 2790
    },
    {
      "epoch": 0.14924975347138936,
      "grad_norm": 1.928646439015856,
      "learning_rate": 2.985074626865672e-06,
      "loss": 0.4722,
      "step": 2800
    },
    {
      "epoch": 0.14978278830521574,
      "grad_norm": 1.6957165307693083,
      "learning_rate": 2.9957356076759065e-06,
      "loss": 0.4819,
      "step": 2810
    },
    {
      "epoch": 0.15031582313904213,
      "grad_norm": 1.8031655894710157,
      "learning_rate": 2.999999906780211e-06,
      "loss": 0.4747,
      "step": 2820
    },
    {
      "epoch": 0.15084885797286854,
      "grad_norm": 1.6750833780100653,
      "learning_rate": 2.999999337103763e-06,
      "loss": 0.4689,
      "step": 2830
    },
    {
      "epoch": 0.15138189280669492,
      "grad_norm": 1.612344140009288,
      "learning_rate": 2.9999982495398344e-06,
      "loss": 0.4739,
      "step": 2840
    },
    {
      "epoch": 0.1519149276405213,
      "grad_norm": 1.795313546791791,
      "learning_rate": 2.9999966440888023e-06,
      "loss": 0.4792,
      "step": 2850
    },
    {
      "epoch": 0.1524479624743477,
      "grad_norm": 1.831543063210286,
      "learning_rate": 2.9999945207512196e-06,
      "loss": 0.4564,
      "step": 2860
    },
    {
      "epoch": 0.15298099730817408,
      "grad_norm": 1.850784587056977,
      "learning_rate": 2.99999187952782e-06,
      "loss": 0.4723,
      "step": 2870
    },
    {
      "epoch": 0.1535140321420005,
      "grad_norm": 1.6376500469394626,
      "learning_rate": 2.9999887204195153e-06,
      "loss": 0.4752,
      "step": 2880
    },
    {
      "epoch": 0.15404706697582687,
      "grad_norm": 1.6880735125963033,
      "learning_rate": 2.9999850434273964e-06,
      "loss": 0.4822,
      "step": 2890
    },
    {
      "epoch": 0.15458010180965326,
      "grad_norm": 1.7296444039455219,
      "learning_rate": 2.9999808485527327e-06,
      "loss": 0.4674,
      "step": 2900
    },
    {
      "epoch": 0.15511313664347964,
      "grad_norm": 1.6623189317763998,
      "learning_rate": 2.9999761357969725e-06,
      "loss": 0.4731,
      "step": 2910
    },
    {
      "epoch": 0.15564617147730606,
      "grad_norm": 1.6810682813609874,
      "learning_rate": 2.9999709051617425e-06,
      "loss": 0.4841,
      "step": 2920
    },
    {
      "epoch": 0.15617920631113244,
      "grad_norm": 1.7816054365878717,
      "learning_rate": 2.999965156648849e-06,
      "loss": 0.4824,
      "step": 2930
    },
    {
      "epoch": 0.15671224114495882,
      "grad_norm": 1.7533582395129694,
      "learning_rate": 2.999958890260277e-06,
      "loss": 0.4707,
      "step": 2940
    },
    {
      "epoch": 0.1572452759787852,
      "grad_norm": 1.6974204301749585,
      "learning_rate": 2.9999521059981897e-06,
      "loss": 0.4604,
      "step": 2950
    },
    {
      "epoch": 0.1577783108126116,
      "grad_norm": 1.7570577121286866,
      "learning_rate": 2.9999448038649293e-06,
      "loss": 0.4561,
      "step": 2960
    },
    {
      "epoch": 0.158311345646438,
      "grad_norm": 1.5112858903325956,
      "learning_rate": 2.9999369838630167e-06,
      "loss": 0.4607,
      "step": 2970
    },
    {
      "epoch": 0.1588443804802644,
      "grad_norm": 1.5787465254453812,
      "learning_rate": 2.9999286459951527e-06,
      "loss": 0.4598,
      "step": 2980
    },
    {
      "epoch": 0.15937741531409078,
      "grad_norm": 1.864721779879429,
      "learning_rate": 2.9999197902642153e-06,
      "loss": 0.4632,
      "step": 2990
    },
    {
      "epoch": 0.15991045014791716,
      "grad_norm": 1.7829518435710483,
      "learning_rate": 2.999910416673263e-06,
      "loss": 0.4659,
      "step": 3000
    },
    {
      "epoch": 0.16044348498174354,
      "grad_norm": 1.6079863952299247,
      "learning_rate": 2.9999005252255305e-06,
      "loss": 0.4564,
      "step": 3010
    },
    {
      "epoch": 0.16097651981556996,
      "grad_norm": 1.756347760567654,
      "learning_rate": 2.999890115924434e-06,
      "loss": 0.4867,
      "step": 3020
    },
    {
      "epoch": 0.16150955464939634,
      "grad_norm": 1.6040713307058265,
      "learning_rate": 2.9998791887735674e-06,
      "loss": 0.4656,
      "step": 3030
    },
    {
      "epoch": 0.16204258948322273,
      "grad_norm": 1.548518872333851,
      "learning_rate": 2.999867743776703e-06,
      "loss": 0.463,
      "step": 3040
    },
    {
      "epoch": 0.1625756243170491,
      "grad_norm": 1.6072392237096114,
      "learning_rate": 2.999855780937792e-06,
      "loss": 0.4604,
      "step": 3050
    },
    {
      "epoch": 0.16310865915087552,
      "grad_norm": 1.7739583912751717,
      "learning_rate": 2.9998433002609654e-06,
      "loss": 0.4647,
      "step": 3060
    },
    {
      "epoch": 0.1636416939847019,
      "grad_norm": 1.6223359684588095,
      "learning_rate": 2.9998303017505324e-06,
      "loss": 0.4816,
      "step": 3070
    },
    {
      "epoch": 0.1641747288185283,
      "grad_norm": 1.5145051243584609,
      "learning_rate": 2.9998167854109794e-06,
      "loss": 0.4616,
      "step": 3080
    },
    {
      "epoch": 0.16470776365235468,
      "grad_norm": 1.7462618302450663,
      "learning_rate": 2.999802751246975e-06,
      "loss": 0.4547,
      "step": 3090
    },
    {
      "epoch": 0.16524079848618106,
      "grad_norm": 1.63459286289502,
      "learning_rate": 2.999788199263363e-06,
      "loss": 0.4702,
      "step": 3100
    },
    {
      "epoch": 0.16577383332000747,
      "grad_norm": 1.6066524785723333,
      "learning_rate": 2.9997731294651688e-06,
      "loss": 0.4636,
      "step": 3110
    },
    {
      "epoch": 0.16630686815383386,
      "grad_norm": 1.6988577352511753,
      "learning_rate": 2.9997575418575943e-06,
      "loss": 0.4759,
      "step": 3120
    },
    {
      "epoch": 0.16683990298766024,
      "grad_norm": 1.7921629641000747,
      "learning_rate": 2.9997414364460224e-06,
      "loss": 0.4684,
      "step": 3130
    },
    {
      "epoch": 0.16737293782148663,
      "grad_norm": 1.7007876391018237,
      "learning_rate": 2.999724813236012e-06,
      "loss": 0.4585,
      "step": 3140
    },
    {
      "epoch": 0.167905972655313,
      "grad_norm": 1.5536161177891106,
      "learning_rate": 2.9997076722333044e-06,
      "loss": 0.4672,
      "step": 3150
    },
    {
      "epoch": 0.16843900748913943,
      "grad_norm": 1.6697459933103784,
      "learning_rate": 2.9996900134438164e-06,
      "loss": 0.4662,
      "step": 3160
    },
    {
      "epoch": 0.1689720423229658,
      "grad_norm": 1.647010387805045,
      "learning_rate": 2.999671836873645e-06,
      "loss": 0.4544,
      "step": 3170
    },
    {
      "epoch": 0.1695050771567922,
      "grad_norm": 1.5098449280976487,
      "learning_rate": 2.999653142529066e-06,
      "loss": 0.4734,
      "step": 3180
    },
    {
      "epoch": 0.17003811199061858,
      "grad_norm": 1.645930940874626,
      "learning_rate": 2.999633930416534e-06,
      "loss": 0.4536,
      "step": 3190
    },
    {
      "epoch": 0.170571146824445,
      "grad_norm": 1.3842232473056548,
      "learning_rate": 2.999614200542681e-06,
      "loss": 0.4669,
      "step": 3200
    },
    {
      "epoch": 0.17110418165827138,
      "grad_norm": 1.5909963769780455,
      "learning_rate": 2.9995939529143205e-06,
      "loss": 0.4677,
      "step": 3210
    },
    {
      "epoch": 0.17163721649209776,
      "grad_norm": 1.6452226092287423,
      "learning_rate": 2.9995731875384426e-06,
      "loss": 0.4623,
      "step": 3220
    },
    {
      "epoch": 0.17217025132592415,
      "grad_norm": 1.6399656056241063,
      "learning_rate": 2.9995519044222163e-06,
      "loss": 0.4731,
      "step": 3230
    },
    {
      "epoch": 0.17270328615975053,
      "grad_norm": 1.6762569466713584,
      "learning_rate": 2.99953010357299e-06,
      "loss": 0.4506,
      "step": 3240
    },
    {
      "epoch": 0.17323632099357694,
      "grad_norm": 1.6167067522288208,
      "learning_rate": 2.9995077849982907e-06,
      "loss": 0.4613,
      "step": 3250
    },
    {
      "epoch": 0.17376935582740333,
      "grad_norm": 1.627533327437292,
      "learning_rate": 2.9994849487058244e-06,
      "loss": 0.468,
      "step": 3260
    },
    {
      "epoch": 0.1743023906612297,
      "grad_norm": 1.6686721709677732,
      "learning_rate": 2.9994615947034745e-06,
      "loss": 0.4756,
      "step": 3270
    },
    {
      "epoch": 0.1748354254950561,
      "grad_norm": 1.5324487409521619,
      "learning_rate": 2.9994377229993056e-06,
      "loss": 0.4647,
      "step": 3280
    },
    {
      "epoch": 0.17536846032888248,
      "grad_norm": 1.5869013859847092,
      "learning_rate": 2.999413333601558e-06,
      "loss": 0.4621,
      "step": 3290
    },
    {
      "epoch": 0.1759014951627089,
      "grad_norm": 1.5278219436297606,
      "learning_rate": 2.9993884265186537e-06,
      "loss": 0.4651,
      "step": 3300
    },
    {
      "epoch": 0.17643452999653528,
      "grad_norm": 1.3878421460176371,
      "learning_rate": 2.9993630017591918e-06,
      "loss": 0.4629,
      "step": 3310
    },
    {
      "epoch": 0.17696756483036166,
      "grad_norm": 1.6146096998380026,
      "learning_rate": 2.9993370593319498e-06,
      "loss": 0.4681,
      "step": 3320
    },
    {
      "epoch": 0.17750059966418805,
      "grad_norm": 1.5566443040707618,
      "learning_rate": 2.9993105992458846e-06,
      "loss": 0.4561,
      "step": 3330
    },
    {
      "epoch": 0.17803363449801443,
      "grad_norm": 1.4577621441660769,
      "learning_rate": 2.999283621510133e-06,
      "loss": 0.4662,
      "step": 3340
    },
    {
      "epoch": 0.17856666933184084,
      "grad_norm": 1.4085624064590525,
      "learning_rate": 2.999256126134008e-06,
      "loss": 0.4561,
      "step": 3350
    },
    {
      "epoch": 0.17909970416566723,
      "grad_norm": 1.620228539762752,
      "learning_rate": 2.9992281131270027e-06,
      "loss": 0.4673,
      "step": 3360
    },
    {
      "epoch": 0.1796327389994936,
      "grad_norm": 1.5831400842402863,
      "learning_rate": 2.9991995824987893e-06,
      "loss": 0.4865,
      "step": 3370
    },
    {
      "epoch": 0.18016577383332,
      "grad_norm": 1.8363371878280756,
      "learning_rate": 2.999170534259218e-06,
      "loss": 0.4554,
      "step": 3380
    },
    {
      "epoch": 0.1806988086671464,
      "grad_norm": 1.5466565191656683,
      "learning_rate": 2.999140968418318e-06,
      "loss": 0.4591,
      "step": 3390
    },
    {
      "epoch": 0.1812318435009728,
      "grad_norm": 1.5707390027381525,
      "learning_rate": 2.9991108849862973e-06,
      "loss": 0.4704,
      "step": 3400
    },
    {
      "epoch": 0.18176487833479918,
      "grad_norm": 1.6366916399312674,
      "learning_rate": 2.9990802839735428e-06,
      "loss": 0.4692,
      "step": 3410
    },
    {
      "epoch": 0.18229791316862556,
      "grad_norm": 1.7019086412840565,
      "learning_rate": 2.9990491653906185e-06,
      "loss": 0.4649,
      "step": 3420
    },
    {
      "epoch": 0.18283094800245195,
      "grad_norm": 1.6396315172684046,
      "learning_rate": 2.99901752924827e-06,
      "loss": 0.4681,
      "step": 3430
    },
    {
      "epoch": 0.18336398283627836,
      "grad_norm": 1.6613298906787837,
      "learning_rate": 2.9989853755574184e-06,
      "loss": 0.4418,
      "step": 3440
    },
    {
      "epoch": 0.18389701767010475,
      "grad_norm": 1.6646590609464886,
      "learning_rate": 2.998952704329166e-06,
      "loss": 0.4471,
      "step": 3450
    },
    {
      "epoch": 0.18443005250393113,
      "grad_norm": 1.5013075178238962,
      "learning_rate": 2.9989195155747923e-06,
      "loss": 0.4651,
      "step": 3460
    },
    {
      "epoch": 0.18496308733775751,
      "grad_norm": 1.662972481207715,
      "learning_rate": 2.9988858093057563e-06,
      "loss": 0.4508,
      "step": 3470
    },
    {
      "epoch": 0.1854961221715839,
      "grad_norm": 1.5961630317747317,
      "learning_rate": 2.9988515855336954e-06,
      "loss": 0.4681,
      "step": 3480
    },
    {
      "epoch": 0.1860291570054103,
      "grad_norm": 1.618814745656187,
      "learning_rate": 2.9988168442704255e-06,
      "loss": 0.4792,
      "step": 3490
    },
    {
      "epoch": 0.1865621918392367,
      "grad_norm": 1.5353643377113313,
      "learning_rate": 2.9987815855279417e-06,
      "loss": 0.4551,
      "step": 3500
    },
    {
      "epoch": 0.18709522667306308,
      "grad_norm": 1.6703325650389744,
      "learning_rate": 2.998745809318417e-06,
      "loss": 0.4568,
      "step": 3510
    },
    {
      "epoch": 0.18762826150688947,
      "grad_norm": 1.6319317545687613,
      "learning_rate": 2.9987095156542034e-06,
      "loss": 0.4587,
      "step": 3520
    },
    {
      "epoch": 0.18816129634071588,
      "grad_norm": 1.4890644677542084,
      "learning_rate": 2.9986727045478315e-06,
      "loss": 0.4582,
      "step": 3530
    },
    {
      "epoch": 0.18869433117454226,
      "grad_norm": 1.494555208176799,
      "learning_rate": 2.9986353760120113e-06,
      "loss": 0.4459,
      "step": 3540
    },
    {
      "epoch": 0.18922736600836865,
      "grad_norm": 2.0564914458540016,
      "learning_rate": 2.99859753005963e-06,
      "loss": 0.4638,
      "step": 3550
    },
    {
      "epoch": 0.18976040084219503,
      "grad_norm": 1.7241381527855608,
      "learning_rate": 2.9985591667037546e-06,
      "loss": 0.4743,
      "step": 3560
    },
    {
      "epoch": 0.19029343567602142,
      "grad_norm": 1.6711262755757803,
      "learning_rate": 2.9985202859576304e-06,
      "loss": 0.4585,
      "step": 3570
    },
    {
      "epoch": 0.19082647050984783,
      "grad_norm": 1.7132918739150431,
      "learning_rate": 2.998480887834681e-06,
      "loss": 0.4431,
      "step": 3580
    },
    {
      "epoch": 0.1913595053436742,
      "grad_norm": 1.5313033903680584,
      "learning_rate": 2.99844097234851e-06,
      "loss": 0.4606,
      "step": 3590
    },
    {
      "epoch": 0.1918925401775006,
      "grad_norm": 1.7254901092253947,
      "learning_rate": 2.9984005395128974e-06,
      "loss": 0.4559,
      "step": 3600
    },
    {
      "epoch": 0.19242557501132698,
      "grad_norm": 1.645097433635555,
      "learning_rate": 2.998359589341803e-06,
      "loss": 0.4374,
      "step": 3610
    },
    {
      "epoch": 0.19295860984515337,
      "grad_norm": 1.5395304911207355,
      "learning_rate": 2.9983181218493654e-06,
      "loss": 0.4498,
      "step": 3620
    },
    {
      "epoch": 0.19349164467897978,
      "grad_norm": 1.7567010962136618,
      "learning_rate": 2.9982761370499024e-06,
      "loss": 0.4658,
      "step": 3630
    },
    {
      "epoch": 0.19402467951280616,
      "grad_norm": 1.8571868955078175,
      "learning_rate": 2.9982336349579085e-06,
      "loss": 0.4649,
      "step": 3640
    },
    {
      "epoch": 0.19455771434663255,
      "grad_norm": 1.637593677791729,
      "learning_rate": 2.998190615588058e-06,
      "loss": 0.4484,
      "step": 3650
    },
    {
      "epoch": 0.19509074918045893,
      "grad_norm": 1.6434315395657313,
      "learning_rate": 2.9981470789552046e-06,
      "loss": 0.4642,
      "step": 3660
    },
    {
      "epoch": 0.19562378401428535,
      "grad_norm": 1.815340805041272,
      "learning_rate": 2.998103025074379e-06,
      "loss": 0.4484,
      "step": 3670
    },
    {
      "epoch": 0.19615681884811173,
      "grad_norm": 1.5389948675670018,
      "learning_rate": 2.9980584539607913e-06,
      "loss": 0.4634,
      "step": 3680
    },
    {
      "epoch": 0.19668985368193811,
      "grad_norm": 1.5489883254493602,
      "learning_rate": 2.99801336562983e-06,
      "loss": 0.4706,
      "step": 3690
    },
    {
      "epoch": 0.1972228885157645,
      "grad_norm": 1.6274345202406608,
      "learning_rate": 2.9979677600970622e-06,
      "loss": 0.47,
      "step": 3700
    },
    {
      "epoch": 0.19775592334959088,
      "grad_norm": 1.5760744328735794,
      "learning_rate": 2.9979216373782343e-06,
      "loss": 0.4628,
      "step": 3710
    },
    {
      "epoch": 0.1982889581834173,
      "grad_norm": 1.4956785159988102,
      "learning_rate": 2.997874997489269e-06,
      "loss": 0.4492,
      "step": 3720
    },
    {
      "epoch": 0.19882199301724368,
      "grad_norm": 1.7080924559166886,
      "learning_rate": 2.997827840446271e-06,
      "loss": 0.4529,
      "step": 3730
    },
    {
      "epoch": 0.19935502785107007,
      "grad_norm": 1.5915206618874034,
      "learning_rate": 2.99778016626552e-06,
      "loss": 0.4633,
      "step": 3740
    },
    {
      "epoch": 0.19988806268489645,
      "grad_norm": 1.5048489860129535,
      "learning_rate": 2.9977319749634774e-06,
      "loss": 0.4526,
      "step": 3750
    },
    {
      "epoch": 0.20042109751872283,
      "grad_norm": 1.5041280839206856,
      "learning_rate": 2.99768326655678e-06,
      "loss": 0.4641,
      "step": 3760
    },
    {
      "epoch": 0.20095413235254925,
      "grad_norm": 1.466087678237714,
      "learning_rate": 2.9976340410622464e-06,
      "loss": 0.4396,
      "step": 3770
    },
    {
      "epoch": 0.20148716718637563,
      "grad_norm": 1.7743885987789316,
      "learning_rate": 2.997584298496871e-06,
      "loss": 0.4665,
      "step": 3780
    },
    {
      "epoch": 0.20202020202020202,
      "grad_norm": 1.603992714080962,
      "learning_rate": 2.9975340388778286e-06,
      "loss": 0.457,
      "step": 3790
    },
    {
      "epoch": 0.2025532368540284,
      "grad_norm": 1.543570274830021,
      "learning_rate": 2.9974832622224712e-06,
      "loss": 0.4451,
      "step": 3800
    },
    {
      "epoch": 0.2030862716878548,
      "grad_norm": 1.5287734491563887,
      "learning_rate": 2.9974319685483304e-06,
      "loss": 0.4463,
      "step": 3810
    },
    {
      "epoch": 0.2036193065216812,
      "grad_norm": 1.5868593930462622,
      "learning_rate": 2.997380157873115e-06,
      "loss": 0.4253,
      "step": 3820
    },
    {
      "epoch": 0.20415234135550758,
      "grad_norm": 1.651829324425076,
      "learning_rate": 2.997327830214714e-06,
      "loss": 0.4306,
      "step": 3830
    },
    {
      "epoch": 0.20468537618933397,
      "grad_norm": 1.8722112674795424,
      "learning_rate": 2.997274985591193e-06,
      "loss": 0.4612,
      "step": 3840
    },
    {
      "epoch": 0.20521841102316035,
      "grad_norm": 1.586839450943598,
      "learning_rate": 2.9972216240207976e-06,
      "loss": 0.4613,
      "step": 3850
    },
    {
      "epoch": 0.20575144585698676,
      "grad_norm": 1.65561725732763,
      "learning_rate": 2.9971677455219515e-06,
      "loss": 0.4753,
      "step": 3860
    },
    {
      "epoch": 0.20628448069081315,
      "grad_norm": 1.6654730662916013,
      "learning_rate": 2.9971133501132568e-06,
      "loss": 0.4293,
      "step": 3870
    },
    {
      "epoch": 0.20681751552463953,
      "grad_norm": 1.7221959528288986,
      "learning_rate": 2.9970584378134934e-06,
      "loss": 0.4348,
      "step": 3880
    },
    {
      "epoch": 0.20735055035846592,
      "grad_norm": 1.5910407161008908,
      "learning_rate": 2.9970030086416204e-06,
      "loss": 0.4511,
      "step": 3890
    },
    {
      "epoch": 0.2078835851922923,
      "grad_norm": 1.6141970451791834,
      "learning_rate": 2.9969470626167754e-06,
      "loss": 0.4414,
      "step": 3900
    },
    {
      "epoch": 0.20841662002611872,
      "grad_norm": 1.7867460769140482,
      "learning_rate": 2.996890599758274e-06,
      "loss": 0.445,
      "step": 3910
    },
    {
      "epoch": 0.2089496548599451,
      "grad_norm": 1.595718696260754,
      "learning_rate": 2.9968336200856107e-06,
      "loss": 0.4353,
      "step": 3920
    },
    {
      "epoch": 0.20948268969377148,
      "grad_norm": 1.676254990466111,
      "learning_rate": 2.9967761236184587e-06,
      "loss": 0.452,
      "step": 3930
    },
    {
      "epoch": 0.21001572452759787,
      "grad_norm": 1.5269354127652917,
      "learning_rate": 2.9967181103766682e-06,
      "loss": 0.4371,
      "step": 3940
    },
    {
      "epoch": 0.21054875936142428,
      "grad_norm": 1.7744320905768576,
      "learning_rate": 2.9966595803802686e-06,
      "loss": 0.4502,
      "step": 3950
    },
    {
      "epoch": 0.21108179419525067,
      "grad_norm": 1.4757823489660078,
      "learning_rate": 2.9966005336494692e-06,
      "loss": 0.4679,
      "step": 3960
    },
    {
      "epoch": 0.21161482902907705,
      "grad_norm": 1.4989736270803655,
      "learning_rate": 2.9965409702046552e-06,
      "loss": 0.4606,
      "step": 3970
    },
    {
      "epoch": 0.21214786386290344,
      "grad_norm": 1.704780250787622,
      "learning_rate": 2.9964808900663917e-06,
      "loss": 0.4323,
      "step": 3980
    },
    {
      "epoch": 0.21268089869672982,
      "grad_norm": 1.5334460115950397,
      "learning_rate": 2.9964202932554224e-06,
      "loss": 0.4456,
      "step": 3990
    },
    {
      "epoch": 0.21321393353055623,
      "grad_norm": 1.7155779812422818,
      "learning_rate": 2.9963591797926683e-06,
      "loss": 0.4332,
      "step": 4000
    },
    {
      "epoch": 0.21374696836438262,
      "grad_norm": 1.5465873446480514,
      "learning_rate": 2.996297549699229e-06,
      "loss": 0.4542,
      "step": 4010
    },
    {
      "epoch": 0.214280003198209,
      "grad_norm": 1.5309415649407119,
      "learning_rate": 2.9962354029963836e-06,
      "loss": 0.4527,
      "step": 4020
    },
    {
      "epoch": 0.21481303803203539,
      "grad_norm": 1.7228835864420653,
      "learning_rate": 2.9961727397055887e-06,
      "loss": 0.4423,
      "step": 4030
    },
    {
      "epoch": 0.21534607286586177,
      "grad_norm": 1.4923330650311342,
      "learning_rate": 2.996109559848479e-06,
      "loss": 0.4575,
      "step": 4040
    },
    {
      "epoch": 0.21587910769968818,
      "grad_norm": 1.5728391343623689,
      "learning_rate": 2.9960458634468673e-06,
      "loss": 0.4594,
      "step": 4050
    },
    {
      "epoch": 0.21641214253351457,
      "grad_norm": 1.708177585976328,
      "learning_rate": 2.9959816505227467e-06,
      "loss": 0.4394,
      "step": 4060
    },
    {
      "epoch": 0.21694517736734095,
      "grad_norm": 1.6196290389455337,
      "learning_rate": 2.9959169210982866e-06,
      "loss": 0.45,
      "step": 4070
    },
    {
      "epoch": 0.21747821220116734,
      "grad_norm": 1.6243128885757592,
      "learning_rate": 2.9958516751958354e-06,
      "loss": 0.4427,
      "step": 4080
    },
    {
      "epoch": 0.21801124703499375,
      "grad_norm": 1.6786587625364946,
      "learning_rate": 2.995785912837919e-06,
      "loss": 0.4522,
      "step": 4090
    },
    {
      "epoch": 0.21854428186882013,
      "grad_norm": 1.5079021102365517,
      "learning_rate": 2.995719634047244e-06,
      "loss": 0.4559,
      "step": 4100
    },
    {
      "epoch": 0.21907731670264652,
      "grad_norm": 1.6724642613778926,
      "learning_rate": 2.9956528388466922e-06,
      "loss": 0.4484,
      "step": 4110
    },
    {
      "epoch": 0.2196103515364729,
      "grad_norm": 1.8030881030305077,
      "learning_rate": 2.995585527259326e-06,
      "loss": 0.4567,
      "step": 4120
    },
    {
      "epoch": 0.2201433863702993,
      "grad_norm": 1.6748168171675852,
      "learning_rate": 2.9955176993083855e-06,
      "loss": 0.4387,
      "step": 4130
    },
    {
      "epoch": 0.2206764212041257,
      "grad_norm": 1.648500665231352,
      "learning_rate": 2.9954493550172885e-06,
      "loss": 0.4384,
      "step": 4140
    },
    {
      "epoch": 0.22120945603795208,
      "grad_norm": 1.6012217561678899,
      "learning_rate": 2.995380494409631e-06,
      "loss": 0.4548,
      "step": 4150
    },
    {
      "epoch": 0.22174249087177847,
      "grad_norm": 1.6559612740139638,
      "learning_rate": 2.9953111175091887e-06,
      "loss": 0.4647,
      "step": 4160
    },
    {
      "epoch": 0.22227552570560485,
      "grad_norm": 1.6208015918349856,
      "learning_rate": 2.995241224339914e-06,
      "loss": 0.4442,
      "step": 4170
    },
    {
      "epoch": 0.22280856053943124,
      "grad_norm": 1.8070273610482273,
      "learning_rate": 2.995170814925938e-06,
      "loss": 0.4518,
      "step": 4180
    },
    {
      "epoch": 0.22334159537325765,
      "grad_norm": 2.08914606239165,
      "learning_rate": 2.99509988929157e-06,
      "loss": 0.4485,
      "step": 4190
    },
    {
      "epoch": 0.22387463020708404,
      "grad_norm": 1.579705453532804,
      "learning_rate": 2.9950284474612985e-06,
      "loss": 0.4445,
      "step": 4200
    },
    {
      "epoch": 0.22440766504091042,
      "grad_norm": 1.6207813632705006,
      "learning_rate": 2.9949564894597887e-06,
      "loss": 0.4431,
      "step": 4210
    },
    {
      "epoch": 0.2249406998747368,
      "grad_norm": 1.5802741216513316,
      "learning_rate": 2.994884015311885e-06,
      "loss": 0.4547,
      "step": 4220
    },
    {
      "epoch": 0.22547373470856322,
      "grad_norm": 1.6297823572798982,
      "learning_rate": 2.9948110250426093e-06,
      "loss": 0.4523,
      "step": 4230
    },
    {
      "epoch": 0.2260067695423896,
      "grad_norm": 1.8281204535912232,
      "learning_rate": 2.994737518677163e-06,
      "loss": 0.4485,
      "step": 4240
    },
    {
      "epoch": 0.226539804376216,
      "grad_norm": 1.5153505884941239,
      "learning_rate": 2.9946634962409237e-06,
      "loss": 0.4512,
      "step": 4250
    },
    {
      "epoch": 0.22707283921004237,
      "grad_norm": 1.9042971464522553,
      "learning_rate": 2.994588957759449e-06,
      "loss": 0.4583,
      "step": 4260
    },
    {
      "epoch": 0.22760587404386876,
      "grad_norm": 1.552844823810127,
      "learning_rate": 2.994513903258474e-06,
      "loss": 0.4529,
      "step": 4270
    },
    {
      "epoch": 0.22813890887769517,
      "grad_norm": 1.551437621496883,
      "learning_rate": 2.9944383327639113e-06,
      "loss": 0.4648,
      "step": 4280
    },
    {
      "epoch": 0.22867194371152155,
      "grad_norm": 1.5646925445220212,
      "learning_rate": 2.994362246301852e-06,
      "loss": 0.4472,
      "step": 4290
    },
    {
      "epoch": 0.22920497854534794,
      "grad_norm": 1.6304750433301312,
      "learning_rate": 2.9942856438985673e-06,
      "loss": 0.4499,
      "step": 4300
    },
    {
      "epoch": 0.22973801337917432,
      "grad_norm": 1.6472194227955004,
      "learning_rate": 2.994208525580503e-06,
      "loss": 0.4442,
      "step": 4310
    },
    {
      "epoch": 0.2302710482130007,
      "grad_norm": 1.4959734612979503,
      "learning_rate": 2.9941308913742854e-06,
      "loss": 0.4572,
      "step": 4320
    },
    {
      "epoch": 0.23080408304682712,
      "grad_norm": 1.810912152153463,
      "learning_rate": 2.994052741306719e-06,
      "loss": 0.4306,
      "step": 4330
    },
    {
      "epoch": 0.2313371178806535,
      "grad_norm": 1.5520143659917245,
      "learning_rate": 2.993974075404785e-06,
      "loss": 0.441,
      "step": 4340
    },
    {
      "epoch": 0.2318701527144799,
      "grad_norm": 1.6303210751414012,
      "learning_rate": 2.9938948936956443e-06,
      "loss": 0.4414,
      "step": 4350
    },
    {
      "epoch": 0.23240318754830627,
      "grad_norm": 1.5863261623737483,
      "learning_rate": 2.9938151962066338e-06,
      "loss": 0.4489,
      "step": 4360
    },
    {
      "epoch": 0.23293622238213268,
      "grad_norm": 1.5723471828450144,
      "learning_rate": 2.9937349829652706e-06,
      "loss": 0.455,
      "step": 4370
    },
    {
      "epoch": 0.23346925721595907,
      "grad_norm": 1.720463131126141,
      "learning_rate": 2.993654253999249e-06,
      "loss": 0.4405,
      "step": 4380
    },
    {
      "epoch": 0.23400229204978545,
      "grad_norm": 1.5772912374693178,
      "learning_rate": 2.9935730093364417e-06,
      "loss": 0.4402,
      "step": 4390
    },
    {
      "epoch": 0.23453532688361184,
      "grad_norm": 1.6987453469415965,
      "learning_rate": 2.993491249004898e-06,
      "loss": 0.4598,
      "step": 4400
    },
    {
      "epoch": 0.23506836171743822,
      "grad_norm": 1.5128089543682928,
      "learning_rate": 2.9934089730328474e-06,
      "loss": 0.441,
      "step": 4410
    },
    {
      "epoch": 0.23560139655126464,
      "grad_norm": 1.785023375618966,
      "learning_rate": 2.9933261814486956e-06,
      "loss": 0.4537,
      "step": 4420
    },
    {
      "epoch": 0.23613443138509102,
      "grad_norm": 1.567988135815809,
      "learning_rate": 2.9932428742810276e-06,
      "loss": 0.4439,
      "step": 4430
    },
    {
      "epoch": 0.2366674662189174,
      "grad_norm": 1.8560104630453909,
      "learning_rate": 2.9931590515586057e-06,
      "loss": 0.4392,
      "step": 4440
    },
    {
      "epoch": 0.2372005010527438,
      "grad_norm": 1.714821979884251,
      "learning_rate": 2.9930747133103705e-06,
      "loss": 0.4429,
      "step": 4450
    },
    {
      "epoch": 0.23773353588657017,
      "grad_norm": 1.5661715064623511,
      "learning_rate": 2.99298985956544e-06,
      "loss": 0.4616,
      "step": 4460
    },
    {
      "epoch": 0.2382665707203966,
      "grad_norm": 1.6532985857875688,
      "learning_rate": 2.9929044903531117e-06,
      "loss": 0.4494,
      "step": 4470
    },
    {
      "epoch": 0.23879960555422297,
      "grad_norm": 1.5534915918437062,
      "learning_rate": 2.9928186057028595e-06,
      "loss": 0.4468,
      "step": 4480
    },
    {
      "epoch": 0.23933264038804936,
      "grad_norm": 1.7257876945892578,
      "learning_rate": 2.9927322056443354e-06,
      "loss": 0.433,
      "step": 4490
    },
    {
      "epoch": 0.23986567522187574,
      "grad_norm": 1.7097647077243794,
      "learning_rate": 2.9926452902073707e-06,
      "loss": 0.4346,
      "step": 4500
    },
    {
      "epoch": 0.24039871005570215,
      "grad_norm": 1.7245345610777312,
      "learning_rate": 2.9925578594219724e-06,
      "loss": 0.4433,
      "step": 4510
    },
    {
      "epoch": 0.24093174488952854,
      "grad_norm": 1.94640057153672,
      "learning_rate": 2.9924699133183283e-06,
      "loss": 0.4314,
      "step": 4520
    },
    {
      "epoch": 0.24146477972335492,
      "grad_norm": 1.5639012383831647,
      "learning_rate": 2.992381451926801e-06,
      "loss": 0.4424,
      "step": 4530
    },
    {
      "epoch": 0.2419978145571813,
      "grad_norm": 1.5119201408008984,
      "learning_rate": 2.992292475277933e-06,
      "loss": 0.4499,
      "step": 4540
    },
    {
      "epoch": 0.2425308493910077,
      "grad_norm": 1.5661485849848098,
      "learning_rate": 2.992202983402445e-06,
      "loss": 0.4517,
      "step": 4550
    },
    {
      "epoch": 0.2430638842248341,
      "grad_norm": 1.7286594213343105,
      "learning_rate": 2.992112976331234e-06,
      "loss": 0.4335,
      "step": 4560
    },
    {
      "epoch": 0.2435969190586605,
      "grad_norm": 1.65235104393594,
      "learning_rate": 2.992022454095376e-06,
      "loss": 0.4532,
      "step": 4570
    },
    {
      "epoch": 0.24412995389248687,
      "grad_norm": 1.649496576108496,
      "learning_rate": 2.991931416726125e-06,
      "loss": 0.4502,
      "step": 4580
    },
    {
      "epoch": 0.24466298872631326,
      "grad_norm": 1.671076820394612,
      "learning_rate": 2.9918398642549112e-06,
      "loss": 0.455,
      "step": 4590
    },
    {
      "epoch": 0.24519602356013964,
      "grad_norm": 1.5682819727388604,
      "learning_rate": 2.9917477967133456e-06,
      "loss": 0.4507,
      "step": 4600
    },
    {
      "epoch": 0.24572905839396605,
      "grad_norm": 1.6792524816544012,
      "learning_rate": 2.9916552141332138e-06,
      "loss": 0.4318,
      "step": 4610
    },
    {
      "epoch": 0.24626209322779244,
      "grad_norm": 1.7151732533704052,
      "learning_rate": 2.9915621165464816e-06,
      "loss": 0.4334,
      "step": 4620
    },
    {
      "epoch": 0.24679512806161882,
      "grad_norm": 1.6050371790176439,
      "learning_rate": 2.9914685039852906e-06,
      "loss": 0.4441,
      "step": 4630
    },
    {
      "epoch": 0.2473281628954452,
      "grad_norm": 1.6422256225869465,
      "learning_rate": 2.991374376481963e-06,
      "loss": 0.4362,
      "step": 4640
    },
    {
      "epoch": 0.24786119772927162,
      "grad_norm": 1.5457940878568954,
      "learning_rate": 2.9912797340689955e-06,
      "loss": 0.4355,
      "step": 4650
    },
    {
      "epoch": 0.248394232563098,
      "grad_norm": 1.69589490670946,
      "learning_rate": 2.9911845767790655e-06,
      "loss": 0.4207,
      "step": 4660
    },
    {
      "epoch": 0.2489272673969244,
      "grad_norm": 1.6914609782763563,
      "learning_rate": 2.991088904645026e-06,
      "loss": 0.4559,
      "step": 4670
    },
    {
      "epoch": 0.24946030223075077,
      "grad_norm": 1.574786275327489,
      "learning_rate": 2.9909927176999097e-06,
      "loss": 0.4634,
      "step": 4680
    },
    {
      "epoch": 0.24999333706457716,
      "grad_norm": 1.6916817167859033,
      "learning_rate": 2.9908960159769243e-06,
      "loss": 0.4409,
      "step": 4690
    },
    {
      "epoch": 0.25052637189840354,
      "grad_norm": 1.6346762114793119,
      "learning_rate": 2.990798799509458e-06,
      "loss": 0.4315,
      "step": 4700
    },
    {
      "epoch": 0.25105940673222993,
      "grad_norm": 1.6446348908077835,
      "learning_rate": 2.990701068331075e-06,
      "loss": 0.438,
      "step": 4710
    },
    {
      "epoch": 0.25159244156605637,
      "grad_norm": 1.8249833266882203,
      "learning_rate": 2.9906028224755193e-06,
      "loss": 0.4295,
      "step": 4720
    },
    {
      "epoch": 0.25212547639988275,
      "grad_norm": 1.7045165285792323,
      "learning_rate": 2.990504061976709e-06,
      "loss": 0.4274,
      "step": 4730
    },
    {
      "epoch": 0.25265851123370914,
      "grad_norm": 1.5419183954887759,
      "learning_rate": 2.9904047868687432e-06,
      "loss": 0.4424,
      "step": 4740
    },
    {
      "epoch": 0.2531915460675355,
      "grad_norm": 1.5816311318418992,
      "learning_rate": 2.9903049971858977e-06,
      "loss": 0.4435,
      "step": 4750
    },
    {
      "epoch": 0.2537245809013619,
      "grad_norm": 1.4922048988802745,
      "learning_rate": 2.990204692962625e-06,
      "loss": 0.4411,
      "step": 4760
    },
    {
      "epoch": 0.2542576157351883,
      "grad_norm": 1.6124055351644564,
      "learning_rate": 2.9901038742335556e-06,
      "loss": 0.4254,
      "step": 4770
    },
    {
      "epoch": 0.2547906505690147,
      "grad_norm": 1.562286493859035,
      "learning_rate": 2.9900025410334997e-06,
      "loss": 0.4473,
      "step": 4780
    },
    {
      "epoch": 0.25532368540284106,
      "grad_norm": 1.6138582673418314,
      "learning_rate": 2.9899006933974425e-06,
      "loss": 0.4477,
      "step": 4790
    },
    {
      "epoch": 0.25585672023666745,
      "grad_norm": 1.6158155095198716,
      "learning_rate": 2.9897983313605473e-06,
      "loss": 0.4458,
      "step": 4800
    },
    {
      "epoch": 0.25638975507049383,
      "grad_norm": 1.6217362897850136,
      "learning_rate": 2.989695454958156e-06,
      "loss": 0.4454,
      "step": 4810
    },
    {
      "epoch": 0.25692278990432027,
      "grad_norm": 1.4371598050105188,
      "learning_rate": 2.989592064225787e-06,
      "loss": 0.4192,
      "step": 4820
    },
    {
      "epoch": 0.25745582473814665,
      "grad_norm": 1.7616371896755234,
      "learning_rate": 2.9894881591991377e-06,
      "loss": 0.4425,
      "step": 4830
    },
    {
      "epoch": 0.25798885957197304,
      "grad_norm": 1.6407524182793443,
      "learning_rate": 2.9893837399140816e-06,
      "loss": 0.4427,
      "step": 4840
    },
    {
      "epoch": 0.2585218944057994,
      "grad_norm": 1.557889807385073,
      "learning_rate": 2.989278806406671e-06,
      "loss": 0.4446,
      "step": 4850
    },
    {
      "epoch": 0.2590549292396258,
      "grad_norm": 1.8828968253643723,
      "learning_rate": 2.9891733587131342e-06,
      "loss": 0.4489,
      "step": 4860
    },
    {
      "epoch": 0.2595879640734522,
      "grad_norm": 1.4840557364856535,
      "learning_rate": 2.989067396869878e-06,
      "loss": 0.4386,
      "step": 4870
    },
    {
      "epoch": 0.2601209989072786,
      "grad_norm": 1.6513095004826863,
      "learning_rate": 2.988960920913487e-06,
      "loss": 0.4395,
      "step": 4880
    },
    {
      "epoch": 0.26065403374110496,
      "grad_norm": 1.594827034124819,
      "learning_rate": 2.988853930880723e-06,
      "loss": 0.4385,
      "step": 4890
    },
    {
      "epoch": 0.26118706857493135,
      "grad_norm": 1.522785754866806,
      "learning_rate": 2.988746426808525e-06,
      "loss": 0.4385,
      "step": 4900
    },
    {
      "epoch": 0.2617201034087578,
      "grad_norm": 1.6824409079130382,
      "learning_rate": 2.98863840873401e-06,
      "loss": 0.4338,
      "step": 4910
    },
    {
      "epoch": 0.26225313824258417,
      "grad_norm": 1.5677604985204645,
      "learning_rate": 2.9885298766944715e-06,
      "loss": 0.43,
      "step": 4920
    },
    {
      "epoch": 0.26278617307641056,
      "grad_norm": 1.7230004988468286,
      "learning_rate": 2.9884208307273815e-06,
      "loss": 0.4392,
      "step": 4930
    },
    {
      "epoch": 0.26331920791023694,
      "grad_norm": 1.6514760693400496,
      "learning_rate": 2.988311270870389e-06,
      "loss": 0.4458,
      "step": 4940
    },
    {
      "epoch": 0.2638522427440633,
      "grad_norm": 1.702580415570755,
      "learning_rate": 2.988201197161321e-06,
      "loss": 0.4327,
      "step": 4950
    },
    {
      "epoch": 0.2643852775778897,
      "grad_norm": 1.6407046943697028,
      "learning_rate": 2.9880906096381804e-06,
      "loss": 0.4417,
      "step": 4960
    },
    {
      "epoch": 0.2649183124117161,
      "grad_norm": 1.9061693114597238,
      "learning_rate": 2.9879795083391493e-06,
      "loss": 0.4429,
      "step": 4970
    },
    {
      "epoch": 0.2654513472455425,
      "grad_norm": 1.484711555285733,
      "learning_rate": 2.987867893302586e-06,
      "loss": 0.4344,
      "step": 4980
    },
    {
      "epoch": 0.26598438207936886,
      "grad_norm": 1.5815148651198652,
      "learning_rate": 2.9877557645670264e-06,
      "loss": 0.4335,
      "step": 4990
    },
    {
      "epoch": 0.2665174169131953,
      "grad_norm": 1.7081891699885108,
      "learning_rate": 2.987643122171184e-06,
      "loss": 0.4439,
      "step": 5000
    },
    {
      "epoch": 0.2670504517470217,
      "grad_norm": 1.5648904132118493,
      "learning_rate": 2.9875299661539498e-06,
      "loss": 0.4442,
      "step": 5010
    },
    {
      "epoch": 0.2675834865808481,
      "grad_norm": 1.7643629619545491,
      "learning_rate": 2.9874162965543913e-06,
      "loss": 0.4484,
      "step": 5020
    },
    {
      "epoch": 0.26811652141467446,
      "grad_norm": 1.6310592892945788,
      "learning_rate": 2.987302113411755e-06,
      "loss": 0.4577,
      "step": 5030
    },
    {
      "epoch": 0.26864955624850084,
      "grad_norm": 1.4743085963469467,
      "learning_rate": 2.9871874167654625e-06,
      "loss": 0.4377,
      "step": 5040
    },
    {
      "epoch": 0.2691825910823272,
      "grad_norm": 1.7405038010027776,
      "learning_rate": 2.987072206655114e-06,
      "loss": 0.445,
      "step": 5050
    },
    {
      "epoch": 0.2697156259161536,
      "grad_norm": 1.7380779059647404,
      "learning_rate": 2.9869564831204872e-06,
      "loss": 0.4582,
      "step": 5060
    },
    {
      "epoch": 0.27024866074998,
      "grad_norm": 1.5999970454550323,
      "learning_rate": 2.9868402462015363e-06,
      "loss": 0.4376,
      "step": 5070
    },
    {
      "epoch": 0.2707816955838064,
      "grad_norm": 1.4936113294612414,
      "learning_rate": 2.986723495938393e-06,
      "loss": 0.4299,
      "step": 5080
    },
    {
      "epoch": 0.27131473041763277,
      "grad_norm": 1.6148393782948467,
      "learning_rate": 2.9866062323713664e-06,
      "loss": 0.4367,
      "step": 5090
    },
    {
      "epoch": 0.2718477652514592,
      "grad_norm": 1.3821227021650941,
      "learning_rate": 2.986488455540943e-06,
      "loss": 0.4159,
      "step": 5100
    },
    {
      "epoch": 0.2723808000852856,
      "grad_norm": 1.787216851872511,
      "learning_rate": 2.9863701654877864e-06,
      "loss": 0.4411,
      "step": 5110
    },
    {
      "epoch": 0.272913834919112,
      "grad_norm": 1.4371472591681986,
      "learning_rate": 2.9862513622527365e-06,
      "loss": 0.4388,
      "step": 5120
    },
    {
      "epoch": 0.27344686975293836,
      "grad_norm": 1.8033827313505777,
      "learning_rate": 2.986132045876811e-06,
      "loss": 0.452,
      "step": 5130
    },
    {
      "epoch": 0.27397990458676474,
      "grad_norm": 1.7330518783082376,
      "learning_rate": 2.986012216401206e-06,
      "loss": 0.444,
      "step": 5140
    },
    {
      "epoch": 0.27451293942059113,
      "grad_norm": 1.6235747832288199,
      "learning_rate": 2.9858918738672933e-06,
      "loss": 0.45,
      "step": 5150
    },
    {
      "epoch": 0.2750459742544175,
      "grad_norm": 1.5854865507159814,
      "learning_rate": 2.9857710183166213e-06,
      "loss": 0.434,
      "step": 5160
    },
    {
      "epoch": 0.2755790090882439,
      "grad_norm": 1.594168237598214,
      "learning_rate": 2.9856496497909177e-06,
      "loss": 0.4301,
      "step": 5170
    },
    {
      "epoch": 0.2761120439220703,
      "grad_norm": 1.5168110000684847,
      "learning_rate": 2.9855277683320847e-06,
      "loss": 0.4383,
      "step": 5180
    },
    {
      "epoch": 0.2766450787558967,
      "grad_norm": 1.5571375377001808,
      "learning_rate": 2.985405373982204e-06,
      "loss": 0.4524,
      "step": 5190
    },
    {
      "epoch": 0.2771781135897231,
      "grad_norm": 1.687513312399059,
      "learning_rate": 2.9852824667835323e-06,
      "loss": 0.4493,
      "step": 5200
    },
    {
      "epoch": 0.2777111484235495,
      "grad_norm": 1.6539174041403841,
      "learning_rate": 2.9851590467785056e-06,
      "loss": 0.4417,
      "step": 5210
    },
    {
      "epoch": 0.2782441832573759,
      "grad_norm": 1.5525801128997774,
      "learning_rate": 2.9850351140097344e-06,
      "loss": 0.4476,
      "step": 5220
    },
    {
      "epoch": 0.27877721809120226,
      "grad_norm": 1.6108736999393107,
      "learning_rate": 2.9849106685200085e-06,
      "loss": 0.4337,
      "step": 5230
    },
    {
      "epoch": 0.27931025292502865,
      "grad_norm": 1.6967534154613837,
      "learning_rate": 2.984785710352293e-06,
      "loss": 0.455,
      "step": 5240
    },
    {
      "epoch": 0.27984328775885503,
      "grad_norm": 1.5422861418485796,
      "learning_rate": 2.9846602395497314e-06,
      "loss": 0.4369,
      "step": 5250
    },
    {
      "epoch": 0.2803763225926814,
      "grad_norm": 1.5532485024698688,
      "learning_rate": 2.9845342561556433e-06,
      "loss": 0.4166,
      "step": 5260
    },
    {
      "epoch": 0.2809093574265078,
      "grad_norm": 1.5971608274093063,
      "learning_rate": 2.9844077602135255e-06,
      "loss": 0.4531,
      "step": 5270
    },
    {
      "epoch": 0.28144239226033424,
      "grad_norm": 1.698567222653232,
      "learning_rate": 2.9842807517670524e-06,
      "loss": 0.4284,
      "step": 5280
    },
    {
      "epoch": 0.2819754270941606,
      "grad_norm": 1.5620879819533362,
      "learning_rate": 2.984153230860073e-06,
      "loss": 0.4339,
      "step": 5290
    },
    {
      "epoch": 0.282508461927987,
      "grad_norm": 1.9091119340297336,
      "learning_rate": 2.984025197536617e-06,
      "loss": 0.4467,
      "step": 5300
    },
    {
      "epoch": 0.2830414967618134,
      "grad_norm": 1.5726783217485447,
      "learning_rate": 2.983896651840888e-06,
      "loss": 0.4257,
      "step": 5310
    },
    {
      "epoch": 0.2835745315956398,
      "grad_norm": 1.5237104351967512,
      "learning_rate": 2.9837675938172676e-06,
      "loss": 0.4316,
      "step": 5320
    },
    {
      "epoch": 0.28410756642946616,
      "grad_norm": 1.6206240469613897,
      "learning_rate": 2.983638023510314e-06,
      "loss": 0.4405,
      "step": 5330
    },
    {
      "epoch": 0.28464060126329255,
      "grad_norm": 1.8390155270157214,
      "learning_rate": 2.983507940964763e-06,
      "loss": 0.4296,
      "step": 5340
    },
    {
      "epoch": 0.28517363609711893,
      "grad_norm": 1.5588583841602022,
      "learning_rate": 2.983377346225526e-06,
      "loss": 0.4359,
      "step": 5350
    },
    {
      "epoch": 0.2857066709309453,
      "grad_norm": 1.6656545852820501,
      "learning_rate": 2.9832462393376928e-06,
      "loss": 0.4506,
      "step": 5360
    },
    {
      "epoch": 0.2862397057647717,
      "grad_norm": 1.4371177693550143,
      "learning_rate": 2.9831146203465288e-06,
      "loss": 0.4294,
      "step": 5370
    },
    {
      "epoch": 0.28677274059859814,
      "grad_norm": 1.6977410718840877,
      "learning_rate": 2.982982489297476e-06,
      "loss": 0.4308,
      "step": 5380
    },
    {
      "epoch": 0.2873057754324245,
      "grad_norm": 1.6767286162428785,
      "learning_rate": 2.982849846236154e-06,
      "loss": 0.4405,
      "step": 5390
    },
    {
      "epoch": 0.2878388102662509,
      "grad_norm": 1.670718264926009,
      "learning_rate": 2.9827166912083594e-06,
      "loss": 0.44,
      "step": 5400
    },
    {
      "epoch": 0.2883718451000773,
      "grad_norm": 1.5229507938665452,
      "learning_rate": 2.982583024260065e-06,
      "loss": 0.444,
      "step": 5410
    },
    {
      "epoch": 0.2889048799339037,
      "grad_norm": 1.5235669298931718,
      "learning_rate": 2.9824488454374197e-06,
      "loss": 0.4299,
      "step": 5420
    },
    {
      "epoch": 0.28943791476773006,
      "grad_norm": 1.6380394830128373,
      "learning_rate": 2.9823141547867506e-06,
      "loss": 0.4298,
      "step": 5430
    },
    {
      "epoch": 0.28997094960155645,
      "grad_norm": 1.6540862277102244,
      "learning_rate": 2.982178952354561e-06,
      "loss": 0.4331,
      "step": 5440
    },
    {
      "epoch": 0.29050398443538283,
      "grad_norm": 1.677499163271086,
      "learning_rate": 2.9820432381875305e-06,
      "loss": 0.4176,
      "step": 5450
    },
    {
      "epoch": 0.2910370192692092,
      "grad_norm": 1.6015713385649804,
      "learning_rate": 2.981907012332515e-06,
      "loss": 0.4277,
      "step": 5460
    },
    {
      "epoch": 0.29157005410303566,
      "grad_norm": 1.4559848203782635,
      "learning_rate": 2.981770274836548e-06,
      "loss": 0.433,
      "step": 5470
    },
    {
      "epoch": 0.29210308893686204,
      "grad_norm": 1.664718914012717,
      "learning_rate": 2.9816330257468398e-06,
      "loss": 0.4385,
      "step": 5480
    },
    {
      "epoch": 0.2926361237706884,
      "grad_norm": 1.8270615353833766,
      "learning_rate": 2.981495265110776e-06,
      "loss": 0.4252,
      "step": 5490
    },
    {
      "epoch": 0.2931691586045148,
      "grad_norm": 1.530178399595827,
      "learning_rate": 2.9813569929759202e-06,
      "loss": 0.4336,
      "step": 5500
    },
    {
      "epoch": 0.2937021934383412,
      "grad_norm": 1.6294386231734062,
      "learning_rate": 2.9812182093900115e-06,
      "loss": 0.4283,
      "step": 5510
    },
    {
      "epoch": 0.2942352282721676,
      "grad_norm": 1.4422882137266106,
      "learning_rate": 2.981078914400966e-06,
      "loss": 0.4325,
      "step": 5520
    },
    {
      "epoch": 0.29476826310599397,
      "grad_norm": 1.677668765720838,
      "learning_rate": 2.9809391080568775e-06,
      "loss": 0.4362,
      "step": 5530
    },
    {
      "epoch": 0.29530129793982035,
      "grad_norm": 1.513264583171719,
      "learning_rate": 2.9807987904060146e-06,
      "loss": 0.4271,
      "step": 5540
    },
    {
      "epoch": 0.29583433277364674,
      "grad_norm": 1.6348742300180483,
      "learning_rate": 2.980657961496823e-06,
      "loss": 0.4361,
      "step": 5550
    },
    {
      "epoch": 0.2963673676074732,
      "grad_norm": 1.5907611755718645,
      "learning_rate": 2.9805166213779255e-06,
      "loss": 0.4386,
      "step": 5560
    },
    {
      "epoch": 0.29690040244129956,
      "grad_norm": 1.5621826881478291,
      "learning_rate": 2.9803747700981203e-06,
      "loss": 0.4259,
      "step": 5570
    },
    {
      "epoch": 0.29743343727512594,
      "grad_norm": 1.5964323571177013,
      "learning_rate": 2.980232407706384e-06,
      "loss": 0.441,
      "step": 5580
    },
    {
      "epoch": 0.29796647210895233,
      "grad_norm": 1.6482586306304368,
      "learning_rate": 2.9800895342518674e-06,
      "loss": 0.4339,
      "step": 5590
    },
    {
      "epoch": 0.2984995069427787,
      "grad_norm": 1.7961368679894048,
      "learning_rate": 2.9799461497838986e-06,
      "loss": 0.4321,
      "step": 5600
    },
    {
      "epoch": 0.2990325417766051,
      "grad_norm": 1.4773957185360056,
      "learning_rate": 2.9798022543519834e-06,
      "loss": 0.4178,
      "step": 5610
    },
    {
      "epoch": 0.2995655766104315,
      "grad_norm": 1.6191512300986841,
      "learning_rate": 2.9796578480058018e-06,
      "loss": 0.4298,
      "step": 5620
    },
    {
      "epoch": 0.30009861144425787,
      "grad_norm": 1.5817940874948935,
      "learning_rate": 2.9795129307952124e-06,
      "loss": 0.4337,
      "step": 5630
    },
    {
      "epoch": 0.30063164627808425,
      "grad_norm": 1.648708283015776,
      "learning_rate": 2.979367502770248e-06,
      "loss": 0.4401,
      "step": 5640
    },
    {
      "epoch": 0.30116468111191064,
      "grad_norm": 1.787315503059311,
      "learning_rate": 2.9792215639811193e-06,
      "loss": 0.4456,
      "step": 5650
    },
    {
      "epoch": 0.3016977159457371,
      "grad_norm": 1.6223314451023445,
      "learning_rate": 2.9790751144782125e-06,
      "loss": 0.4351,
      "step": 5660
    },
    {
      "epoch": 0.30223075077956346,
      "grad_norm": 1.431844924729601,
      "learning_rate": 2.9789281543120912e-06,
      "loss": 0.4079,
      "step": 5670
    },
    {
      "epoch": 0.30276378561338985,
      "grad_norm": 1.6867517408716806,
      "learning_rate": 2.978780683533495e-06,
      "loss": 0.4316,
      "step": 5680
    },
    {
      "epoch": 0.30329682044721623,
      "grad_norm": 1.6110745018766008,
      "learning_rate": 2.9786327021933382e-06,
      "loss": 0.4344,
      "step": 5690
    },
    {
      "epoch": 0.3038298552810426,
      "grad_norm": 1.542450302892925,
      "learning_rate": 2.978484210342714e-06,
      "loss": 0.4345,
      "step": 5700
    },
    {
      "epoch": 0.304362890114869,
      "grad_norm": 1.563208533037698,
      "learning_rate": 2.9783352080328888e-06,
      "loss": 0.4216,
      "step": 5710
    },
    {
      "epoch": 0.3048959249486954,
      "grad_norm": 1.6242438533234171,
      "learning_rate": 2.9781856953153085e-06,
      "loss": 0.4292,
      "step": 5720
    },
    {
      "epoch": 0.30542895978252177,
      "grad_norm": 1.493895541562595,
      "learning_rate": 2.9780356722415928e-06,
      "loss": 0.4229,
      "step": 5730
    },
    {
      "epoch": 0.30596199461634815,
      "grad_norm": 1.6046335270551686,
      "learning_rate": 2.9778851388635382e-06,
      "loss": 0.4376,
      "step": 5740
    },
    {
      "epoch": 0.3064950294501746,
      "grad_norm": 1.7383356432998933,
      "learning_rate": 2.9777340952331187e-06,
      "loss": 0.436,
      "step": 5750
    },
    {
      "epoch": 0.307028064284001,
      "grad_norm": 1.6825344903169306,
      "learning_rate": 2.9775825414024823e-06,
      "loss": 0.4342,
      "step": 5760
    },
    {
      "epoch": 0.30756109911782736,
      "grad_norm": 1.6270778665885282,
      "learning_rate": 2.977430477423955e-06,
      "loss": 0.4401,
      "step": 5770
    },
    {
      "epoch": 0.30809413395165375,
      "grad_norm": 1.5179321119960736,
      "learning_rate": 2.9772779033500375e-06,
      "loss": 0.4424,
      "step": 5780
    },
    {
      "epoch": 0.30862716878548013,
      "grad_norm": 1.7945348767052742,
      "learning_rate": 2.9771248192334077e-06,
      "loss": 0.4428,
      "step": 5790
    },
    {
      "epoch": 0.3091602036193065,
      "grad_norm": 1.800422032662795,
      "learning_rate": 2.976971225126919e-06,
      "loss": 0.4229,
      "step": 5800
    },
    {
      "epoch": 0.3096932384531329,
      "grad_norm": 1.681592093147011,
      "learning_rate": 2.9768171210836015e-06,
      "loss": 0.4365,
      "step": 5810
    },
    {
      "epoch": 0.3102262732869593,
      "grad_norm": 1.6179092632145686,
      "learning_rate": 2.97666250715666e-06,
      "loss": 0.4355,
      "step": 5820
    },
    {
      "epoch": 0.31075930812078567,
      "grad_norm": 1.4760324448495825,
      "learning_rate": 2.9765073833994777e-06,
      "loss": 0.4359,
      "step": 5830
    },
    {
      "epoch": 0.3112923429546121,
      "grad_norm": 1.5392970132003494,
      "learning_rate": 2.9763517498656107e-06,
      "loss": 0.4196,
      "step": 5840
    },
    {
      "epoch": 0.3118253777884385,
      "grad_norm": 1.7311081029836572,
      "learning_rate": 2.9761956066087947e-06,
      "loss": 0.4139,
      "step": 5850
    },
    {
      "epoch": 0.3123584126222649,
      "grad_norm": 1.661098479249934,
      "learning_rate": 2.9760389536829374e-06,
      "loss": 0.4304,
      "step": 5860
    },
    {
      "epoch": 0.31289144745609127,
      "grad_norm": 1.4948310451170144,
      "learning_rate": 2.975881791142126e-06,
      "loss": 0.4164,
      "step": 5870
    },
    {
      "epoch": 0.31342448228991765,
      "grad_norm": 1.5948477699862678,
      "learning_rate": 2.975724119040622e-06,
      "loss": 0.4244,
      "step": 5880
    },
    {
      "epoch": 0.31395751712374403,
      "grad_norm": 1.405655037969923,
      "learning_rate": 2.9755659374328626e-06,
      "loss": 0.423,
      "step": 5890
    },
    {
      "epoch": 0.3144905519575704,
      "grad_norm": 1.751875042948595,
      "learning_rate": 2.975407246373462e-06,
      "loss": 0.4199,
      "step": 5900
    },
    {
      "epoch": 0.3150235867913968,
      "grad_norm": 1.6307903578015341,
      "learning_rate": 2.975248045917209e-06,
      "loss": 0.4355,
      "step": 5910
    },
    {
      "epoch": 0.3155566216252232,
      "grad_norm": 1.6219989949867764,
      "learning_rate": 2.975088336119069e-06,
      "loss": 0.4223,
      "step": 5920
    },
    {
      "epoch": 0.3160896564590496,
      "grad_norm": 1.5536610478037198,
      "learning_rate": 2.974928117034183e-06,
      "loss": 0.4347,
      "step": 5930
    },
    {
      "epoch": 0.316622691292876,
      "grad_norm": 1.641102860114325,
      "learning_rate": 2.9747673887178686e-06,
      "loss": 0.4384,
      "step": 5940
    },
    {
      "epoch": 0.3171557261267024,
      "grad_norm": 1.7237900940901483,
      "learning_rate": 2.974606151225618e-06,
      "loss": 0.4302,
      "step": 5950
    },
    {
      "epoch": 0.3176887609605288,
      "grad_norm": 1.6168013083983608,
      "learning_rate": 2.9744444046131002e-06,
      "loss": 0.4472,
      "step": 5960
    },
    {
      "epoch": 0.31822179579435517,
      "grad_norm": 1.5470949319976988,
      "learning_rate": 2.97428214893616e-06,
      "loss": 0.4418,
      "step": 5970
    },
    {
      "epoch": 0.31875483062818155,
      "grad_norm": 1.4250231775624402,
      "learning_rate": 2.9741193842508162e-06,
      "loss": 0.4311,
      "step": 5980
    },
    {
      "epoch": 0.31928786546200794,
      "grad_norm": 1.6151878496603973,
      "learning_rate": 2.973956110613266e-06,
      "loss": 0.4318,
      "step": 5990
    },
    {
      "epoch": 0.3198209002958343,
      "grad_norm": 1.5052112012216967,
      "learning_rate": 2.9737923280798806e-06,
      "loss": 0.4339,
      "step": 6000
    },
    {
      "epoch": 0.3203539351296607,
      "grad_norm": 1.5141431267749252,
      "learning_rate": 2.9736280367072067e-06,
      "loss": 0.4384,
      "step": 6010
    },
    {
      "epoch": 0.3208869699634871,
      "grad_norm": 1.5849168683837396,
      "learning_rate": 2.9734632365519678e-06,
      "loss": 0.4345,
      "step": 6020
    },
    {
      "epoch": 0.32142000479731353,
      "grad_norm": 1.5358308949250419,
      "learning_rate": 2.9732979276710633e-06,
      "loss": 0.423,
      "step": 6030
    },
    {
      "epoch": 0.3219530396311399,
      "grad_norm": 1.5538071002008005,
      "learning_rate": 2.973132110121566e-06,
      "loss": 0.4109,
      "step": 6040
    },
    {
      "epoch": 0.3224860744649663,
      "grad_norm": 1.5521549886236965,
      "learning_rate": 2.972965783960727e-06,
      "loss": 0.4385,
      "step": 6050
    },
    {
      "epoch": 0.3230191092987927,
      "grad_norm": 1.5698741983940294,
      "learning_rate": 2.972798949245971e-06,
      "loss": 0.4332,
      "step": 6060
    },
    {
      "epoch": 0.32355214413261907,
      "grad_norm": 1.6140370750755964,
      "learning_rate": 2.9726316060349005e-06,
      "loss": 0.4193,
      "step": 6070
    },
    {
      "epoch": 0.32408517896644545,
      "grad_norm": 1.464931396391657,
      "learning_rate": 2.9724637543852902e-06,
      "loss": 0.4364,
      "step": 6080
    },
    {
      "epoch": 0.32461821380027184,
      "grad_norm": 1.8637631354477113,
      "learning_rate": 2.9722953943550935e-06,
      "loss": 0.43,
      "step": 6090
    },
    {
      "epoch": 0.3251512486340982,
      "grad_norm": 1.549463043157198,
      "learning_rate": 2.9721265260024375e-06,
      "loss": 0.4284,
      "step": 6100
    },
    {
      "epoch": 0.3256842834679246,
      "grad_norm": 1.5134739981809822,
      "learning_rate": 2.9719571493856265e-06,
      "loss": 0.4274,
      "step": 6110
    },
    {
      "epoch": 0.32621731830175105,
      "grad_norm": 1.6343886349274972,
      "learning_rate": 2.971787264563138e-06,
      "loss": 0.4396,
      "step": 6120
    },
    {
      "epoch": 0.32675035313557743,
      "grad_norm": 1.6616364072107324,
      "learning_rate": 2.971616871593627e-06,
      "loss": 0.4334,
      "step": 6130
    },
    {
      "epoch": 0.3272833879694038,
      "grad_norm": 1.678041669454987,
      "learning_rate": 2.9714459705359226e-06,
      "loss": 0.428,
      "step": 6140
    },
    {
      "epoch": 0.3278164228032302,
      "grad_norm": 1.6647908349147245,
      "learning_rate": 2.97127456144903e-06,
      "loss": 0.4276,
      "step": 6150
    },
    {
      "epoch": 0.3283494576370566,
      "grad_norm": 1.8462658095224402,
      "learning_rate": 2.9711026443921298e-06,
      "loss": 0.4302,
      "step": 6160
    },
    {
      "epoch": 0.32888249247088297,
      "grad_norm": 1.5878897605642863,
      "learning_rate": 2.9709302194245776e-06,
      "loss": 0.4227,
      "step": 6170
    },
    {
      "epoch": 0.32941552730470935,
      "grad_norm": 1.706024186802193,
      "learning_rate": 2.970757286605905e-06,
      "loss": 0.4259,
      "step": 6180
    },
    {
      "epoch": 0.32994856213853574,
      "grad_norm": 1.7177152431099663,
      "learning_rate": 2.970583845995818e-06,
      "loss": 0.4139,
      "step": 6190
    },
    {
      "epoch": 0.3304815969723621,
      "grad_norm": 1.527582656257654,
      "learning_rate": 2.970409897654198e-06,
      "loss": 0.4187,
      "step": 6200
    },
    {
      "epoch": 0.3310146318061885,
      "grad_norm": 1.6131152731146399,
      "learning_rate": 2.970235441641104e-06,
      "loss": 0.4278,
      "step": 6210
    },
    {
      "epoch": 0.33154766664001495,
      "grad_norm": 1.9354559182132087,
      "learning_rate": 2.9700604780167667e-06,
      "loss": 0.438,
      "step": 6220
    },
    {
      "epoch": 0.33208070147384133,
      "grad_norm": 1.749048160771119,
      "learning_rate": 2.969885006841594e-06,
      "loss": 0.4481,
      "step": 6230
    },
    {
      "epoch": 0.3326137363076677,
      "grad_norm": 1.645947297635966,
      "learning_rate": 2.9697090281761696e-06,
      "loss": 0.4265,
      "step": 6240
    },
    {
      "epoch": 0.3331467711414941,
      "grad_norm": 1.6999800045470679,
      "learning_rate": 2.9695325420812506e-06,
      "loss": 0.4182,
      "step": 6250
    },
    {
      "epoch": 0.3336798059753205,
      "grad_norm": 1.7444989233639856,
      "learning_rate": 2.969355548617771e-06,
      "loss": 0.4438,
      "step": 6260
    },
    {
      "epoch": 0.33421284080914687,
      "grad_norm": 1.5392374214355635,
      "learning_rate": 2.969178047846839e-06,
      "loss": 0.4362,
      "step": 6270
    },
    {
      "epoch": 0.33474587564297326,
      "grad_norm": 1.4600387222080204,
      "learning_rate": 2.9690000398297383e-06,
      "loss": 0.4274,
      "step": 6280
    },
    {
      "epoch": 0.33527891047679964,
      "grad_norm": 1.5442277557133366,
      "learning_rate": 2.9688215246279284e-06,
      "loss": 0.423,
      "step": 6290
    },
    {
      "epoch": 0.335811945310626,
      "grad_norm": 1.501160320948494,
      "learning_rate": 2.9686425023030418e-06,
      "loss": 0.4089,
      "step": 6300
    },
    {
      "epoch": 0.33634498014445247,
      "grad_norm": 1.5531044566240355,
      "learning_rate": 2.9684629729168886e-06,
      "loss": 0.4383,
      "step": 6310
    },
    {
      "epoch": 0.33687801497827885,
      "grad_norm": 1.720673900539268,
      "learning_rate": 2.968282936531452e-06,
      "loss": 0.4306,
      "step": 6320
    },
    {
      "epoch": 0.33741104981210523,
      "grad_norm": 1.7350909267725416,
      "learning_rate": 2.968102393208892e-06,
      "loss": 0.4485,
      "step": 6330
    },
    {
      "epoch": 0.3379440846459316,
      "grad_norm": 1.5377549780201976,
      "learning_rate": 2.9679213430115425e-06,
      "loss": 0.4363,
      "step": 6340
    },
    {
      "epoch": 0.338477119479758,
      "grad_norm": 1.5606358850945456,
      "learning_rate": 2.967739786001912e-06,
      "loss": 0.441,
      "step": 6350
    },
    {
      "epoch": 0.3390101543135844,
      "grad_norm": 1.616705804620765,
      "learning_rate": 2.9675577222426846e-06,
      "loss": 0.435,
      "step": 6360
    },
    {
      "epoch": 0.3395431891474108,
      "grad_norm": 1.7178853286696927,
      "learning_rate": 2.96737515179672e-06,
      "loss": 0.4267,
      "step": 6370
    },
    {
      "epoch": 0.34007622398123716,
      "grad_norm": 1.6173103553224393,
      "learning_rate": 2.967192074727052e-06,
      "loss": 0.4235,
      "step": 6380
    },
    {
      "epoch": 0.34060925881506354,
      "grad_norm": 1.6221072249479735,
      "learning_rate": 2.9670084910968893e-06,
      "loss": 0.4413,
      "step": 6390
    },
    {
      "epoch": 0.34114229364889,
      "grad_norm": 1.49390225963427,
      "learning_rate": 2.966824400969616e-06,
      "loss": 0.4346,
      "step": 6400
    },
    {
      "epoch": 0.34167532848271637,
      "grad_norm": 1.825090247515385,
      "learning_rate": 2.96663980440879e-06,
      "loss": 0.4218,
      "step": 6410
    },
    {
      "epoch": 0.34220836331654275,
      "grad_norm": 1.7278060711820127,
      "learning_rate": 2.966454701478146e-06,
      "loss": 0.4207,
      "step": 6420
    },
    {
      "epoch": 0.34274139815036914,
      "grad_norm": 1.5568723067877832,
      "learning_rate": 2.966269092241592e-06,
      "loss": 0.4337,
      "step": 6430
    },
    {
      "epoch": 0.3432744329841955,
      "grad_norm": 1.6662446944834262,
      "learning_rate": 2.96608297676321e-06,
      "loss": 0.4208,
      "step": 6440
    },
    {
      "epoch": 0.3438074678180219,
      "grad_norm": 1.614855733975428,
      "learning_rate": 2.96589635510726e-06,
      "loss": 0.4136,
      "step": 6450
    },
    {
      "epoch": 0.3443405026518483,
      "grad_norm": 1.5933718003086859,
      "learning_rate": 2.9657092273381728e-06,
      "loss": 0.4203,
      "step": 6460
    },
    {
      "epoch": 0.3448735374856747,
      "grad_norm": 1.7060118616852338,
      "learning_rate": 2.965521593520557e-06,
      "loss": 0.4253,
      "step": 6470
    },
    {
      "epoch": 0.34540657231950106,
      "grad_norm": 1.6310484172430664,
      "learning_rate": 2.9653334537191944e-06,
      "loss": 0.4334,
      "step": 6480
    },
    {
      "epoch": 0.34593960715332744,
      "grad_norm": 1.466491307785643,
      "learning_rate": 2.9651448079990417e-06,
      "loss": 0.418,
      "step": 6490
    },
    {
      "epoch": 0.3464726419871539,
      "grad_norm": 1.5697743652363598,
      "learning_rate": 2.964955656425231e-06,
      "loss": 0.4306,
      "step": 6500
    },
    {
      "epoch": 0.34700567682098027,
      "grad_norm": 1.5972880783969272,
      "learning_rate": 2.9647659990630675e-06,
      "loss": 0.4185,
      "step": 6510
    },
    {
      "epoch": 0.34753871165480665,
      "grad_norm": 1.507300869963076,
      "learning_rate": 2.9645758359780326e-06,
      "loss": 0.4264,
      "step": 6520
    },
    {
      "epoch": 0.34807174648863304,
      "grad_norm": 1.5407374887590906,
      "learning_rate": 2.9643851672357822e-06,
      "loss": 0.425,
      "step": 6530
    },
    {
      "epoch": 0.3486047813224594,
      "grad_norm": 1.638447449847881,
      "learning_rate": 2.9641939929021457e-06,
      "loss": 0.4246,
      "step": 6540
    },
    {
      "epoch": 0.3491378161562858,
      "grad_norm": 1.6997729045022967,
      "learning_rate": 2.9640023130431273e-06,
      "loss": 0.446,
      "step": 6550
    },
    {
      "epoch": 0.3496708509901122,
      "grad_norm": 1.6622712567874156,
      "learning_rate": 2.963810127724907e-06,
      "loss": 0.4286,
      "step": 6560
    },
    {
      "epoch": 0.3502038858239386,
      "grad_norm": 1.5191238753665401,
      "learning_rate": 2.963617437013838e-06,
      "loss": 0.4094,
      "step": 6570
    },
    {
      "epoch": 0.35073692065776496,
      "grad_norm": 1.6061869639350372,
      "learning_rate": 2.9634242409764483e-06,
      "loss": 0.4241,
      "step": 6580
    },
    {
      "epoch": 0.3512699554915914,
      "grad_norm": 1.6408748077375865,
      "learning_rate": 2.96323053967944e-06,
      "loss": 0.4515,
      "step": 6590
    },
    {
      "epoch": 0.3518029903254178,
      "grad_norm": 1.5519063034535103,
      "learning_rate": 2.963036333189691e-06,
      "loss": 0.4272,
      "step": 6600
    },
    {
      "epoch": 0.35233602515924417,
      "grad_norm": 1.7048124018824924,
      "learning_rate": 2.962841621574253e-06,
      "loss": 0.4301,
      "step": 6610
    },
    {
      "epoch": 0.35286905999307056,
      "grad_norm": 1.633526500844667,
      "learning_rate": 2.9626464049003507e-06,
      "loss": 0.4432,
      "step": 6620
    },
    {
      "epoch": 0.35340209482689694,
      "grad_norm": 1.4921691584219783,
      "learning_rate": 2.962450683235385e-06,
      "loss": 0.4149,
      "step": 6630
    },
    {
      "epoch": 0.3539351296607233,
      "grad_norm": 1.766329625149137,
      "learning_rate": 2.9622544566469303e-06,
      "loss": 0.4264,
      "step": 6640
    },
    {
      "epoch": 0.3544681644945497,
      "grad_norm": 1.5875140369840268,
      "learning_rate": 2.9620577252027346e-06,
      "loss": 0.4376,
      "step": 6650
    },
    {
      "epoch": 0.3550011993283761,
      "grad_norm": 1.4506568654201621,
      "learning_rate": 2.9618604889707233e-06,
      "loss": 0.4296,
      "step": 6660
    },
    {
      "epoch": 0.3555342341622025,
      "grad_norm": 1.6275768055961044,
      "learning_rate": 2.9616627480189917e-06,
      "loss": 0.4236,
      "step": 6670
    },
    {
      "epoch": 0.35606726899602886,
      "grad_norm": 1.5301223863021702,
      "learning_rate": 2.961464502415813e-06,
      "loss": 0.4236,
      "step": 6680
    },
    {
      "epoch": 0.3566003038298553,
      "grad_norm": 1.5631224569949131,
      "learning_rate": 2.961265752229632e-06,
      "loss": 0.4322,
      "step": 6690
    },
    {
      "epoch": 0.3571333386636817,
      "grad_norm": 1.67832387761322,
      "learning_rate": 2.9610664975290697e-06,
      "loss": 0.4318,
      "step": 6700
    },
    {
      "epoch": 0.35766637349750807,
      "grad_norm": 1.5596583006537503,
      "learning_rate": 2.9608667383829204e-06,
      "loss": 0.4156,
      "step": 6710
    },
    {
      "epoch": 0.35819940833133446,
      "grad_norm": 1.7075693422562126,
      "learning_rate": 2.9606664748601527e-06,
      "loss": 0.4269,
      "step": 6720
    },
    {
      "epoch": 0.35873244316516084,
      "grad_norm": 1.707483865521375,
      "learning_rate": 2.9604657070299084e-06,
      "loss": 0.4156,
      "step": 6730
    },
    {
      "epoch": 0.3592654779989872,
      "grad_norm": 1.6991749329911974,
      "learning_rate": 2.9602644349615054e-06,
      "loss": 0.4376,
      "step": 6740
    },
    {
      "epoch": 0.3597985128328136,
      "grad_norm": 1.617189763978349,
      "learning_rate": 2.9600626587244344e-06,
      "loss": 0.4201,
      "step": 6750
    },
    {
      "epoch": 0.36033154766664,
      "grad_norm": 1.6928129337056317,
      "learning_rate": 2.9598603783883594e-06,
      "loss": 0.4264,
      "step": 6760
    },
    {
      "epoch": 0.3608645825004664,
      "grad_norm": 1.6973673604437498,
      "learning_rate": 2.9596575940231205e-06,
      "loss": 0.426,
      "step": 6770
    },
    {
      "epoch": 0.3613976173342928,
      "grad_norm": 1.6584859170233988,
      "learning_rate": 2.95945430569873e-06,
      "loss": 0.4243,
      "step": 6780
    },
    {
      "epoch": 0.3619306521681192,
      "grad_norm": 1.69323846893125,
      "learning_rate": 2.959250513485376e-06,
      "loss": 0.4212,
      "step": 6790
    },
    {
      "epoch": 0.3624636870019456,
      "grad_norm": 1.6544006172694286,
      "learning_rate": 2.959046217453418e-06,
      "loss": 0.425,
      "step": 6800
    },
    {
      "epoch": 0.362996721835772,
      "grad_norm": 1.7845355812314003,
      "learning_rate": 2.9588414176733913e-06,
      "loss": 0.4082,
      "step": 6810
    },
    {
      "epoch": 0.36352975666959836,
      "grad_norm": 1.6598532140728652,
      "learning_rate": 2.9586361142160055e-06,
      "loss": 0.433,
      "step": 6820
    },
    {
      "epoch": 0.36406279150342474,
      "grad_norm": 1.5263144632372705,
      "learning_rate": 2.958430307152143e-06,
      "loss": 0.4144,
      "step": 6830
    },
    {
      "epoch": 0.3645958263372511,
      "grad_norm": 1.4939023195672378,
      "learning_rate": 2.9582239965528606e-06,
      "loss": 0.4332,
      "step": 6840
    },
    {
      "epoch": 0.3651288611710775,
      "grad_norm": 1.5667616611848862,
      "learning_rate": 2.9580171824893887e-06,
      "loss": 0.4229,
      "step": 6850
    },
    {
      "epoch": 0.3656618960049039,
      "grad_norm": 1.5669429935745147,
      "learning_rate": 2.9578098650331312e-06,
      "loss": 0.4065,
      "step": 6860
    },
    {
      "epoch": 0.36619493083873034,
      "grad_norm": 1.433683483555676,
      "learning_rate": 2.957602044255667e-06,
      "loss": 0.403,
      "step": 6870
    },
    {
      "epoch": 0.3667279656725567,
      "grad_norm": 1.4912206002853188,
      "learning_rate": 2.957393720228747e-06,
      "loss": 0.4035,
      "step": 6880
    },
    {
      "epoch": 0.3672610005063831,
      "grad_norm": 1.504232794105487,
      "learning_rate": 2.9571848930242975e-06,
      "loss": 0.4161,
      "step": 6890
    },
    {
      "epoch": 0.3677940353402095,
      "grad_norm": 1.650561754017296,
      "learning_rate": 2.9569755627144175e-06,
      "loss": 0.4337,
      "step": 6900
    },
    {
      "epoch": 0.3683270701740359,
      "grad_norm": 1.7591672315611653,
      "learning_rate": 2.95676572937138e-06,
      "loss": 0.4285,
      "step": 6910
    },
    {
      "epoch": 0.36886010500786226,
      "grad_norm": 1.60301610622303,
      "learning_rate": 2.9565553930676323e-06,
      "loss": 0.421,
      "step": 6920
    },
    {
      "epoch": 0.36939313984168864,
      "grad_norm": 1.5168789019932878,
      "learning_rate": 2.956344553875795e-06,
      "loss": 0.4228,
      "step": 6930
    },
    {
      "epoch": 0.36992617467551503,
      "grad_norm": 1.5940718104914382,
      "learning_rate": 2.956133211868661e-06,
      "loss": 0.4144,
      "step": 6940
    },
    {
      "epoch": 0.3704592095093414,
      "grad_norm": 1.6099433387984643,
      "learning_rate": 2.9559213671191983e-06,
      "loss": 0.4129,
      "step": 6950
    },
    {
      "epoch": 0.3709922443431678,
      "grad_norm": 1.5992118581782357,
      "learning_rate": 2.955709019700549e-06,
      "loss": 0.4266,
      "step": 6960
    },
    {
      "epoch": 0.37152527917699424,
      "grad_norm": 2.0938413496975716,
      "learning_rate": 2.9554961696860263e-06,
      "loss": 0.426,
      "step": 6970
    },
    {
      "epoch": 0.3720583140108206,
      "grad_norm": 1.579150885321858,
      "learning_rate": 2.95528281714912e-06,
      "loss": 0.4125,
      "step": 6980
    },
    {
      "epoch": 0.372591348844647,
      "grad_norm": 1.3469684322843098,
      "learning_rate": 2.9550689621634906e-06,
      "loss": 0.4124,
      "step": 6990
    },
    {
      "epoch": 0.3731243836784734,
      "grad_norm": 1.4689342197584156,
      "learning_rate": 2.9548546048029745e-06,
      "loss": 0.3875,
      "step": 7000
    },
    {
      "epoch": 0.3736574185122998,
      "grad_norm": 1.3959784585326207,
      "learning_rate": 2.9546397451415796e-06,
      "loss": 0.4246,
      "step": 7010
    },
    {
      "epoch": 0.37419045334612616,
      "grad_norm": 1.6699340776550209,
      "learning_rate": 2.9544243832534882e-06,
      "loss": 0.4168,
      "step": 7020
    },
    {
      "epoch": 0.37472348817995255,
      "grad_norm": 1.7791707413975577,
      "learning_rate": 2.954208519213056e-06,
      "loss": 0.4167,
      "step": 7030
    },
    {
      "epoch": 0.37525652301377893,
      "grad_norm": 1.5243558486319142,
      "learning_rate": 2.953992153094812e-06,
      "loss": 0.4231,
      "step": 7040
    },
    {
      "epoch": 0.3757895578476053,
      "grad_norm": 1.652255628611805,
      "learning_rate": 2.953775284973458e-06,
      "loss": 0.4427,
      "step": 7050
    },
    {
      "epoch": 0.37632259268143176,
      "grad_norm": 1.6337680171819196,
      "learning_rate": 2.9535579149238694e-06,
      "loss": 0.4164,
      "step": 7060
    },
    {
      "epoch": 0.37685562751525814,
      "grad_norm": 1.4421046818351917,
      "learning_rate": 2.9533400430210956e-06,
      "loss": 0.4221,
      "step": 7070
    },
    {
      "epoch": 0.3773886623490845,
      "grad_norm": 1.5573453388988545,
      "learning_rate": 2.9531216693403585e-06,
      "loss": 0.4279,
      "step": 7080
    },
    {
      "epoch": 0.3779216971829109,
      "grad_norm": 1.5800406967377374,
      "learning_rate": 2.9529027939570536e-06,
      "loss": 0.4197,
      "step": 7090
    },
    {
      "epoch": 0.3784547320167373,
      "grad_norm": 1.5852799254910612,
      "learning_rate": 2.9526834169467496e-06,
      "loss": 0.4353,
      "step": 7100
    },
    {
      "epoch": 0.3789877668505637,
      "grad_norm": 1.4653958075983655,
      "learning_rate": 2.952463538385188e-06,
      "loss": 0.4287,
      "step": 7110
    },
    {
      "epoch": 0.37952080168439006,
      "grad_norm": 1.5555537517929727,
      "learning_rate": 2.9522431583482834e-06,
      "loss": 0.437,
      "step": 7120
    },
    {
      "epoch": 0.38005383651821645,
      "grad_norm": 1.5173481914084632,
      "learning_rate": 2.9520222769121242e-06,
      "loss": 0.4103,
      "step": 7130
    },
    {
      "epoch": 0.38058687135204283,
      "grad_norm": 1.5005103614495325,
      "learning_rate": 2.9518008941529715e-06,
      "loss": 0.4204,
      "step": 7140
    },
    {
      "epoch": 0.3811199061858693,
      "grad_norm": 1.5431517066178606,
      "learning_rate": 2.95157901014726e-06,
      "loss": 0.4161,
      "step": 7150
    },
    {
      "epoch": 0.38165294101969566,
      "grad_norm": 1.5104382693845257,
      "learning_rate": 2.951356624971597e-06,
      "loss": 0.4111,
      "step": 7160
    },
    {
      "epoch": 0.38218597585352204,
      "grad_norm": 1.854950512809969,
      "learning_rate": 2.9511337387027623e-06,
      "loss": 0.4396,
      "step": 7170
    },
    {
      "epoch": 0.3827190106873484,
      "grad_norm": 1.5589332316311124,
      "learning_rate": 2.95091035141771e-06,
      "loss": 0.4146,
      "step": 7180
    },
    {
      "epoch": 0.3832520455211748,
      "grad_norm": 1.5159826140520984,
      "learning_rate": 2.9506864631935658e-06,
      "loss": 0.4174,
      "step": 7190
    },
    {
      "epoch": 0.3837850803550012,
      "grad_norm": 1.5324743603681843,
      "learning_rate": 2.950462074107629e-06,
      "loss": 0.4019,
      "step": 7200
    },
    {
      "epoch": 0.3843181151888276,
      "grad_norm": 1.6573051677286261,
      "learning_rate": 2.950237184237372e-06,
      "loss": 0.4148,
      "step": 7210
    },
    {
      "epoch": 0.38485115002265396,
      "grad_norm": 1.7077462727011432,
      "learning_rate": 2.9500117936604405e-06,
      "loss": 0.4136,
      "step": 7220
    },
    {
      "epoch": 0.38538418485648035,
      "grad_norm": 1.7161341232616143,
      "learning_rate": 2.9497859024546517e-06,
      "loss": 0.4194,
      "step": 7230
    },
    {
      "epoch": 0.38591721969030673,
      "grad_norm": 1.642902871832163,
      "learning_rate": 2.9495595106979966e-06,
      "loss": 0.4224,
      "step": 7240
    },
    {
      "epoch": 0.3864502545241332,
      "grad_norm": 2.0018461107067735,
      "learning_rate": 2.9493326184686394e-06,
      "loss": 0.4141,
      "step": 7250
    },
    {
      "epoch": 0.38698328935795956,
      "grad_norm": 1.5791876695803535,
      "learning_rate": 2.949105225844916e-06,
      "loss": 0.4292,
      "step": 7260
    },
    {
      "epoch": 0.38751632419178594,
      "grad_norm": 1.578545563864404,
      "learning_rate": 2.9488773329053357e-06,
      "loss": 0.4032,
      "step": 7270
    },
    {
      "epoch": 0.38804935902561233,
      "grad_norm": 1.6503153659541419,
      "learning_rate": 2.9486489397285807e-06,
      "loss": 0.4246,
      "step": 7280
    },
    {
      "epoch": 0.3885823938594387,
      "grad_norm": 1.6620855149679803,
      "learning_rate": 2.9484200463935058e-06,
      "loss": 0.4058,
      "step": 7290
    },
    {
      "epoch": 0.3891154286932651,
      "grad_norm": 1.4629456758669772,
      "learning_rate": 2.948190652979138e-06,
      "loss": 0.4098,
      "step": 7300
    },
    {
      "epoch": 0.3896484635270915,
      "grad_norm": 1.6222819828694544,
      "learning_rate": 2.9479607595646775e-06,
      "loss": 0.4194,
      "step": 7310
    },
    {
      "epoch": 0.39018149836091787,
      "grad_norm": 1.62215954812823,
      "learning_rate": 2.947730366229497e-06,
      "loss": 0.408,
      "step": 7320
    },
    {
      "epoch": 0.39071453319474425,
      "grad_norm": 1.5370123610447388,
      "learning_rate": 2.9474994730531414e-06,
      "loss": 0.4178,
      "step": 7330
    },
    {
      "epoch": 0.3912475680285707,
      "grad_norm": 1.5638120823993014,
      "learning_rate": 2.947268080115329e-06,
      "loss": 0.4094,
      "step": 7340
    },
    {
      "epoch": 0.3917806028623971,
      "grad_norm": 1.5482151621238354,
      "learning_rate": 2.9470361874959496e-06,
      "loss": 0.4308,
      "step": 7350
    },
    {
      "epoch": 0.39231363769622346,
      "grad_norm": 1.6338482214479055,
      "learning_rate": 2.946803795275067e-06,
      "loss": 0.4167,
      "step": 7360
    },
    {
      "epoch": 0.39284667253004985,
      "grad_norm": 1.7199532895827334,
      "learning_rate": 2.9465709035329157e-06,
      "loss": 0.4089,
      "step": 7370
    },
    {
      "epoch": 0.39337970736387623,
      "grad_norm": 1.4099147454250378,
      "learning_rate": 2.9463375123499036e-06,
      "loss": 0.416,
      "step": 7380
    },
    {
      "epoch": 0.3939127421977026,
      "grad_norm": 1.398138412374819,
      "learning_rate": 2.946103621806612e-06,
      "loss": 0.4078,
      "step": 7390
    },
    {
      "epoch": 0.394445777031529,
      "grad_norm": 1.5471801294339103,
      "learning_rate": 2.9458692319837927e-06,
      "loss": 0.4185,
      "step": 7400
    },
    {
      "epoch": 0.3949788118653554,
      "grad_norm": 1.4259788957885504,
      "learning_rate": 2.9456343429623705e-06,
      "loss": 0.4272,
      "step": 7410
    },
    {
      "epoch": 0.39551184669918177,
      "grad_norm": 1.5519588853612656,
      "learning_rate": 2.9453989548234435e-06,
      "loss": 0.4133,
      "step": 7420
    },
    {
      "epoch": 0.3960448815330082,
      "grad_norm": 1.5652354183527466,
      "learning_rate": 2.945163067648281e-06,
      "loss": 0.4345,
      "step": 7430
    },
    {
      "epoch": 0.3965779163668346,
      "grad_norm": 1.6600246160799446,
      "learning_rate": 2.9449266815183254e-06,
      "loss": 0.426,
      "step": 7440
    },
    {
      "epoch": 0.397110951200661,
      "grad_norm": 1.6229649686580363,
      "learning_rate": 2.9446897965151906e-06,
      "loss": 0.4022,
      "step": 7450
    },
    {
      "epoch": 0.39764398603448736,
      "grad_norm": 1.6743591863411296,
      "learning_rate": 2.9444524127206638e-06,
      "loss": 0.4252,
      "step": 7460
    },
    {
      "epoch": 0.39817702086831375,
      "grad_norm": 1.4605448278499182,
      "learning_rate": 2.9442145302167033e-06,
      "loss": 0.4316,
      "step": 7470
    },
    {
      "epoch": 0.39871005570214013,
      "grad_norm": 1.5880113421344213,
      "learning_rate": 2.94397614908544e-06,
      "loss": 0.4264,
      "step": 7480
    },
    {
      "epoch": 0.3992430905359665,
      "grad_norm": 1.6206534767222953,
      "learning_rate": 2.943737269409177e-06,
      "loss": 0.398,
      "step": 7490
    },
    {
      "epoch": 0.3997761253697929,
      "grad_norm": 1.5889574564873155,
      "learning_rate": 2.9434978912703902e-06,
      "loss": 0.4281,
      "step": 7500
    },
    {
      "epoch": 0.4003091602036193,
      "grad_norm": 1.8813772021245978,
      "learning_rate": 2.943258014751726e-06,
      "loss": 0.4253,
      "step": 7510
    },
    {
      "epoch": 0.40084219503744567,
      "grad_norm": 1.6030100273647854,
      "learning_rate": 2.943017639936004e-06,
      "loss": 0.4198,
      "step": 7520
    },
    {
      "epoch": 0.4013752298712721,
      "grad_norm": 1.5762066109514417,
      "learning_rate": 2.942776766906216e-06,
      "loss": 0.4103,
      "step": 7530
    },
    {
      "epoch": 0.4019082647050985,
      "grad_norm": 1.5983110610277405,
      "learning_rate": 2.9425353957455257e-06,
      "loss": 0.4218,
      "step": 7540
    },
    {
      "epoch": 0.4024412995389249,
      "grad_norm": 1.723442508143728,
      "learning_rate": 2.9422935265372677e-06,
      "loss": 0.4211,
      "step": 7550
    },
    {
      "epoch": 0.40297433437275126,
      "grad_norm": 1.844308496148667,
      "learning_rate": 2.9420511593649502e-06,
      "loss": 0.4297,
      "step": 7560
    },
    {
      "epoch": 0.40350736920657765,
      "grad_norm": 1.648711969216378,
      "learning_rate": 2.941808294312252e-06,
      "loss": 0.4226,
      "step": 7570
    },
    {
      "epoch": 0.40404040404040403,
      "grad_norm": 1.679764401534548,
      "learning_rate": 2.9415649314630245e-06,
      "loss": 0.4052,
      "step": 7580
    },
    {
      "epoch": 0.4045734388742304,
      "grad_norm": 1.7239791386058514,
      "learning_rate": 2.9413210709012906e-06,
      "loss": 0.4175,
      "step": 7590
    },
    {
      "epoch": 0.4051064737080568,
      "grad_norm": 1.5555799360749905,
      "learning_rate": 2.9410767127112454e-06,
      "loss": 0.413,
      "step": 7600
    },
    {
      "epoch": 0.4056395085418832,
      "grad_norm": 1.546525208250271,
      "learning_rate": 2.9408318569772556e-06,
      "loss": 0.4283,
      "step": 7610
    },
    {
      "epoch": 0.4061725433757096,
      "grad_norm": 1.5047584974914554,
      "learning_rate": 2.94058650378386e-06,
      "loss": 0.4174,
      "step": 7620
    },
    {
      "epoch": 0.406705578209536,
      "grad_norm": 1.80382041917988,
      "learning_rate": 2.9403406532157685e-06,
      "loss": 0.4041,
      "step": 7630
    },
    {
      "epoch": 0.4072386130433624,
      "grad_norm": 1.5916222482663263,
      "learning_rate": 2.940094305357863e-06,
      "loss": 0.4084,
      "step": 7640
    },
    {
      "epoch": 0.4077716478771888,
      "grad_norm": 1.699528886818383,
      "learning_rate": 2.9398474602951973e-06,
      "loss": 0.4205,
      "step": 7650
    },
    {
      "epoch": 0.40830468271101517,
      "grad_norm": 1.5201243647034341,
      "learning_rate": 2.939600118112997e-06,
      "loss": 0.412,
      "step": 7660
    },
    {
      "epoch": 0.40883771754484155,
      "grad_norm": 1.647707954049428,
      "learning_rate": 2.939352278896659e-06,
      "loss": 0.4258,
      "step": 7670
    },
    {
      "epoch": 0.40937075237866793,
      "grad_norm": 1.5593164657936662,
      "learning_rate": 2.9391039427317514e-06,
      "loss": 0.4119,
      "step": 7680
    },
    {
      "epoch": 0.4099037872124943,
      "grad_norm": 1.6064711590170297,
      "learning_rate": 2.938855109704015e-06,
      "loss": 0.4077,
      "step": 7690
    },
    {
      "epoch": 0.4104368220463207,
      "grad_norm": 1.5577253551905574,
      "learning_rate": 2.938605779899362e-06,
      "loss": 0.4137,
      "step": 7700
    },
    {
      "epoch": 0.41096985688014714,
      "grad_norm": 1.803277144438798,
      "learning_rate": 2.9383559534038737e-06,
      "loss": 0.4312,
      "step": 7710
    },
    {
      "epoch": 0.41150289171397353,
      "grad_norm": 1.4249147439088885,
      "learning_rate": 2.9381056303038068e-06,
      "loss": 0.4081,
      "step": 7720
    },
    {
      "epoch": 0.4120359265477999,
      "grad_norm": 1.485248645330218,
      "learning_rate": 2.9378548106855864e-06,
      "loss": 0.419,
      "step": 7730
    },
    {
      "epoch": 0.4125689613816263,
      "grad_norm": 1.6704245434888088,
      "learning_rate": 2.9376034946358103e-06,
      "loss": 0.4234,
      "step": 7740
    },
    {
      "epoch": 0.4131019962154527,
      "grad_norm": 1.4781501841453533,
      "learning_rate": 2.9373516822412476e-06,
      "loss": 0.4194,
      "step": 7750
    },
    {
      "epoch": 0.41363503104927907,
      "grad_norm": 1.5847656324884918,
      "learning_rate": 2.9370993735888386e-06,
      "loss": 0.4036,
      "step": 7760
    },
    {
      "epoch": 0.41416806588310545,
      "grad_norm": 1.4994033270444906,
      "learning_rate": 2.9368465687656953e-06,
      "loss": 0.4126,
      "step": 7770
    },
    {
      "epoch": 0.41470110071693184,
      "grad_norm": 1.6230128273969524,
      "learning_rate": 2.9365932678591004e-06,
      "loss": 0.4097,
      "step": 7780
    },
    {
      "epoch": 0.4152341355507582,
      "grad_norm": 1.4459593408420588,
      "learning_rate": 2.936339470956508e-06,
      "loss": 0.4182,
      "step": 7790
    },
    {
      "epoch": 0.4157671703845846,
      "grad_norm": 1.6286273125696396,
      "learning_rate": 2.936085178145544e-06,
      "loss": 0.4233,
      "step": 7800
    },
    {
      "epoch": 0.41630020521841105,
      "grad_norm": 1.4223471036220727,
      "learning_rate": 2.935830389514005e-06,
      "loss": 0.4178,
      "step": 7810
    },
    {
      "epoch": 0.41683324005223743,
      "grad_norm": 1.5905822093442257,
      "learning_rate": 2.9355751051498593e-06,
      "loss": 0.3976,
      "step": 7820
    },
    {
      "epoch": 0.4173662748860638,
      "grad_norm": 1.5301889702526843,
      "learning_rate": 2.935319325141245e-06,
      "loss": 0.4043,
      "step": 7830
    },
    {
      "epoch": 0.4178993097198902,
      "grad_norm": 1.5904661838595013,
      "learning_rate": 2.9350630495764734e-06,
      "loss": 0.4048,
      "step": 7840
    },
    {
      "epoch": 0.4184323445537166,
      "grad_norm": 1.4784304422367702,
      "learning_rate": 2.934806278544025e-06,
      "loss": 0.4138,
      "step": 7850
    },
    {
      "epoch": 0.41896537938754297,
      "grad_norm": 1.4167844735407589,
      "learning_rate": 2.934549012132553e-06,
      "loss": 0.4232,
      "step": 7860
    },
    {
      "epoch": 0.41949841422136935,
      "grad_norm": 1.6454819116212354,
      "learning_rate": 2.93429125043088e-06,
      "loss": 0.4163,
      "step": 7870
    },
    {
      "epoch": 0.42003144905519574,
      "grad_norm": 1.593421438965332,
      "learning_rate": 2.934032993528001e-06,
      "loss": 0.4199,
      "step": 7880
    },
    {
      "epoch": 0.4205644838890221,
      "grad_norm": 1.514826322961756,
      "learning_rate": 2.9337742415130807e-06,
      "loss": 0.4216,
      "step": 7890
    },
    {
      "epoch": 0.42109751872284856,
      "grad_norm": 1.546654329029193,
      "learning_rate": 2.933514994475456e-06,
      "loss": 0.4156,
      "step": 7900
    },
    {
      "epoch": 0.42163055355667495,
      "grad_norm": 1.682864628752226,
      "learning_rate": 2.9332552525046345e-06,
      "loss": 0.4154,
      "step": 7910
    },
    {
      "epoch": 0.42216358839050133,
      "grad_norm": 1.7701263773277038,
      "learning_rate": 2.9329950156902932e-06,
      "loss": 0.4105,
      "step": 7920
    },
    {
      "epoch": 0.4226966232243277,
      "grad_norm": 1.6851463631217232,
      "learning_rate": 2.932734284122282e-06,
      "loss": 0.4095,
      "step": 7930
    },
    {
      "epoch": 0.4232296580581541,
      "grad_norm": 1.5079217751467697,
      "learning_rate": 2.9324730578906203e-06,
      "loss": 0.4128,
      "step": 7940
    },
    {
      "epoch": 0.4237626928919805,
      "grad_norm": 1.7300414513140812,
      "learning_rate": 2.9322113370854987e-06,
      "loss": 0.4221,
      "step": 7950
    },
    {
      "epoch": 0.42429572772580687,
      "grad_norm": 1.53899885108711,
      "learning_rate": 2.9319491217972787e-06,
      "loss": 0.4109,
      "step": 7960
    },
    {
      "epoch": 0.42482876255963326,
      "grad_norm": 1.5558466269927365,
      "learning_rate": 2.931686412116492e-06,
      "loss": 0.4267,
      "step": 7970
    },
    {
      "epoch": 0.42536179739345964,
      "grad_norm": 1.5017860916884838,
      "learning_rate": 2.9314232081338417e-06,
      "loss": 0.4007,
      "step": 7980
    },
    {
      "epoch": 0.4258948322272861,
      "grad_norm": 1.6343789482636109,
      "learning_rate": 2.931159509940201e-06,
      "loss": 0.409,
      "step": 7990
    },
    {
      "epoch": 0.42642786706111246,
      "grad_norm": 1.6222545648960967,
      "learning_rate": 2.930895317626614e-06,
      "loss": 0.4188,
      "step": 8000
    },
    {
      "epoch": 0.42696090189493885,
      "grad_norm": 1.6155586118015453,
      "learning_rate": 2.9306306312842954e-06,
      "loss": 0.4109,
      "step": 8010
    },
    {
      "epoch": 0.42749393672876523,
      "grad_norm": 1.5765458288307308,
      "learning_rate": 2.9303654510046304e-06,
      "loss": 0.4053,
      "step": 8020
    },
    {
      "epoch": 0.4280269715625916,
      "grad_norm": 1.6413995649684712,
      "learning_rate": 2.9300997768791746e-06,
      "loss": 0.4127,
      "step": 8030
    },
    {
      "epoch": 0.428560006396418,
      "grad_norm": 1.5874700448655936,
      "learning_rate": 2.9298336089996537e-06,
      "loss": 0.41,
      "step": 8040
    },
    {
      "epoch": 0.4290930412302444,
      "grad_norm": 1.6844801959264637,
      "learning_rate": 2.9295669474579657e-06,
      "loss": 0.409,
      "step": 8050
    },
    {
      "epoch": 0.42962607606407077,
      "grad_norm": 1.460845751006175,
      "learning_rate": 2.929299792346177e-06,
      "loss": 0.4012,
      "step": 8060
    },
    {
      "epoch": 0.43015911089789716,
      "grad_norm": 1.6786113157387517,
      "learning_rate": 2.9290321437565254e-06,
      "loss": 0.4078,
      "step": 8070
    },
    {
      "epoch": 0.43069214573172354,
      "grad_norm": 1.5880080840732538,
      "learning_rate": 2.9287640017814184e-06,
      "loss": 0.3992,
      "step": 8080
    },
    {
      "epoch": 0.43122518056555,
      "grad_norm": 1.691837892650592,
      "learning_rate": 2.928495366513435e-06,
      "loss": 0.4232,
      "step": 8090
    },
    {
      "epoch": 0.43175821539937637,
      "grad_norm": 1.5769955779968576,
      "learning_rate": 2.928226238045323e-06,
      "loss": 0.403,
      "step": 8100
    },
    {
      "epoch": 0.43229125023320275,
      "grad_norm": 1.5119353223803156,
      "learning_rate": 2.9279566164700016e-06,
      "loss": 0.3914,
      "step": 8110
    },
    {
      "epoch": 0.43282428506702914,
      "grad_norm": 1.6975993903696593,
      "learning_rate": 2.9276865018805598e-06,
      "loss": 0.4236,
      "step": 8120
    },
    {
      "epoch": 0.4333573199008555,
      "grad_norm": 1.5882894145215583,
      "learning_rate": 2.9274158943702576e-06,
      "loss": 0.4296,
      "step": 8130
    },
    {
      "epoch": 0.4338903547346819,
      "grad_norm": 1.5448781175456678,
      "learning_rate": 2.927144794032524e-06,
      "loss": 0.4251,
      "step": 8140
    },
    {
      "epoch": 0.4344233895685083,
      "grad_norm": 1.729775236818066,
      "learning_rate": 2.926873200960959e-06,
      "loss": 0.4297,
      "step": 8150
    },
    {
      "epoch": 0.4349564244023347,
      "grad_norm": 1.5906976534767527,
      "learning_rate": 2.9266011152493318e-06,
      "loss": 0.4136,
      "step": 8160
    },
    {
      "epoch": 0.43548945923616106,
      "grad_norm": 1.7401715177125827,
      "learning_rate": 2.9263285369915825e-06,
      "loss": 0.4167,
      "step": 8170
    },
    {
      "epoch": 0.4360224940699875,
      "grad_norm": 1.6079276841774246,
      "learning_rate": 2.9260554662818217e-06,
      "loss": 0.4105,
      "step": 8180
    },
    {
      "epoch": 0.4365555289038139,
      "grad_norm": 1.6881425967310364,
      "learning_rate": 2.925781903214329e-06,
      "loss": 0.411,
      "step": 8190
    },
    {
      "epoch": 0.43708856373764027,
      "grad_norm": 1.8368120422345755,
      "learning_rate": 2.9255078478835536e-06,
      "loss": 0.4162,
      "step": 8200
    },
    {
      "epoch": 0.43762159857146665,
      "grad_norm": 1.4735301058229453,
      "learning_rate": 2.9252333003841164e-06,
      "loss": 0.4039,
      "step": 8210
    },
    {
      "epoch": 0.43815463340529304,
      "grad_norm": 1.4785167027135069,
      "learning_rate": 2.9249582608108076e-06,
      "loss": 0.4252,
      "step": 8220
    },
    {
      "epoch": 0.4386876682391194,
      "grad_norm": 1.4928343612572648,
      "learning_rate": 2.9246827292585852e-06,
      "loss": 0.4213,
      "step": 8230
    },
    {
      "epoch": 0.4392207030729458,
      "grad_norm": 1.4389330256335633,
      "learning_rate": 2.9244067058225808e-06,
      "loss": 0.4165,
      "step": 8240
    },
    {
      "epoch": 0.4397537379067722,
      "grad_norm": 1.6293801431128347,
      "learning_rate": 2.9241301905980926e-06,
      "loss": 0.4188,
      "step": 8250
    },
    {
      "epoch": 0.4402867727405986,
      "grad_norm": 1.6015578269306125,
      "learning_rate": 2.92385318368059e-06,
      "loss": 0.4157,
      "step": 8260
    },
    {
      "epoch": 0.440819807574425,
      "grad_norm": 1.435916236499711,
      "learning_rate": 2.9235756851657123e-06,
      "loss": 0.4132,
      "step": 8270
    },
    {
      "epoch": 0.4413528424082514,
      "grad_norm": 1.5676414055445547,
      "learning_rate": 2.9232976951492678e-06,
      "loss": 0.4117,
      "step": 8280
    },
    {
      "epoch": 0.4418858772420778,
      "grad_norm": 1.479414067746698,
      "learning_rate": 2.9230192137272357e-06,
      "loss": 0.408,
      "step": 8290
    },
    {
      "epoch": 0.44241891207590417,
      "grad_norm": 1.652095027984649,
      "learning_rate": 2.922740240995763e-06,
      "loss": 0.3991,
      "step": 8300
    },
    {
      "epoch": 0.44295194690973055,
      "grad_norm": 1.497777155913364,
      "learning_rate": 2.9224607770511683e-06,
      "loss": 0.4195,
      "step": 8310
    },
    {
      "epoch": 0.44348498174355694,
      "grad_norm": 1.7255631646054368,
      "learning_rate": 2.9221808219899384e-06,
      "loss": 0.4253,
      "step": 8320
    },
    {
      "epoch": 0.4440180165773833,
      "grad_norm": 1.5921587932125698,
      "learning_rate": 2.92190037590873e-06,
      "loss": 0.4072,
      "step": 8330
    },
    {
      "epoch": 0.4445510514112097,
      "grad_norm": 1.4915530417684664,
      "learning_rate": 2.9216194389043704e-06,
      "loss": 0.4116,
      "step": 8340
    },
    {
      "epoch": 0.4450840862450361,
      "grad_norm": 1.6958537172297319,
      "learning_rate": 2.921338011073854e-06,
      "loss": 0.4107,
      "step": 8350
    },
    {
      "epoch": 0.4456171210788625,
      "grad_norm": 1.5904646503087096,
      "learning_rate": 2.9210560925143475e-06,
      "loss": 0.4006,
      "step": 8360
    },
    {
      "epoch": 0.4461501559126889,
      "grad_norm": 1.573718123214918,
      "learning_rate": 2.920773683323185e-06,
      "loss": 0.4208,
      "step": 8370
    },
    {
      "epoch": 0.4466831907465153,
      "grad_norm": 1.5765420572072621,
      "learning_rate": 2.920490783597871e-06,
      "loss": 0.408,
      "step": 8380
    },
    {
      "epoch": 0.4472162255803417,
      "grad_norm": 1.388783808975152,
      "learning_rate": 2.920207393436079e-06,
      "loss": 0.4045,
      "step": 8390
    },
    {
      "epoch": 0.44774926041416807,
      "grad_norm": 1.5560064769009971,
      "learning_rate": 2.9199235129356516e-06,
      "loss": 0.4191,
      "step": 8400
    },
    {
      "epoch": 0.44828229524799446,
      "grad_norm": 1.572600763217562,
      "learning_rate": 2.9196391421946006e-06,
      "loss": 0.4056,
      "step": 8410
    },
    {
      "epoch": 0.44881533008182084,
      "grad_norm": 1.534456816067655,
      "learning_rate": 2.9193542813111084e-06,
      "loss": 0.403,
      "step": 8420
    },
    {
      "epoch": 0.4493483649156472,
      "grad_norm": 1.574436258731142,
      "learning_rate": 2.9190689303835245e-06,
      "loss": 0.4231,
      "step": 8430
    },
    {
      "epoch": 0.4498813997494736,
      "grad_norm": 1.687153536583124,
      "learning_rate": 2.91878308951037e-06,
      "loss": 0.4223,
      "step": 8440
    },
    {
      "epoch": 0.4504144345833,
      "grad_norm": 1.5588068876245815,
      "learning_rate": 2.918496758790333e-06,
      "loss": 0.4132,
      "step": 8450
    },
    {
      "epoch": 0.45094746941712643,
      "grad_norm": 1.733942782998014,
      "learning_rate": 2.9182099383222716e-06,
      "loss": 0.4073,
      "step": 8460
    },
    {
      "epoch": 0.4514805042509528,
      "grad_norm": 1.6026277202619796,
      "learning_rate": 2.917922628205213e-06,
      "loss": 0.4183,
      "step": 8470
    },
    {
      "epoch": 0.4520135390847792,
      "grad_norm": 1.471843191590328,
      "learning_rate": 2.9176348285383546e-06,
      "loss": 0.4171,
      "step": 8480
    },
    {
      "epoch": 0.4525465739186056,
      "grad_norm": 1.5120739698155388,
      "learning_rate": 2.91734653942106e-06,
      "loss": 0.4075,
      "step": 8490
    },
    {
      "epoch": 0.453079608752432,
      "grad_norm": 1.638814100821955,
      "learning_rate": 2.9170577609528643e-06,
      "loss": 0.3936,
      "step": 8500
    },
    {
      "epoch": 0.45361264358625836,
      "grad_norm": 1.578675272530091,
      "learning_rate": 2.9167684932334704e-06,
      "loss": 0.4023,
      "step": 8510
    },
    {
      "epoch": 0.45414567842008474,
      "grad_norm": 1.7097708663287463,
      "learning_rate": 2.916478736362751e-06,
      "loss": 0.4281,
      "step": 8520
    },
    {
      "epoch": 0.4546787132539111,
      "grad_norm": 1.4068358574230455,
      "learning_rate": 2.9161884904407465e-06,
      "loss": 0.4081,
      "step": 8530
    },
    {
      "epoch": 0.4552117480877375,
      "grad_norm": 1.6917007396890231,
      "learning_rate": 2.9158977555676674e-06,
      "loss": 0.4134,
      "step": 8540
    },
    {
      "epoch": 0.45574478292156395,
      "grad_norm": 1.4742286194606795,
      "learning_rate": 2.915606531843892e-06,
      "loss": 0.4128,
      "step": 8550
    },
    {
      "epoch": 0.45627781775539034,
      "grad_norm": 1.4630514633399585,
      "learning_rate": 2.9153148193699674e-06,
      "loss": 0.4103,
      "step": 8560
    },
    {
      "epoch": 0.4568108525892167,
      "grad_norm": 1.7006752475123899,
      "learning_rate": 2.91502261824661e-06,
      "loss": 0.4152,
      "step": 8570
    },
    {
      "epoch": 0.4573438874230431,
      "grad_norm": 1.7474348651195561,
      "learning_rate": 2.9147299285747055e-06,
      "loss": 0.4223,
      "step": 8580
    },
    {
      "epoch": 0.4578769222568695,
      "grad_norm": 1.6107835714965402,
      "learning_rate": 2.914436750455306e-06,
      "loss": 0.4119,
      "step": 8590
    },
    {
      "epoch": 0.4584099570906959,
      "grad_norm": 1.9125128017514217,
      "learning_rate": 2.914143083989635e-06,
      "loss": 0.4097,
      "step": 8600
    },
    {
      "epoch": 0.45894299192452226,
      "grad_norm": 1.4319198063050853,
      "learning_rate": 2.9138489292790833e-06,
      "loss": 0.4178,
      "step": 8610
    },
    {
      "epoch": 0.45947602675834864,
      "grad_norm": 1.7328201137135673,
      "learning_rate": 2.913554286425209e-06,
      "loss": 0.4287,
      "step": 8620
    },
    {
      "epoch": 0.46000906159217503,
      "grad_norm": 1.447257843385008,
      "learning_rate": 2.9132591555297423e-06,
      "loss": 0.4059,
      "step": 8630
    },
    {
      "epoch": 0.4605420964260014,
      "grad_norm": 1.505222329441332,
      "learning_rate": 2.9129635366945774e-06,
      "loss": 0.3954,
      "step": 8640
    },
    {
      "epoch": 0.46107513125982785,
      "grad_norm": 1.5656522742020575,
      "learning_rate": 2.91266743002178e-06,
      "loss": 0.3953,
      "step": 8650
    },
    {
      "epoch": 0.46160816609365424,
      "grad_norm": 1.416237604591824,
      "learning_rate": 2.9123708356135844e-06,
      "loss": 0.4066,
      "step": 8660
    },
    {
      "epoch": 0.4621412009274806,
      "grad_norm": 1.5846909544632926,
      "learning_rate": 2.912073753572391e-06,
      "loss": 0.4104,
      "step": 8670
    },
    {
      "epoch": 0.462674235761307,
      "grad_norm": 1.6736703787065854,
      "learning_rate": 2.91177618400077e-06,
      "loss": 0.4065,
      "step": 8680
    },
    {
      "epoch": 0.4632072705951334,
      "grad_norm": 1.5122104576155988,
      "learning_rate": 2.9114781270014607e-06,
      "loss": 0.4063,
      "step": 8690
    },
    {
      "epoch": 0.4637403054289598,
      "grad_norm": 1.4959558072520527,
      "learning_rate": 2.9111795826773693e-06,
      "loss": 0.4042,
      "step": 8700
    },
    {
      "epoch": 0.46427334026278616,
      "grad_norm": 1.5126427115522083,
      "learning_rate": 2.9108805511315703e-06,
      "loss": 0.4156,
      "step": 8710
    },
    {
      "epoch": 0.46480637509661255,
      "grad_norm": 1.6018787294791146,
      "learning_rate": 2.9105810324673076e-06,
      "loss": 0.4134,
      "step": 8720
    },
    {
      "epoch": 0.46533940993043893,
      "grad_norm": 1.759863015005145,
      "learning_rate": 2.9102810267879925e-06,
      "loss": 0.4138,
      "step": 8730
    },
    {
      "epoch": 0.46587244476426537,
      "grad_norm": 1.805514369757183,
      "learning_rate": 2.9099805341972045e-06,
      "loss": 0.4078,
      "step": 8740
    },
    {
      "epoch": 0.46640547959809175,
      "grad_norm": 1.5148198702600317,
      "learning_rate": 2.9096795547986903e-06,
      "loss": 0.4069,
      "step": 8750
    },
    {
      "epoch": 0.46693851443191814,
      "grad_norm": 1.8852231296025546,
      "learning_rate": 2.9093780886963667e-06,
      "loss": 0.4083,
      "step": 8760
    },
    {
      "epoch": 0.4674715492657445,
      "grad_norm": 1.4539023783041651,
      "learning_rate": 2.909076135994317e-06,
      "loss": 0.4233,
      "step": 8770
    },
    {
      "epoch": 0.4680045840995709,
      "grad_norm": 1.5313647620887976,
      "learning_rate": 2.9087736967967928e-06,
      "loss": 0.4082,
      "step": 8780
    },
    {
      "epoch": 0.4685376189333973,
      "grad_norm": 1.5542347545163724,
      "learning_rate": 2.908470771208214e-06,
      "loss": 0.4213,
      "step": 8790
    },
    {
      "epoch": 0.4690706537672237,
      "grad_norm": 1.5199089239696877,
      "learning_rate": 2.908167359333168e-06,
      "loss": 0.4162,
      "step": 8800
    },
    {
      "epoch": 0.46960368860105006,
      "grad_norm": 1.759628928213846,
      "learning_rate": 2.9078634612764106e-06,
      "loss": 0.408,
      "step": 8810
    },
    {
      "epoch": 0.47013672343487645,
      "grad_norm": 1.5557512397654005,
      "learning_rate": 2.9075590771428657e-06,
      "loss": 0.397,
      "step": 8820
    },
    {
      "epoch": 0.4706697582687029,
      "grad_norm": 1.5479658711487387,
      "learning_rate": 2.907254207037623e-06,
      "loss": 0.3984,
      "step": 8830
    },
    {
      "epoch": 0.47120279310252927,
      "grad_norm": 1.5140068535227431,
      "learning_rate": 2.9069488510659427e-06,
      "loss": 0.4017,
      "step": 8840
    },
    {
      "epoch": 0.47173582793635566,
      "grad_norm": 1.5897960704006835,
      "learning_rate": 2.906643009333251e-06,
      "loss": 0.4143,
      "step": 8850
    },
    {
      "epoch": 0.47226886277018204,
      "grad_norm": 1.7285592168271886,
      "learning_rate": 2.906336681945143e-06,
      "loss": 0.4142,
      "step": 8860
    },
    {
      "epoch": 0.4728018976040084,
      "grad_norm": 1.653535089139761,
      "learning_rate": 2.9060298690073802e-06,
      "loss": 0.3957,
      "step": 8870
    },
    {
      "epoch": 0.4733349324378348,
      "grad_norm": 1.472867310685367,
      "learning_rate": 2.905722570625893e-06,
      "loss": 0.4155,
      "step": 8880
    },
    {
      "epoch": 0.4738679672716612,
      "grad_norm": 1.7057156878110258,
      "learning_rate": 2.9054147869067774e-06,
      "loss": 0.4094,
      "step": 8890
    },
    {
      "epoch": 0.4744010021054876,
      "grad_norm": 1.6670466156944517,
      "learning_rate": 2.905106517956299e-06,
      "loss": 0.396,
      "step": 8900
    },
    {
      "epoch": 0.47493403693931396,
      "grad_norm": 1.5978373853824697,
      "learning_rate": 2.9047977638808916e-06,
      "loss": 0.4043,
      "step": 8910
    },
    {
      "epoch": 0.47546707177314035,
      "grad_norm": 1.5194113205110502,
      "learning_rate": 2.9044885247871534e-06,
      "loss": 0.4108,
      "step": 8920
    },
    {
      "epoch": 0.4760001066069668,
      "grad_norm": 1.543181279894468,
      "learning_rate": 2.904178800781853e-06,
      "loss": 0.4118,
      "step": 8930
    },
    {
      "epoch": 0.4765331414407932,
      "grad_norm": 1.5496527131823437,
      "learning_rate": 2.9038685919719245e-06,
      "loss": 0.4082,
      "step": 8940
    },
    {
      "epoch": 0.47706617627461956,
      "grad_norm": 1.5488129943677427,
      "learning_rate": 2.9035578984644697e-06,
      "loss": 0.3993,
      "step": 8950
    },
    {
      "epoch": 0.47759921110844594,
      "grad_norm": 1.4794880202048881,
      "learning_rate": 2.9032467203667594e-06,
      "loss": 0.4122,
      "step": 8960
    },
    {
      "epoch": 0.4781322459422723,
      "grad_norm": 1.5167802714318308,
      "learning_rate": 2.90293505778623e-06,
      "loss": 0.4067,
      "step": 8970
    },
    {
      "epoch": 0.4786652807760987,
      "grad_norm": 1.546092391364774,
      "learning_rate": 2.902622910830485e-06,
      "loss": 0.4071,
      "step": 8980
    },
    {
      "epoch": 0.4791983156099251,
      "grad_norm": 1.6234210410818577,
      "learning_rate": 2.902310279607297e-06,
      "loss": 0.407,
      "step": 8990
    },
    {
      "epoch": 0.4797313504437515,
      "grad_norm": 1.4703342878927876,
      "learning_rate": 2.9019971642246037e-06,
      "loss": 0.4117,
      "step": 9000
    },
    {
      "epoch": 0.48026438527757787,
      "grad_norm": 1.6655778748112098,
      "learning_rate": 2.901683564790511e-06,
      "loss": 0.4131,
      "step": 9010
    },
    {
      "epoch": 0.4807974201114043,
      "grad_norm": 1.4377216238803339,
      "learning_rate": 2.901369481413291e-06,
      "loss": 0.4143,
      "step": 9020
    },
    {
      "epoch": 0.4813304549452307,
      "grad_norm": 1.5870004932408421,
      "learning_rate": 2.9010549142013857e-06,
      "loss": 0.4098,
      "step": 9030
    },
    {
      "epoch": 0.4818634897790571,
      "grad_norm": 1.7810825428351413,
      "learning_rate": 2.9007398632634e-06,
      "loss": 0.4071,
      "step": 9040
    },
    {
      "epoch": 0.48239652461288346,
      "grad_norm": 1.7734200644758729,
      "learning_rate": 2.9004243287081093e-06,
      "loss": 0.4123,
      "step": 9050
    },
    {
      "epoch": 0.48292955944670984,
      "grad_norm": 1.4639650208073236,
      "learning_rate": 2.9001083106444534e-06,
      "loss": 0.3949,
      "step": 9060
    },
    {
      "epoch": 0.48346259428053623,
      "grad_norm": 1.5984334547779417,
      "learning_rate": 2.8997918091815408e-06,
      "loss": 0.3931,
      "step": 9070
    },
    {
      "epoch": 0.4839956291143626,
      "grad_norm": 1.5279463839184906,
      "learning_rate": 2.899474824428647e-06,
      "loss": 0.4003,
      "step": 9080
    },
    {
      "epoch": 0.484528663948189,
      "grad_norm": 1.3729773971165875,
      "learning_rate": 2.8991573564952123e-06,
      "loss": 0.4098,
      "step": 9090
    },
    {
      "epoch": 0.4850616987820154,
      "grad_norm": 1.5837278152853682,
      "learning_rate": 2.8988394054908465e-06,
      "loss": 0.4024,
      "step": 9100
    },
    {
      "epoch": 0.4855947336158418,
      "grad_norm": 1.621571263150768,
      "learning_rate": 2.898520971525324e-06,
      "loss": 0.4111,
      "step": 9110
    },
    {
      "epoch": 0.4861277684496682,
      "grad_norm": 1.4334226400605534,
      "learning_rate": 2.898202054708587e-06,
      "loss": 0.4004,
      "step": 9120
    },
    {
      "epoch": 0.4866608032834946,
      "grad_norm": 1.7329660280862744,
      "learning_rate": 2.8978826551507453e-06,
      "loss": 0.412,
      "step": 9130
    },
    {
      "epoch": 0.487193838117321,
      "grad_norm": 1.5738326821537885,
      "learning_rate": 2.8975627729620724e-06,
      "loss": 0.4082,
      "step": 9140
    },
    {
      "epoch": 0.48772687295114736,
      "grad_norm": 1.5010788169288483,
      "learning_rate": 2.8972424082530114e-06,
      "loss": 0.3957,
      "step": 9150
    },
    {
      "epoch": 0.48825990778497375,
      "grad_norm": 1.4915431525703808,
      "learning_rate": 2.8969215611341708e-06,
      "loss": 0.3961,
      "step": 9160
    },
    {
      "epoch": 0.48879294261880013,
      "grad_norm": 1.6784132066746704,
      "learning_rate": 2.8966002317163256e-06,
      "loss": 0.4007,
      "step": 9170
    },
    {
      "epoch": 0.4893259774526265,
      "grad_norm": 1.6421114541747448,
      "learning_rate": 2.896278420110418e-06,
      "loss": 0.4058,
      "step": 9180
    },
    {
      "epoch": 0.4898590122864529,
      "grad_norm": 1.5909288229672987,
      "learning_rate": 2.8959561264275557e-06,
      "loss": 0.4099,
      "step": 9190
    },
    {
      "epoch": 0.4903920471202793,
      "grad_norm": 1.3819604352287718,
      "learning_rate": 2.895633350779014e-06,
      "loss": 0.3903,
      "step": 9200
    },
    {
      "epoch": 0.4909250819541057,
      "grad_norm": 1.7378949142298985,
      "learning_rate": 2.8953100932762332e-06,
      "loss": 0.404,
      "step": 9210
    },
    {
      "epoch": 0.4914581167879321,
      "grad_norm": 1.7313372132857148,
      "learning_rate": 2.8949863540308206e-06,
      "loss": 0.3939,
      "step": 9220
    },
    {
      "epoch": 0.4919911516217585,
      "grad_norm": 1.657200380205512,
      "learning_rate": 2.8946621331545502e-06,
      "loss": 0.4156,
      "step": 9230
    },
    {
      "epoch": 0.4925241864555849,
      "grad_norm": 1.6936440183309365,
      "learning_rate": 2.894337430759363e-06,
      "loss": 0.4096,
      "step": 9240
    },
    {
      "epoch": 0.49305722128941126,
      "grad_norm": 1.4933169673193765,
      "learning_rate": 2.894012246957363e-06,
      "loss": 0.4097,
      "step": 9250
    },
    {
      "epoch": 0.49359025612323765,
      "grad_norm": 1.6031539399954224,
      "learning_rate": 2.893686581860825e-06,
      "loss": 0.3934,
      "step": 9260
    },
    {
      "epoch": 0.49412329095706403,
      "grad_norm": 1.57481672676039,
      "learning_rate": 2.8933604355821863e-06,
      "loss": 0.3975,
      "step": 9270
    },
    {
      "epoch": 0.4946563257908904,
      "grad_norm": 1.6604083625961876,
      "learning_rate": 2.893033808234052e-06,
      "loss": 0.3993,
      "step": 9280
    },
    {
      "epoch": 0.4951893606247168,
      "grad_norm": 1.336039210116603,
      "learning_rate": 2.8927066999291927e-06,
      "loss": 0.3986,
      "step": 9290
    },
    {
      "epoch": 0.49572239545854324,
      "grad_norm": 1.5058883724743022,
      "learning_rate": 2.892379110780546e-06,
      "loss": 0.3996,
      "step": 9300
    },
    {
      "epoch": 0.4962554302923696,
      "grad_norm": 1.5092913581208995,
      "learning_rate": 2.8920510409012135e-06,
      "loss": 0.4034,
      "step": 9310
    },
    {
      "epoch": 0.496788465126196,
      "grad_norm": 1.698502350366665,
      "learning_rate": 2.8917224904044653e-06,
      "loss": 0.3967,
      "step": 9320
    },
    {
      "epoch": 0.4973214999600224,
      "grad_norm": 1.6794594752979148,
      "learning_rate": 2.891393459403736e-06,
      "loss": 0.4005,
      "step": 9330
    },
    {
      "epoch": 0.4978545347938488,
      "grad_norm": 1.6575559516729883,
      "learning_rate": 2.8910639480126265e-06,
      "loss": 0.4082,
      "step": 9340
    },
    {
      "epoch": 0.49838756962767516,
      "grad_norm": 1.5836818547664802,
      "learning_rate": 2.8907339563449034e-06,
      "loss": 0.4145,
      "step": 9350
    },
    {
      "epoch": 0.49892060446150155,
      "grad_norm": 1.5492688677923814,
      "learning_rate": 2.8904034845144982e-06,
      "loss": 0.4081,
      "step": 9360
    },
    {
      "epoch": 0.49945363929532793,
      "grad_norm": 1.5151683645537453,
      "learning_rate": 2.89007253263551e-06,
      "loss": 0.3979,
      "step": 9370
    },
    {
      "epoch": 0.4999866741291543,
      "grad_norm": 1.6403046958185434,
      "learning_rate": 2.8897411008222027e-06,
      "loss": 0.4054,
      "step": 9380
    },
    {
      "epoch": 0.5005197089629807,
      "grad_norm": 1.7066554864694206,
      "learning_rate": 2.8894091891890057e-06,
      "loss": 0.4079,
      "step": 9390
    },
    {
      "epoch": 0.5010527437968071,
      "grad_norm": 1.415185528732737,
      "learning_rate": 2.8890767978505143e-06,
      "loss": 0.4147,
      "step": 9400
    },
    {
      "epoch": 0.5015857786306335,
      "grad_norm": 1.664150509088447,
      "learning_rate": 2.888743926921489e-06,
      "loss": 0.4099,
      "step": 9410
    },
    {
      "epoch": 0.5021188134644599,
      "grad_norm": 1.553847379787356,
      "learning_rate": 2.8884105765168582e-06,
      "loss": 0.4269,
      "step": 9420
    },
    {
      "epoch": 0.5026518482982862,
      "grad_norm": 1.526068972526621,
      "learning_rate": 2.8880767467517113e-06,
      "loss": 0.412,
      "step": 9430
    },
    {
      "epoch": 0.5031848831321127,
      "grad_norm": 1.6446946775590638,
      "learning_rate": 2.8877424377413076e-06,
      "loss": 0.4123,
      "step": 9440
    },
    {
      "epoch": 0.5037179179659391,
      "grad_norm": 1.7282185506409873,
      "learning_rate": 2.887407649601069e-06,
      "loss": 0.4156,
      "step": 9450
    },
    {
      "epoch": 0.5042509527997655,
      "grad_norm": 1.678648478014172,
      "learning_rate": 2.887072382446585e-06,
      "loss": 0.4092,
      "step": 9460
    },
    {
      "epoch": 0.5047839876335919,
      "grad_norm": 1.6388324219631059,
      "learning_rate": 2.8867366363936096e-06,
      "loss": 0.4221,
      "step": 9470
    },
    {
      "epoch": 0.5053170224674183,
      "grad_norm": 1.4352677320818634,
      "learning_rate": 2.8864004115580604e-06,
      "loss": 0.4078,
      "step": 9480
    },
    {
      "epoch": 0.5058500573012447,
      "grad_norm": 1.5946313421328584,
      "learning_rate": 2.8860637080560234e-06,
      "loss": 0.404,
      "step": 9490
    },
    {
      "epoch": 0.506383092135071,
      "grad_norm": 1.6394255732614298,
      "learning_rate": 2.8857265260037477e-06,
      "loss": 0.4115,
      "step": 9500
    },
    {
      "epoch": 0.5069161269688974,
      "grad_norm": 1.4922328394211986,
      "learning_rate": 2.8853888655176478e-06,
      "loss": 0.4136,
      "step": 9510
    },
    {
      "epoch": 0.5074491618027238,
      "grad_norm": 1.7176331599168186,
      "learning_rate": 2.885050726714305e-06,
      "loss": 0.389,
      "step": 9520
    },
    {
      "epoch": 0.5079821966365502,
      "grad_norm": 1.7143563046989718,
      "learning_rate": 2.8847121097104636e-06,
      "loss": 0.4076,
      "step": 9530
    },
    {
      "epoch": 0.5085152314703766,
      "grad_norm": 1.6454780687145094,
      "learning_rate": 2.8843730146230347e-06,
      "loss": 0.3969,
      "step": 9540
    },
    {
      "epoch": 0.509048266304203,
      "grad_norm": 1.4689363468454348,
      "learning_rate": 2.884033441569093e-06,
      "loss": 0.4064,
      "step": 9550
    },
    {
      "epoch": 0.5095813011380294,
      "grad_norm": 1.5044312829945745,
      "learning_rate": 2.8836933906658797e-06,
      "loss": 0.3976,
      "step": 9560
    },
    {
      "epoch": 0.5101143359718557,
      "grad_norm": 1.4836392218032295,
      "learning_rate": 2.8833528620307995e-06,
      "loss": 0.3901,
      "step": 9570
    },
    {
      "epoch": 0.5106473708056821,
      "grad_norm": 1.5766636297460137,
      "learning_rate": 2.8830118557814236e-06,
      "loss": 0.4079,
      "step": 9580
    },
    {
      "epoch": 0.5111804056395085,
      "grad_norm": 1.6254674879634141,
      "learning_rate": 2.8826703720354865e-06,
      "loss": 0.405,
      "step": 9590
    },
    {
      "epoch": 0.5117134404733349,
      "grad_norm": 1.4133846767074656,
      "learning_rate": 2.8823284109108893e-06,
      "loss": 0.4023,
      "step": 9600
    },
    {
      "epoch": 0.5122464753071613,
      "grad_norm": 1.7031582596260313,
      "learning_rate": 2.8819859725256962e-06,
      "loss": 0.3947,
      "step": 9610
    },
    {
      "epoch": 0.5127795101409877,
      "grad_norm": 1.5624163066184436,
      "learning_rate": 2.8816430569981373e-06,
      "loss": 0.4122,
      "step": 9620
    },
    {
      "epoch": 0.5133125449748142,
      "grad_norm": 1.7328336859833728,
      "learning_rate": 2.881299664446607e-06,
      "loss": 0.3944,
      "step": 9630
    },
    {
      "epoch": 0.5138455798086405,
      "grad_norm": 1.6153217714134185,
      "learning_rate": 2.8809557949896644e-06,
      "loss": 0.4089,
      "step": 9640
    },
    {
      "epoch": 0.5143786146424669,
      "grad_norm": 1.515318833380608,
      "learning_rate": 2.8806114487460336e-06,
      "loss": 0.4089,
      "step": 9650
    },
    {
      "epoch": 0.5149116494762933,
      "grad_norm": 1.4466205077618761,
      "learning_rate": 2.8802666258346025e-06,
      "loss": 0.4136,
      "step": 9660
    },
    {
      "epoch": 0.5154446843101197,
      "grad_norm": 1.6161321640816924,
      "learning_rate": 2.8799213263744252e-06,
      "loss": 0.4083,
      "step": 9670
    },
    {
      "epoch": 0.5159777191439461,
      "grad_norm": 1.4885332756276555,
      "learning_rate": 2.8795755504847175e-06,
      "loss": 0.4076,
      "step": 9680
    },
    {
      "epoch": 0.5165107539777725,
      "grad_norm": 1.6471482339865045,
      "learning_rate": 2.8792292982848636e-06,
      "loss": 0.3956,
      "step": 9690
    },
    {
      "epoch": 0.5170437888115988,
      "grad_norm": 1.568301485808753,
      "learning_rate": 2.8788825698944086e-06,
      "loss": 0.4132,
      "step": 9700
    },
    {
      "epoch": 0.5175768236454252,
      "grad_norm": 1.4389512590562439,
      "learning_rate": 2.8785353654330637e-06,
      "loss": 0.4058,
      "step": 9710
    },
    {
      "epoch": 0.5181098584792516,
      "grad_norm": 1.6866785287908799,
      "learning_rate": 2.8781876850207044e-06,
      "loss": 0.4005,
      "step": 9720
    },
    {
      "epoch": 0.518642893313078,
      "grad_norm": 1.6458968524252657,
      "learning_rate": 2.8778395287773696e-06,
      "loss": 0.4005,
      "step": 9730
    },
    {
      "epoch": 0.5191759281469044,
      "grad_norm": 1.7150434456610066,
      "learning_rate": 2.8774908968232643e-06,
      "loss": 0.395,
      "step": 9740
    },
    {
      "epoch": 0.5197089629807308,
      "grad_norm": 1.8132235876417566,
      "learning_rate": 2.877141789278756e-06,
      "loss": 0.4084,
      "step": 9750
    },
    {
      "epoch": 0.5202419978145572,
      "grad_norm": 1.4785716305633303,
      "learning_rate": 2.876792206264377e-06,
      "loss": 0.3914,
      "step": 9760
    },
    {
      "epoch": 0.5207750326483835,
      "grad_norm": 1.6596947462689444,
      "learning_rate": 2.876442147900824e-06,
      "loss": 0.4026,
      "step": 9770
    },
    {
      "epoch": 0.5213080674822099,
      "grad_norm": 1.406171622880347,
      "learning_rate": 2.8760916143089575e-06,
      "loss": 0.394,
      "step": 9780
    },
    {
      "epoch": 0.5218411023160363,
      "grad_norm": 1.571862309547308,
      "learning_rate": 2.8757406056098024e-06,
      "loss": 0.3953,
      "step": 9790
    },
    {
      "epoch": 0.5223741371498627,
      "grad_norm": 1.7362524083313324,
      "learning_rate": 2.875389121924547e-06,
      "loss": 0.4071,
      "step": 9800
    },
    {
      "epoch": 0.5229071719836891,
      "grad_norm": 1.6167194815084442,
      "learning_rate": 2.8750371633745442e-06,
      "loss": 0.4051,
      "step": 9810
    },
    {
      "epoch": 0.5234402068175156,
      "grad_norm": 1.5816653347720229,
      "learning_rate": 2.8746847300813106e-06,
      "loss": 0.4129,
      "step": 9820
    },
    {
      "epoch": 0.523973241651342,
      "grad_norm": 1.6458069252330907,
      "learning_rate": 2.8743318221665273e-06,
      "loss": 0.3885,
      "step": 9830
    },
    {
      "epoch": 0.5245062764851683,
      "grad_norm": 1.4772944455091512,
      "learning_rate": 2.8739784397520384e-06,
      "loss": 0.3793,
      "step": 9840
    },
    {
      "epoch": 0.5250393113189947,
      "grad_norm": 1.3845055958148909,
      "learning_rate": 2.8736245829598517e-06,
      "loss": 0.4003,
      "step": 9850
    },
    {
      "epoch": 0.5255723461528211,
      "grad_norm": 1.6157949763742652,
      "learning_rate": 2.8732702519121404e-06,
      "loss": 0.396,
      "step": 9860
    },
    {
      "epoch": 0.5261053809866475,
      "grad_norm": 1.6330823052705719,
      "learning_rate": 2.872915446731239e-06,
      "loss": 0.3987,
      "step": 9870
    },
    {
      "epoch": 0.5266384158204739,
      "grad_norm": 1.711796487941707,
      "learning_rate": 2.8725601675396475e-06,
      "loss": 0.4113,
      "step": 9880
    },
    {
      "epoch": 0.5271714506543003,
      "grad_norm": 1.5848820299463564,
      "learning_rate": 2.87220441446003e-06,
      "loss": 0.3991,
      "step": 9890
    },
    {
      "epoch": 0.5277044854881267,
      "grad_norm": 1.3866219205029024,
      "learning_rate": 2.8718481876152116e-06,
      "loss": 0.4253,
      "step": 9900
    },
    {
      "epoch": 0.528237520321953,
      "grad_norm": 1.4503470440045874,
      "learning_rate": 2.8714914871281833e-06,
      "loss": 0.4135,
      "step": 9910
    },
    {
      "epoch": 0.5287705551557794,
      "grad_norm": 1.433593960550652,
      "learning_rate": 2.8711343131220996e-06,
      "loss": 0.3949,
      "step": 9920
    },
    {
      "epoch": 0.5293035899896058,
      "grad_norm": 1.531094824111099,
      "learning_rate": 2.870776665720277e-06,
      "loss": 0.4014,
      "step": 9930
    },
    {
      "epoch": 0.5298366248234322,
      "grad_norm": 1.6274275512213063,
      "learning_rate": 2.870418545046197e-06,
      "loss": 0.4032,
      "step": 9940
    },
    {
      "epoch": 0.5303696596572586,
      "grad_norm": 1.5505587034864172,
      "learning_rate": 2.870059951223503e-06,
      "loss": 0.3941,
      "step": 9950
    },
    {
      "epoch": 0.530902694491085,
      "grad_norm": 1.618968848772058,
      "learning_rate": 2.8697008843760035e-06,
      "loss": 0.3993,
      "step": 9960
    },
    {
      "epoch": 0.5314357293249113,
      "grad_norm": 1.6120874627959603,
      "learning_rate": 2.869341344627669e-06,
      "loss": 0.4058,
      "step": 9970
    },
    {
      "epoch": 0.5319687641587377,
      "grad_norm": 1.6356430931524362,
      "learning_rate": 2.8689813321026333e-06,
      "loss": 0.4121,
      "step": 9980
    },
    {
      "epoch": 0.5325017989925641,
      "grad_norm": 1.5785457872866275,
      "learning_rate": 2.868620846925194e-06,
      "loss": 0.3986,
      "step": 9990
    },
    {
      "epoch": 0.5330348338263906,
      "grad_norm": 1.4965603722100826,
      "learning_rate": 2.868259889219811e-06,
      "loss": 0.3938,
      "step": 10000
    },
    {
      "epoch": 0.533567868660217,
      "grad_norm": 1.5867025101395884,
      "learning_rate": 2.8678984591111097e-06,
      "loss": 0.386,
      "step": 10010
    },
    {
      "epoch": 0.5341009034940434,
      "grad_norm": 1.5534019324161734,
      "learning_rate": 2.8675365567238753e-06,
      "loss": 0.3961,
      "step": 10020
    },
    {
      "epoch": 0.5346339383278698,
      "grad_norm": 1.8061862740226802,
      "learning_rate": 2.8671741821830582e-06,
      "loss": 0.4127,
      "step": 10030
    },
    {
      "epoch": 0.5351669731616961,
      "grad_norm": 1.4579109951997724,
      "learning_rate": 2.866811335613772e-06,
      "loss": 0.4177,
      "step": 10040
    },
    {
      "epoch": 0.5357000079955225,
      "grad_norm": 1.6642463591867394,
      "learning_rate": 2.866448017141291e-06,
      "loss": 0.3907,
      "step": 10050
    },
    {
      "epoch": 0.5362330428293489,
      "grad_norm": 1.4892110596304102,
      "learning_rate": 2.866084226891055e-06,
      "loss": 0.3935,
      "step": 10060
    },
    {
      "epoch": 0.5367660776631753,
      "grad_norm": 1.5359037672697213,
      "learning_rate": 2.8657199649886657e-06,
      "loss": 0.3951,
      "step": 10070
    },
    {
      "epoch": 0.5372991124970017,
      "grad_norm": 1.5144731486207554,
      "learning_rate": 2.8653552315598875e-06,
      "loss": 0.3954,
      "step": 10080
    },
    {
      "epoch": 0.5378321473308281,
      "grad_norm": 1.5215931325124186,
      "learning_rate": 2.8649900267306478e-06,
      "loss": 0.3957,
      "step": 10090
    },
    {
      "epoch": 0.5383651821646545,
      "grad_norm": 1.6155875390867562,
      "learning_rate": 2.864624350627036e-06,
      "loss": 0.3909,
      "step": 10100
    },
    {
      "epoch": 0.5388982169984808,
      "grad_norm": 1.6428343439318636,
      "learning_rate": 2.864258203375306e-06,
      "loss": 0.4025,
      "step": 10110
    },
    {
      "epoch": 0.5394312518323072,
      "grad_norm": 1.5654541597805702,
      "learning_rate": 2.863891585101872e-06,
      "loss": 0.3969,
      "step": 10120
    },
    {
      "epoch": 0.5399642866661336,
      "grad_norm": 1.5265120674901116,
      "learning_rate": 2.863524495933313e-06,
      "loss": 0.3938,
      "step": 10130
    },
    {
      "epoch": 0.54049732149996,
      "grad_norm": 1.4736365199067178,
      "learning_rate": 2.863156935996369e-06,
      "loss": 0.3912,
      "step": 10140
    },
    {
      "epoch": 0.5410303563337864,
      "grad_norm": 1.7192348969896323,
      "learning_rate": 2.8627889054179437e-06,
      "loss": 0.3988,
      "step": 10150
    },
    {
      "epoch": 0.5415633911676128,
      "grad_norm": 1.7429638348817058,
      "learning_rate": 2.8624204043251025e-06,
      "loss": 0.3906,
      "step": 10160
    },
    {
      "epoch": 0.5420964260014391,
      "grad_norm": 1.5919744294896851,
      "learning_rate": 2.8620514328450734e-06,
      "loss": 0.4018,
      "step": 10170
    },
    {
      "epoch": 0.5426294608352655,
      "grad_norm": 1.6527949272175446,
      "learning_rate": 2.8616819911052477e-06,
      "loss": 0.3846,
      "step": 10180
    },
    {
      "epoch": 0.543162495669092,
      "grad_norm": 1.3926071798106674,
      "learning_rate": 2.8613120792331775e-06,
      "loss": 0.3956,
      "step": 10190
    },
    {
      "epoch": 0.5436955305029184,
      "grad_norm": 1.6568790431469391,
      "learning_rate": 2.860941697356578e-06,
      "loss": 0.4014,
      "step": 10200
    },
    {
      "epoch": 0.5442285653367448,
      "grad_norm": 1.636416760466028,
      "learning_rate": 2.8605708456033274e-06,
      "loss": 0.3872,
      "step": 10210
    },
    {
      "epoch": 0.5447616001705712,
      "grad_norm": 1.4547880434668894,
      "learning_rate": 2.860199524101464e-06,
      "loss": 0.3884,
      "step": 10220
    },
    {
      "epoch": 0.5452946350043976,
      "grad_norm": 1.74912827875163,
      "learning_rate": 2.859827732979191e-06,
      "loss": 0.4126,
      "step": 10230
    },
    {
      "epoch": 0.545827669838224,
      "grad_norm": 1.5875232908515233,
      "learning_rate": 2.8594554723648723e-06,
      "loss": 0.3972,
      "step": 10240
    },
    {
      "epoch": 0.5463607046720503,
      "grad_norm": 1.5339760028458476,
      "learning_rate": 2.859082742387034e-06,
      "loss": 0.393,
      "step": 10250
    },
    {
      "epoch": 0.5468937395058767,
      "grad_norm": 1.4114830634948567,
      "learning_rate": 2.8587095431743637e-06,
      "loss": 0.3913,
      "step": 10260
    },
    {
      "epoch": 0.5474267743397031,
      "grad_norm": 1.3990399103587143,
      "learning_rate": 2.858335874855712e-06,
      "loss": 0.4046,
      "step": 10270
    },
    {
      "epoch": 0.5479598091735295,
      "grad_norm": 1.514447667085036,
      "learning_rate": 2.8579617375600912e-06,
      "loss": 0.3998,
      "step": 10280
    },
    {
      "epoch": 0.5484928440073559,
      "grad_norm": 1.3963669858759415,
      "learning_rate": 2.857587131416675e-06,
      "loss": 0.3994,
      "step": 10290
    },
    {
      "epoch": 0.5490258788411823,
      "grad_norm": 1.486639377515058,
      "learning_rate": 2.8572120565547994e-06,
      "loss": 0.4085,
      "step": 10300
    },
    {
      "epoch": 0.5495589136750086,
      "grad_norm": 1.5333184988023538,
      "learning_rate": 2.8568365131039624e-06,
      "loss": 0.4029,
      "step": 10310
    },
    {
      "epoch": 0.550091948508835,
      "grad_norm": 1.3969208013298322,
      "learning_rate": 2.8564605011938234e-06,
      "loss": 0.3912,
      "step": 10320
    },
    {
      "epoch": 0.5506249833426614,
      "grad_norm": 1.5779766323495066,
      "learning_rate": 2.856084020954204e-06,
      "loss": 0.3984,
      "step": 10330
    },
    {
      "epoch": 0.5511580181764878,
      "grad_norm": 1.5721736227160277,
      "learning_rate": 2.855707072515087e-06,
      "loss": 0.3938,
      "step": 10340
    },
    {
      "epoch": 0.5516910530103142,
      "grad_norm": 1.4232754576981241,
      "learning_rate": 2.8553296560066166e-06,
      "loss": 0.3952,
      "step": 10350
    },
    {
      "epoch": 0.5522240878441406,
      "grad_norm": 1.5407831072897882,
      "learning_rate": 2.854951771559099e-06,
      "loss": 0.3951,
      "step": 10360
    },
    {
      "epoch": 0.552757122677967,
      "grad_norm": 1.44353905855885,
      "learning_rate": 2.854573419303003e-06,
      "loss": 0.4079,
      "step": 10370
    },
    {
      "epoch": 0.5532901575117934,
      "grad_norm": 1.502073545758752,
      "learning_rate": 2.854194599368958e-06,
      "loss": 0.3972,
      "step": 10380
    },
    {
      "epoch": 0.5538231923456198,
      "grad_norm": 1.4819012616625045,
      "learning_rate": 2.853815311887753e-06,
      "loss": 0.3887,
      "step": 10390
    },
    {
      "epoch": 0.5543562271794462,
      "grad_norm": 1.5078684627676604,
      "learning_rate": 2.8534355569903423e-06,
      "loss": 0.3866,
      "step": 10400
    },
    {
      "epoch": 0.5548892620132726,
      "grad_norm": 1.5126708128236444,
      "learning_rate": 2.853055334807838e-06,
      "loss": 0.4035,
      "step": 10410
    },
    {
      "epoch": 0.555422296847099,
      "grad_norm": 1.6044363976388567,
      "learning_rate": 2.8526746454715158e-06,
      "loss": 0.4021,
      "step": 10420
    },
    {
      "epoch": 0.5559553316809254,
      "grad_norm": 1.6704315725220897,
      "learning_rate": 2.8522934891128114e-06,
      "loss": 0.3908,
      "step": 10430
    },
    {
      "epoch": 0.5564883665147518,
      "grad_norm": 1.662836301288045,
      "learning_rate": 2.851911865863323e-06,
      "loss": 0.4005,
      "step": 10440
    },
    {
      "epoch": 0.5570214013485781,
      "grad_norm": 1.601158339393024,
      "learning_rate": 2.8515297758548085e-06,
      "loss": 0.4062,
      "step": 10450
    },
    {
      "epoch": 0.5575544361824045,
      "grad_norm": 1.6237095561230108,
      "learning_rate": 2.8511472192191884e-06,
      "loss": 0.3974,
      "step": 10460
    },
    {
      "epoch": 0.5580874710162309,
      "grad_norm": 1.5793596654689936,
      "learning_rate": 2.850764196088543e-06,
      "loss": 0.3966,
      "step": 10470
    },
    {
      "epoch": 0.5586205058500573,
      "grad_norm": 1.5605363535407812,
      "learning_rate": 2.8503807065951147e-06,
      "loss": 0.3881,
      "step": 10480
    },
    {
      "epoch": 0.5591535406838837,
      "grad_norm": 1.6702675353241125,
      "learning_rate": 2.849996750871306e-06,
      "loss": 0.4118,
      "step": 10490
    },
    {
      "epoch": 0.5596865755177101,
      "grad_norm": 1.643604603389349,
      "learning_rate": 2.849612329049682e-06,
      "loss": 0.4049,
      "step": 10500
    },
    {
      "epoch": 0.5602196103515364,
      "grad_norm": 1.6561848706502336,
      "learning_rate": 2.849227441262966e-06,
      "loss": 0.3982,
      "step": 10510
    },
    {
      "epoch": 0.5607526451853628,
      "grad_norm": 1.590531903879771,
      "learning_rate": 2.848842087644045e-06,
      "loss": 0.391,
      "step": 10520
    },
    {
      "epoch": 0.5612856800191892,
      "grad_norm": 1.6889155758124705,
      "learning_rate": 2.8484562683259646e-06,
      "loss": 0.4042,
      "step": 10530
    },
    {
      "epoch": 0.5618187148530156,
      "grad_norm": 1.641126267683817,
      "learning_rate": 2.848069983441934e-06,
      "loss": 0.4022,
      "step": 10540
    },
    {
      "epoch": 0.562351749686842,
      "grad_norm": 1.5798488813596172,
      "learning_rate": 2.847683233125319e-06,
      "loss": 0.389,
      "step": 10550
    },
    {
      "epoch": 0.5628847845206685,
      "grad_norm": 1.7486106855174088,
      "learning_rate": 2.8472960175096492e-06,
      "loss": 0.4088,
      "step": 10560
    },
    {
      "epoch": 0.5634178193544949,
      "grad_norm": 1.6875542372746397,
      "learning_rate": 2.8469083367286156e-06,
      "loss": 0.3923,
      "step": 10570
    },
    {
      "epoch": 0.5639508541883212,
      "grad_norm": 1.602928739183086,
      "learning_rate": 2.846520190916066e-06,
      "loss": 0.3904,
      "step": 10580
    },
    {
      "epoch": 0.5644838890221476,
      "grad_norm": 1.5715388461988955,
      "learning_rate": 2.8461315802060128e-06,
      "loss": 0.4133,
      "step": 10590
    },
    {
      "epoch": 0.565016923855974,
      "grad_norm": 1.6412992031891698,
      "learning_rate": 2.845742504732626e-06,
      "loss": 0.4186,
      "step": 10600
    },
    {
      "epoch": 0.5655499586898004,
      "grad_norm": 3.6979589778440727,
      "learning_rate": 2.845352964630238e-06,
      "loss": 0.3924,
      "step": 10610
    },
    {
      "epoch": 0.5660829935236268,
      "grad_norm": 1.6569160402984553,
      "learning_rate": 2.8449629600333402e-06,
      "loss": 0.3913,
      "step": 10620
    },
    {
      "epoch": 0.5666160283574532,
      "grad_norm": 1.7075699277783256,
      "learning_rate": 2.8445724910765854e-06,
      "loss": 0.4074,
      "step": 10630
    },
    {
      "epoch": 0.5671490631912796,
      "grad_norm": 1.613437597893956,
      "learning_rate": 2.8441815578947863e-06,
      "loss": 0.4077,
      "step": 10640
    },
    {
      "epoch": 0.5676820980251059,
      "grad_norm": 1.560621673960656,
      "learning_rate": 2.8437901606229157e-06,
      "loss": 0.407,
      "step": 10650
    },
    {
      "epoch": 0.5682151328589323,
      "grad_norm": 1.6156946155944378,
      "learning_rate": 2.843398299396106e-06,
      "loss": 0.3958,
      "step": 10660
    },
    {
      "epoch": 0.5687481676927587,
      "grad_norm": 1.3741828683826942,
      "learning_rate": 2.8430059743496527e-06,
      "loss": 0.3976,
      "step": 10670
    },
    {
      "epoch": 0.5692812025265851,
      "grad_norm": 1.5938642039216717,
      "learning_rate": 2.8426131856190074e-06,
      "loss": 0.3998,
      "step": 10680
    },
    {
      "epoch": 0.5698142373604115,
      "grad_norm": 1.6185251189724847,
      "learning_rate": 2.842219933339785e-06,
      "loss": 0.387,
      "step": 10690
    },
    {
      "epoch": 0.5703472721942379,
      "grad_norm": 1.721631309827698,
      "learning_rate": 2.8418262176477585e-06,
      "loss": 0.4054,
      "step": 10700
    },
    {
      "epoch": 0.5708803070280642,
      "grad_norm": 1.5312620396476804,
      "learning_rate": 2.8414320386788613e-06,
      "loss": 0.3848,
      "step": 10710
    },
    {
      "epoch": 0.5714133418618906,
      "grad_norm": 1.6466118131579388,
      "learning_rate": 2.841037396569188e-06,
      "loss": 0.3935,
      "step": 10720
    },
    {
      "epoch": 0.571946376695717,
      "grad_norm": 1.6032642631175114,
      "learning_rate": 2.8406422914549915e-06,
      "loss": 0.3995,
      "step": 10730
    },
    {
      "epoch": 0.5724794115295434,
      "grad_norm": 1.677288561601342,
      "learning_rate": 2.8402467234726847e-06,
      "loss": 0.4024,
      "step": 10740
    },
    {
      "epoch": 0.5730124463633699,
      "grad_norm": 1.7205492469098584,
      "learning_rate": 2.839850692758842e-06,
      "loss": 0.3802,
      "step": 10750
    },
    {
      "epoch": 0.5735454811971963,
      "grad_norm": 1.60233188086089,
      "learning_rate": 2.8394541994501955e-06,
      "loss": 0.3848,
      "step": 10760
    },
    {
      "epoch": 0.5740785160310227,
      "grad_norm": 1.5647105400654844,
      "learning_rate": 2.8390572436836382e-06,
      "loss": 0.3886,
      "step": 10770
    },
    {
      "epoch": 0.574611550864849,
      "grad_norm": 1.4561842724504217,
      "learning_rate": 2.8386598255962226e-06,
      "loss": 0.3931,
      "step": 10780
    },
    {
      "epoch": 0.5751445856986754,
      "grad_norm": 1.6279145145893323,
      "learning_rate": 2.8382619453251598e-06,
      "loss": 0.3981,
      "step": 10790
    },
    {
      "epoch": 0.5756776205325018,
      "grad_norm": 1.4857469776656202,
      "learning_rate": 2.8378636030078225e-06,
      "loss": 0.3799,
      "step": 10800
    },
    {
      "epoch": 0.5762106553663282,
      "grad_norm": 1.4461093301470354,
      "learning_rate": 2.837464798781741e-06,
      "loss": 0.4011,
      "step": 10810
    },
    {
      "epoch": 0.5767436902001546,
      "grad_norm": 1.4643713873836488,
      "learning_rate": 2.8370655327846064e-06,
      "loss": 0.3965,
      "step": 10820
    },
    {
      "epoch": 0.577276725033981,
      "grad_norm": 1.6397069192723366,
      "learning_rate": 2.836665805154268e-06,
      "loss": 0.3904,
      "step": 10830
    },
    {
      "epoch": 0.5778097598678074,
      "grad_norm": 1.6153524399193817,
      "learning_rate": 2.8362656160287352e-06,
      "loss": 0.393,
      "step": 10840
    },
    {
      "epoch": 0.5783427947016337,
      "grad_norm": 1.533128077411933,
      "learning_rate": 2.835864965546177e-06,
      "loss": 0.3881,
      "step": 10850
    },
    {
      "epoch": 0.5788758295354601,
      "grad_norm": 1.506981038459625,
      "learning_rate": 2.835463853844922e-06,
      "loss": 0.3881,
      "step": 10860
    },
    {
      "epoch": 0.5794088643692865,
      "grad_norm": 1.3957981422127663,
      "learning_rate": 2.835062281063456e-06,
      "loss": 0.3939,
      "step": 10870
    },
    {
      "epoch": 0.5799418992031129,
      "grad_norm": 1.556890609180311,
      "learning_rate": 2.834660247340426e-06,
      "loss": 0.382,
      "step": 10880
    },
    {
      "epoch": 0.5804749340369393,
      "grad_norm": 1.5439773883675694,
      "learning_rate": 2.8342577528146384e-06,
      "loss": 0.388,
      "step": 10890
    },
    {
      "epoch": 0.5810079688707657,
      "grad_norm": 1.6327680129307451,
      "learning_rate": 2.8338547976250564e-06,
      "loss": 0.3955,
      "step": 10900
    },
    {
      "epoch": 0.581541003704592,
      "grad_norm": 1.600281933147193,
      "learning_rate": 2.8334513819108054e-06,
      "loss": 0.4014,
      "step": 10910
    },
    {
      "epoch": 0.5820740385384184,
      "grad_norm": 1.621057959094922,
      "learning_rate": 2.8330475058111667e-06,
      "loss": 0.3872,
      "step": 10920
    },
    {
      "epoch": 0.5826070733722448,
      "grad_norm": 1.601534627497534,
      "learning_rate": 2.8326431694655822e-06,
      "loss": 0.3856,
      "step": 10930
    },
    {
      "epoch": 0.5831401082060713,
      "grad_norm": 1.6904303075461933,
      "learning_rate": 2.8322383730136524e-06,
      "loss": 0.4069,
      "step": 10940
    },
    {
      "epoch": 0.5836731430398977,
      "grad_norm": 1.6290138862314838,
      "learning_rate": 2.8318331165951376e-06,
      "loss": 0.4096,
      "step": 10950
    },
    {
      "epoch": 0.5842061778737241,
      "grad_norm": 1.40145473356515,
      "learning_rate": 2.831427400349955e-06,
      "loss": 0.3917,
      "step": 10960
    },
    {
      "epoch": 0.5847392127075505,
      "grad_norm": 1.4930505550808872,
      "learning_rate": 2.831021224418182e-06,
      "loss": 0.3878,
      "step": 10970
    },
    {
      "epoch": 0.5852722475413769,
      "grad_norm": 1.730859699973955,
      "learning_rate": 2.8306145889400537e-06,
      "loss": 0.3844,
      "step": 10980
    },
    {
      "epoch": 0.5858052823752032,
      "grad_norm": 1.6731858008414877,
      "learning_rate": 2.830207494055965e-06,
      "loss": 0.396,
      "step": 10990
    },
    {
      "epoch": 0.5863383172090296,
      "grad_norm": 1.860870918490304,
      "learning_rate": 2.8297999399064696e-06,
      "loss": 0.3836,
      "step": 11000
    },
    {
      "epoch": 0.586871352042856,
      "grad_norm": 1.6065116215870288,
      "learning_rate": 2.8293919266322776e-06,
      "loss": 0.3946,
      "step": 11010
    },
    {
      "epoch": 0.5874043868766824,
      "grad_norm": 1.7196307359790566,
      "learning_rate": 2.8289834543742597e-06,
      "loss": 0.3921,
      "step": 11020
    },
    {
      "epoch": 0.5879374217105088,
      "grad_norm": 1.4683301802490478,
      "learning_rate": 2.8285745232734446e-06,
      "loss": 0.3972,
      "step": 11030
    },
    {
      "epoch": 0.5884704565443352,
      "grad_norm": 1.58314055523465,
      "learning_rate": 2.828165133471019e-06,
      "loss": 0.3904,
      "step": 11040
    },
    {
      "epoch": 0.5890034913781615,
      "grad_norm": 1.3428032769263376,
      "learning_rate": 2.8277552851083276e-06,
      "loss": 0.3957,
      "step": 11050
    },
    {
      "epoch": 0.5895365262119879,
      "grad_norm": 1.6980158256452536,
      "learning_rate": 2.827344978326875e-06,
      "loss": 0.3956,
      "step": 11060
    },
    {
      "epoch": 0.5900695610458143,
      "grad_norm": 1.4273036074140575,
      "learning_rate": 2.826934213268323e-06,
      "loss": 0.4123,
      "step": 11070
    },
    {
      "epoch": 0.5906025958796407,
      "grad_norm": 1.4760804808151409,
      "learning_rate": 2.826522990074491e-06,
      "loss": 0.3933,
      "step": 11080
    },
    {
      "epoch": 0.5911356307134671,
      "grad_norm": 1.5592700800967758,
      "learning_rate": 2.826111308887358e-06,
      "loss": 0.3901,
      "step": 11090
    },
    {
      "epoch": 0.5916686655472935,
      "grad_norm": 1.6755229269602052,
      "learning_rate": 2.82569916984906e-06,
      "loss": 0.3887,
      "step": 11100
    },
    {
      "epoch": 0.5922017003811199,
      "grad_norm": 1.5754410324386798,
      "learning_rate": 2.8252865731018915e-06,
      "loss": 0.3926,
      "step": 11110
    },
    {
      "epoch": 0.5927347352149464,
      "grad_norm": 1.689941680605763,
      "learning_rate": 2.8248735187883053e-06,
      "loss": 0.403,
      "step": 11120
    },
    {
      "epoch": 0.5932677700487727,
      "grad_norm": 1.4887919059673544,
      "learning_rate": 2.824460007050912e-06,
      "loss": 0.412,
      "step": 11130
    },
    {
      "epoch": 0.5938008048825991,
      "grad_norm": 1.4112522194521475,
      "learning_rate": 2.824046038032479e-06,
      "loss": 0.393,
      "step": 11140
    },
    {
      "epoch": 0.5943338397164255,
      "grad_norm": 1.5321257088523927,
      "learning_rate": 2.8236316118759337e-06,
      "loss": 0.3945,
      "step": 11150
    },
    {
      "epoch": 0.5948668745502519,
      "grad_norm": 1.5716652350993627,
      "learning_rate": 2.82321672872436e-06,
      "loss": 0.4042,
      "step": 11160
    },
    {
      "epoch": 0.5953999093840783,
      "grad_norm": 1.618259304671504,
      "learning_rate": 2.822801388721e-06,
      "loss": 0.3931,
      "step": 11170
    },
    {
      "epoch": 0.5959329442179047,
      "grad_norm": 1.589493000573229,
      "learning_rate": 2.8223855920092527e-06,
      "loss": 0.4032,
      "step": 11180
    },
    {
      "epoch": 0.596465979051731,
      "grad_norm": 1.637701216793911,
      "learning_rate": 2.821969338732676e-06,
      "loss": 0.3916,
      "step": 11190
    },
    {
      "epoch": 0.5969990138855574,
      "grad_norm": 1.6518253397411737,
      "learning_rate": 2.8215526290349852e-06,
      "loss": 0.3974,
      "step": 11200
    },
    {
      "epoch": 0.5975320487193838,
      "grad_norm": 1.708003531491653,
      "learning_rate": 2.8211354630600516e-06,
      "loss": 0.3787,
      "step": 11210
    },
    {
      "epoch": 0.5980650835532102,
      "grad_norm": 1.6079405003350675,
      "learning_rate": 2.8207178409519065e-06,
      "loss": 0.3952,
      "step": 11220
    },
    {
      "epoch": 0.5985981183870366,
      "grad_norm": 1.6239283535281774,
      "learning_rate": 2.8202997628547372e-06,
      "loss": 0.4077,
      "step": 11230
    },
    {
      "epoch": 0.599131153220863,
      "grad_norm": 1.6018865470980452,
      "learning_rate": 2.819881228912888e-06,
      "loss": 0.3956,
      "step": 11240
    },
    {
      "epoch": 0.5996641880546894,
      "grad_norm": 1.5010263983394176,
      "learning_rate": 2.8194622392708623e-06,
      "loss": 0.3975,
      "step": 11250
    },
    {
      "epoch": 0.6001972228885157,
      "grad_norm": 1.5002876031292254,
      "learning_rate": 2.819042794073319e-06,
      "loss": 0.3939,
      "step": 11260
    },
    {
      "epoch": 0.6007302577223421,
      "grad_norm": 1.4924650200651106,
      "learning_rate": 2.8186228934650757e-06,
      "loss": 0.3998,
      "step": 11270
    },
    {
      "epoch": 0.6012632925561685,
      "grad_norm": 1.8183182565396483,
      "learning_rate": 2.8182025375911064e-06,
      "loss": 0.4073,
      "step": 11280
    },
    {
      "epoch": 0.6017963273899949,
      "grad_norm": 1.542696349302032,
      "learning_rate": 2.8177817265965424e-06,
      "loss": 0.4024,
      "step": 11290
    },
    {
      "epoch": 0.6023293622238213,
      "grad_norm": 1.7645257426266767,
      "learning_rate": 2.8173604606266723e-06,
      "loss": 0.3937,
      "step": 11300
    },
    {
      "epoch": 0.6028623970576478,
      "grad_norm": 1.5206299790783435,
      "learning_rate": 2.816938739826942e-06,
      "loss": 0.4063,
      "step": 11310
    },
    {
      "epoch": 0.6033954318914742,
      "grad_norm": 1.5435971182398083,
      "learning_rate": 2.816516564342954e-06,
      "loss": 0.3863,
      "step": 11320
    },
    {
      "epoch": 0.6039284667253005,
      "grad_norm": 1.6734834601567221,
      "learning_rate": 2.8160939343204675e-06,
      "loss": 0.3961,
      "step": 11330
    },
    {
      "epoch": 0.6044615015591269,
      "grad_norm": 1.658222991081644,
      "learning_rate": 2.8156708499054003e-06,
      "loss": 0.3838,
      "step": 11340
    },
    {
      "epoch": 0.6049945363929533,
      "grad_norm": 1.5771183053159084,
      "learning_rate": 2.815247311243825e-06,
      "loss": 0.4024,
      "step": 11350
    },
    {
      "epoch": 0.6055275712267797,
      "grad_norm": 1.578314296310473,
      "learning_rate": 2.8148233184819717e-06,
      "loss": 0.3935,
      "step": 11360
    },
    {
      "epoch": 0.6060606060606061,
      "grad_norm": 1.6897340844857647,
      "learning_rate": 2.814398871766228e-06,
      "loss": 0.4054,
      "step": 11370
    },
    {
      "epoch": 0.6065936408944325,
      "grad_norm": 1.6566983053221538,
      "learning_rate": 2.813973971243138e-06,
      "loss": 0.3946,
      "step": 11380
    },
    {
      "epoch": 0.6071266757282588,
      "grad_norm": 1.5285278034309233,
      "learning_rate": 2.813548617059401e-06,
      "loss": 0.3991,
      "step": 11390
    },
    {
      "epoch": 0.6076597105620852,
      "grad_norm": 1.6301930066901391,
      "learning_rate": 2.8131228093618756e-06,
      "loss": 0.3985,
      "step": 11400
    },
    {
      "epoch": 0.6081927453959116,
      "grad_norm": 1.562617132118841,
      "learning_rate": 2.8126965482975745e-06,
      "loss": 0.3931,
      "step": 11410
    },
    {
      "epoch": 0.608725780229738,
      "grad_norm": 1.6545662683126874,
      "learning_rate": 2.8122698340136684e-06,
      "loss": 0.3931,
      "step": 11420
    },
    {
      "epoch": 0.6092588150635644,
      "grad_norm": 1.557998796185988,
      "learning_rate": 2.811842666657484e-06,
      "loss": 0.3955,
      "step": 11430
    },
    {
      "epoch": 0.6097918498973908,
      "grad_norm": 1.7368039494892533,
      "learning_rate": 2.811415046376504e-06,
      "loss": 0.39,
      "step": 11440
    },
    {
      "epoch": 0.6103248847312172,
      "grad_norm": 1.4119999043698248,
      "learning_rate": 2.8109869733183683e-06,
      "loss": 0.3812,
      "step": 11450
    },
    {
      "epoch": 0.6108579195650435,
      "grad_norm": 1.5442733322001594,
      "learning_rate": 2.8105584476308726e-06,
      "loss": 0.3971,
      "step": 11460
    },
    {
      "epoch": 0.6113909543988699,
      "grad_norm": 1.6011152090696201,
      "learning_rate": 2.8101294694619697e-06,
      "loss": 0.3979,
      "step": 11470
    },
    {
      "epoch": 0.6119239892326963,
      "grad_norm": 1.7118684012085856,
      "learning_rate": 2.809700038959767e-06,
      "loss": 0.4046,
      "step": 11480
    },
    {
      "epoch": 0.6124570240665227,
      "grad_norm": 1.5630829040472216,
      "learning_rate": 2.80927015627253e-06,
      "loss": 0.3885,
      "step": 11490
    },
    {
      "epoch": 0.6129900589003492,
      "grad_norm": 1.4463955260745038,
      "learning_rate": 2.808839821548678e-06,
      "loss": 0.387,
      "step": 11500
    },
    {
      "epoch": 0.6135230937341756,
      "grad_norm": 1.559307508910976,
      "learning_rate": 2.8084090349367884e-06,
      "loss": 0.3924,
      "step": 11510
    },
    {
      "epoch": 0.614056128568002,
      "grad_norm": 1.4961447241233707,
      "learning_rate": 2.807977796585595e-06,
      "loss": 0.3983,
      "step": 11520
    },
    {
      "epoch": 0.6145891634018283,
      "grad_norm": 1.5905112762805313,
      "learning_rate": 2.8075461066439847e-06,
      "loss": 0.3978,
      "step": 11530
    },
    {
      "epoch": 0.6151221982356547,
      "grad_norm": 1.594145707426966,
      "learning_rate": 2.807113965261003e-06,
      "loss": 0.3906,
      "step": 11540
    },
    {
      "epoch": 0.6156552330694811,
      "grad_norm": 1.5166910573016834,
      "learning_rate": 2.8066813725858503e-06,
      "loss": 0.3941,
      "step": 11550
    },
    {
      "epoch": 0.6161882679033075,
      "grad_norm": 1.550322330198248,
      "learning_rate": 2.8062483287678827e-06,
      "loss": 0.3825,
      "step": 11560
    },
    {
      "epoch": 0.6167213027371339,
      "grad_norm": 1.5409823337605422,
      "learning_rate": 2.8058148339566126e-06,
      "loss": 0.3885,
      "step": 11570
    },
    {
      "epoch": 0.6172543375709603,
      "grad_norm": 1.54912229475863,
      "learning_rate": 2.8053808883017076e-06,
      "loss": 0.3976,
      "step": 11580
    },
    {
      "epoch": 0.6177873724047866,
      "grad_norm": 1.6634766519483923,
      "learning_rate": 2.804946491952991e-06,
      "loss": 0.402,
      "step": 11590
    },
    {
      "epoch": 0.618320407238613,
      "grad_norm": 1.5752399891749032,
      "learning_rate": 2.8045116450604415e-06,
      "loss": 0.3974,
      "step": 11600
    },
    {
      "epoch": 0.6188534420724394,
      "grad_norm": 1.4769240667947436,
      "learning_rate": 2.8040763477741943e-06,
      "loss": 0.3881,
      "step": 11610
    },
    {
      "epoch": 0.6193864769062658,
      "grad_norm": 1.4996019797104767,
      "learning_rate": 2.8036406002445397e-06,
      "loss": 0.3867,
      "step": 11620
    },
    {
      "epoch": 0.6199195117400922,
      "grad_norm": 1.6892343067338462,
      "learning_rate": 2.8032044026219223e-06,
      "loss": 0.3916,
      "step": 11630
    },
    {
      "epoch": 0.6204525465739186,
      "grad_norm": 1.499296099145915,
      "learning_rate": 2.802767755056944e-06,
      "loss": 0.3843,
      "step": 11640
    },
    {
      "epoch": 0.620985581407745,
      "grad_norm": 1.6891078099206098,
      "learning_rate": 2.8023306577003603e-06,
      "loss": 0.3802,
      "step": 11650
    },
    {
      "epoch": 0.6215186162415713,
      "grad_norm": 1.5359532620906358,
      "learning_rate": 2.8018931107030827e-06,
      "loss": 0.3974,
      "step": 11660
    },
    {
      "epoch": 0.6220516510753977,
      "grad_norm": 1.551650860156057,
      "learning_rate": 2.8014551142161787e-06,
      "loss": 0.3895,
      "step": 11670
    },
    {
      "epoch": 0.6225846859092242,
      "grad_norm": 1.3962965306756638,
      "learning_rate": 2.8010166683908696e-06,
      "loss": 0.3881,
      "step": 11680
    },
    {
      "epoch": 0.6231177207430506,
      "grad_norm": 1.5671334600304856,
      "learning_rate": 2.800577773378533e-06,
      "loss": 0.4011,
      "step": 11690
    },
    {
      "epoch": 0.623650755576877,
      "grad_norm": 1.6220708730378142,
      "learning_rate": 2.800138429330701e-06,
      "loss": 0.3995,
      "step": 11700
    },
    {
      "epoch": 0.6241837904107034,
      "grad_norm": 1.580132090675789,
      "learning_rate": 2.799698636399061e-06,
      "loss": 0.3936,
      "step": 11710
    },
    {
      "epoch": 0.6247168252445298,
      "grad_norm": 1.4441344481236609,
      "learning_rate": 2.7992583947354555e-06,
      "loss": 0.3971,
      "step": 11720
    },
    {
      "epoch": 0.6252498600783561,
      "grad_norm": 1.8916747056041694,
      "learning_rate": 2.7988177044918806e-06,
      "loss": 0.4003,
      "step": 11730
    },
    {
      "epoch": 0.6257828949121825,
      "grad_norm": 1.6477323388946823,
      "learning_rate": 2.7983765658204892e-06,
      "loss": 0.3939,
      "step": 11740
    },
    {
      "epoch": 0.6263159297460089,
      "grad_norm": 1.5218455501090793,
      "learning_rate": 2.797934978873588e-06,
      "loss": 0.3939,
      "step": 11750
    },
    {
      "epoch": 0.6268489645798353,
      "grad_norm": 1.4701590918490661,
      "learning_rate": 2.7974929438036384e-06,
      "loss": 0.3896,
      "step": 11760
    },
    {
      "epoch": 0.6273819994136617,
      "grad_norm": 1.5835516193376344,
      "learning_rate": 2.797050460763257e-06,
      "loss": 0.3922,
      "step": 11770
    },
    {
      "epoch": 0.6279150342474881,
      "grad_norm": 1.5041646873772463,
      "learning_rate": 2.7966075299052145e-06,
      "loss": 0.3892,
      "step": 11780
    },
    {
      "epoch": 0.6284480690813145,
      "grad_norm": 1.6303887715028764,
      "learning_rate": 2.7961641513824376e-06,
      "loss": 0.3887,
      "step": 11790
    },
    {
      "epoch": 0.6289811039151408,
      "grad_norm": 1.5349812867480792,
      "learning_rate": 2.7957203253480044e-06,
      "loss": 0.3917,
      "step": 11800
    },
    {
      "epoch": 0.6295141387489672,
      "grad_norm": 1.6465981440360993,
      "learning_rate": 2.7952760519551523e-06,
      "loss": 0.3738,
      "step": 11810
    },
    {
      "epoch": 0.6300471735827936,
      "grad_norm": 1.6584516900753647,
      "learning_rate": 2.794831331357268e-06,
      "loss": 0.4002,
      "step": 11820
    },
    {
      "epoch": 0.63058020841662,
      "grad_norm": 1.5859618642595954,
      "learning_rate": 2.7943861637078965e-06,
      "loss": 0.4007,
      "step": 11830
    },
    {
      "epoch": 0.6311132432504464,
      "grad_norm": 1.7511956751160198,
      "learning_rate": 2.793940549160735e-06,
      "loss": 0.4001,
      "step": 11840
    },
    {
      "epoch": 0.6316462780842728,
      "grad_norm": 1.4246756095863924,
      "learning_rate": 2.7934944878696365e-06,
      "loss": 0.3909,
      "step": 11850
    },
    {
      "epoch": 0.6321793129180991,
      "grad_norm": 1.646032214208831,
      "learning_rate": 2.793047979988607e-06,
      "loss": 0.3961,
      "step": 11860
    },
    {
      "epoch": 0.6327123477519256,
      "grad_norm": 1.5620031004110237,
      "learning_rate": 2.7926010256718064e-06,
      "loss": 0.3898,
      "step": 11870
    },
    {
      "epoch": 0.633245382585752,
      "grad_norm": 1.7086478858528293,
      "learning_rate": 2.7921536250735508e-06,
      "loss": 0.3971,
      "step": 11880
    },
    {
      "epoch": 0.6337784174195784,
      "grad_norm": 1.6443157640472792,
      "learning_rate": 2.791705778348308e-06,
      "loss": 0.3984,
      "step": 11890
    },
    {
      "epoch": 0.6343114522534048,
      "grad_norm": 1.584048024256836,
      "learning_rate": 2.7912574856507014e-06,
      "loss": 0.3923,
      "step": 11900
    },
    {
      "epoch": 0.6348444870872312,
      "grad_norm": 1.510070722855611,
      "learning_rate": 2.7908087471355076e-06,
      "loss": 0.3888,
      "step": 11910
    },
    {
      "epoch": 0.6353775219210576,
      "grad_norm": 1.6268634317390713,
      "learning_rate": 2.790359562957657e-06,
      "loss": 0.3895,
      "step": 11920
    },
    {
      "epoch": 0.635910556754884,
      "grad_norm": 1.6236884243463365,
      "learning_rate": 2.7899099332722352e-06,
      "loss": 0.3973,
      "step": 11930
    },
    {
      "epoch": 0.6364435915887103,
      "grad_norm": 1.6338186141778763,
      "learning_rate": 2.7894598582344803e-06,
      "loss": 0.3866,
      "step": 11940
    },
    {
      "epoch": 0.6369766264225367,
      "grad_norm": 1.5622984679180052,
      "learning_rate": 2.789009337999784e-06,
      "loss": 0.3763,
      "step": 11950
    },
    {
      "epoch": 0.6375096612563631,
      "grad_norm": 1.673952296983321,
      "learning_rate": 2.788558372723693e-06,
      "loss": 0.3925,
      "step": 11960
    },
    {
      "epoch": 0.6380426960901895,
      "grad_norm": 1.5932902944192429,
      "learning_rate": 2.7881069625619065e-06,
      "loss": 0.3848,
      "step": 11970
    },
    {
      "epoch": 0.6385757309240159,
      "grad_norm": 1.4751362343477044,
      "learning_rate": 2.7876551076702774e-06,
      "loss": 0.3869,
      "step": 11980
    },
    {
      "epoch": 0.6391087657578423,
      "grad_norm": 1.4843454969109606,
      "learning_rate": 2.7872028082048128e-06,
      "loss": 0.3819,
      "step": 11990
    },
    {
      "epoch": 0.6396418005916686,
      "grad_norm": 1.5677105104818196,
      "learning_rate": 2.7867500643216733e-06,
      "loss": 0.3892,
      "step": 12000
    },
    {
      "epoch": 0.640174835425495,
      "grad_norm": 1.5718146747780057,
      "learning_rate": 2.7862968761771713e-06,
      "loss": 0.3888,
      "step": 12010
    },
    {
      "epoch": 0.6407078702593214,
      "grad_norm": 1.5279642054407692,
      "learning_rate": 2.7858432439277754e-06,
      "loss": 0.3968,
      "step": 12020
    },
    {
      "epoch": 0.6412409050931478,
      "grad_norm": 1.6503034433177566,
      "learning_rate": 2.7853891677301045e-06,
      "loss": 0.3834,
      "step": 12030
    },
    {
      "epoch": 0.6417739399269742,
      "grad_norm": 1.7350616613879568,
      "learning_rate": 2.784934647740934e-06,
      "loss": 0.4051,
      "step": 12040
    },
    {
      "epoch": 0.6423069747608006,
      "grad_norm": 1.7500245634075582,
      "learning_rate": 2.784479684117189e-06,
      "loss": 0.3959,
      "step": 12050
    },
    {
      "epoch": 0.6428400095946271,
      "grad_norm": 1.6174581393382492,
      "learning_rate": 2.7840242770159503e-06,
      "loss": 0.3949,
      "step": 12060
    },
    {
      "epoch": 0.6433730444284534,
      "grad_norm": 1.4739002685686147,
      "learning_rate": 2.783568426594451e-06,
      "loss": 0.3791,
      "step": 12070
    },
    {
      "epoch": 0.6439060792622798,
      "grad_norm": 1.6727367696671884,
      "learning_rate": 2.7831121330100775e-06,
      "loss": 0.3863,
      "step": 12080
    },
    {
      "epoch": 0.6444391140961062,
      "grad_norm": 1.5022474982429521,
      "learning_rate": 2.7826553964203694e-06,
      "loss": 0.3845,
      "step": 12090
    },
    {
      "epoch": 0.6449721489299326,
      "grad_norm": 1.5254855001374756,
      "learning_rate": 2.782198216983018e-06,
      "loss": 0.3992,
      "step": 12100
    },
    {
      "epoch": 0.645505183763759,
      "grad_norm": 1.6587868380084196,
      "learning_rate": 2.7817405948558684e-06,
      "loss": 0.3871,
      "step": 12110
    },
    {
      "epoch": 0.6460382185975854,
      "grad_norm": 1.5555910493220377,
      "learning_rate": 2.7812825301969194e-06,
      "loss": 0.3994,
      "step": 12120
    },
    {
      "epoch": 0.6465712534314118,
      "grad_norm": 1.5755611511995655,
      "learning_rate": 2.7808240231643207e-06,
      "loss": 0.3913,
      "step": 12130
    },
    {
      "epoch": 0.6471042882652381,
      "grad_norm": 1.6869162094152887,
      "learning_rate": 2.780365073916377e-06,
      "loss": 0.3932,
      "step": 12140
    },
    {
      "epoch": 0.6476373230990645,
      "grad_norm": 1.519299268086837,
      "learning_rate": 2.779905682611543e-06,
      "loss": 0.3978,
      "step": 12150
    },
    {
      "epoch": 0.6481703579328909,
      "grad_norm": 1.4803838347284977,
      "learning_rate": 2.779445849408428e-06,
      "loss": 0.3783,
      "step": 12160
    },
    {
      "epoch": 0.6487033927667173,
      "grad_norm": 1.7044629885259677,
      "learning_rate": 2.7789855744657935e-06,
      "loss": 0.3766,
      "step": 12170
    },
    {
      "epoch": 0.6492364276005437,
      "grad_norm": 1.5840843113701972,
      "learning_rate": 2.7785248579425537e-06,
      "loss": 0.3749,
      "step": 12180
    },
    {
      "epoch": 0.6497694624343701,
      "grad_norm": 1.5211073198637621,
      "learning_rate": 2.7780636999977737e-06,
      "loss": 0.3908,
      "step": 12190
    },
    {
      "epoch": 0.6503024972681964,
      "grad_norm": 1.5858033266004543,
      "learning_rate": 2.7776021007906734e-06,
      "loss": 0.3901,
      "step": 12200
    },
    {
      "epoch": 0.6508355321020228,
      "grad_norm": 1.6076389777867792,
      "learning_rate": 2.7771400604806227e-06,
      "loss": 0.3923,
      "step": 12210
    },
    {
      "epoch": 0.6513685669358492,
      "grad_norm": 1.7696672716687496,
      "learning_rate": 2.7766775792271464e-06,
      "loss": 0.3976,
      "step": 12220
    },
    {
      "epoch": 0.6519016017696756,
      "grad_norm": 1.7391643004557578,
      "learning_rate": 2.7762146571899187e-06,
      "loss": 0.3777,
      "step": 12230
    },
    {
      "epoch": 0.6524346366035021,
      "grad_norm": 1.6984356458111702,
      "learning_rate": 2.775751294528768e-06,
      "loss": 0.3855,
      "step": 12240
    },
    {
      "epoch": 0.6529676714373285,
      "grad_norm": 1.7935011590016339,
      "learning_rate": 2.775287491403674e-06,
      "loss": 0.3936,
      "step": 12250
    },
    {
      "epoch": 0.6535007062711549,
      "grad_norm": 1.6694000704578658,
      "learning_rate": 2.7748232479747683e-06,
      "loss": 0.3895,
      "step": 12260
    },
    {
      "epoch": 0.6540337411049812,
      "grad_norm": 1.499603418456414,
      "learning_rate": 2.774358564402336e-06,
      "loss": 0.3925,
      "step": 12270
    },
    {
      "epoch": 0.6545667759388076,
      "grad_norm": 1.6250028271238461,
      "learning_rate": 2.773893440846812e-06,
      "loss": 0.3975,
      "step": 12280
    },
    {
      "epoch": 0.655099810772634,
      "grad_norm": 1.481337339713028,
      "learning_rate": 2.7734278774687842e-06,
      "loss": 0.3887,
      "step": 12290
    },
    {
      "epoch": 0.6556328456064604,
      "grad_norm": 1.5277235272770724,
      "learning_rate": 2.7729618744289933e-06,
      "loss": 0.3902,
      "step": 12300
    },
    {
      "epoch": 0.6561658804402868,
      "grad_norm": 1.6356182023504948,
      "learning_rate": 2.7724954318883294e-06,
      "loss": 0.3938,
      "step": 12310
    },
    {
      "epoch": 0.6566989152741132,
      "grad_norm": 1.5844459219660234,
      "learning_rate": 2.772028550007837e-06,
      "loss": 0.3879,
      "step": 12320
    },
    {
      "epoch": 0.6572319501079396,
      "grad_norm": 1.74063027247512,
      "learning_rate": 2.77156122894871e-06,
      "loss": 0.3794,
      "step": 12330
    },
    {
      "epoch": 0.6577649849417659,
      "grad_norm": 1.6460156163706472,
      "learning_rate": 2.7710934688722958e-06,
      "loss": 0.3958,
      "step": 12340
    },
    {
      "epoch": 0.6582980197755923,
      "grad_norm": 1.4862972578445002,
      "learning_rate": 2.7706252699400917e-06,
      "loss": 0.3901,
      "step": 12350
    },
    {
      "epoch": 0.6588310546094187,
      "grad_norm": 1.6798943550451348,
      "learning_rate": 2.7701566323137478e-06,
      "loss": 0.3924,
      "step": 12360
    },
    {
      "epoch": 0.6593640894432451,
      "grad_norm": 1.7390232253098628,
      "learning_rate": 2.7696875561550656e-06,
      "loss": 0.397,
      "step": 12370
    },
    {
      "epoch": 0.6598971242770715,
      "grad_norm": 1.6570460319815885,
      "learning_rate": 2.769218041625997e-06,
      "loss": 0.3744,
      "step": 12380
    },
    {
      "epoch": 0.6604301591108979,
      "grad_norm": 1.7009473756868425,
      "learning_rate": 2.768748088888646e-06,
      "loss": 0.3823,
      "step": 12390
    },
    {
      "epoch": 0.6609631939447242,
      "grad_norm": 1.633786993834928,
      "learning_rate": 2.768277698105268e-06,
      "loss": 0.3915,
      "step": 12400
    },
    {
      "epoch": 0.6614962287785506,
      "grad_norm": 1.6082217898062294,
      "learning_rate": 2.767806869438269e-06,
      "loss": 0.3939,
      "step": 12410
    },
    {
      "epoch": 0.662029263612377,
      "grad_norm": 1.8336157331627372,
      "learning_rate": 2.767335603050207e-06,
      "loss": 0.3989,
      "step": 12420
    },
    {
      "epoch": 0.6625622984462035,
      "grad_norm": 1.5944938834193176,
      "learning_rate": 2.7668638991037905e-06,
      "loss": 0.3828,
      "step": 12430
    },
    {
      "epoch": 0.6630953332800299,
      "grad_norm": 1.6435895951256687,
      "learning_rate": 2.7663917577618794e-06,
      "loss": 0.3866,
      "step": 12440
    },
    {
      "epoch": 0.6636283681138563,
      "grad_norm": 1.6459012383545821,
      "learning_rate": 2.7659191791874847e-06,
      "loss": 0.3867,
      "step": 12450
    },
    {
      "epoch": 0.6641614029476827,
      "grad_norm": 1.6654858926813907,
      "learning_rate": 2.7654461635437677e-06,
      "loss": 0.386,
      "step": 12460
    },
    {
      "epoch": 0.664694437781509,
      "grad_norm": 1.6554821904539627,
      "learning_rate": 2.7649727109940407e-06,
      "loss": 0.3967,
      "step": 12470
    },
    {
      "epoch": 0.6652274726153354,
      "grad_norm": 1.4834347250957596,
      "learning_rate": 2.7644988217017683e-06,
      "loss": 0.3742,
      "step": 12480
    },
    {
      "epoch": 0.6657605074491618,
      "grad_norm": 1.3693380798575572,
      "learning_rate": 2.7640244958305642e-06,
      "loss": 0.3847,
      "step": 12490
    },
    {
      "epoch": 0.6662935422829882,
      "grad_norm": 1.5588271991203428,
      "learning_rate": 2.763549733544193e-06,
      "loss": 0.3939,
      "step": 12500
    },
    {
      "epoch": 0.6668265771168146,
      "grad_norm": 1.5094609414840297,
      "learning_rate": 2.7630745350065716e-06,
      "loss": 0.3948,
      "step": 12510
    },
    {
      "epoch": 0.667359611950641,
      "grad_norm": 1.6096562139624555,
      "learning_rate": 2.7625989003817652e-06,
      "loss": 0.3873,
      "step": 12520
    },
    {
      "epoch": 0.6678926467844674,
      "grad_norm": 1.3720150144708103,
      "learning_rate": 2.762122829833991e-06,
      "loss": 0.4006,
      "step": 12530
    },
    {
      "epoch": 0.6684256816182937,
      "grad_norm": 1.5973622483367065,
      "learning_rate": 2.761646323527616e-06,
      "loss": 0.3935,
      "step": 12540
    },
    {
      "epoch": 0.6689587164521201,
      "grad_norm": 1.670572258949521,
      "learning_rate": 2.7611693816271594e-06,
      "loss": 0.3831,
      "step": 12550
    },
    {
      "epoch": 0.6694917512859465,
      "grad_norm": 1.5466619369265269,
      "learning_rate": 2.760692004297288e-06,
      "loss": 0.3962,
      "step": 12560
    },
    {
      "epoch": 0.6700247861197729,
      "grad_norm": 1.5784082936759711,
      "learning_rate": 2.7602141917028215e-06,
      "loss": 0.3893,
      "step": 12570
    },
    {
      "epoch": 0.6705578209535993,
      "grad_norm": 1.3770342295151579,
      "learning_rate": 2.7597359440087274e-06,
      "loss": 0.3785,
      "step": 12580
    },
    {
      "epoch": 0.6710908557874257,
      "grad_norm": 1.5267884625368433,
      "learning_rate": 2.7592572613801262e-06,
      "loss": 0.3958,
      "step": 12590
    },
    {
      "epoch": 0.671623890621252,
      "grad_norm": 1.55731717516025,
      "learning_rate": 2.7587781439822857e-06,
      "loss": 0.3944,
      "step": 12600
    },
    {
      "epoch": 0.6721569254550784,
      "grad_norm": 1.5908625787217203,
      "learning_rate": 2.758298591980626e-06,
      "loss": 0.4036,
      "step": 12610
    },
    {
      "epoch": 0.6726899602889049,
      "grad_norm": 1.7357248306716566,
      "learning_rate": 2.7578186055407176e-06,
      "loss": 0.394,
      "step": 12620
    },
    {
      "epoch": 0.6732229951227313,
      "grad_norm": 1.595642554910627,
      "learning_rate": 2.757338184828278e-06,
      "loss": 0.3869,
      "step": 12630
    },
    {
      "epoch": 0.6737560299565577,
      "grad_norm": 1.5204323027139102,
      "learning_rate": 2.756857330009177e-06,
      "loss": 0.3994,
      "step": 12640
    },
    {
      "epoch": 0.6742890647903841,
      "grad_norm": 1.4689370333113319,
      "learning_rate": 2.7563760412494345e-06,
      "loss": 0.3902,
      "step": 12650
    },
    {
      "epoch": 0.6748220996242105,
      "grad_norm": 1.5979756040887692,
      "learning_rate": 2.755894318715219e-06,
      "loss": 0.4041,
      "step": 12660
    },
    {
      "epoch": 0.6753551344580369,
      "grad_norm": 1.5231933200814187,
      "learning_rate": 2.7554121625728494e-06,
      "loss": 0.376,
      "step": 12670
    },
    {
      "epoch": 0.6758881692918632,
      "grad_norm": 1.5699045488484404,
      "learning_rate": 2.7549295729887943e-06,
      "loss": 0.394,
      "step": 12680
    },
    {
      "epoch": 0.6764212041256896,
      "grad_norm": 1.4778739839337431,
      "learning_rate": 2.7544465501296718e-06,
      "loss": 0.3901,
      "step": 12690
    },
    {
      "epoch": 0.676954238959516,
      "grad_norm": 1.6906897466958548,
      "learning_rate": 2.753963094162249e-06,
      "loss": 0.3907,
      "step": 12700
    },
    {
      "epoch": 0.6774872737933424,
      "grad_norm": 1.6157173022924884,
      "learning_rate": 2.7534792052534444e-06,
      "loss": 0.3981,
      "step": 12710
    },
    {
      "epoch": 0.6780203086271688,
      "grad_norm": 1.4746277625120372,
      "learning_rate": 2.752994883570324e-06,
      "loss": 0.3818,
      "step": 12720
    },
    {
      "epoch": 0.6785533434609952,
      "grad_norm": 1.4980146475409215,
      "learning_rate": 2.7525101292801038e-06,
      "loss": 0.3947,
      "step": 12730
    },
    {
      "epoch": 0.6790863782948215,
      "grad_norm": 1.7288711020539997,
      "learning_rate": 2.7520249425501493e-06,
      "loss": 0.3874,
      "step": 12740
    },
    {
      "epoch": 0.6796194131286479,
      "grad_norm": 1.487037201541101,
      "learning_rate": 2.7515393235479764e-06,
      "loss": 0.3682,
      "step": 12750
    },
    {
      "epoch": 0.6801524479624743,
      "grad_norm": 1.5890556211452727,
      "learning_rate": 2.751053272441248e-06,
      "loss": 0.3829,
      "step": 12760
    },
    {
      "epoch": 0.6806854827963007,
      "grad_norm": 1.4327425499391107,
      "learning_rate": 2.7505667893977774e-06,
      "loss": 0.3935,
      "step": 12770
    },
    {
      "epoch": 0.6812185176301271,
      "grad_norm": 1.691058617531051,
      "learning_rate": 2.750079874585528e-06,
      "loss": 0.4008,
      "step": 12780
    },
    {
      "epoch": 0.6817515524639535,
      "grad_norm": 1.520556124456952,
      "learning_rate": 2.74959252817261e-06,
      "loss": 0.3705,
      "step": 12790
    },
    {
      "epoch": 0.68228458729778,
      "grad_norm": 1.7311682938052233,
      "learning_rate": 2.7491047503272848e-06,
      "loss": 0.3881,
      "step": 12800
    },
    {
      "epoch": 0.6828176221316063,
      "grad_norm": 1.6002491186645775,
      "learning_rate": 2.748616541217961e-06,
      "loss": 0.3981,
      "step": 12810
    },
    {
      "epoch": 0.6833506569654327,
      "grad_norm": 1.8889318875427128,
      "learning_rate": 2.7481279010131975e-06,
      "loss": 0.3976,
      "step": 12820
    },
    {
      "epoch": 0.6838836917992591,
      "grad_norm": 1.5127717189862142,
      "learning_rate": 2.747638829881702e-06,
      "loss": 0.3881,
      "step": 12830
    },
    {
      "epoch": 0.6844167266330855,
      "grad_norm": 1.4793281533153604,
      "learning_rate": 2.74714932799233e-06,
      "loss": 0.3851,
      "step": 12840
    },
    {
      "epoch": 0.6849497614669119,
      "grad_norm": 1.693718557764651,
      "learning_rate": 2.746659395514085e-06,
      "loss": 0.3949,
      "step": 12850
    },
    {
      "epoch": 0.6854827963007383,
      "grad_norm": 1.7812046830091415,
      "learning_rate": 2.746169032616122e-06,
      "loss": 0.3862,
      "step": 12860
    },
    {
      "epoch": 0.6860158311345647,
      "grad_norm": 1.559640382471567,
      "learning_rate": 2.7456782394677428e-06,
      "loss": 0.3866,
      "step": 12870
    },
    {
      "epoch": 0.686548865968391,
      "grad_norm": 1.574501810795579,
      "learning_rate": 2.745187016238397e-06,
      "loss": 0.3915,
      "step": 12880
    },
    {
      "epoch": 0.6870819008022174,
      "grad_norm": 1.5767675304948638,
      "learning_rate": 2.7446953630976836e-06,
      "loss": 0.3839,
      "step": 12890
    },
    {
      "epoch": 0.6876149356360438,
      "grad_norm": 1.5216465100054701,
      "learning_rate": 2.744203280215351e-06,
      "loss": 0.3852,
      "step": 12900
    },
    {
      "epoch": 0.6881479704698702,
      "grad_norm": 1.7421711843910284,
      "learning_rate": 2.743710767761294e-06,
      "loss": 0.3938,
      "step": 12910
    },
    {
      "epoch": 0.6886810053036966,
      "grad_norm": 1.5065946298557589,
      "learning_rate": 2.7432178259055574e-06,
      "loss": 0.3816,
      "step": 12920
    },
    {
      "epoch": 0.689214040137523,
      "grad_norm": 1.4825540407382936,
      "learning_rate": 2.742724454818333e-06,
      "loss": 0.3889,
      "step": 12930
    },
    {
      "epoch": 0.6897470749713493,
      "grad_norm": 1.640329470277106,
      "learning_rate": 2.7422306546699618e-06,
      "loss": 0.3915,
      "step": 12940
    },
    {
      "epoch": 0.6902801098051757,
      "grad_norm": 1.5228227884510819,
      "learning_rate": 2.741736425630932e-06,
      "loss": 0.3837,
      "step": 12950
    },
    {
      "epoch": 0.6908131446390021,
      "grad_norm": 1.5943598376766,
      "learning_rate": 2.7412417678718803e-06,
      "loss": 0.3936,
      "step": 12960
    },
    {
      "epoch": 0.6913461794728285,
      "grad_norm": 1.6640939943309663,
      "learning_rate": 2.7407466815635925e-06,
      "loss": 0.389,
      "step": 12970
    },
    {
      "epoch": 0.6918792143066549,
      "grad_norm": 1.5914265221935253,
      "learning_rate": 2.740251166877001e-06,
      "loss": 0.3937,
      "step": 12980
    },
    {
      "epoch": 0.6924122491404814,
      "grad_norm": 1.4778763138575604,
      "learning_rate": 2.739755223983185e-06,
      "loss": 0.3947,
      "step": 12990
    },
    {
      "epoch": 0.6929452839743078,
      "grad_norm": 1.5526855020948873,
      "learning_rate": 2.7392588530533744e-06,
      "loss": 0.373,
      "step": 13000
    },
    {
      "epoch": 0.6934783188081342,
      "grad_norm": 1.6275445729544191,
      "learning_rate": 2.7387620542589457e-06,
      "loss": 0.3891,
      "step": 13010
    },
    {
      "epoch": 0.6940113536419605,
      "grad_norm": 1.5013424776270292,
      "learning_rate": 2.738264827771422e-06,
      "loss": 0.387,
      "step": 13020
    },
    {
      "epoch": 0.6945443884757869,
      "grad_norm": 1.6394412410844779,
      "learning_rate": 2.7377671737624754e-06,
      "loss": 0.3848,
      "step": 13030
    },
    {
      "epoch": 0.6950774233096133,
      "grad_norm": 1.6087724219258532,
      "learning_rate": 2.737269092403925e-06,
      "loss": 0.3866,
      "step": 13040
    },
    {
      "epoch": 0.6956104581434397,
      "grad_norm": 1.499480313581256,
      "learning_rate": 2.7367705838677377e-06,
      "loss": 0.3872,
      "step": 13050
    },
    {
      "epoch": 0.6961434929772661,
      "grad_norm": 1.4249272150967984,
      "learning_rate": 2.736271648326028e-06,
      "loss": 0.3861,
      "step": 13060
    },
    {
      "epoch": 0.6966765278110925,
      "grad_norm": 1.5347293157426203,
      "learning_rate": 2.735772285951057e-06,
      "loss": 0.3866,
      "step": 13070
    },
    {
      "epoch": 0.6972095626449188,
      "grad_norm": 1.4839464584619286,
      "learning_rate": 2.7352724969152344e-06,
      "loss": 0.3792,
      "step": 13080
    },
    {
      "epoch": 0.6977425974787452,
      "grad_norm": 1.660798590531833,
      "learning_rate": 2.7347722813911164e-06,
      "loss": 0.3786,
      "step": 13090
    },
    {
      "epoch": 0.6982756323125716,
      "grad_norm": 1.615091682066585,
      "learning_rate": 2.734271639551407e-06,
      "loss": 0.3978,
      "step": 13100
    },
    {
      "epoch": 0.698808667146398,
      "grad_norm": 1.6289233735933937,
      "learning_rate": 2.733770571568956e-06,
      "loss": 0.3816,
      "step": 13110
    },
    {
      "epoch": 0.6993417019802244,
      "grad_norm": 1.5640971774206687,
      "learning_rate": 2.733269077616762e-06,
      "loss": 0.3855,
      "step": 13120
    },
    {
      "epoch": 0.6998747368140508,
      "grad_norm": 3.1169071364406746,
      "learning_rate": 2.732767157867971e-06,
      "loss": 0.3996,
      "step": 13130
    },
    {
      "epoch": 0.7004077716478772,
      "grad_norm": 1.637535585462466,
      "learning_rate": 2.7322648124958733e-06,
      "loss": 0.3877,
      "step": 13140
    },
    {
      "epoch": 0.7009408064817035,
      "grad_norm": 1.338365834276967,
      "learning_rate": 2.731762041673909e-06,
      "loss": 0.3782,
      "step": 13150
    },
    {
      "epoch": 0.7014738413155299,
      "grad_norm": 1.6075817656975602,
      "learning_rate": 2.731258845575664e-06,
      "loss": 0.388,
      "step": 13160
    },
    {
      "epoch": 0.7020068761493563,
      "grad_norm": 1.5489653356145214,
      "learning_rate": 2.73075522437487e-06,
      "loss": 0.4131,
      "step": 13170
    },
    {
      "epoch": 0.7025399109831828,
      "grad_norm": 1.5015209058029906,
      "learning_rate": 2.7302511782454075e-06,
      "loss": 0.3825,
      "step": 13180
    },
    {
      "epoch": 0.7030729458170092,
      "grad_norm": 1.4912536277176622,
      "learning_rate": 2.7297467073613023e-06,
      "loss": 0.3821,
      "step": 13190
    },
    {
      "epoch": 0.7036059806508356,
      "grad_norm": 1.5127147010007216,
      "learning_rate": 2.729241811896728e-06,
      "loss": 0.3934,
      "step": 13200
    },
    {
      "epoch": 0.704139015484662,
      "grad_norm": 1.522773318601683,
      "learning_rate": 2.728736492026003e-06,
      "loss": 0.3881,
      "step": 13210
    },
    {
      "epoch": 0.7046720503184883,
      "grad_norm": 1.622986666839565,
      "learning_rate": 2.7282307479235935e-06,
      "loss": 0.3878,
      "step": 13220
    },
    {
      "epoch": 0.7052050851523147,
      "grad_norm": 1.57857404434383,
      "learning_rate": 2.7277245797641123e-06,
      "loss": 0.3736,
      "step": 13230
    },
    {
      "epoch": 0.7057381199861411,
      "grad_norm": 1.6580360295681327,
      "learning_rate": 2.727217987722318e-06,
      "loss": 0.3828,
      "step": 13240
    },
    {
      "epoch": 0.7062711548199675,
      "grad_norm": 1.577739831378232,
      "learning_rate": 2.7267109719731157e-06,
      "loss": 0.4002,
      "step": 13250
    },
    {
      "epoch": 0.7068041896537939,
      "grad_norm": 1.6991613464206654,
      "learning_rate": 2.726203532691557e-06,
      "loss": 0.3832,
      "step": 13260
    },
    {
      "epoch": 0.7073372244876203,
      "grad_norm": 1.50920445086471,
      "learning_rate": 2.72569567005284e-06,
      "loss": 0.3562,
      "step": 13270
    },
    {
      "epoch": 0.7078702593214466,
      "grad_norm": 1.594758541333458,
      "learning_rate": 2.725187384232308e-06,
      "loss": 0.3794,
      "step": 13280
    },
    {
      "epoch": 0.708403294155273,
      "grad_norm": 1.4690326485687684,
      "learning_rate": 2.7246786754054507e-06,
      "loss": 0.3982,
      "step": 13290
    },
    {
      "epoch": 0.7089363289890994,
      "grad_norm": 1.5137623870548904,
      "learning_rate": 2.7241695437479054e-06,
      "loss": 0.3742,
      "step": 13300
    },
    {
      "epoch": 0.7094693638229258,
      "grad_norm": 1.5745854341801928,
      "learning_rate": 2.723659989435453e-06,
      "loss": 0.3829,
      "step": 13310
    },
    {
      "epoch": 0.7100023986567522,
      "grad_norm": 1.8141976136973568,
      "learning_rate": 2.7231500126440216e-06,
      "loss": 0.3775,
      "step": 13320
    },
    {
      "epoch": 0.7105354334905786,
      "grad_norm": 1.3874108647877745,
      "learning_rate": 2.7226396135496854e-06,
      "loss": 0.3751,
      "step": 13330
    },
    {
      "epoch": 0.711068468324405,
      "grad_norm": 1.525279426750466,
      "learning_rate": 2.722128792328664e-06,
      "loss": 0.3713,
      "step": 13340
    },
    {
      "epoch": 0.7116015031582313,
      "grad_norm": 1.4276440398999402,
      "learning_rate": 2.721617549157322e-06,
      "loss": 0.3883,
      "step": 13350
    },
    {
      "epoch": 0.7121345379920577,
      "grad_norm": 1.3819094931839373,
      "learning_rate": 2.721105884212172e-06,
      "loss": 0.3917,
      "step": 13360
    },
    {
      "epoch": 0.7126675728258842,
      "grad_norm": 1.6112203117077113,
      "learning_rate": 2.720593797669869e-06,
      "loss": 0.377,
      "step": 13370
    },
    {
      "epoch": 0.7132006076597106,
      "grad_norm": 1.5913240291688735,
      "learning_rate": 2.7200812897072156e-06,
      "loss": 0.3837,
      "step": 13380
    },
    {
      "epoch": 0.713733642493537,
      "grad_norm": 1.4725209599416555,
      "learning_rate": 2.71956836050116e-06,
      "loss": 0.3893,
      "step": 13390
    },
    {
      "epoch": 0.7142666773273634,
      "grad_norm": 1.648601587773875,
      "learning_rate": 2.719055010228795e-06,
      "loss": 0.3815,
      "step": 13400
    },
    {
      "epoch": 0.7147997121611898,
      "grad_norm": 1.6433557817933246,
      "learning_rate": 2.7185412390673595e-06,
      "loss": 0.3896,
      "step": 13410
    },
    {
      "epoch": 0.7153327469950161,
      "grad_norm": 1.4819913856175566,
      "learning_rate": 2.718027047194237e-06,
      "loss": 0.3716,
      "step": 13420
    },
    {
      "epoch": 0.7158657818288425,
      "grad_norm": 1.5042676159233608,
      "learning_rate": 2.717512434786957e-06,
      "loss": 0.3951,
      "step": 13430
    },
    {
      "epoch": 0.7163988166626689,
      "grad_norm": 1.7504478372677417,
      "learning_rate": 2.716997402023193e-06,
      "loss": 0.3901,
      "step": 13440
    },
    {
      "epoch": 0.7169318514964953,
      "grad_norm": 1.6553920045258095,
      "learning_rate": 2.7164819490807652e-06,
      "loss": 0.3883,
      "step": 13450
    },
    {
      "epoch": 0.7174648863303217,
      "grad_norm": 1.462738438758308,
      "learning_rate": 2.7159660761376383e-06,
      "loss": 0.3743,
      "step": 13460
    },
    {
      "epoch": 0.7179979211641481,
      "grad_norm": 1.6043869019545287,
      "learning_rate": 2.715449783371921e-06,
      "loss": 0.3945,
      "step": 13470
    },
    {
      "epoch": 0.7185309559979745,
      "grad_norm": 1.5124295693475063,
      "learning_rate": 2.7149330709618677e-06,
      "loss": 0.3887,
      "step": 13480
    },
    {
      "epoch": 0.7190639908318008,
      "grad_norm": 1.699222381556251,
      "learning_rate": 2.7144159390858787e-06,
      "loss": 0.4007,
      "step": 13490
    },
    {
      "epoch": 0.7195970256656272,
      "grad_norm": 1.5741560149187015,
      "learning_rate": 2.7138983879224974e-06,
      "loss": 0.3723,
      "step": 13500
    },
    {
      "epoch": 0.7201300604994536,
      "grad_norm": 1.6301904876684477,
      "learning_rate": 2.7133804176504126e-06,
      "loss": 0.3787,
      "step": 13510
    },
    {
      "epoch": 0.72066309533328,
      "grad_norm": 1.6197016339526287,
      "learning_rate": 2.712862028448459e-06,
      "loss": 0.3785,
      "step": 13520
    },
    {
      "epoch": 0.7211961301671064,
      "grad_norm": 1.5527484867333805,
      "learning_rate": 2.7123432204956133e-06,
      "loss": 0.4004,
      "step": 13530
    },
    {
      "epoch": 0.7217291650009328,
      "grad_norm": 1.7191320204858784,
      "learning_rate": 2.7118239939709993e-06,
      "loss": 0.3976,
      "step": 13540
    },
    {
      "epoch": 0.7222621998347593,
      "grad_norm": 1.7710265285463023,
      "learning_rate": 2.711304349053884e-06,
      "loss": 0.3714,
      "step": 13550
    },
    {
      "epoch": 0.7227952346685856,
      "grad_norm": 1.7964362812482046,
      "learning_rate": 2.7107842859236805e-06,
      "loss": 0.3914,
      "step": 13560
    },
    {
      "epoch": 0.723328269502412,
      "grad_norm": 1.5334407354457222,
      "learning_rate": 2.710263804759943e-06,
      "loss": 0.3757,
      "step": 13570
    },
    {
      "epoch": 0.7238613043362384,
      "grad_norm": 1.5419702064547496,
      "learning_rate": 2.709742905742373e-06,
      "loss": 0.383,
      "step": 13580
    },
    {
      "epoch": 0.7243943391700648,
      "grad_norm": 1.5701350411320052,
      "learning_rate": 2.7092215890508153e-06,
      "loss": 0.3864,
      "step": 13590
    },
    {
      "epoch": 0.7249273740038912,
      "grad_norm": 1.6177726224412516,
      "learning_rate": 2.708699854865259e-06,
      "loss": 0.3751,
      "step": 13600
    },
    {
      "epoch": 0.7254604088377176,
      "grad_norm": 1.696569539318935,
      "learning_rate": 2.708177703365837e-06,
      "loss": 0.3903,
      "step": 13610
    },
    {
      "epoch": 0.725993443671544,
      "grad_norm": 1.7301027838412937,
      "learning_rate": 2.7076551347328267e-06,
      "loss": 0.3931,
      "step": 13620
    },
    {
      "epoch": 0.7265264785053703,
      "grad_norm": 1.4018042335658578,
      "learning_rate": 2.7071321491466493e-06,
      "loss": 0.3822,
      "step": 13630
    },
    {
      "epoch": 0.7270595133391967,
      "grad_norm": 1.6852557666125316,
      "learning_rate": 2.7066087467878696e-06,
      "loss": 0.3854,
      "step": 13640
    },
    {
      "epoch": 0.7275925481730231,
      "grad_norm": 1.621367972057748,
      "learning_rate": 2.7060849278371975e-06,
      "loss": 0.4003,
      "step": 13650
    },
    {
      "epoch": 0.7281255830068495,
      "grad_norm": 1.50999693876737,
      "learning_rate": 2.705560692475485e-06,
      "loss": 0.3823,
      "step": 13660
    },
    {
      "epoch": 0.7286586178406759,
      "grad_norm": 1.5439011651767698,
      "learning_rate": 2.70503604088373e-06,
      "loss": 0.3845,
      "step": 13670
    },
    {
      "epoch": 0.7291916526745023,
      "grad_norm": 1.5693718207431124,
      "learning_rate": 2.7045109732430718e-06,
      "loss": 0.3928,
      "step": 13680
    },
    {
      "epoch": 0.7297246875083286,
      "grad_norm": 1.7260676201328702,
      "learning_rate": 2.703985489734795e-06,
      "loss": 0.3924,
      "step": 13690
    },
    {
      "epoch": 0.730257722342155,
      "grad_norm": 1.5912490301842381,
      "learning_rate": 2.703459590540327e-06,
      "loss": 0.3908,
      "step": 13700
    },
    {
      "epoch": 0.7307907571759814,
      "grad_norm": 1.4819689534606348,
      "learning_rate": 2.7029332758412393e-06,
      "loss": 0.3797,
      "step": 13710
    },
    {
      "epoch": 0.7313237920098078,
      "grad_norm": 1.573359263649508,
      "learning_rate": 2.702406545819246e-06,
      "loss": 0.3893,
      "step": 13720
    },
    {
      "epoch": 0.7318568268436342,
      "grad_norm": 1.515514883608717,
      "learning_rate": 2.7018794006562055e-06,
      "loss": 0.3803,
      "step": 13730
    },
    {
      "epoch": 0.7323898616774607,
      "grad_norm": 1.4455124863133146,
      "learning_rate": 2.701351840534119e-06,
      "loss": 0.3879,
      "step": 13740
    },
    {
      "epoch": 0.7329228965112871,
      "grad_norm": 1.553992008773466,
      "learning_rate": 2.7008238656351312e-06,
      "loss": 0.3958,
      "step": 13750
    },
    {
      "epoch": 0.7334559313451134,
      "grad_norm": 1.514945479265164,
      "learning_rate": 2.70029547614153e-06,
      "loss": 0.3887,
      "step": 13760
    },
    {
      "epoch": 0.7339889661789398,
      "grad_norm": 2.0475782789851737,
      "learning_rate": 2.6997666722357456e-06,
      "loss": 0.3834,
      "step": 13770
    },
    {
      "epoch": 0.7345220010127662,
      "grad_norm": 1.5685919373167614,
      "learning_rate": 2.6992374541003523e-06,
      "loss": 0.3757,
      "step": 13780
    },
    {
      "epoch": 0.7350550358465926,
      "grad_norm": 1.5066461477919797,
      "learning_rate": 2.698707821918068e-06,
      "loss": 0.367,
      "step": 13790
    },
    {
      "epoch": 0.735588070680419,
      "grad_norm": 1.439312999748882,
      "learning_rate": 2.698177775871752e-06,
      "loss": 0.3838,
      "step": 13800
    },
    {
      "epoch": 0.7361211055142454,
      "grad_norm": 1.507905304836201,
      "learning_rate": 2.6976473161444076e-06,
      "loss": 0.382,
      "step": 13810
    },
    {
      "epoch": 0.7366541403480718,
      "grad_norm": 1.8021993262354592,
      "learning_rate": 2.69711644291918e-06,
      "loss": 0.3858,
      "step": 13820
    },
    {
      "epoch": 0.7371871751818981,
      "grad_norm": 1.5815430459684612,
      "learning_rate": 2.6965851563793578e-06,
      "loss": 0.3865,
      "step": 13830
    },
    {
      "epoch": 0.7377202100157245,
      "grad_norm": 1.6980970729787377,
      "learning_rate": 2.696053456708373e-06,
      "loss": 0.3851,
      "step": 13840
    },
    {
      "epoch": 0.7382532448495509,
      "grad_norm": 1.5879026673067482,
      "learning_rate": 2.6955213440897976e-06,
      "loss": 0.3777,
      "step": 13850
    },
    {
      "epoch": 0.7387862796833773,
      "grad_norm": 1.706498150529188,
      "learning_rate": 2.6949888187073495e-06,
      "loss": 0.3874,
      "step": 13860
    },
    {
      "epoch": 0.7393193145172037,
      "grad_norm": 1.6244462033848353,
      "learning_rate": 2.6944558807448874e-06,
      "loss": 0.3899,
      "step": 13870
    },
    {
      "epoch": 0.7398523493510301,
      "grad_norm": 1.4863210698554998,
      "learning_rate": 2.693922530386412e-06,
      "loss": 0.3784,
      "step": 13880
    },
    {
      "epoch": 0.7403853841848564,
      "grad_norm": 1.5687367500223122,
      "learning_rate": 2.6933887678160683e-06,
      "loss": 0.3818,
      "step": 13890
    },
    {
      "epoch": 0.7409184190186828,
      "grad_norm": 1.4901809941082058,
      "learning_rate": 2.6928545932181406e-06,
      "loss": 0.3816,
      "step": 13900
    },
    {
      "epoch": 0.7414514538525092,
      "grad_norm": 1.579608611906349,
      "learning_rate": 2.692320006777059e-06,
      "loss": 0.3907,
      "step": 13910
    },
    {
      "epoch": 0.7419844886863356,
      "grad_norm": 1.4820958708779177,
      "learning_rate": 2.6917850086773926e-06,
      "loss": 0.3697,
      "step": 13920
    },
    {
      "epoch": 0.7425175235201621,
      "grad_norm": 1.4872939118579305,
      "learning_rate": 2.691249599103855e-06,
      "loss": 0.3759,
      "step": 13930
    },
    {
      "epoch": 0.7430505583539885,
      "grad_norm": 1.4542868805457698,
      "learning_rate": 2.6907137782413e-06,
      "loss": 0.3825,
      "step": 13940
    },
    {
      "epoch": 0.7435835931878149,
      "grad_norm": 1.4380016676050995,
      "learning_rate": 2.6901775462747243e-06,
      "loss": 0.3937,
      "step": 13950
    },
    {
      "epoch": 0.7441166280216412,
      "grad_norm": 1.524173149205176,
      "learning_rate": 2.689640903389268e-06,
      "loss": 0.3707,
      "step": 13960
    },
    {
      "epoch": 0.7446496628554676,
      "grad_norm": 1.6651272277498275,
      "learning_rate": 2.68910384977021e-06,
      "loss": 0.3986,
      "step": 13970
    },
    {
      "epoch": 0.745182697689294,
      "grad_norm": 1.8018108869900669,
      "learning_rate": 2.688566385602973e-06,
      "loss": 0.3993,
      "step": 13980
    },
    {
      "epoch": 0.7457157325231204,
      "grad_norm": 1.4878680397164445,
      "learning_rate": 2.6880285110731215e-06,
      "loss": 0.3804,
      "step": 13990
    },
    {
      "epoch": 0.7462487673569468,
      "grad_norm": 1.4537101709160996,
      "learning_rate": 2.6874902263663612e-06,
      "loss": 0.3781,
      "step": 14000
    },
    {
      "epoch": 0.7467818021907732,
      "grad_norm": 1.5481186542559633,
      "learning_rate": 2.686951531668539e-06,
      "loss": 0.3767,
      "step": 14010
    },
    {
      "epoch": 0.7473148370245996,
      "grad_norm": 1.545481683502921,
      "learning_rate": 2.6864124271656437e-06,
      "loss": 0.3747,
      "step": 14020
    },
    {
      "epoch": 0.7478478718584259,
      "grad_norm": 1.6328173089179736,
      "learning_rate": 2.6858729130438067e-06,
      "loss": 0.3864,
      "step": 14030
    },
    {
      "epoch": 0.7483809066922523,
      "grad_norm": 1.558402507260832,
      "learning_rate": 2.6853329894892992e-06,
      "loss": 0.392,
      "step": 14040
    },
    {
      "epoch": 0.7489139415260787,
      "grad_norm": 1.705139255518345,
      "learning_rate": 2.6847926566885338e-06,
      "loss": 0.3778,
      "step": 14050
    },
    {
      "epoch": 0.7494469763599051,
      "grad_norm": 1.6721644047160122,
      "learning_rate": 2.684251914828066e-06,
      "loss": 0.3693,
      "step": 14060
    },
    {
      "epoch": 0.7499800111937315,
      "grad_norm": 1.5516489095124752,
      "learning_rate": 2.6837107640945906e-06,
      "loss": 0.3712,
      "step": 14070
    },
    {
      "epoch": 0.7505130460275579,
      "grad_norm": 1.5983957786606795,
      "learning_rate": 2.683169204674945e-06,
      "loss": 0.378,
      "step": 14080
    },
    {
      "epoch": 0.7510460808613842,
      "grad_norm": 1.565366324946499,
      "learning_rate": 2.682627236756107e-06,
      "loss": 0.3864,
      "step": 14090
    },
    {
      "epoch": 0.7515791156952106,
      "grad_norm": 1.5239895960312135,
      "learning_rate": 2.682084860525196e-06,
      "loss": 0.3703,
      "step": 14100
    },
    {
      "epoch": 0.7521121505290371,
      "grad_norm": 1.4760910387103512,
      "learning_rate": 2.6815420761694716e-06,
      "loss": 0.3633,
      "step": 14110
    },
    {
      "epoch": 0.7526451853628635,
      "grad_norm": 1.6231749377585034,
      "learning_rate": 2.680998883876335e-06,
      "loss": 0.3853,
      "step": 14120
    },
    {
      "epoch": 0.7531782201966899,
      "grad_norm": 1.431045760213454,
      "learning_rate": 2.680455283833327e-06,
      "loss": 0.3826,
      "step": 14130
    },
    {
      "epoch": 0.7537112550305163,
      "grad_norm": 1.5416390952620564,
      "learning_rate": 2.679911276228131e-06,
      "loss": 0.3688,
      "step": 14140
    },
    {
      "epoch": 0.7542442898643427,
      "grad_norm": 1.5148282528028743,
      "learning_rate": 2.679366861248571e-06,
      "loss": 0.3649,
      "step": 14150
    },
    {
      "epoch": 0.754777324698169,
      "grad_norm": 1.5817808971069427,
      "learning_rate": 2.678822039082609e-06,
      "loss": 0.3799,
      "step": 14160
    },
    {
      "epoch": 0.7553103595319954,
      "grad_norm": 1.4906424703900187,
      "learning_rate": 2.678276809918351e-06,
      "loss": 0.38,
      "step": 14170
    },
    {
      "epoch": 0.7558433943658218,
      "grad_norm": 1.610919235890603,
      "learning_rate": 2.677731173944041e-06,
      "loss": 0.3864,
      "step": 14180
    },
    {
      "epoch": 0.7563764291996482,
      "grad_norm": 1.7412703916621453,
      "learning_rate": 2.6771851313480652e-06,
      "loss": 0.3825,
      "step": 14190
    },
    {
      "epoch": 0.7569094640334746,
      "grad_norm": 1.4859213488164094,
      "learning_rate": 2.676638682318949e-06,
      "loss": 0.3681,
      "step": 14200
    },
    {
      "epoch": 0.757442498867301,
      "grad_norm": 1.4486163249900998,
      "learning_rate": 2.676091827045359e-06,
      "loss": 0.3802,
      "step": 14210
    },
    {
      "epoch": 0.7579755337011274,
      "grad_norm": 1.9040289677671705,
      "learning_rate": 2.6755445657161005e-06,
      "loss": 0.3855,
      "step": 14220
    },
    {
      "epoch": 0.7585085685349537,
      "grad_norm": 1.610102578431111,
      "learning_rate": 2.674996898520121e-06,
      "loss": 0.3743,
      "step": 14230
    },
    {
      "epoch": 0.7590416033687801,
      "grad_norm": 1.532273974811401,
      "learning_rate": 2.674448825646508e-06,
      "loss": 0.3845,
      "step": 14240
    },
    {
      "epoch": 0.7595746382026065,
      "grad_norm": 1.622277955554152,
      "learning_rate": 2.6739003472844866e-06,
      "loss": 0.3898,
      "step": 14250
    },
    {
      "epoch": 0.7601076730364329,
      "grad_norm": 1.677938381989656,
      "learning_rate": 2.6733514636234246e-06,
      "loss": 0.3821,
      "step": 14260
    },
    {
      "epoch": 0.7606407078702593,
      "grad_norm": 1.5323608443982975,
      "learning_rate": 2.6728021748528285e-06,
      "loss": 0.377,
      "step": 14270
    },
    {
      "epoch": 0.7611737427040857,
      "grad_norm": 1.6028970491668473,
      "learning_rate": 2.672252481162345e-06,
      "loss": 0.3774,
      "step": 14280
    },
    {
      "epoch": 0.761706777537912,
      "grad_norm": 1.7027384185760273,
      "learning_rate": 2.671702382741761e-06,
      "loss": 0.3878,
      "step": 14290
    },
    {
      "epoch": 0.7622398123717385,
      "grad_norm": 1.7698752611373998,
      "learning_rate": 2.6711518797810016e-06,
      "loss": 0.3818,
      "step": 14300
    },
    {
      "epoch": 0.7627728472055649,
      "grad_norm": 1.651857115872705,
      "learning_rate": 2.670600972470133e-06,
      "loss": 0.3747,
      "step": 14310
    },
    {
      "epoch": 0.7633058820393913,
      "grad_norm": 1.5117780004384307,
      "learning_rate": 2.670049660999361e-06,
      "loss": 0.3803,
      "step": 14320
    },
    {
      "epoch": 0.7638389168732177,
      "grad_norm": 1.6762370842755865,
      "learning_rate": 2.66949794555903e-06,
      "loss": 0.372,
      "step": 14330
    },
    {
      "epoch": 0.7643719517070441,
      "grad_norm": 1.5275310309304424,
      "learning_rate": 2.668945826339625e-06,
      "loss": 0.3788,
      "step": 14340
    },
    {
      "epoch": 0.7649049865408705,
      "grad_norm": 1.5285612240160436,
      "learning_rate": 2.6683933035317695e-06,
      "loss": 0.3753,
      "step": 14350
    },
    {
      "epoch": 0.7654380213746969,
      "grad_norm": 1.5206932444236243,
      "learning_rate": 2.667840377326227e-06,
      "loss": 0.3801,
      "step": 14360
    },
    {
      "epoch": 0.7659710562085232,
      "grad_norm": 1.5537614704481293,
      "learning_rate": 2.667287047913899e-06,
      "loss": 0.3835,
      "step": 14370
    },
    {
      "epoch": 0.7665040910423496,
      "grad_norm": 1.5682640700687065,
      "learning_rate": 2.6667333154858285e-06,
      "loss": 0.3779,
      "step": 14380
    },
    {
      "epoch": 0.767037125876176,
      "grad_norm": 1.705467179337518,
      "learning_rate": 2.6661791802331956e-06,
      "loss": 0.3856,
      "step": 14390
    },
    {
      "epoch": 0.7675701607100024,
      "grad_norm": 1.636218762545239,
      "learning_rate": 2.66562464234732e-06,
      "loss": 0.3656,
      "step": 14400
    },
    {
      "epoch": 0.7681031955438288,
      "grad_norm": 1.3767479995706844,
      "learning_rate": 2.665069702019661e-06,
      "loss": 0.3914,
      "step": 14410
    },
    {
      "epoch": 0.7686362303776552,
      "grad_norm": 1.8589785835593269,
      "learning_rate": 2.664514359441817e-06,
      "loss": 0.3753,
      "step": 14420
    },
    {
      "epoch": 0.7691692652114815,
      "grad_norm": 1.7060659719896527,
      "learning_rate": 2.663958614805523e-06,
      "loss": 0.3935,
      "step": 14430
    },
    {
      "epoch": 0.7697023000453079,
      "grad_norm": 1.591105907921803,
      "learning_rate": 2.663402468302656e-06,
      "loss": 0.3724,
      "step": 14440
    },
    {
      "epoch": 0.7702353348791343,
      "grad_norm": 1.6738017723436225,
      "learning_rate": 2.66284592012523e-06,
      "loss": 0.3955,
      "step": 14450
    },
    {
      "epoch": 0.7707683697129607,
      "grad_norm": 1.7085572652224879,
      "learning_rate": 2.6622889704653975e-06,
      "loss": 0.396,
      "step": 14460
    },
    {
      "epoch": 0.7713014045467871,
      "grad_norm": 1.6441537457697075,
      "learning_rate": 2.6617316195154508e-06,
      "loss": 0.367,
      "step": 14470
    },
    {
      "epoch": 0.7718344393806135,
      "grad_norm": 1.5215714048435565,
      "learning_rate": 2.6611738674678195e-06,
      "loss": 0.3861,
      "step": 14480
    },
    {
      "epoch": 0.77236747421444,
      "grad_norm": 1.5405281400606958,
      "learning_rate": 2.6606157145150727e-06,
      "loss": 0.3836,
      "step": 14490
    },
    {
      "epoch": 0.7729005090482663,
      "grad_norm": 1.6955724201748459,
      "learning_rate": 2.660057160849917e-06,
      "loss": 0.3712,
      "step": 14500
    },
    {
      "epoch": 0.7734335438820927,
      "grad_norm": 1.623092111117168,
      "learning_rate": 2.6594982066651976e-06,
      "loss": 0.3766,
      "step": 14510
    },
    {
      "epoch": 0.7739665787159191,
      "grad_norm": 1.6379731630258436,
      "learning_rate": 2.658938852153899e-06,
      "loss": 0.3895,
      "step": 14520
    },
    {
      "epoch": 0.7744996135497455,
      "grad_norm": 1.5002087878104542,
      "learning_rate": 2.6583790975091423e-06,
      "loss": 0.387,
      "step": 14530
    },
    {
      "epoch": 0.7750326483835719,
      "grad_norm": 1.530748386011778,
      "learning_rate": 2.6578189429241876e-06,
      "loss": 0.3861,
      "step": 14540
    },
    {
      "epoch": 0.7755656832173983,
      "grad_norm": 1.7659549894503823,
      "learning_rate": 2.6572583885924337e-06,
      "loss": 0.3842,
      "step": 14550
    },
    {
      "epoch": 0.7760987180512247,
      "grad_norm": 1.5865874524892567,
      "learning_rate": 2.656697434707416e-06,
      "loss": 0.3748,
      "step": 14560
    },
    {
      "epoch": 0.776631752885051,
      "grad_norm": 1.5595482234022222,
      "learning_rate": 2.6561360814628093e-06,
      "loss": 0.3839,
      "step": 14570
    },
    {
      "epoch": 0.7771647877188774,
      "grad_norm": 1.6152141636044386,
      "learning_rate": 2.6555743290524245e-06,
      "loss": 0.3926,
      "step": 14580
    },
    {
      "epoch": 0.7776978225527038,
      "grad_norm": 1.658589427232685,
      "learning_rate": 2.6550121776702123e-06,
      "loss": 0.3777,
      "step": 14590
    },
    {
      "epoch": 0.7782308573865302,
      "grad_norm": 1.5946282310880198,
      "learning_rate": 2.65444962751026e-06,
      "loss": 0.3703,
      "step": 14600
    },
    {
      "epoch": 0.7787638922203566,
      "grad_norm": 1.583960561867612,
      "learning_rate": 2.6538866787667925e-06,
      "loss": 0.3824,
      "step": 14610
    },
    {
      "epoch": 0.779296927054183,
      "grad_norm": 1.5401933646754378,
      "learning_rate": 2.653323331634173e-06,
      "loss": 0.3722,
      "step": 14620
    },
    {
      "epoch": 0.7798299618880093,
      "grad_norm": 1.4717065312762139,
      "learning_rate": 2.652759586306901e-06,
      "loss": 0.3761,
      "step": 14630
    },
    {
      "epoch": 0.7803629967218357,
      "grad_norm": 1.542125287119514,
      "learning_rate": 2.652195442979616e-06,
      "loss": 0.3702,
      "step": 14640
    },
    {
      "epoch": 0.7808960315556621,
      "grad_norm": 1.709435641369509,
      "learning_rate": 2.651630901847092e-06,
      "loss": 0.3807,
      "step": 14650
    },
    {
      "epoch": 0.7814290663894885,
      "grad_norm": 1.6756778931666965,
      "learning_rate": 2.651065963104242e-06,
      "loss": 0.3727,
      "step": 14660
    },
    {
      "epoch": 0.781962101223315,
      "grad_norm": 1.6617455201194067,
      "learning_rate": 2.650500626946116e-06,
      "loss": 0.3821,
      "step": 14670
    },
    {
      "epoch": 0.7824951360571414,
      "grad_norm": 1.5889826510780378,
      "learning_rate": 2.649934893567901e-06,
      "loss": 0.3906,
      "step": 14680
    },
    {
      "epoch": 0.7830281708909678,
      "grad_norm": 1.543896943124956,
      "learning_rate": 2.649368763164921e-06,
      "loss": 0.3774,
      "step": 14690
    },
    {
      "epoch": 0.7835612057247942,
      "grad_norm": 1.458772394809297,
      "learning_rate": 2.648802235932637e-06,
      "loss": 0.3733,
      "step": 14700
    },
    {
      "epoch": 0.7840942405586205,
      "grad_norm": 1.4064454865750395,
      "learning_rate": 2.6482353120666485e-06,
      "loss": 0.3814,
      "step": 14710
    },
    {
      "epoch": 0.7846272753924469,
      "grad_norm": 1.6576605975658734,
      "learning_rate": 2.6476679917626898e-06,
      "loss": 0.3785,
      "step": 14720
    },
    {
      "epoch": 0.7851603102262733,
      "grad_norm": 1.5398003798072037,
      "learning_rate": 2.647100275216633e-06,
      "loss": 0.3826,
      "step": 14730
    },
    {
      "epoch": 0.7856933450600997,
      "grad_norm": 1.6009980451790458,
      "learning_rate": 2.646532162624488e-06,
      "loss": 0.3839,
      "step": 14740
    },
    {
      "epoch": 0.7862263798939261,
      "grad_norm": 1.5974359764117594,
      "learning_rate": 2.6459636541823995e-06,
      "loss": 0.3773,
      "step": 14750
    },
    {
      "epoch": 0.7867594147277525,
      "grad_norm": 1.526211852989671,
      "learning_rate": 2.6453947500866505e-06,
      "loss": 0.3907,
      "step": 14760
    },
    {
      "epoch": 0.7872924495615788,
      "grad_norm": 1.7564690904552895,
      "learning_rate": 2.644825450533659e-06,
      "loss": 0.3828,
      "step": 14770
    },
    {
      "epoch": 0.7878254843954052,
      "grad_norm": 1.4530410472836608,
      "learning_rate": 2.6442557557199818e-06,
      "loss": 0.3698,
      "step": 14780
    },
    {
      "epoch": 0.7883585192292316,
      "grad_norm": 1.5889706646970847,
      "learning_rate": 2.6436856658423097e-06,
      "loss": 0.3698,
      "step": 14790
    },
    {
      "epoch": 0.788891554063058,
      "grad_norm": 1.499026000053821,
      "learning_rate": 2.6431151810974724e-06,
      "loss": 0.3811,
      "step": 14800
    },
    {
      "epoch": 0.7894245888968844,
      "grad_norm": 1.5459216488342085,
      "learning_rate": 2.6425443016824335e-06,
      "loss": 0.3805,
      "step": 14810
    },
    {
      "epoch": 0.7899576237307108,
      "grad_norm": 1.5909499923729846,
      "learning_rate": 2.641973027794294e-06,
      "loss": 0.3611,
      "step": 14820
    },
    {
      "epoch": 0.7904906585645372,
      "grad_norm": 1.6968052394583941,
      "learning_rate": 2.6414013596302914e-06,
      "loss": 0.372,
      "step": 14830
    },
    {
      "epoch": 0.7910236933983635,
      "grad_norm": 1.691090314111364,
      "learning_rate": 2.6408292973877984e-06,
      "loss": 0.3875,
      "step": 14840
    },
    {
      "epoch": 0.7915567282321899,
      "grad_norm": 1.6026699152846986,
      "learning_rate": 2.6402568412643255e-06,
      "loss": 0.3766,
      "step": 14850
    },
    {
      "epoch": 0.7920897630660164,
      "grad_norm": 1.5919582698850068,
      "learning_rate": 2.6396839914575168e-06,
      "loss": 0.3686,
      "step": 14860
    },
    {
      "epoch": 0.7926227978998428,
      "grad_norm": 1.6596544716727593,
      "learning_rate": 2.6391107481651544e-06,
      "loss": 0.3782,
      "step": 14870
    },
    {
      "epoch": 0.7931558327336692,
      "grad_norm": 1.456029829493708,
      "learning_rate": 2.6385371115851548e-06,
      "loss": 0.3537,
      "step": 14880
    },
    {
      "epoch": 0.7936888675674956,
      "grad_norm": 1.4015542912795187,
      "learning_rate": 2.6379630819155713e-06,
      "loss": 0.3608,
      "step": 14890
    },
    {
      "epoch": 0.794221902401322,
      "grad_norm": 1.548085500573455,
      "learning_rate": 2.6373886593545925e-06,
      "loss": 0.3649,
      "step": 14900
    },
    {
      "epoch": 0.7947549372351483,
      "grad_norm": 1.4242272018327753,
      "learning_rate": 2.6368138441005425e-06,
      "loss": 0.3908,
      "step": 14910
    },
    {
      "epoch": 0.7952879720689747,
      "grad_norm": 1.9033306068739955,
      "learning_rate": 2.6362386363518814e-06,
      "loss": 0.3807,
      "step": 14920
    },
    {
      "epoch": 0.7958210069028011,
      "grad_norm": 1.801444482296597,
      "learning_rate": 2.635663036307204e-06,
      "loss": 0.381,
      "step": 14930
    },
    {
      "epoch": 0.7963540417366275,
      "grad_norm": 1.5576521323179306,
      "learning_rate": 2.6350870441652413e-06,
      "loss": 0.3798,
      "step": 14940
    },
    {
      "epoch": 0.7968870765704539,
      "grad_norm": 1.571417644855115,
      "learning_rate": 2.63451066012486e-06,
      "loss": 0.3842,
      "step": 14950
    },
    {
      "epoch": 0.7974201114042803,
      "grad_norm": 1.531332735013395,
      "learning_rate": 2.6339338843850607e-06,
      "loss": 0.3797,
      "step": 14960
    },
    {
      "epoch": 0.7979531462381066,
      "grad_norm": 1.5441614344992205,
      "learning_rate": 2.6333567171449803e-06,
      "loss": 0.3641,
      "step": 14970
    },
    {
      "epoch": 0.798486181071933,
      "grad_norm": 1.407836934841589,
      "learning_rate": 2.6327791586038917e-06,
      "loss": 0.3758,
      "step": 14980
    },
    {
      "epoch": 0.7990192159057594,
      "grad_norm": 1.4204714634793922,
      "learning_rate": 2.632201208961201e-06,
      "loss": 0.3717,
      "step": 14990
    },
    {
      "epoch": 0.7995522507395858,
      "grad_norm": 1.6047150043523342,
      "learning_rate": 2.63162286841645e-06,
      "loss": 0.375,
      "step": 15000
    },
    {
      "epoch": 0.8000852855734122,
      "grad_norm": 1.6087766601697615,
      "learning_rate": 2.631044137169316e-06,
      "loss": 0.3733,
      "step": 15010
    },
    {
      "epoch": 0.8006183204072386,
      "grad_norm": 1.605954450636087,
      "learning_rate": 2.630465015419611e-06,
      "loss": 0.3779,
      "step": 15020
    },
    {
      "epoch": 0.801151355241065,
      "grad_norm": 1.6558969809918074,
      "learning_rate": 2.6298855033672816e-06,
      "loss": 0.386,
      "step": 15030
    },
    {
      "epoch": 0.8016843900748913,
      "grad_norm": 1.6257344700118077,
      "learning_rate": 2.629305601212409e-06,
      "loss": 0.3801,
      "step": 15040
    },
    {
      "epoch": 0.8022174249087178,
      "grad_norm": 1.5377736833006272,
      "learning_rate": 2.62872530915521e-06,
      "loss": 0.3842,
      "step": 15050
    },
    {
      "epoch": 0.8027504597425442,
      "grad_norm": 1.5732148435230946,
      "learning_rate": 2.628144627396034e-06,
      "loss": 0.379,
      "step": 15060
    },
    {
      "epoch": 0.8032834945763706,
      "grad_norm": 1.4165809386610702,
      "learning_rate": 2.6275635561353686e-06,
      "loss": 0.3624,
      "step": 15070
    },
    {
      "epoch": 0.803816529410197,
      "grad_norm": 1.54934296562986,
      "learning_rate": 2.626982095573831e-06,
      "loss": 0.3806,
      "step": 15080
    },
    {
      "epoch": 0.8043495642440234,
      "grad_norm": 1.6292097271628858,
      "learning_rate": 2.6264002459121773e-06,
      "loss": 0.3828,
      "step": 15090
    },
    {
      "epoch": 0.8048825990778498,
      "grad_norm": 1.6270654164154734,
      "learning_rate": 2.6258180073512943e-06,
      "loss": 0.3893,
      "step": 15100
    },
    {
      "epoch": 0.8054156339116761,
      "grad_norm": 1.760893696539679,
      "learning_rate": 2.6252353800922066e-06,
      "loss": 0.3744,
      "step": 15110
    },
    {
      "epoch": 0.8059486687455025,
      "grad_norm": 1.542118720290548,
      "learning_rate": 2.62465236433607e-06,
      "loss": 0.3707,
      "step": 15120
    },
    {
      "epoch": 0.8064817035793289,
      "grad_norm": 1.587410532519548,
      "learning_rate": 2.6240689602841762e-06,
      "loss": 0.3708,
      "step": 15130
    },
    {
      "epoch": 0.8070147384131553,
      "grad_norm": 1.6615269663652716,
      "learning_rate": 2.62348516813795e-06,
      "loss": 0.3714,
      "step": 15140
    },
    {
      "epoch": 0.8075477732469817,
      "grad_norm": 1.6162044168857292,
      "learning_rate": 2.622900988098951e-06,
      "loss": 0.3724,
      "step": 15150
    },
    {
      "epoch": 0.8080808080808081,
      "grad_norm": 1.6059293626523352,
      "learning_rate": 2.6223164203688715e-06,
      "loss": 0.3754,
      "step": 15160
    },
    {
      "epoch": 0.8086138429146345,
      "grad_norm": 1.6827122531337826,
      "learning_rate": 2.6217314651495398e-06,
      "loss": 0.3842,
      "step": 15170
    },
    {
      "epoch": 0.8091468777484608,
      "grad_norm": 1.629828723261033,
      "learning_rate": 2.6211461226429154e-06,
      "loss": 0.3763,
      "step": 15180
    },
    {
      "epoch": 0.8096799125822872,
      "grad_norm": 1.4945387461282473,
      "learning_rate": 2.620560393051093e-06,
      "loss": 0.3736,
      "step": 15190
    },
    {
      "epoch": 0.8102129474161136,
      "grad_norm": 1.5745774243136732,
      "learning_rate": 2.6199742765763015e-06,
      "loss": 0.3695,
      "step": 15200
    },
    {
      "epoch": 0.81074598224994,
      "grad_norm": 1.6168392082579486,
      "learning_rate": 2.6193877734209024e-06,
      "loss": 0.3665,
      "step": 15210
    },
    {
      "epoch": 0.8112790170837664,
      "grad_norm": 1.668775283846866,
      "learning_rate": 2.6188008837873897e-06,
      "loss": 0.3849,
      "step": 15220
    },
    {
      "epoch": 0.8118120519175929,
      "grad_norm": 1.5811239058275377,
      "learning_rate": 2.6182136078783928e-06,
      "loss": 0.375,
      "step": 15230
    },
    {
      "epoch": 0.8123450867514193,
      "grad_norm": 1.616824865935622,
      "learning_rate": 2.6176259458966743e-06,
      "loss": 0.3732,
      "step": 15240
    },
    {
      "epoch": 0.8128781215852456,
      "grad_norm": 1.6154223509146513,
      "learning_rate": 2.6170378980451284e-06,
      "loss": 0.3734,
      "step": 15250
    },
    {
      "epoch": 0.813411156419072,
      "grad_norm": 1.6879889522393021,
      "learning_rate": 2.6164494645267845e-06,
      "loss": 0.3686,
      "step": 15260
    },
    {
      "epoch": 0.8139441912528984,
      "grad_norm": 1.610260987642283,
      "learning_rate": 2.6158606455448036e-06,
      "loss": 0.3784,
      "step": 15270
    },
    {
      "epoch": 0.8144772260867248,
      "grad_norm": 1.5035690608274144,
      "learning_rate": 2.615271441302481e-06,
      "loss": 0.3822,
      "step": 15280
    },
    {
      "epoch": 0.8150102609205512,
      "grad_norm": 1.6677448426405344,
      "learning_rate": 2.6146818520032437e-06,
      "loss": 0.3742,
      "step": 15290
    },
    {
      "epoch": 0.8155432957543776,
      "grad_norm": 1.4544448963978922,
      "learning_rate": 2.6140918778506533e-06,
      "loss": 0.375,
      "step": 15300
    },
    {
      "epoch": 0.816076330588204,
      "grad_norm": 1.5192376991238918,
      "learning_rate": 2.6135015190484026e-06,
      "loss": 0.3804,
      "step": 15310
    },
    {
      "epoch": 0.8166093654220303,
      "grad_norm": 1.6237236655698981,
      "learning_rate": 2.612910775800318e-06,
      "loss": 0.3831,
      "step": 15320
    },
    {
      "epoch": 0.8171424002558567,
      "grad_norm": 1.5335464491926976,
      "learning_rate": 2.6123196483103597e-06,
      "loss": 0.384,
      "step": 15330
    },
    {
      "epoch": 0.8176754350896831,
      "grad_norm": 1.5343674002288097,
      "learning_rate": 2.611728136782618e-06,
      "loss": 0.3872,
      "step": 15340
    },
    {
      "epoch": 0.8182084699235095,
      "grad_norm": 1.499365496692385,
      "learning_rate": 2.6111362414213184e-06,
      "loss": 0.3658,
      "step": 15350
    },
    {
      "epoch": 0.8187415047573359,
      "grad_norm": 1.5453789470920598,
      "learning_rate": 2.6105439624308173e-06,
      "loss": 0.3729,
      "step": 15360
    },
    {
      "epoch": 0.8192745395911623,
      "grad_norm": 1.5945461006809831,
      "learning_rate": 2.609951300015604e-06,
      "loss": 0.3682,
      "step": 15370
    },
    {
      "epoch": 0.8198075744249886,
      "grad_norm": 1.6214924921885283,
      "learning_rate": 2.6093582543803e-06,
      "loss": 0.3852,
      "step": 15380
    },
    {
      "epoch": 0.820340609258815,
      "grad_norm": 1.653306574781278,
      "learning_rate": 2.60876482572966e-06,
      "loss": 0.3747,
      "step": 15390
    },
    {
      "epoch": 0.8208736440926414,
      "grad_norm": 1.652697903243615,
      "learning_rate": 2.6081710142685704e-06,
      "loss": 0.3844,
      "step": 15400
    },
    {
      "epoch": 0.8214066789264678,
      "grad_norm": 1.5806590223261436,
      "learning_rate": 2.6075768202020483e-06,
      "loss": 0.3762,
      "step": 15410
    },
    {
      "epoch": 0.8219397137602943,
      "grad_norm": 1.6276971132412037,
      "learning_rate": 2.6069822437352456e-06,
      "loss": 0.3821,
      "step": 15420
    },
    {
      "epoch": 0.8224727485941207,
      "grad_norm": 1.4094612102213824,
      "learning_rate": 2.6063872850734438e-06,
      "loss": 0.3671,
      "step": 15430
    },
    {
      "epoch": 0.8230057834279471,
      "grad_norm": 1.6332986254206006,
      "learning_rate": 2.605791944422058e-06,
      "loss": 0.3877,
      "step": 15440
    },
    {
      "epoch": 0.8235388182617734,
      "grad_norm": 1.6790547002385467,
      "learning_rate": 2.605196221986635e-06,
      "loss": 0.362,
      "step": 15450
    },
    {
      "epoch": 0.8240718530955998,
      "grad_norm": 1.5907722567882416,
      "learning_rate": 2.6046001179728524e-06,
      "loss": 0.3737,
      "step": 15460
    },
    {
      "epoch": 0.8246048879294262,
      "grad_norm": 1.6107971218364705,
      "learning_rate": 2.6040036325865208e-06,
      "loss": 0.3809,
      "step": 15470
    },
    {
      "epoch": 0.8251379227632526,
      "grad_norm": 1.532492314013442,
      "learning_rate": 2.6034067660335803e-06,
      "loss": 0.3794,
      "step": 15480
    },
    {
      "epoch": 0.825670957597079,
      "grad_norm": 1.6443022286253832,
      "learning_rate": 2.602809518520106e-06,
      "loss": 0.3765,
      "step": 15490
    },
    {
      "epoch": 0.8262039924309054,
      "grad_norm": 1.5919629885529851,
      "learning_rate": 2.602211890252302e-06,
      "loss": 0.3986,
      "step": 15500
    },
    {
      "epoch": 0.8267370272647317,
      "grad_norm": 1.5154786873485722,
      "learning_rate": 2.601613881436504e-06,
      "loss": 0.3773,
      "step": 15510
    },
    {
      "epoch": 0.8272700620985581,
      "grad_norm": 1.4776845993666656,
      "learning_rate": 2.6010154922791804e-06,
      "loss": 0.3697,
      "step": 15520
    },
    {
      "epoch": 0.8278030969323845,
      "grad_norm": 1.6374945759086548,
      "learning_rate": 2.6004167229869293e-06,
      "loss": 0.3716,
      "step": 15530
    },
    {
      "epoch": 0.8283361317662109,
      "grad_norm": 1.5549057219632585,
      "learning_rate": 2.5998175737664816e-06,
      "loss": 0.3715,
      "step": 15540
    },
    {
      "epoch": 0.8288691666000373,
      "grad_norm": 1.5094224164605325,
      "learning_rate": 2.5992180448246982e-06,
      "loss": 0.3733,
      "step": 15550
    },
    {
      "epoch": 0.8294022014338637,
      "grad_norm": 1.7336169609341017,
      "learning_rate": 2.5986181363685713e-06,
      "loss": 0.3728,
      "step": 15560
    },
    {
      "epoch": 0.8299352362676901,
      "grad_norm": 1.8428380304146343,
      "learning_rate": 2.5980178486052257e-06,
      "loss": 0.3815,
      "step": 15570
    },
    {
      "epoch": 0.8304682711015164,
      "grad_norm": 1.6699268292173932,
      "learning_rate": 2.5974171817419143e-06,
      "loss": 0.3856,
      "step": 15580
    },
    {
      "epoch": 0.8310013059353428,
      "grad_norm": 1.7862875544807921,
      "learning_rate": 2.596816135986023e-06,
      "loss": 0.3725,
      "step": 15590
    },
    {
      "epoch": 0.8315343407691692,
      "grad_norm": 1.7560587533461487,
      "learning_rate": 2.596214711545068e-06,
      "loss": 0.394,
      "step": 15600
    },
    {
      "epoch": 0.8320673756029957,
      "grad_norm": 1.5871906646477703,
      "learning_rate": 2.5956129086266964e-06,
      "loss": 0.3744,
      "step": 15610
    },
    {
      "epoch": 0.8326004104368221,
      "grad_norm": 1.6228239940567555,
      "learning_rate": 2.5950107274386853e-06,
      "loss": 0.379,
      "step": 15620
    },
    {
      "epoch": 0.8331334452706485,
      "grad_norm": 1.4800779794282513,
      "learning_rate": 2.5944081681889435e-06,
      "loss": 0.3758,
      "step": 15630
    },
    {
      "epoch": 0.8336664801044749,
      "grad_norm": 1.529981787566232,
      "learning_rate": 2.5938052310855085e-06,
      "loss": 0.3964,
      "step": 15640
    },
    {
      "epoch": 0.8341995149383012,
      "grad_norm": 1.5146217835555718,
      "learning_rate": 2.5932019163365505e-06,
      "loss": 0.3652,
      "step": 15650
    },
    {
      "epoch": 0.8347325497721276,
      "grad_norm": 1.6121703106601808,
      "learning_rate": 2.592598224150369e-06,
      "loss": 0.3684,
      "step": 15660
    },
    {
      "epoch": 0.835265584605954,
      "grad_norm": 1.5833644705515921,
      "learning_rate": 2.5919941547353933e-06,
      "loss": 0.3815,
      "step": 15670
    },
    {
      "epoch": 0.8357986194397804,
      "grad_norm": 1.6844655081422828,
      "learning_rate": 2.591389708300184e-06,
      "loss": 0.3811,
      "step": 15680
    },
    {
      "epoch": 0.8363316542736068,
      "grad_norm": 1.5220854699172737,
      "learning_rate": 2.5907848850534308e-06,
      "loss": 0.3757,
      "step": 15690
    },
    {
      "epoch": 0.8368646891074332,
      "grad_norm": 1.386063400713167,
      "learning_rate": 2.5901796852039546e-06,
      "loss": 0.3778,
      "step": 15700
    },
    {
      "epoch": 0.8373977239412596,
      "grad_norm": 1.4196135201780693,
      "learning_rate": 2.5895741089607054e-06,
      "loss": 0.3667,
      "step": 15710
    },
    {
      "epoch": 0.8379307587750859,
      "grad_norm": 1.523409425044886,
      "learning_rate": 2.588968156532764e-06,
      "loss": 0.3682,
      "step": 15720
    },
    {
      "epoch": 0.8384637936089123,
      "grad_norm": 1.5576176343821933,
      "learning_rate": 2.58836182812934e-06,
      "loss": 0.3819,
      "step": 15730
    },
    {
      "epoch": 0.8389968284427387,
      "grad_norm": 1.5725986020128702,
      "learning_rate": 2.587755123959774e-06,
      "loss": 0.3791,
      "step": 15740
    },
    {
      "epoch": 0.8395298632765651,
      "grad_norm": 1.3689228427535598,
      "learning_rate": 2.587148044233535e-06,
      "loss": 0.3756,
      "step": 15750
    },
    {
      "epoch": 0.8400628981103915,
      "grad_norm": 1.778739762848662,
      "learning_rate": 2.5865405891602234e-06,
      "loss": 0.3838,
      "step": 15760
    },
    {
      "epoch": 0.8405959329442179,
      "grad_norm": 1.549534558837446,
      "learning_rate": 2.585932758949567e-06,
      "loss": 0.3647,
      "step": 15770
    },
    {
      "epoch": 0.8411289677780442,
      "grad_norm": 1.6465677234811233,
      "learning_rate": 2.585324553811425e-06,
      "loss": 0.3542,
      "step": 15780
    },
    {
      "epoch": 0.8416620026118707,
      "grad_norm": 1.5172515613551347,
      "learning_rate": 2.584715973955786e-06,
      "loss": 0.3785,
      "step": 15790
    },
    {
      "epoch": 0.8421950374456971,
      "grad_norm": 1.521408041097268,
      "learning_rate": 2.5841070195927665e-06,
      "loss": 0.3863,
      "step": 15800
    },
    {
      "epoch": 0.8427280722795235,
      "grad_norm": 1.6940708584064232,
      "learning_rate": 2.583497690932613e-06,
      "loss": 0.3751,
      "step": 15810
    },
    {
      "epoch": 0.8432611071133499,
      "grad_norm": 1.5649364796648757,
      "learning_rate": 2.582887988185702e-06,
      "loss": 0.3761,
      "step": 15820
    },
    {
      "epoch": 0.8437941419471763,
      "grad_norm": 1.5321600487979177,
      "learning_rate": 2.582277911562538e-06,
      "loss": 0.3641,
      "step": 15830
    },
    {
      "epoch": 0.8443271767810027,
      "grad_norm": 1.5282635965656433,
      "learning_rate": 2.5816674612737566e-06,
      "loss": 0.3695,
      "step": 15840
    },
    {
      "epoch": 0.844860211614829,
      "grad_norm": 1.6004399333232189,
      "learning_rate": 2.581056637530119e-06,
      "loss": 0.3692,
      "step": 15850
    },
    {
      "epoch": 0.8453932464486554,
      "grad_norm": 1.5888084113911927,
      "learning_rate": 2.580445440542518e-06,
      "loss": 0.3832,
      "step": 15860
    },
    {
      "epoch": 0.8459262812824818,
      "grad_norm": 1.5962224594629717,
      "learning_rate": 2.579833870521974e-06,
      "loss": 0.3787,
      "step": 15870
    },
    {
      "epoch": 0.8464593161163082,
      "grad_norm": 1.3847873658756509,
      "learning_rate": 2.5792219276796383e-06,
      "loss": 0.3735,
      "step": 15880
    },
    {
      "epoch": 0.8469923509501346,
      "grad_norm": 1.6443905605117313,
      "learning_rate": 2.578609612226788e-06,
      "loss": 0.3722,
      "step": 15890
    },
    {
      "epoch": 0.847525385783961,
      "grad_norm": 1.346841177561081,
      "learning_rate": 2.5779969243748307e-06,
      "loss": 0.3733,
      "step": 15900
    },
    {
      "epoch": 0.8480584206177874,
      "grad_norm": 1.6341427788319907,
      "learning_rate": 2.577383864335302e-06,
      "loss": 0.3771,
      "step": 15910
    },
    {
      "epoch": 0.8485914554516137,
      "grad_norm": 1.6817339865397696,
      "learning_rate": 2.576770432319865e-06,
      "loss": 0.3773,
      "step": 15920
    },
    {
      "epoch": 0.8491244902854401,
      "grad_norm": 1.5871104413822126,
      "learning_rate": 2.5761566285403145e-06,
      "loss": 0.3793,
      "step": 15930
    },
    {
      "epoch": 0.8496575251192665,
      "grad_norm": 1.6519170350433356,
      "learning_rate": 2.5755424532085695e-06,
      "loss": 0.3677,
      "step": 15940
    },
    {
      "epoch": 0.8501905599530929,
      "grad_norm": 1.7186682458139666,
      "learning_rate": 2.5749279065366804e-06,
      "loss": 0.3812,
      "step": 15950
    },
    {
      "epoch": 0.8507235947869193,
      "grad_norm": 1.5822487152956444,
      "learning_rate": 2.574312988736824e-06,
      "loss": 0.378,
      "step": 15960
    },
    {
      "epoch": 0.8512566296207457,
      "grad_norm": 1.6510525798610263,
      "learning_rate": 2.5736977000213064e-06,
      "loss": 0.3837,
      "step": 15970
    },
    {
      "epoch": 0.8517896644545722,
      "grad_norm": 1.5310094266109477,
      "learning_rate": 2.573082040602561e-06,
      "loss": 0.3698,
      "step": 15980
    },
    {
      "epoch": 0.8523226992883985,
      "grad_norm": 1.4529211881849582,
      "learning_rate": 2.572466010693149e-06,
      "loss": 0.3749,
      "step": 15990
    },
    {
      "epoch": 0.8528557341222249,
      "grad_norm": 1.5101692805802407,
      "learning_rate": 2.5718496105057607e-06,
      "loss": 0.364,
      "step": 16000
    },
    {
      "epoch": 0.8533887689560513,
      "grad_norm": 1.4688126022706927,
      "learning_rate": 2.5712328402532126e-06,
      "loss": 0.3667,
      "step": 16010
    },
    {
      "epoch": 0.8539218037898777,
      "grad_norm": 1.4942247042528989,
      "learning_rate": 2.570615700148451e-06,
      "loss": 0.3691,
      "step": 16020
    },
    {
      "epoch": 0.8544548386237041,
      "grad_norm": 1.6603712157923245,
      "learning_rate": 2.569998190404548e-06,
      "loss": 0.375,
      "step": 16030
    },
    {
      "epoch": 0.8549878734575305,
      "grad_norm": 1.5588096332047554,
      "learning_rate": 2.569380311234704e-06,
      "loss": 0.381,
      "step": 16040
    },
    {
      "epoch": 0.8555209082913569,
      "grad_norm": 1.4659923184327874,
      "learning_rate": 2.568762062852248e-06,
      "loss": 0.3778,
      "step": 16050
    },
    {
      "epoch": 0.8560539431251832,
      "grad_norm": 1.3979362089825116,
      "learning_rate": 2.5681434454706335e-06,
      "loss": 0.3755,
      "step": 16060
    },
    {
      "epoch": 0.8565869779590096,
      "grad_norm": 1.752895491513466,
      "learning_rate": 2.567524459303445e-06,
      "loss": 0.3615,
      "step": 16070
    },
    {
      "epoch": 0.857120012792836,
      "grad_norm": 1.7547747372537255,
      "learning_rate": 2.566905104564393e-06,
      "loss": 0.3838,
      "step": 16080
    },
    {
      "epoch": 0.8576530476266624,
      "grad_norm": 1.6129268055446493,
      "learning_rate": 2.5662853814673137e-06,
      "loss": 0.3667,
      "step": 16090
    },
    {
      "epoch": 0.8581860824604888,
      "grad_norm": 1.6319171638649352,
      "learning_rate": 2.565665290226172e-06,
      "loss": 0.376,
      "step": 16100
    },
    {
      "epoch": 0.8587191172943152,
      "grad_norm": 1.5177051176078427,
      "learning_rate": 2.5650448310550606e-06,
      "loss": 0.3705,
      "step": 16110
    },
    {
      "epoch": 0.8592521521281415,
      "grad_norm": 1.73445074077126,
      "learning_rate": 2.5644240041681977e-06,
      "loss": 0.3647,
      "step": 16120
    },
    {
      "epoch": 0.8597851869619679,
      "grad_norm": 1.6452702383339917,
      "learning_rate": 2.5638028097799287e-06,
      "loss": 0.3729,
      "step": 16130
    },
    {
      "epoch": 0.8603182217957943,
      "grad_norm": 1.7696969414639545,
      "learning_rate": 2.5631812481047268e-06,
      "loss": 0.3661,
      "step": 16140
    },
    {
      "epoch": 0.8608512566296207,
      "grad_norm": 1.546585335395,
      "learning_rate": 2.562559319357191e-06,
      "loss": 0.3634,
      "step": 16150
    },
    {
      "epoch": 0.8613842914634471,
      "grad_norm": 1.491649513938422,
      "learning_rate": 2.5619370237520477e-06,
      "loss": 0.3719,
      "step": 16160
    },
    {
      "epoch": 0.8619173262972736,
      "grad_norm": 1.5068967664791548,
      "learning_rate": 2.5613143615041496e-06,
      "loss": 0.3724,
      "step": 16170
    },
    {
      "epoch": 0.8624503611311,
      "grad_norm": 1.5853384603878022,
      "learning_rate": 2.5606913328284764e-06,
      "loss": 0.3708,
      "step": 16180
    },
    {
      "epoch": 0.8629833959649263,
      "grad_norm": 1.5099798844478494,
      "learning_rate": 2.5600679379401336e-06,
      "loss": 0.3783,
      "step": 16190
    },
    {
      "epoch": 0.8635164307987527,
      "grad_norm": 1.7102450189589395,
      "learning_rate": 2.559444177054354e-06,
      "loss": 0.3876,
      "step": 16200
    },
    {
      "epoch": 0.8640494656325791,
      "grad_norm": 1.6677562151293002,
      "learning_rate": 2.5588200503864957e-06,
      "loss": 0.3804,
      "step": 16210
    },
    {
      "epoch": 0.8645825004664055,
      "grad_norm": 1.626738274622516,
      "learning_rate": 2.5581955581520446e-06,
      "loss": 0.3792,
      "step": 16220
    },
    {
      "epoch": 0.8651155353002319,
      "grad_norm": 1.5608963153656066,
      "learning_rate": 2.5575707005666116e-06,
      "loss": 0.3788,
      "step": 16230
    },
    {
      "epoch": 0.8656485701340583,
      "grad_norm": 1.4639992337473382,
      "learning_rate": 2.5569454778459336e-06,
      "loss": 0.3665,
      "step": 16240
    },
    {
      "epoch": 0.8661816049678847,
      "grad_norm": 1.5077559027975775,
      "learning_rate": 2.5563198902058747e-06,
      "loss": 0.3734,
      "step": 16250
    },
    {
      "epoch": 0.866714639801711,
      "grad_norm": 1.6879167166085143,
      "learning_rate": 2.555693937862424e-06,
      "loss": 0.3801,
      "step": 16260
    },
    {
      "epoch": 0.8672476746355374,
      "grad_norm": 1.638295981773267,
      "learning_rate": 2.5550676210316965e-06,
      "loss": 0.3642,
      "step": 16270
    },
    {
      "epoch": 0.8677807094693638,
      "grad_norm": 1.7287336334949517,
      "learning_rate": 2.554440939929934e-06,
      "loss": 0.3895,
      "step": 16280
    },
    {
      "epoch": 0.8683137443031902,
      "grad_norm": 1.603657332634575,
      "learning_rate": 2.5538138947735035e-06,
      "loss": 0.3847,
      "step": 16290
    },
    {
      "epoch": 0.8688467791370166,
      "grad_norm": 1.6615849368066204,
      "learning_rate": 2.553186485778897e-06,
      "loss": 0.3745,
      "step": 16300
    },
    {
      "epoch": 0.869379813970843,
      "grad_norm": 1.598232462242146,
      "learning_rate": 2.5525587131627336e-06,
      "loss": 0.3749,
      "step": 16310
    },
    {
      "epoch": 0.8699128488046693,
      "grad_norm": 1.638500175883556,
      "learning_rate": 2.5519305771417564e-06,
      "loss": 0.3524,
      "step": 16320
    },
    {
      "epoch": 0.8704458836384957,
      "grad_norm": 1.5447728314436266,
      "learning_rate": 2.551302077932835e-06,
      "loss": 0.3778,
      "step": 16330
    },
    {
      "epoch": 0.8709789184723221,
      "grad_norm": 1.6940851981408045,
      "learning_rate": 2.550673215752964e-06,
      "loss": 0.3586,
      "step": 16340
    },
    {
      "epoch": 0.8715119533061485,
      "grad_norm": 1.6388450343210867,
      "learning_rate": 2.5500439908192637e-06,
      "loss": 0.374,
      "step": 16350
    },
    {
      "epoch": 0.872044988139975,
      "grad_norm": 1.6200539154563658,
      "learning_rate": 2.5494144033489784e-06,
      "loss": 0.3784,
      "step": 16360
    },
    {
      "epoch": 0.8725780229738014,
      "grad_norm": 1.3792852368666522,
      "learning_rate": 2.54878445355948e-06,
      "loss": 0.3721,
      "step": 16370
    },
    {
      "epoch": 0.8731110578076278,
      "grad_norm": 1.6614456476798383,
      "learning_rate": 2.5481541416682624e-06,
      "loss": 0.3716,
      "step": 16380
    },
    {
      "epoch": 0.8736440926414542,
      "grad_norm": 1.871795895368198,
      "learning_rate": 2.5475234678929475e-06,
      "loss": 0.3746,
      "step": 16390
    },
    {
      "epoch": 0.8741771274752805,
      "grad_norm": 1.5939911537858293,
      "learning_rate": 2.54689243245128e-06,
      "loss": 0.3719,
      "step": 16400
    },
    {
      "epoch": 0.8747101623091069,
      "grad_norm": 1.5813498650047086,
      "learning_rate": 2.54626103556113e-06,
      "loss": 0.4017,
      "step": 16410
    },
    {
      "epoch": 0.8752431971429333,
      "grad_norm": 1.6391277452202946,
      "learning_rate": 2.5456292774404936e-06,
      "loss": 0.3822,
      "step": 16420
    },
    {
      "epoch": 0.8757762319767597,
      "grad_norm": 1.585544556941181,
      "learning_rate": 2.5449971583074897e-06,
      "loss": 0.3682,
      "step": 16430
    },
    {
      "epoch": 0.8763092668105861,
      "grad_norm": 1.3774672461817439,
      "learning_rate": 2.5443646783803637e-06,
      "loss": 0.3701,
      "step": 16440
    },
    {
      "epoch": 0.8768423016444125,
      "grad_norm": 1.5370211442878157,
      "learning_rate": 2.5437318378774834e-06,
      "loss": 0.3633,
      "step": 16450
    },
    {
      "epoch": 0.8773753364782388,
      "grad_norm": 1.4835213341033064,
      "learning_rate": 2.5430986370173433e-06,
      "loss": 0.3704,
      "step": 16460
    },
    {
      "epoch": 0.8779083713120652,
      "grad_norm": 1.4503348420392632,
      "learning_rate": 2.542465076018561e-06,
      "loss": 0.362,
      "step": 16470
    },
    {
      "epoch": 0.8784414061458916,
      "grad_norm": 1.7153479897425834,
      "learning_rate": 2.541831155099879e-06,
      "loss": 0.3718,
      "step": 16480
    },
    {
      "epoch": 0.878974440979718,
      "grad_norm": 1.6442307406855792,
      "learning_rate": 2.5411968744801646e-06,
      "loss": 0.3726,
      "step": 16490
    },
    {
      "epoch": 0.8795074758135444,
      "grad_norm": 1.6064051120357314,
      "learning_rate": 2.5405622343784068e-06,
      "loss": 0.375,
      "step": 16500
    },
    {
      "epoch": 0.8800405106473708,
      "grad_norm": 1.8617132070149183,
      "learning_rate": 2.5399272350137213e-06,
      "loss": 0.3705,
      "step": 16510
    },
    {
      "epoch": 0.8805735454811972,
      "grad_norm": 1.6586012144764883,
      "learning_rate": 2.5392918766053475e-06,
      "loss": 0.3768,
      "step": 16520
    },
    {
      "epoch": 0.8811065803150235,
      "grad_norm": 1.7688549617684624,
      "learning_rate": 2.5386561593726477e-06,
      "loss": 0.3914,
      "step": 16530
    },
    {
      "epoch": 0.88163961514885,
      "grad_norm": 1.5140260813796467,
      "learning_rate": 2.5380200835351095e-06,
      "loss": 0.3727,
      "step": 16540
    },
    {
      "epoch": 0.8821726499826764,
      "grad_norm": 1.5847791420617885,
      "learning_rate": 2.537383649312342e-06,
      "loss": 0.3656,
      "step": 16550
    },
    {
      "epoch": 0.8827056848165028,
      "grad_norm": 1.5786263398800064,
      "learning_rate": 2.5367468569240807e-06,
      "loss": 0.3732,
      "step": 16560
    },
    {
      "epoch": 0.8832387196503292,
      "grad_norm": 1.5010672497747122,
      "learning_rate": 2.5361097065901833e-06,
      "loss": 0.3537,
      "step": 16570
    },
    {
      "epoch": 0.8837717544841556,
      "grad_norm": 1.6741605612831219,
      "learning_rate": 2.5354721985306313e-06,
      "loss": 0.3653,
      "step": 16580
    },
    {
      "epoch": 0.884304789317982,
      "grad_norm": 1.3879840429853558,
      "learning_rate": 2.5348343329655298e-06,
      "loss": 0.3742,
      "step": 16590
    },
    {
      "epoch": 0.8848378241518083,
      "grad_norm": 1.5094327415354354,
      "learning_rate": 2.534196110115107e-06,
      "loss": 0.378,
      "step": 16600
    },
    {
      "epoch": 0.8853708589856347,
      "grad_norm": 1.430564918210685,
      "learning_rate": 2.5335575301997148e-06,
      "loss": 0.3638,
      "step": 16610
    },
    {
      "epoch": 0.8859038938194611,
      "grad_norm": 1.7528605349800008,
      "learning_rate": 2.5329185934398286e-06,
      "loss": 0.3752,
      "step": 16620
    },
    {
      "epoch": 0.8864369286532875,
      "grad_norm": 1.7256482527304138,
      "learning_rate": 2.5322793000560467e-06,
      "loss": 0.3767,
      "step": 16630
    },
    {
      "epoch": 0.8869699634871139,
      "grad_norm": 1.6233691311033993,
      "learning_rate": 2.5316396502690906e-06,
      "loss": 0.3705,
      "step": 16640
    },
    {
      "epoch": 0.8875029983209403,
      "grad_norm": 1.5534173200233747,
      "learning_rate": 2.5309996442998043e-06,
      "loss": 0.3518,
      "step": 16650
    },
    {
      "epoch": 0.8880360331547666,
      "grad_norm": 1.474485031730477,
      "learning_rate": 2.530359282369156e-06,
      "loss": 0.3863,
      "step": 16660
    },
    {
      "epoch": 0.888569067988593,
      "grad_norm": 1.5660077731666442,
      "learning_rate": 2.529718564698235e-06,
      "loss": 0.3682,
      "step": 16670
    },
    {
      "epoch": 0.8891021028224194,
      "grad_norm": 1.5987562218011735,
      "learning_rate": 2.529077491508256e-06,
      "loss": 0.3708,
      "step": 16680
    },
    {
      "epoch": 0.8896351376562458,
      "grad_norm": 1.464437429640059,
      "learning_rate": 2.528436063020554e-06,
      "loss": 0.3753,
      "step": 16690
    },
    {
      "epoch": 0.8901681724900722,
      "grad_norm": 1.7208027187125727,
      "learning_rate": 2.527794279456588e-06,
      "loss": 0.3753,
      "step": 16700
    },
    {
      "epoch": 0.8907012073238986,
      "grad_norm": 1.7898384322917824,
      "learning_rate": 2.5271521410379387e-06,
      "loss": 0.3795,
      "step": 16710
    },
    {
      "epoch": 0.891234242157725,
      "grad_norm": 1.5138370911923953,
      "learning_rate": 2.5265096479863104e-06,
      "loss": 0.363,
      "step": 16720
    },
    {
      "epoch": 0.8917672769915514,
      "grad_norm": 1.6610679993188082,
      "learning_rate": 2.525866800523529e-06,
      "loss": 0.3714,
      "step": 16730
    },
    {
      "epoch": 0.8923003118253778,
      "grad_norm": 1.6235260763561166,
      "learning_rate": 2.5252235988715434e-06,
      "loss": 0.3621,
      "step": 16740
    },
    {
      "epoch": 0.8928333466592042,
      "grad_norm": 1.5866802026765918,
      "learning_rate": 2.5245800432524235e-06,
      "loss": 0.3564,
      "step": 16750
    },
    {
      "epoch": 0.8933663814930306,
      "grad_norm": 1.522207694721726,
      "learning_rate": 2.523936133888363e-06,
      "loss": 0.3688,
      "step": 16760
    },
    {
      "epoch": 0.893899416326857,
      "grad_norm": 1.505365445553488,
      "learning_rate": 2.5232918710016766e-06,
      "loss": 0.3772,
      "step": 16770
    },
    {
      "epoch": 0.8944324511606834,
      "grad_norm": 1.4937826924163504,
      "learning_rate": 2.5226472548148024e-06,
      "loss": 0.367,
      "step": 16780
    },
    {
      "epoch": 0.8949654859945098,
      "grad_norm": 1.6835416780129235,
      "learning_rate": 2.5220022855502994e-06,
      "loss": 0.3756,
      "step": 16790
    },
    {
      "epoch": 0.8954985208283361,
      "grad_norm": 1.5492489149071564,
      "learning_rate": 2.521356963430848e-06,
      "loss": 0.3838,
      "step": 16800
    },
    {
      "epoch": 0.8960315556621625,
      "grad_norm": 1.470886389362583,
      "learning_rate": 2.5207112886792515e-06,
      "loss": 0.3661,
      "step": 16810
    },
    {
      "epoch": 0.8965645904959889,
      "grad_norm": 1.7466458774817337,
      "learning_rate": 2.520065261518435e-06,
      "loss": 0.367,
      "step": 16820
    },
    {
      "epoch": 0.8970976253298153,
      "grad_norm": 1.5430961351416572,
      "learning_rate": 2.519418882171444e-06,
      "loss": 0.3775,
      "step": 16830
    },
    {
      "epoch": 0.8976306601636417,
      "grad_norm": 1.4481540710904404,
      "learning_rate": 2.5187721508614472e-06,
      "loss": 0.3692,
      "step": 16840
    },
    {
      "epoch": 0.8981636949974681,
      "grad_norm": 1.4749951731980901,
      "learning_rate": 2.5181250678117338e-06,
      "loss": 0.3602,
      "step": 16850
    },
    {
      "epoch": 0.8986967298312944,
      "grad_norm": 1.3894301359762353,
      "learning_rate": 2.517477633245714e-06,
      "loss": 0.3797,
      "step": 16860
    },
    {
      "epoch": 0.8992297646651208,
      "grad_norm": 1.5390883022574198,
      "learning_rate": 2.5168298473869214e-06,
      "loss": 0.3724,
      "step": 16870
    },
    {
      "epoch": 0.8997627994989472,
      "grad_norm": 1.5873898724450495,
      "learning_rate": 2.516181710459009e-06,
      "loss": 0.3888,
      "step": 16880
    },
    {
      "epoch": 0.9002958343327736,
      "grad_norm": 1.462613791560266,
      "learning_rate": 2.5155332226857514e-06,
      "loss": 0.3698,
      "step": 16890
    },
    {
      "epoch": 0.9008288691666,
      "grad_norm": 1.6735106794062748,
      "learning_rate": 2.5148843842910443e-06,
      "loss": 0.3777,
      "step": 16900
    },
    {
      "epoch": 0.9013619040004264,
      "grad_norm": 1.4762980101025525,
      "learning_rate": 2.514235195498905e-06,
      "loss": 0.3745,
      "step": 16910
    },
    {
      "epoch": 0.9018949388342529,
      "grad_norm": 1.698960198130694,
      "learning_rate": 2.513585656533471e-06,
      "loss": 0.3749,
      "step": 16920
    },
    {
      "epoch": 0.9024279736680793,
      "grad_norm": 1.7180561980762914,
      "learning_rate": 2.512935767619002e-06,
      "loss": 0.3664,
      "step": 16930
    },
    {
      "epoch": 0.9029610085019056,
      "grad_norm": 1.4304462538158722,
      "learning_rate": 2.512285528979877e-06,
      "loss": 0.3585,
      "step": 16940
    },
    {
      "epoch": 0.903494043335732,
      "grad_norm": 1.9039798130211056,
      "learning_rate": 2.5116349408405963e-06,
      "loss": 0.3665,
      "step": 16950
    },
    {
      "epoch": 0.9040270781695584,
      "grad_norm": 1.487438368131928,
      "learning_rate": 2.5109840034257813e-06,
      "loss": 0.3704,
      "step": 16960
    },
    {
      "epoch": 0.9045601130033848,
      "grad_norm": 1.610107122753202,
      "learning_rate": 2.5103327169601737e-06,
      "loss": 0.3659,
      "step": 16970
    },
    {
      "epoch": 0.9050931478372112,
      "grad_norm": 1.6894004250974037,
      "learning_rate": 2.5096810816686353e-06,
      "loss": 0.3778,
      "step": 16980
    },
    {
      "epoch": 0.9056261826710376,
      "grad_norm": 1.6339970168391242,
      "learning_rate": 2.509029097776149e-06,
      "loss": 0.3689,
      "step": 16990
    },
    {
      "epoch": 0.906159217504864,
      "grad_norm": 1.6852246573107863,
      "learning_rate": 2.508376765507818e-06,
      "loss": 0.3682,
      "step": 17000
    },
    {
      "epoch": 0.9066922523386903,
      "grad_norm": 1.4630275327257258,
      "learning_rate": 2.507724085088865e-06,
      "loss": 0.3787,
      "step": 17010
    },
    {
      "epoch": 0.9072252871725167,
      "grad_norm": 1.6517431522186092,
      "learning_rate": 2.507071056744633e-06,
      "loss": 0.3676,
      "step": 17020
    },
    {
      "epoch": 0.9077583220063431,
      "grad_norm": 1.701800878945533,
      "learning_rate": 2.5064176807005873e-06,
      "loss": 0.3632,
      "step": 17030
    },
    {
      "epoch": 0.9082913568401695,
      "grad_norm": 1.692234126968275,
      "learning_rate": 2.50576395718231e-06,
      "loss": 0.3512,
      "step": 17040
    },
    {
      "epoch": 0.9088243916739959,
      "grad_norm": 1.5124708655596437,
      "learning_rate": 2.5051098864155057e-06,
      "loss": 0.3612,
      "step": 17050
    },
    {
      "epoch": 0.9093574265078223,
      "grad_norm": 1.6165180533973313,
      "learning_rate": 2.504455468625997e-06,
      "loss": 0.3739,
      "step": 17060
    },
    {
      "epoch": 0.9098904613416486,
      "grad_norm": 1.6833039741171256,
      "learning_rate": 2.5038007040397268e-06,
      "loss": 0.3695,
      "step": 17070
    },
    {
      "epoch": 0.910423496175475,
      "grad_norm": 1.4967775873948406,
      "learning_rate": 2.5031455928827595e-06,
      "loss": 0.3786,
      "step": 17080
    },
    {
      "epoch": 0.9109565310093014,
      "grad_norm": 1.6169457528277469,
      "learning_rate": 2.502490135381277e-06,
      "loss": 0.3624,
      "step": 17090
    },
    {
      "epoch": 0.9114895658431279,
      "grad_norm": 1.4820685636734916,
      "learning_rate": 2.5018343317615818e-06,
      "loss": 0.3655,
      "step": 17100
    },
    {
      "epoch": 0.9120226006769543,
      "grad_norm": 1.6196579726566909,
      "learning_rate": 2.5011781822500955e-06,
      "loss": 0.375,
      "step": 17110
    },
    {
      "epoch": 0.9125556355107807,
      "grad_norm": 1.508273970983703,
      "learning_rate": 2.500521687073359e-06,
      "loss": 0.3654,
      "step": 17120
    },
    {
      "epoch": 0.9130886703446071,
      "grad_norm": 1.5030710300776673,
      "learning_rate": 2.4998648464580332e-06,
      "loss": 0.3646,
      "step": 17130
    },
    {
      "epoch": 0.9136217051784334,
      "grad_norm": 1.620283186805845,
      "learning_rate": 2.499207660630898e-06,
      "loss": 0.379,
      "step": 17140
    },
    {
      "epoch": 0.9141547400122598,
      "grad_norm": 1.5628084945231049,
      "learning_rate": 2.498550129818852e-06,
      "loss": 0.3687,
      "step": 17150
    },
    {
      "epoch": 0.9146877748460862,
      "grad_norm": 1.696540509817431,
      "learning_rate": 2.4978922542489137e-06,
      "loss": 0.3653,
      "step": 17160
    },
    {
      "epoch": 0.9152208096799126,
      "grad_norm": 1.7455171273145023,
      "learning_rate": 2.4972340341482198e-06,
      "loss": 0.3717,
      "step": 17170
    },
    {
      "epoch": 0.915753844513739,
      "grad_norm": 1.6338894695615422,
      "learning_rate": 2.4965754697440263e-06,
      "loss": 0.3721,
      "step": 17180
    },
    {
      "epoch": 0.9162868793475654,
      "grad_norm": 1.8131928426266846,
      "learning_rate": 2.495916561263709e-06,
      "loss": 0.3572,
      "step": 17190
    },
    {
      "epoch": 0.9168199141813917,
      "grad_norm": 1.5300294807613743,
      "learning_rate": 2.49525730893476e-06,
      "loss": 0.3655,
      "step": 17200
    },
    {
      "epoch": 0.9173529490152181,
      "grad_norm": 1.6216814078782733,
      "learning_rate": 2.494597712984794e-06,
      "loss": 0.367,
      "step": 17210
    },
    {
      "epoch": 0.9178859838490445,
      "grad_norm": 1.4364420541354284,
      "learning_rate": 2.49393777364154e-06,
      "loss": 0.3716,
      "step": 17220
    },
    {
      "epoch": 0.9184190186828709,
      "grad_norm": 1.7022997921564593,
      "learning_rate": 2.4932774911328488e-06,
      "loss": 0.3716,
      "step": 17230
    },
    {
      "epoch": 0.9189520535166973,
      "grad_norm": 1.7891312385508846,
      "learning_rate": 2.492616865686688e-06,
      "loss": 0.3752,
      "step": 17240
    },
    {
      "epoch": 0.9194850883505237,
      "grad_norm": 1.5373504256347597,
      "learning_rate": 2.4919558975311447e-06,
      "loss": 0.3602,
      "step": 17250
    },
    {
      "epoch": 0.9200181231843501,
      "grad_norm": 1.5604028028552743,
      "learning_rate": 2.491294586894424e-06,
      "loss": 0.3666,
      "step": 17260
    },
    {
      "epoch": 0.9205511580181764,
      "grad_norm": 1.4687632862878885,
      "learning_rate": 2.4906329340048473e-06,
      "loss": 0.3721,
      "step": 17270
    },
    {
      "epoch": 0.9210841928520028,
      "grad_norm": 1.5554092704522868,
      "learning_rate": 2.489970939090858e-06,
      "loss": 0.3715,
      "step": 17280
    },
    {
      "epoch": 0.9216172276858293,
      "grad_norm": 1.5603746558545535,
      "learning_rate": 2.4893086023810137e-06,
      "loss": 0.369,
      "step": 17290
    },
    {
      "epoch": 0.9221502625196557,
      "grad_norm": 1.9214917456509688,
      "learning_rate": 2.4886459241039927e-06,
      "loss": 0.3682,
      "step": 17300
    },
    {
      "epoch": 0.9226832973534821,
      "grad_norm": 1.5814369586103256,
      "learning_rate": 2.48798290448859e-06,
      "loss": 0.3619,
      "step": 17310
    },
    {
      "epoch": 0.9232163321873085,
      "grad_norm": 1.633097754000609,
      "learning_rate": 2.487319543763719e-06,
      "loss": 0.3706,
      "step": 17320
    },
    {
      "epoch": 0.9237493670211349,
      "grad_norm": 1.4030161367094047,
      "learning_rate": 2.4866558421584104e-06,
      "loss": 0.3677,
      "step": 17330
    },
    {
      "epoch": 0.9242824018549612,
      "grad_norm": 1.443605802267004,
      "learning_rate": 2.485991799901813e-06,
      "loss": 0.3627,
      "step": 17340
    },
    {
      "epoch": 0.9248154366887876,
      "grad_norm": 1.5282972689662562,
      "learning_rate": 2.4853274172231927e-06,
      "loss": 0.3573,
      "step": 17350
    },
    {
      "epoch": 0.925348471522614,
      "grad_norm": 1.7561577627016143,
      "learning_rate": 2.484662694351933e-06,
      "loss": 0.3641,
      "step": 17360
    },
    {
      "epoch": 0.9258815063564404,
      "grad_norm": 1.415277556429084,
      "learning_rate": 2.483997631517536e-06,
      "loss": 0.3649,
      "step": 17370
    },
    {
      "epoch": 0.9264145411902668,
      "grad_norm": 1.5985875375566883,
      "learning_rate": 2.48333222894962e-06,
      "loss": 0.372,
      "step": 17380
    },
    {
      "epoch": 0.9269475760240932,
      "grad_norm": 1.6241152316961869,
      "learning_rate": 2.48266648687792e-06,
      "loss": 0.3763,
      "step": 17390
    },
    {
      "epoch": 0.9274806108579196,
      "grad_norm": 1.6356146736491557,
      "learning_rate": 2.4820004055322905e-06,
      "loss": 0.378,
      "step": 17400
    },
    {
      "epoch": 0.9280136456917459,
      "grad_norm": 1.681918348554039,
      "learning_rate": 2.4813339851427007e-06,
      "loss": 0.3698,
      "step": 17410
    },
    {
      "epoch": 0.9285466805255723,
      "grad_norm": 1.6842573868607527,
      "learning_rate": 2.480667225939238e-06,
      "loss": 0.3762,
      "step": 17420
    },
    {
      "epoch": 0.9290797153593987,
      "grad_norm": 1.5264317352005894,
      "learning_rate": 2.4800001281521072e-06,
      "loss": 0.3761,
      "step": 17430
    },
    {
      "epoch": 0.9296127501932251,
      "grad_norm": 1.476755271296365,
      "learning_rate": 2.479332692011629e-06,
      "loss": 0.358,
      "step": 17440
    },
    {
      "epoch": 0.9301457850270515,
      "grad_norm": 1.4805263758967948,
      "learning_rate": 2.478664917748241e-06,
      "loss": 0.3603,
      "step": 17450
    },
    {
      "epoch": 0.9306788198608779,
      "grad_norm": 1.525436465381181,
      "learning_rate": 2.477996805592499e-06,
      "loss": 0.376,
      "step": 17460
    },
    {
      "epoch": 0.9312118546947042,
      "grad_norm": 1.648566810118355,
      "learning_rate": 2.4773283557750737e-06,
      "loss": 0.3695,
      "step": 17470
    },
    {
      "epoch": 0.9317448895285307,
      "grad_norm": 1.7113367370825123,
      "learning_rate": 2.476659568526753e-06,
      "loss": 0.374,
      "step": 17480
    },
    {
      "epoch": 0.9322779243623571,
      "grad_norm": 1.4904209524537468,
      "learning_rate": 2.4759904440784418e-06,
      "loss": 0.3687,
      "step": 17490
    },
    {
      "epoch": 0.9328109591961835,
      "grad_norm": 1.5561071215329212,
      "learning_rate": 2.4753209826611604e-06,
      "loss": 0.376,
      "step": 17500
    },
    {
      "epoch": 0.9333439940300099,
      "grad_norm": 1.6039022614531266,
      "learning_rate": 2.474651184506046e-06,
      "loss": 0.3726,
      "step": 17510
    },
    {
      "epoch": 0.9338770288638363,
      "grad_norm": 1.7460051562511008,
      "learning_rate": 2.473981049844353e-06,
      "loss": 0.3661,
      "step": 17520
    },
    {
      "epoch": 0.9344100636976627,
      "grad_norm": 1.4938989063829358,
      "learning_rate": 2.4733105789074503e-06,
      "loss": 0.3686,
      "step": 17530
    },
    {
      "epoch": 0.934943098531489,
      "grad_norm": 1.58904564233612,
      "learning_rate": 2.472639771926824e-06,
      "loss": 0.3656,
      "step": 17540
    },
    {
      "epoch": 0.9354761333653154,
      "grad_norm": 1.4791366539508526,
      "learning_rate": 2.4719686291340754e-06,
      "loss": 0.3663,
      "step": 17550
    },
    {
      "epoch": 0.9360091681991418,
      "grad_norm": 1.6234568159798726,
      "learning_rate": 2.471297150760923e-06,
      "loss": 0.3686,
      "step": 17560
    },
    {
      "epoch": 0.9365422030329682,
      "grad_norm": 1.5914629135678837,
      "learning_rate": 2.4706253370392e-06,
      "loss": 0.3644,
      "step": 17570
    },
    {
      "epoch": 0.9370752378667946,
      "grad_norm": 1.3192337859178085,
      "learning_rate": 2.4699531882008553e-06,
      "loss": 0.3672,
      "step": 17580
    },
    {
      "epoch": 0.937608272700621,
      "grad_norm": 1.4983221392346286,
      "learning_rate": 2.4692807044779545e-06,
      "loss": 0.3603,
      "step": 17590
    },
    {
      "epoch": 0.9381413075344474,
      "grad_norm": 1.5973126916992249,
      "learning_rate": 2.4686078861026786e-06,
      "loss": 0.3721,
      "step": 17600
    },
    {
      "epoch": 0.9386743423682737,
      "grad_norm": 1.5523063050352968,
      "learning_rate": 2.467934733307323e-06,
      "loss": 0.3633,
      "step": 17610
    },
    {
      "epoch": 0.9392073772021001,
      "grad_norm": 1.6736114168264564,
      "learning_rate": 2.4672612463243004e-06,
      "loss": 0.371,
      "step": 17620
    },
    {
      "epoch": 0.9397404120359265,
      "grad_norm": 1.9173319357627825,
      "learning_rate": 2.466587425386137e-06,
      "loss": 0.3596,
      "step": 17630
    },
    {
      "epoch": 0.9402734468697529,
      "grad_norm": 1.603323919224505,
      "learning_rate": 2.465913270725476e-06,
      "loss": 0.3617,
      "step": 17640
    },
    {
      "epoch": 0.9408064817035793,
      "grad_norm": 1.6153372751997013,
      "learning_rate": 2.4652387825750744e-06,
      "loss": 0.3666,
      "step": 17650
    },
    {
      "epoch": 0.9413395165374058,
      "grad_norm": 1.476201457804405,
      "learning_rate": 2.4645639611678054e-06,
      "loss": 0.3729,
      "step": 17660
    },
    {
      "epoch": 0.9418725513712322,
      "grad_norm": 1.8343743669833457,
      "learning_rate": 2.4638888067366557e-06,
      "loss": 0.3735,
      "step": 17670
    },
    {
      "epoch": 0.9424055862050585,
      "grad_norm": 1.8501413391469148,
      "learning_rate": 2.4632133195147298e-06,
      "loss": 0.3694,
      "step": 17680
    },
    {
      "epoch": 0.9429386210388849,
      "grad_norm": 1.6882610304258439,
      "learning_rate": 2.4625374997352436e-06,
      "loss": 0.3629,
      "step": 17690
    },
    {
      "epoch": 0.9434716558727113,
      "grad_norm": 1.6062698520592167,
      "learning_rate": 2.4618613476315314e-06,
      "loss": 0.367,
      "step": 17700
    },
    {
      "epoch": 0.9440046907065377,
      "grad_norm": 1.5738805619236607,
      "learning_rate": 2.461184863437039e-06,
      "loss": 0.3555,
      "step": 17710
    },
    {
      "epoch": 0.9445377255403641,
      "grad_norm": 1.6530339092462851,
      "learning_rate": 2.460508047385329e-06,
      "loss": 0.3658,
      "step": 17720
    },
    {
      "epoch": 0.9450707603741905,
      "grad_norm": 1.6161906807110695,
      "learning_rate": 2.4598308997100784e-06,
      "loss": 0.3847,
      "step": 17730
    },
    {
      "epoch": 0.9456037952080169,
      "grad_norm": 1.4604749876874386,
      "learning_rate": 2.4591534206450767e-06,
      "loss": 0.3692,
      "step": 17740
    },
    {
      "epoch": 0.9461368300418432,
      "grad_norm": 1.5185517295183157,
      "learning_rate": 2.45847561042423e-06,
      "loss": 0.3708,
      "step": 17750
    },
    {
      "epoch": 0.9466698648756696,
      "grad_norm": 1.7132156635550504,
      "learning_rate": 2.4577974692815584e-06,
      "loss": 0.3717,
      "step": 17760
    },
    {
      "epoch": 0.947202899709496,
      "grad_norm": 1.543104191911181,
      "learning_rate": 2.457118997451196e-06,
      "loss": 0.3731,
      "step": 17770
    },
    {
      "epoch": 0.9477359345433224,
      "grad_norm": 1.6161894911997858,
      "learning_rate": 2.45644019516739e-06,
      "loss": 0.3696,
      "step": 17780
    },
    {
      "epoch": 0.9482689693771488,
      "grad_norm": 1.4729742964038686,
      "learning_rate": 2.4557610626645035e-06,
      "loss": 0.3748,
      "step": 17790
    },
    {
      "epoch": 0.9488020042109752,
      "grad_norm": 1.5276426885909562,
      "learning_rate": 2.4550816001770123e-06,
      "loss": 0.3617,
      "step": 17800
    },
    {
      "epoch": 0.9493350390448015,
      "grad_norm": 1.7163154329909904,
      "learning_rate": 2.454401807939506e-06,
      "loss": 0.364,
      "step": 17810
    },
    {
      "epoch": 0.9498680738786279,
      "grad_norm": 1.5921794005720344,
      "learning_rate": 2.4537216861866907e-06,
      "loss": 0.3629,
      "step": 17820
    },
    {
      "epoch": 0.9504011087124543,
      "grad_norm": 1.3869546460769775,
      "learning_rate": 2.4530412351533822e-06,
      "loss": 0.3643,
      "step": 17830
    },
    {
      "epoch": 0.9509341435462807,
      "grad_norm": 1.5203927594497786,
      "learning_rate": 2.452360455074512e-06,
      "loss": 0.3719,
      "step": 17840
    },
    {
      "epoch": 0.9514671783801072,
      "grad_norm": 1.545618630544393,
      "learning_rate": 2.4516793461851267e-06,
      "loss": 0.3639,
      "step": 17850
    },
    {
      "epoch": 0.9520002132139336,
      "grad_norm": 1.579502199217928,
      "learning_rate": 2.4509979087203837e-06,
      "loss": 0.364,
      "step": 17860
    },
    {
      "epoch": 0.95253324804776,
      "grad_norm": 1.5573899724930402,
      "learning_rate": 2.4503161429155553e-06,
      "loss": 0.3709,
      "step": 17870
    },
    {
      "epoch": 0.9530662828815863,
      "grad_norm": 1.5166311273174973,
      "learning_rate": 2.4496340490060275e-06,
      "loss": 0.3529,
      "step": 17880
    },
    {
      "epoch": 0.9535993177154127,
      "grad_norm": 1.5785594326300647,
      "learning_rate": 2.4489516272272976e-06,
      "loss": 0.351,
      "step": 17890
    },
    {
      "epoch": 0.9541323525492391,
      "grad_norm": 1.6356705833215939,
      "learning_rate": 2.448268877814979e-06,
      "loss": 0.3673,
      "step": 17900
    },
    {
      "epoch": 0.9546653873830655,
      "grad_norm": 1.405334914648528,
      "learning_rate": 2.447585801004796e-06,
      "loss": 0.373,
      "step": 17910
    },
    {
      "epoch": 0.9551984222168919,
      "grad_norm": 1.5614116417005306,
      "learning_rate": 2.446902397032587e-06,
      "loss": 0.3502,
      "step": 17920
    },
    {
      "epoch": 0.9557314570507183,
      "grad_norm": 1.5052327333711553,
      "learning_rate": 2.4462186661343027e-06,
      "loss": 0.366,
      "step": 17930
    },
    {
      "epoch": 0.9562644918845447,
      "grad_norm": 1.728569145652302,
      "learning_rate": 2.4455346085460066e-06,
      "loss": 0.3642,
      "step": 17940
    },
    {
      "epoch": 0.956797526718371,
      "grad_norm": 1.542268593167441,
      "learning_rate": 2.4448502245038766e-06,
      "loss": 0.3695,
      "step": 17950
    },
    {
      "epoch": 0.9573305615521974,
      "grad_norm": 1.397372313417784,
      "learning_rate": 2.444165514244201e-06,
      "loss": 0.3734,
      "step": 17960
    },
    {
      "epoch": 0.9578635963860238,
      "grad_norm": 1.6251252046449296,
      "learning_rate": 2.4434804780033823e-06,
      "loss": 0.3674,
      "step": 17970
    },
    {
      "epoch": 0.9583966312198502,
      "grad_norm": 1.7163614107110303,
      "learning_rate": 2.442795116017935e-06,
      "loss": 0.3593,
      "step": 17980
    },
    {
      "epoch": 0.9589296660536766,
      "grad_norm": 1.566021206897545,
      "learning_rate": 2.4421094285244855e-06,
      "loss": 0.3722,
      "step": 17990
    },
    {
      "epoch": 0.959462700887503,
      "grad_norm": 1.6304647026237082,
      "learning_rate": 2.441423415759774e-06,
      "loss": 0.3587,
      "step": 18000
    },
    {
      "epoch": 0.9599957357213293,
      "grad_norm": 1.492626828595643,
      "learning_rate": 2.440737077960652e-06,
      "loss": 0.3597,
      "step": 18010
    },
    {
      "epoch": 0.9605287705551557,
      "grad_norm": 1.5001595509604813,
      "learning_rate": 2.440050415364084e-06,
      "loss": 0.3599,
      "step": 18020
    },
    {
      "epoch": 0.9610618053889821,
      "grad_norm": 1.6063213206597977,
      "learning_rate": 2.439363428207145e-06,
      "loss": 0.3631,
      "step": 18030
    },
    {
      "epoch": 0.9615948402228086,
      "grad_norm": 1.6093009715552609,
      "learning_rate": 2.4386761167270234e-06,
      "loss": 0.3614,
      "step": 18040
    },
    {
      "epoch": 0.962127875056635,
      "grad_norm": 1.6308074058789368,
      "learning_rate": 2.43798848116102e-06,
      "loss": 0.3638,
      "step": 18050
    },
    {
      "epoch": 0.9626609098904614,
      "grad_norm": 1.7420568186962917,
      "learning_rate": 2.4373005217465465e-06,
      "loss": 0.371,
      "step": 18060
    },
    {
      "epoch": 0.9631939447242878,
      "grad_norm": 1.4968721393315962,
      "learning_rate": 2.436612238721126e-06,
      "loss": 0.3669,
      "step": 18070
    },
    {
      "epoch": 0.9637269795581141,
      "grad_norm": 1.5665639159164495,
      "learning_rate": 2.435923632322395e-06,
      "loss": 0.378,
      "step": 18080
    },
    {
      "epoch": 0.9642600143919405,
      "grad_norm": 1.6717461177536967,
      "learning_rate": 2.4352347027881005e-06,
      "loss": 0.3632,
      "step": 18090
    },
    {
      "epoch": 0.9647930492257669,
      "grad_norm": 1.5647881759413753,
      "learning_rate": 2.434545450356101e-06,
      "loss": 0.3624,
      "step": 18100
    },
    {
      "epoch": 0.9653260840595933,
      "grad_norm": 1.600302406680916,
      "learning_rate": 2.4338558752643664e-06,
      "loss": 0.3686,
      "step": 18110
    },
    {
      "epoch": 0.9658591188934197,
      "grad_norm": 1.636536800418388,
      "learning_rate": 2.433165977750979e-06,
      "loss": 0.3641,
      "step": 18120
    },
    {
      "epoch": 0.9663921537272461,
      "grad_norm": 1.5731316349695552,
      "learning_rate": 2.4324757580541317e-06,
      "loss": 0.3584,
      "step": 18130
    },
    {
      "epoch": 0.9669251885610725,
      "grad_norm": 1.5097527142077196,
      "learning_rate": 2.431785216412129e-06,
      "loss": 0.3603,
      "step": 18140
    },
    {
      "epoch": 0.9674582233948988,
      "grad_norm": 1.4878256752404058,
      "learning_rate": 2.431094353063385e-06,
      "loss": 0.3724,
      "step": 18150
    },
    {
      "epoch": 0.9679912582287252,
      "grad_norm": 1.5942259210670973,
      "learning_rate": 2.430403168246427e-06,
      "loss": 0.3528,
      "step": 18160
    },
    {
      "epoch": 0.9685242930625516,
      "grad_norm": 1.5265735741518125,
      "learning_rate": 2.4297116621998925e-06,
      "loss": 0.3616,
      "step": 18170
    },
    {
      "epoch": 0.969057327896378,
      "grad_norm": 1.746135241116584,
      "learning_rate": 2.4290198351625295e-06,
      "loss": 0.3705,
      "step": 18180
    },
    {
      "epoch": 0.9695903627302044,
      "grad_norm": 1.5578279963726467,
      "learning_rate": 2.428327687373197e-06,
      "loss": 0.3646,
      "step": 18190
    },
    {
      "epoch": 0.9701233975640308,
      "grad_norm": 1.6358521318726396,
      "learning_rate": 2.427635219070866e-06,
      "loss": 0.3581,
      "step": 18200
    },
    {
      "epoch": 0.9706564323978571,
      "grad_norm": 1.4343881217679273,
      "learning_rate": 2.4269424304946155e-06,
      "loss": 0.3606,
      "step": 18210
    },
    {
      "epoch": 0.9711894672316836,
      "grad_norm": 1.510311478861491,
      "learning_rate": 2.4262493218836377e-06,
      "loss": 0.3648,
      "step": 18220
    },
    {
      "epoch": 0.97172250206551,
      "grad_norm": 1.660404789477891,
      "learning_rate": 2.425555893477234e-06,
      "loss": 0.371,
      "step": 18230
    },
    {
      "epoch": 0.9722555368993364,
      "grad_norm": 1.4212144544897773,
      "learning_rate": 2.4248621455148165e-06,
      "loss": 0.3744,
      "step": 18240
    },
    {
      "epoch": 0.9727885717331628,
      "grad_norm": 1.4489060964325087,
      "learning_rate": 2.4241680782359068e-06,
      "loss": 0.3609,
      "step": 18250
    },
    {
      "epoch": 0.9733216065669892,
      "grad_norm": 1.426970079723212,
      "learning_rate": 2.423473691880138e-06,
      "loss": 0.365,
      "step": 18260
    },
    {
      "epoch": 0.9738546414008156,
      "grad_norm": 1.5536965151741717,
      "learning_rate": 2.422778986687253e-06,
      "loss": 0.364,
      "step": 18270
    },
    {
      "epoch": 0.974387676234642,
      "grad_norm": 1.635580908258095,
      "learning_rate": 2.422083962897105e-06,
      "loss": 0.3566,
      "step": 18280
    },
    {
      "epoch": 0.9749207110684683,
      "grad_norm": 1.4618499087496961,
      "learning_rate": 2.4213886207496564e-06,
      "loss": 0.3771,
      "step": 18290
    },
    {
      "epoch": 0.9754537459022947,
      "grad_norm": 1.6157693302876939,
      "learning_rate": 2.4206929604849796e-06,
      "loss": 0.3627,
      "step": 18300
    },
    {
      "epoch": 0.9759867807361211,
      "grad_norm": 1.4240405488738535,
      "learning_rate": 2.419996982343258e-06,
      "loss": 0.3733,
      "step": 18310
    },
    {
      "epoch": 0.9765198155699475,
      "grad_norm": 1.500814605425278,
      "learning_rate": 2.419300686564784e-06,
      "loss": 0.3623,
      "step": 18320
    },
    {
      "epoch": 0.9770528504037739,
      "grad_norm": 1.5294416558247144,
      "learning_rate": 2.418604073389958e-06,
      "loss": 0.3589,
      "step": 18330
    },
    {
      "epoch": 0.9775858852376003,
      "grad_norm": 1.4592799169375035,
      "learning_rate": 2.4179071430592935e-06,
      "loss": 0.3693,
      "step": 18340
    },
    {
      "epoch": 0.9781189200714266,
      "grad_norm": 1.49931538348573,
      "learning_rate": 2.417209895813411e-06,
      "loss": 0.3542,
      "step": 18350
    },
    {
      "epoch": 0.978651954905253,
      "grad_norm": 1.379484310147667,
      "learning_rate": 2.416512331893041e-06,
      "loss": 0.3668,
      "step": 18360
    },
    {
      "epoch": 0.9791849897390794,
      "grad_norm": 1.7133936154226403,
      "learning_rate": 2.4158144515390228e-06,
      "loss": 0.3631,
      "step": 18370
    },
    {
      "epoch": 0.9797180245729058,
      "grad_norm": 1.5444301178680617,
      "learning_rate": 2.415116254992306e-06,
      "loss": 0.3712,
      "step": 18380
    },
    {
      "epoch": 0.9802510594067322,
      "grad_norm": 1.5739836196563841,
      "learning_rate": 2.4144177424939493e-06,
      "loss": 0.3529,
      "step": 18390
    },
    {
      "epoch": 0.9807840942405586,
      "grad_norm": 1.5059406995094413,
      "learning_rate": 2.413718914285119e-06,
      "loss": 0.351,
      "step": 18400
    },
    {
      "epoch": 0.9813171290743851,
      "grad_norm": 1.6895649086446567,
      "learning_rate": 2.4130197706070926e-06,
      "loss": 0.3612,
      "step": 18410
    },
    {
      "epoch": 0.9818501639082114,
      "grad_norm": 1.7063452143561457,
      "learning_rate": 2.4123203117012546e-06,
      "loss": 0.3817,
      "step": 18420
    },
    {
      "epoch": 0.9823831987420378,
      "grad_norm": 1.6181306893430538,
      "learning_rate": 2.411620537809099e-06,
      "loss": 0.3628,
      "step": 18430
    },
    {
      "epoch": 0.9829162335758642,
      "grad_norm": 1.552597442162855,
      "learning_rate": 2.4109204491722298e-06,
      "loss": 0.3659,
      "step": 18440
    },
    {
      "epoch": 0.9834492684096906,
      "grad_norm": 1.4596416357276856,
      "learning_rate": 2.410220046032357e-06,
      "loss": 0.3532,
      "step": 18450
    },
    {
      "epoch": 0.983982303243517,
      "grad_norm": 1.612795654690574,
      "learning_rate": 2.409519328631302e-06,
      "loss": 0.3676,
      "step": 18460
    },
    {
      "epoch": 0.9845153380773434,
      "grad_norm": 1.5919827568767952,
      "learning_rate": 2.4088182972109924e-06,
      "loss": 0.3529,
      "step": 18470
    },
    {
      "epoch": 0.9850483729111698,
      "grad_norm": 1.7529574427579884,
      "learning_rate": 2.408116952013466e-06,
      "loss": 0.3756,
      "step": 18480
    },
    {
      "epoch": 0.9855814077449961,
      "grad_norm": 1.926275698771631,
      "learning_rate": 2.407415293280868e-06,
      "loss": 0.3785,
      "step": 18490
    },
    {
      "epoch": 0.9861144425788225,
      "grad_norm": 1.6733985301065029,
      "learning_rate": 2.406713321255452e-06,
      "loss": 0.3773,
      "step": 18500
    },
    {
      "epoch": 0.9866474774126489,
      "grad_norm": 1.6789861335387402,
      "learning_rate": 2.4060110361795796e-06,
      "loss": 0.3667,
      "step": 18510
    },
    {
      "epoch": 0.9871805122464753,
      "grad_norm": 1.6123061536924932,
      "learning_rate": 2.4053084382957204e-06,
      "loss": 0.366,
      "step": 18520
    },
    {
      "epoch": 0.9877135470803017,
      "grad_norm": 1.587191956927505,
      "learning_rate": 2.404605527846453e-06,
      "loss": 0.3604,
      "step": 18530
    },
    {
      "epoch": 0.9882465819141281,
      "grad_norm": 1.530623928395734,
      "learning_rate": 2.4039023050744633e-06,
      "loss": 0.3633,
      "step": 18540
    },
    {
      "epoch": 0.9887796167479544,
      "grad_norm": 1.4767438794529086,
      "learning_rate": 2.403198770222543e-06,
      "loss": 0.3663,
      "step": 18550
    },
    {
      "epoch": 0.9893126515817808,
      "grad_norm": 1.6436863081529758,
      "learning_rate": 2.4024949235335962e-06,
      "loss": 0.3746,
      "step": 18560
    },
    {
      "epoch": 0.9898456864156072,
      "grad_norm": 1.6375251689344046,
      "learning_rate": 2.4017907652506306e-06,
      "loss": 0.367,
      "step": 18570
    },
    {
      "epoch": 0.9903787212494336,
      "grad_norm": 1.5685974020849787,
      "learning_rate": 2.4010862956167625e-06,
      "loss": 0.3658,
      "step": 18580
    },
    {
      "epoch": 0.99091175608326,
      "grad_norm": 1.69192421210803,
      "learning_rate": 2.4003815148752167e-06,
      "loss": 0.3783,
      "step": 18590
    },
    {
      "epoch": 0.9914447909170865,
      "grad_norm": 1.5988772782352858,
      "learning_rate": 2.3996764232693237e-06,
      "loss": 0.3706,
      "step": 18600
    },
    {
      "epoch": 0.9919778257509129,
      "grad_norm": 1.6918062581514597,
      "learning_rate": 2.398971021042524e-06,
      "loss": 0.3663,
      "step": 18610
    },
    {
      "epoch": 0.9925108605847393,
      "grad_norm": 1.5502574966642315,
      "learning_rate": 2.398265308438362e-06,
      "loss": 0.3567,
      "step": 18620
    },
    {
      "epoch": 0.9930438954185656,
      "grad_norm": 1.657632474364943,
      "learning_rate": 2.3975592857004926e-06,
      "loss": 0.357,
      "step": 18630
    },
    {
      "epoch": 0.993576930252392,
      "grad_norm": 1.6473630177735619,
      "learning_rate": 2.3968529530726748e-06,
      "loss": 0.3791,
      "step": 18640
    },
    {
      "epoch": 0.9941099650862184,
      "grad_norm": 1.5931069797319237,
      "learning_rate": 2.3961463107987757e-06,
      "loss": 0.3675,
      "step": 18650
    },
    {
      "epoch": 0.9946429999200448,
      "grad_norm": 1.7627140795985976,
      "learning_rate": 2.395439359122771e-06,
      "loss": 0.3612,
      "step": 18660
    },
    {
      "epoch": 0.9951760347538712,
      "grad_norm": 1.6035597605909455,
      "learning_rate": 2.3947320982887403e-06,
      "loss": 0.3692,
      "step": 18670
    },
    {
      "epoch": 0.9957090695876976,
      "grad_norm": 1.533608065769785,
      "learning_rate": 2.394024528540873e-06,
      "loss": 0.3549,
      "step": 18680
    },
    {
      "epoch": 0.9962421044215239,
      "grad_norm": 1.5374536888105628,
      "learning_rate": 2.3933166501234615e-06,
      "loss": 0.3696,
      "step": 18690
    },
    {
      "epoch": 0.9967751392553503,
      "grad_norm": 1.54924377405105,
      "learning_rate": 2.3926084632809084e-06,
      "loss": 0.3672,
      "step": 18700
    },
    {
      "epoch": 0.9973081740891767,
      "grad_norm": 1.565844385793466,
      "learning_rate": 2.391899968257721e-06,
      "loss": 0.3466,
      "step": 18710
    },
    {
      "epoch": 0.9978412089230031,
      "grad_norm": 1.5983914218659367,
      "learning_rate": 2.3911911652985127e-06,
      "loss": 0.3485,
      "step": 18720
    },
    {
      "epoch": 0.9983742437568295,
      "grad_norm": 1.5797941383572287,
      "learning_rate": 2.3904820546480036e-06,
      "loss": 0.3513,
      "step": 18730
    },
    {
      "epoch": 0.9989072785906559,
      "grad_norm": 1.5663481196163236,
      "learning_rate": 2.3897726365510214e-06,
      "loss": 0.3625,
      "step": 18740
    },
    {
      "epoch": 0.9994403134244823,
      "grad_norm": 1.5840773697615038,
      "learning_rate": 2.3890629112524967e-06,
      "loss": 0.3526,
      "step": 18750
    },
    {
      "epoch": 0.9999733482583086,
      "grad_norm": 1.6273337690743201,
      "learning_rate": 2.38835287899747e-06,
      "loss": 0.3572,
      "step": 18760
    },
    {
      "epoch": 1.000506383092135,
      "grad_norm": 1.7185577385814188,
      "learning_rate": 2.3876425400310852e-06,
      "loss": 0.3197,
      "step": 18770
    },
    {
      "epoch": 1.0010394179259614,
      "grad_norm": 1.4672795083447345,
      "learning_rate": 2.3869318945985934e-06,
      "loss": 0.3064,
      "step": 18780
    },
    {
      "epoch": 1.0015724527597878,
      "grad_norm": 1.6578994929501827,
      "learning_rate": 2.38622094294535e-06,
      "loss": 0.3204,
      "step": 18790
    },
    {
      "epoch": 1.0021054875936142,
      "grad_norm": 1.6202311086614796,
      "learning_rate": 2.385509685316817e-06,
      "loss": 0.3102,
      "step": 18800
    },
    {
      "epoch": 1.0026385224274406,
      "grad_norm": 1.7411381826142234,
      "learning_rate": 2.3847981219585633e-06,
      "loss": 0.3166,
      "step": 18810
    },
    {
      "epoch": 1.003171557261267,
      "grad_norm": 1.7111186110108108,
      "learning_rate": 2.3840862531162614e-06,
      "loss": 0.3263,
      "step": 18820
    },
    {
      "epoch": 1.0037045920950933,
      "grad_norm": 1.6358968238976415,
      "learning_rate": 2.3833740790356897e-06,
      "loss": 0.3092,
      "step": 18830
    },
    {
      "epoch": 1.0042376269289197,
      "grad_norm": 1.7713080472771519,
      "learning_rate": 2.3826615999627336e-06,
      "loss": 0.3116,
      "step": 18840
    },
    {
      "epoch": 1.004770661762746,
      "grad_norm": 1.54931472801709,
      "learning_rate": 2.3819488161433816e-06,
      "loss": 0.3107,
      "step": 18850
    },
    {
      "epoch": 1.0053036965965725,
      "grad_norm": 1.6212523398451233,
      "learning_rate": 2.381235727823728e-06,
      "loss": 0.3154,
      "step": 18860
    },
    {
      "epoch": 1.0058367314303989,
      "grad_norm": 1.6009910761940318,
      "learning_rate": 2.3805223352499735e-06,
      "loss": 0.3146,
      "step": 18870
    },
    {
      "epoch": 1.0063697662642255,
      "grad_norm": 1.693460271570902,
      "learning_rate": 2.3798086386684223e-06,
      "loss": 0.3236,
      "step": 18880
    },
    {
      "epoch": 1.0069028010980519,
      "grad_norm": 1.6340783931107634,
      "learning_rate": 2.3790946383254847e-06,
      "loss": 0.3059,
      "step": 18890
    },
    {
      "epoch": 1.0074358359318782,
      "grad_norm": 1.5458288811540937,
      "learning_rate": 2.3783803344676753e-06,
      "loss": 0.3157,
      "step": 18900
    },
    {
      "epoch": 1.0079688707657046,
      "grad_norm": 1.6652450212515932,
      "learning_rate": 2.3776657273416126e-06,
      "loss": 0.3249,
      "step": 18910
    },
    {
      "epoch": 1.008501905599531,
      "grad_norm": 1.593818177263969,
      "learning_rate": 2.3769508171940223e-06,
      "loss": 0.308,
      "step": 18920
    },
    {
      "epoch": 1.0090349404333574,
      "grad_norm": 1.5977494921414706,
      "learning_rate": 2.376235604271732e-06,
      "loss": 0.3096,
      "step": 18930
    },
    {
      "epoch": 1.0095679752671838,
      "grad_norm": 1.5088384012401328,
      "learning_rate": 2.3755200888216753e-06,
      "loss": 0.3099,
      "step": 18940
    },
    {
      "epoch": 1.0101010101010102,
      "grad_norm": 1.7601843230682628,
      "learning_rate": 2.3748042710908905e-06,
      "loss": 0.3118,
      "step": 18950
    },
    {
      "epoch": 1.0106340449348366,
      "grad_norm": 1.5501196543875728,
      "learning_rate": 2.374088151326519e-06,
      "loss": 0.3274,
      "step": 18960
    },
    {
      "epoch": 1.011167079768663,
      "grad_norm": 1.6502996693443244,
      "learning_rate": 2.3733717297758075e-06,
      "loss": 0.3033,
      "step": 18970
    },
    {
      "epoch": 1.0117001146024893,
      "grad_norm": 1.4951842464731067,
      "learning_rate": 2.372655006686107e-06,
      "loss": 0.3145,
      "step": 18980
    },
    {
      "epoch": 1.0122331494363157,
      "grad_norm": 1.5839191501728258,
      "learning_rate": 2.3719379823048718e-06,
      "loss": 0.3238,
      "step": 18990
    },
    {
      "epoch": 1.012766184270142,
      "grad_norm": 1.6237552148984358,
      "learning_rate": 2.37122065687966e-06,
      "loss": 0.306,
      "step": 19000
    },
    {
      "epoch": 1.0132992191039685,
      "grad_norm": 1.6308397511300092,
      "learning_rate": 2.370503030658136e-06,
      "loss": 0.307,
      "step": 19010
    },
    {
      "epoch": 1.0138322539377949,
      "grad_norm": 1.695769860890057,
      "learning_rate": 2.369785103888065e-06,
      "loss": 0.3249,
      "step": 19020
    },
    {
      "epoch": 1.0143652887716212,
      "grad_norm": 1.5964506225425257,
      "learning_rate": 2.369066876817317e-06,
      "loss": 0.304,
      "step": 19030
    },
    {
      "epoch": 1.0148983236054476,
      "grad_norm": 1.8266405020677607,
      "learning_rate": 2.368348349693867e-06,
      "loss": 0.3181,
      "step": 19040
    },
    {
      "epoch": 1.015431358439274,
      "grad_norm": 1.7223808238330438,
      "learning_rate": 2.3676295227657925e-06,
      "loss": 0.3091,
      "step": 19050
    },
    {
      "epoch": 1.0159643932731004,
      "grad_norm": 1.692758669499585,
      "learning_rate": 2.3669103962812744e-06,
      "loss": 0.3138,
      "step": 19060
    },
    {
      "epoch": 1.0164974281069268,
      "grad_norm": 1.5167307661108342,
      "learning_rate": 2.3661909704885965e-06,
      "loss": 0.3147,
      "step": 19070
    },
    {
      "epoch": 1.0170304629407532,
      "grad_norm": 1.5184974066061596,
      "learning_rate": 2.3654712456361474e-06,
      "loss": 0.3069,
      "step": 19080
    },
    {
      "epoch": 1.0175634977745796,
      "grad_norm": 1.751316879862646,
      "learning_rate": 2.364751221972418e-06,
      "loss": 0.3034,
      "step": 19090
    },
    {
      "epoch": 1.018096532608406,
      "grad_norm": 1.7597166150123467,
      "learning_rate": 2.3640308997460028e-06,
      "loss": 0.306,
      "step": 19100
    },
    {
      "epoch": 1.0186295674422323,
      "grad_norm": 1.724748724320993,
      "learning_rate": 2.363310279205599e-06,
      "loss": 0.319,
      "step": 19110
    },
    {
      "epoch": 1.0191626022760587,
      "grad_norm": 1.6481963472818624,
      "learning_rate": 2.362589360600007e-06,
      "loss": 0.3117,
      "step": 19120
    },
    {
      "epoch": 1.019695637109885,
      "grad_norm": 1.7116242445127332,
      "learning_rate": 2.361868144178129e-06,
      "loss": 0.3159,
      "step": 19130
    },
    {
      "epoch": 1.0202286719437115,
      "grad_norm": 1.5986354262012121,
      "learning_rate": 2.361146630188973e-06,
      "loss": 0.3041,
      "step": 19140
    },
    {
      "epoch": 1.0207617067775379,
      "grad_norm": 1.6965889643844392,
      "learning_rate": 2.3604248188816473e-06,
      "loss": 0.3108,
      "step": 19150
    },
    {
      "epoch": 1.0212947416113642,
      "grad_norm": 1.6781525876036727,
      "learning_rate": 2.3597027105053622e-06,
      "loss": 0.3141,
      "step": 19160
    },
    {
      "epoch": 1.0218277764451906,
      "grad_norm": 2.000216796698841,
      "learning_rate": 2.3589803053094324e-06,
      "loss": 0.3113,
      "step": 19170
    },
    {
      "epoch": 1.022360811279017,
      "grad_norm": 1.6150207837547592,
      "learning_rate": 2.3582576035432746e-06,
      "loss": 0.3009,
      "step": 19180
    },
    {
      "epoch": 1.0228938461128434,
      "grad_norm": 1.5035315637901694,
      "learning_rate": 2.3575346054564075e-06,
      "loss": 0.311,
      "step": 19190
    },
    {
      "epoch": 1.0234268809466698,
      "grad_norm": 1.617097767592438,
      "learning_rate": 2.356811311298453e-06,
      "loss": 0.3107,
      "step": 19200
    },
    {
      "epoch": 1.0239599157804962,
      "grad_norm": 1.6109057529637225,
      "learning_rate": 2.3560877213191332e-06,
      "loss": 0.3131,
      "step": 19210
    },
    {
      "epoch": 1.0244929506143226,
      "grad_norm": 1.6197081334960173,
      "learning_rate": 2.3553638357682747e-06,
      "loss": 0.3018,
      "step": 19220
    },
    {
      "epoch": 1.025025985448149,
      "grad_norm": 1.5921502789122441,
      "learning_rate": 2.354639654895805e-06,
      "loss": 0.3012,
      "step": 19230
    },
    {
      "epoch": 1.0255590202819753,
      "grad_norm": 1.6044360167651812,
      "learning_rate": 2.353915178951753e-06,
      "loss": 0.3173,
      "step": 19240
    },
    {
      "epoch": 1.026092055115802,
      "grad_norm": 1.5173977729856618,
      "learning_rate": 2.3531904081862512e-06,
      "loss": 0.3056,
      "step": 19250
    },
    {
      "epoch": 1.0266250899496283,
      "grad_norm": 1.5361598436139567,
      "learning_rate": 2.352465342849532e-06,
      "loss": 0.3089,
      "step": 19260
    },
    {
      "epoch": 1.0271581247834547,
      "grad_norm": 1.7125955590421769,
      "learning_rate": 2.351739983191931e-06,
      "loss": 0.3136,
      "step": 19270
    },
    {
      "epoch": 1.027691159617281,
      "grad_norm": 1.6376777345092481,
      "learning_rate": 2.351014329463884e-06,
      "loss": 0.3031,
      "step": 19280
    },
    {
      "epoch": 1.0282241944511075,
      "grad_norm": 1.6857913781260123,
      "learning_rate": 2.35028838191593e-06,
      "loss": 0.3021,
      "step": 19290
    },
    {
      "epoch": 1.0287572292849338,
      "grad_norm": 1.8233865996276521,
      "learning_rate": 2.349562140798708e-06,
      "loss": 0.3138,
      "step": 19300
    },
    {
      "epoch": 1.0292902641187602,
      "grad_norm": 1.7449481445071577,
      "learning_rate": 2.3488356063629588e-06,
      "loss": 0.3171,
      "step": 19310
    },
    {
      "epoch": 1.0298232989525866,
      "grad_norm": 1.6815181872052598,
      "learning_rate": 2.3481087788595244e-06,
      "loss": 0.3109,
      "step": 19320
    },
    {
      "epoch": 1.030356333786413,
      "grad_norm": 1.6453996080077002,
      "learning_rate": 2.347381658539349e-06,
      "loss": 0.3146,
      "step": 19330
    },
    {
      "epoch": 1.0308893686202394,
      "grad_norm": 1.6689190716130082,
      "learning_rate": 2.3466542456534764e-06,
      "loss": 0.3147,
      "step": 19340
    },
    {
      "epoch": 1.0314224034540658,
      "grad_norm": 1.5376012777142196,
      "learning_rate": 2.345926540453052e-06,
      "loss": 0.3056,
      "step": 19350
    },
    {
      "epoch": 1.0319554382878922,
      "grad_norm": 1.5967960069280163,
      "learning_rate": 2.3451985431893228e-06,
      "loss": 0.3167,
      "step": 19360
    },
    {
      "epoch": 1.0324884731217185,
      "grad_norm": 1.7716987549981884,
      "learning_rate": 2.3444702541136354e-06,
      "loss": 0.3082,
      "step": 19370
    },
    {
      "epoch": 1.033021507955545,
      "grad_norm": 1.74645380911529,
      "learning_rate": 2.343741673477438e-06,
      "loss": 0.3137,
      "step": 19380
    },
    {
      "epoch": 1.0335545427893713,
      "grad_norm": 1.5213147413735495,
      "learning_rate": 2.3430128015322793e-06,
      "loss": 0.3076,
      "step": 19390
    },
    {
      "epoch": 1.0340875776231977,
      "grad_norm": 1.621692540015206,
      "learning_rate": 2.3422836385298083e-06,
      "loss": 0.3136,
      "step": 19400
    },
    {
      "epoch": 1.034620612457024,
      "grad_norm": 1.5393512097869018,
      "learning_rate": 2.3415541847217745e-06,
      "loss": 0.3146,
      "step": 19410
    },
    {
      "epoch": 1.0351536472908505,
      "grad_norm": 1.5225064430863735,
      "learning_rate": 2.340824440360029e-06,
      "loss": 0.3056,
      "step": 19420
    },
    {
      "epoch": 1.0356866821246768,
      "grad_norm": 1.8123159853874484,
      "learning_rate": 2.3400944056965217e-06,
      "loss": 0.3167,
      "step": 19430
    },
    {
      "epoch": 1.0362197169585032,
      "grad_norm": 1.7187955329348583,
      "learning_rate": 2.3393640809833025e-06,
      "loss": 0.3141,
      "step": 19440
    },
    {
      "epoch": 1.0367527517923296,
      "grad_norm": 1.59679186854062,
      "learning_rate": 2.3386334664725237e-06,
      "loss": 0.2993,
      "step": 19450
    },
    {
      "epoch": 1.037285786626156,
      "grad_norm": 1.6695201115464604,
      "learning_rate": 2.337902562416434e-06,
      "loss": 0.3201,
      "step": 19460
    },
    {
      "epoch": 1.0378188214599824,
      "grad_norm": 1.5553969990181506,
      "learning_rate": 2.337171369067386e-06,
      "loss": 0.3011,
      "step": 19470
    },
    {
      "epoch": 1.0383518562938088,
      "grad_norm": 1.6465971085509254,
      "learning_rate": 2.3364398866778303e-06,
      "loss": 0.3101,
      "step": 19480
    },
    {
      "epoch": 1.0388848911276352,
      "grad_norm": 1.7314529715906022,
      "learning_rate": 2.335708115500317e-06,
      "loss": 0.3089,
      "step": 19490
    },
    {
      "epoch": 1.0394179259614615,
      "grad_norm": 1.5774253111142744,
      "learning_rate": 2.334976055787496e-06,
      "loss": 0.3052,
      "step": 19500
    },
    {
      "epoch": 1.039950960795288,
      "grad_norm": 1.6174879057925393,
      "learning_rate": 2.3342437077921175e-06,
      "loss": 0.3205,
      "step": 19510
    },
    {
      "epoch": 1.0404839956291143,
      "grad_norm": 1.809457281242797,
      "learning_rate": 2.333511071767031e-06,
      "loss": 0.3014,
      "step": 19520
    },
    {
      "epoch": 1.0410170304629407,
      "grad_norm": 1.4083336884286384,
      "learning_rate": 2.3327781479651847e-06,
      "loss": 0.2862,
      "step": 19530
    },
    {
      "epoch": 1.041550065296767,
      "grad_norm": 1.5764674355674486,
      "learning_rate": 2.332044936639627e-06,
      "loss": 0.32,
      "step": 19540
    },
    {
      "epoch": 1.0420831001305935,
      "grad_norm": 1.6602003856524954,
      "learning_rate": 2.3313114380435055e-06,
      "loss": 0.315,
      "step": 19550
    },
    {
      "epoch": 1.0426161349644198,
      "grad_norm": 1.6676535847675475,
      "learning_rate": 2.3305776524300666e-06,
      "loss": 0.3058,
      "step": 19560
    },
    {
      "epoch": 1.0431491697982462,
      "grad_norm": 1.8072315645482462,
      "learning_rate": 2.3298435800526565e-06,
      "loss": 0.3038,
      "step": 19570
    },
    {
      "epoch": 1.0436822046320726,
      "grad_norm": 1.7604445798192565,
      "learning_rate": 2.3291092211647193e-06,
      "loss": 0.3078,
      "step": 19580
    },
    {
      "epoch": 1.044215239465899,
      "grad_norm": 1.8702712758079347,
      "learning_rate": 2.328374576019799e-06,
      "loss": 0.3145,
      "step": 19590
    },
    {
      "epoch": 1.0447482742997254,
      "grad_norm": 1.4996388808956578,
      "learning_rate": 2.327639644871538e-06,
      "loss": 0.3096,
      "step": 19600
    },
    {
      "epoch": 1.0452813091335518,
      "grad_norm": 1.6064118336755877,
      "learning_rate": 2.3269044279736773e-06,
      "loss": 0.2993,
      "step": 19610
    },
    {
      "epoch": 1.0458143439673782,
      "grad_norm": 1.6946610163064237,
      "learning_rate": 2.3261689255800567e-06,
      "loss": 0.3162,
      "step": 19620
    },
    {
      "epoch": 1.0463473788012048,
      "grad_norm": 1.5171742988218002,
      "learning_rate": 2.325433137944615e-06,
      "loss": 0.3135,
      "step": 19630
    },
    {
      "epoch": 1.0468804136350311,
      "grad_norm": 1.681952371397534,
      "learning_rate": 2.324697065321389e-06,
      "loss": 0.311,
      "step": 19640
    },
    {
      "epoch": 1.0474134484688575,
      "grad_norm": 1.6643104263190116,
      "learning_rate": 2.323960707964514e-06,
      "loss": 0.3046,
      "step": 19650
    },
    {
      "epoch": 1.047946483302684,
      "grad_norm": 1.6210342981082129,
      "learning_rate": 2.3232240661282237e-06,
      "loss": 0.3028,
      "step": 19660
    },
    {
      "epoch": 1.0484795181365103,
      "grad_norm": 1.9197904912037602,
      "learning_rate": 2.3224871400668493e-06,
      "loss": 0.315,
      "step": 19670
    },
    {
      "epoch": 1.0490125529703367,
      "grad_norm": 1.7093300384936556,
      "learning_rate": 2.3217499300348213e-06,
      "loss": 0.3085,
      "step": 19680
    },
    {
      "epoch": 1.049545587804163,
      "grad_norm": 1.6590430355043468,
      "learning_rate": 2.3210124362866682e-06,
      "loss": 0.3079,
      "step": 19690
    },
    {
      "epoch": 1.0500786226379895,
      "grad_norm": 1.7439279113739472,
      "learning_rate": 2.320274659077015e-06,
      "loss": 0.316,
      "step": 19700
    },
    {
      "epoch": 1.0506116574718158,
      "grad_norm": 1.733935189703632,
      "learning_rate": 2.319536598660586e-06,
      "loss": 0.3076,
      "step": 19710
    },
    {
      "epoch": 1.0511446923056422,
      "grad_norm": 1.6608416566934736,
      "learning_rate": 2.3187982552922033e-06,
      "loss": 0.3063,
      "step": 19720
    },
    {
      "epoch": 1.0516777271394686,
      "grad_norm": 1.8158532269444727,
      "learning_rate": 2.3180596292267853e-06,
      "loss": 0.3189,
      "step": 19730
    },
    {
      "epoch": 1.052210761973295,
      "grad_norm": 1.764521685337587,
      "learning_rate": 2.317320720719349e-06,
      "loss": 0.3152,
      "step": 19740
    },
    {
      "epoch": 1.0527437968071214,
      "grad_norm": 1.756732379108052,
      "learning_rate": 2.31658153002501e-06,
      "loss": 0.3132,
      "step": 19750
    },
    {
      "epoch": 1.0532768316409478,
      "grad_norm": 1.537573893676739,
      "learning_rate": 2.3158420573989783e-06,
      "loss": 0.3108,
      "step": 19760
    },
    {
      "epoch": 1.0538098664747741,
      "grad_norm": 1.68342688875765,
      "learning_rate": 2.3151023030965644e-06,
      "loss": 0.3125,
      "step": 19770
    },
    {
      "epoch": 1.0543429013086005,
      "grad_norm": 1.7489867022687386,
      "learning_rate": 2.3143622673731746e-06,
      "loss": 0.3075,
      "step": 19780
    },
    {
      "epoch": 1.054875936142427,
      "grad_norm": 1.986931380085191,
      "learning_rate": 2.3136219504843122e-06,
      "loss": 0.3093,
      "step": 19790
    },
    {
      "epoch": 1.0554089709762533,
      "grad_norm": 1.5481125556801067,
      "learning_rate": 2.312881352685578e-06,
      "loss": 0.2974,
      "step": 19800
    },
    {
      "epoch": 1.0559420058100797,
      "grad_norm": 1.6705110735201283,
      "learning_rate": 2.3121404742326697e-06,
      "loss": 0.3091,
      "step": 19810
    },
    {
      "epoch": 1.056475040643906,
      "grad_norm": 1.5946087149307544,
      "learning_rate": 2.3113993153813814e-06,
      "loss": 0.3018,
      "step": 19820
    },
    {
      "epoch": 1.0570080754777325,
      "grad_norm": 1.4459937441736979,
      "learning_rate": 2.3106578763876055e-06,
      "loss": 0.3109,
      "step": 19830
    },
    {
      "epoch": 1.0575411103115588,
      "grad_norm": 1.7260057890441056,
      "learning_rate": 2.309916157507329e-06,
      "loss": 0.3067,
      "step": 19840
    },
    {
      "epoch": 1.0580741451453852,
      "grad_norm": 1.7412517385452415,
      "learning_rate": 2.309174158996637e-06,
      "loss": 0.3133,
      "step": 19850
    },
    {
      "epoch": 1.0586071799792116,
      "grad_norm": 1.7765175839440783,
      "learning_rate": 2.3084318811117115e-06,
      "loss": 0.3151,
      "step": 19860
    },
    {
      "epoch": 1.059140214813038,
      "grad_norm": 1.7349533089906835,
      "learning_rate": 2.307689324108829e-06,
      "loss": 0.2978,
      "step": 19870
    },
    {
      "epoch": 1.0596732496468644,
      "grad_norm": 1.6101318274152165,
      "learning_rate": 2.3069464882443643e-06,
      "loss": 0.31,
      "step": 19880
    },
    {
      "epoch": 1.0602062844806908,
      "grad_norm": 1.8430313961496845,
      "learning_rate": 2.306203373774788e-06,
      "loss": 0.3198,
      "step": 19890
    },
    {
      "epoch": 1.0607393193145171,
      "grad_norm": 1.5926851698217748,
      "learning_rate": 2.305459980956667e-06,
      "loss": 0.3249,
      "step": 19900
    },
    {
      "epoch": 1.0612723541483435,
      "grad_norm": 1.750694904810236,
      "learning_rate": 2.304716310046662e-06,
      "loss": 0.3136,
      "step": 19910
    },
    {
      "epoch": 1.06180538898217,
      "grad_norm": 1.6243073105838994,
      "learning_rate": 2.3039723613015335e-06,
      "loss": 0.3089,
      "step": 19920
    },
    {
      "epoch": 1.0623384238159963,
      "grad_norm": 1.5534089185294786,
      "learning_rate": 2.3032281349781355e-06,
      "loss": 0.3065,
      "step": 19930
    },
    {
      "epoch": 1.0628714586498227,
      "grad_norm": 1.824439581254945,
      "learning_rate": 2.302483631333419e-06,
      "loss": 0.3164,
      "step": 19940
    },
    {
      "epoch": 1.063404493483649,
      "grad_norm": 1.6199443334732788,
      "learning_rate": 2.3017388506244293e-06,
      "loss": 0.315,
      "step": 19950
    },
    {
      "epoch": 1.0639375283174755,
      "grad_norm": 1.6000992674164802,
      "learning_rate": 2.3009937931083083e-06,
      "loss": 0.323,
      "step": 19960
    },
    {
      "epoch": 1.0644705631513018,
      "grad_norm": 1.801001843022137,
      "learning_rate": 2.3002484590422946e-06,
      "loss": 0.3061,
      "step": 19970
    },
    {
      "epoch": 1.0650035979851282,
      "grad_norm": 1.5457269985803401,
      "learning_rate": 2.29950284868372e-06,
      "loss": 0.3172,
      "step": 19980
    },
    {
      "epoch": 1.0655366328189548,
      "grad_norm": 1.5899898759268671,
      "learning_rate": 2.2987569622900134e-06,
      "loss": 0.3076,
      "step": 19990
    },
    {
      "epoch": 1.0660696676527812,
      "grad_norm": 1.5605886239159608,
      "learning_rate": 2.2980108001186978e-06,
      "loss": 0.3123,
      "step": 20000
    },
    {
      "epoch": 1.0666027024866076,
      "grad_norm": 1.4545797018638003,
      "learning_rate": 2.297264362427392e-06,
      "loss": 0.3165,
      "step": 20010
    },
    {
      "epoch": 1.067135737320434,
      "grad_norm": 1.586730670029662,
      "learning_rate": 2.2965176494738107e-06,
      "loss": 0.316,
      "step": 20020
    },
    {
      "epoch": 1.0676687721542604,
      "grad_norm": 1.61818578088921,
      "learning_rate": 2.2957706615157625e-06,
      "loss": 0.3105,
      "step": 20030
    },
    {
      "epoch": 1.0682018069880868,
      "grad_norm": 1.6082334666108968,
      "learning_rate": 2.295023398811151e-06,
      "loss": 0.3069,
      "step": 20040
    },
    {
      "epoch": 1.0687348418219131,
      "grad_norm": 1.758159498075471,
      "learning_rate": 2.294275861617975e-06,
      "loss": 0.3024,
      "step": 20050
    },
    {
      "epoch": 1.0692678766557395,
      "grad_norm": 1.5892949055954335,
      "learning_rate": 2.2935280501943285e-06,
      "loss": 0.3193,
      "step": 20060
    },
    {
      "epoch": 1.069800911489566,
      "grad_norm": 1.6830657737207504,
      "learning_rate": 2.2927799647983998e-06,
      "loss": 0.3008,
      "step": 20070
    },
    {
      "epoch": 1.0703339463233923,
      "grad_norm": 1.6309903623130078,
      "learning_rate": 2.292031605688471e-06,
      "loss": 0.3082,
      "step": 20080
    },
    {
      "epoch": 1.0708669811572187,
      "grad_norm": 1.6944918118515484,
      "learning_rate": 2.29128297312292e-06,
      "loss": 0.3055,
      "step": 20090
    },
    {
      "epoch": 1.071400015991045,
      "grad_norm": 1.6289333482834578,
      "learning_rate": 2.2905340673602185e-06,
      "loss": 0.313,
      "step": 20100
    },
    {
      "epoch": 1.0719330508248714,
      "grad_norm": 1.6392344620667187,
      "learning_rate": 2.289784888658932e-06,
      "loss": 0.3005,
      "step": 20110
    },
    {
      "epoch": 1.0724660856586978,
      "grad_norm": 1.8524108200767249,
      "learning_rate": 2.289035437277722e-06,
      "loss": 0.3152,
      "step": 20120
    },
    {
      "epoch": 1.0729991204925242,
      "grad_norm": 1.5796125723740158,
      "learning_rate": 2.2882857134753417e-06,
      "loss": 0.3262,
      "step": 20130
    },
    {
      "epoch": 1.0735321553263506,
      "grad_norm": 1.753728939535228,
      "learning_rate": 2.2875357175106397e-06,
      "loss": 0.3121,
      "step": 20140
    },
    {
      "epoch": 1.074065190160177,
      "grad_norm": 1.5623208561509776,
      "learning_rate": 2.2867854496425595e-06,
      "loss": 0.322,
      "step": 20150
    },
    {
      "epoch": 1.0745982249940034,
      "grad_norm": 1.5325756892143907,
      "learning_rate": 2.286034910130136e-06,
      "loss": 0.3021,
      "step": 20160
    },
    {
      "epoch": 1.0751312598278298,
      "grad_norm": 1.7761690651214683,
      "learning_rate": 2.285284099232501e-06,
      "loss": 0.2988,
      "step": 20170
    },
    {
      "epoch": 1.0756642946616561,
      "grad_norm": 1.5483310944763116,
      "learning_rate": 2.284533017208877e-06,
      "loss": 0.3045,
      "step": 20180
    },
    {
      "epoch": 1.0761973294954825,
      "grad_norm": 1.5443949582015404,
      "learning_rate": 2.283781664318582e-06,
      "loss": 0.3164,
      "step": 20190
    },
    {
      "epoch": 1.076730364329309,
      "grad_norm": 1.6396497040544773,
      "learning_rate": 2.2830300408210266e-06,
      "loss": 0.3058,
      "step": 20200
    },
    {
      "epoch": 1.0772633991631353,
      "grad_norm": 1.554129101309212,
      "learning_rate": 2.2822781469757158e-06,
      "loss": 0.3134,
      "step": 20210
    },
    {
      "epoch": 1.0777964339969617,
      "grad_norm": 1.7526946278918778,
      "learning_rate": 2.281525983042246e-06,
      "loss": 0.3017,
      "step": 20220
    },
    {
      "epoch": 1.078329468830788,
      "grad_norm": 1.744192157689558,
      "learning_rate": 2.28077354928031e-06,
      "loss": 0.3051,
      "step": 20230
    },
    {
      "epoch": 1.0788625036646144,
      "grad_norm": 1.4976672640563824,
      "learning_rate": 2.2800208459496907e-06,
      "loss": 0.3107,
      "step": 20240
    },
    {
      "epoch": 1.0793955384984408,
      "grad_norm": 1.6659791336010934,
      "learning_rate": 2.279267873310266e-06,
      "loss": 0.3091,
      "step": 20250
    },
    {
      "epoch": 1.0799285733322672,
      "grad_norm": 1.5970576389357873,
      "learning_rate": 2.2785146316220048e-06,
      "loss": 0.3143,
      "step": 20260
    },
    {
      "epoch": 1.0804616081660936,
      "grad_norm": 1.6284122566142822,
      "learning_rate": 2.2777611211449713e-06,
      "loss": 0.3177,
      "step": 20270
    },
    {
      "epoch": 1.08099464299992,
      "grad_norm": 1.6316189260615348,
      "learning_rate": 2.2770073421393214e-06,
      "loss": 0.2989,
      "step": 20280
    },
    {
      "epoch": 1.0815276778337464,
      "grad_norm": 1.5986144293097817,
      "learning_rate": 2.2762532948653033e-06,
      "loss": 0.317,
      "step": 20290
    },
    {
      "epoch": 1.0820607126675728,
      "grad_norm": 1.5181692529740152,
      "learning_rate": 2.2754989795832587e-06,
      "loss": 0.3162,
      "step": 20300
    },
    {
      "epoch": 1.0825937475013991,
      "grad_norm": 1.627712541610012,
      "learning_rate": 2.274744396553621e-06,
      "loss": 0.3043,
      "step": 20310
    },
    {
      "epoch": 1.0831267823352255,
      "grad_norm": 1.8331087744547756,
      "learning_rate": 2.2739895460369155e-06,
      "loss": 0.3184,
      "step": 20320
    },
    {
      "epoch": 1.083659817169052,
      "grad_norm": 1.6356187656274184,
      "learning_rate": 2.273234428293762e-06,
      "loss": 0.3033,
      "step": 20330
    },
    {
      "epoch": 1.0841928520028783,
      "grad_norm": 1.489854250295549,
      "learning_rate": 2.2724790435848718e-06,
      "loss": 0.3063,
      "step": 20340
    },
    {
      "epoch": 1.0847258868367047,
      "grad_norm": 1.4351976874950956,
      "learning_rate": 2.271723392171046e-06,
      "loss": 0.3143,
      "step": 20350
    },
    {
      "epoch": 1.085258921670531,
      "grad_norm": 1.703422703327441,
      "learning_rate": 2.2709674743131806e-06,
      "loss": 0.3151,
      "step": 20360
    },
    {
      "epoch": 1.0857919565043574,
      "grad_norm": 1.7045448455070467,
      "learning_rate": 2.270211290272263e-06,
      "loss": 0.3087,
      "step": 20370
    },
    {
      "epoch": 1.086324991338184,
      "grad_norm": 1.7436578222576975,
      "learning_rate": 2.2694548403093717e-06,
      "loss": 0.3244,
      "step": 20380
    },
    {
      "epoch": 1.0868580261720104,
      "grad_norm": 1.6890954265416962,
      "learning_rate": 2.2686981246856778e-06,
      "loss": 0.3111,
      "step": 20390
    },
    {
      "epoch": 1.0873910610058368,
      "grad_norm": 1.7061659818947192,
      "learning_rate": 2.267941143662443e-06,
      "loss": 0.3133,
      "step": 20400
    },
    {
      "epoch": 1.0879240958396632,
      "grad_norm": 1.5839361067641207,
      "learning_rate": 2.2671838975010225e-06,
      "loss": 0.3042,
      "step": 20410
    },
    {
      "epoch": 1.0884571306734896,
      "grad_norm": 1.5942352715656116,
      "learning_rate": 2.266426386462862e-06,
      "loss": 0.3103,
      "step": 20420
    },
    {
      "epoch": 1.088990165507316,
      "grad_norm": 1.6642037967323897,
      "learning_rate": 2.2656686108094967e-06,
      "loss": 0.3032,
      "step": 20430
    },
    {
      "epoch": 1.0895232003411424,
      "grad_norm": 1.7016467213703885,
      "learning_rate": 2.264910570802557e-06,
      "loss": 0.305,
      "step": 20440
    },
    {
      "epoch": 1.0900562351749687,
      "grad_norm": 1.7196687873703265,
      "learning_rate": 2.264152266703762e-06,
      "loss": 0.3137,
      "step": 20450
    },
    {
      "epoch": 1.0905892700087951,
      "grad_norm": 1.6065413791228573,
      "learning_rate": 2.263393698774923e-06,
      "loss": 0.3119,
      "step": 20460
    },
    {
      "epoch": 1.0911223048426215,
      "grad_norm": 1.8610955070531345,
      "learning_rate": 2.2626348672779413e-06,
      "loss": 0.3163,
      "step": 20470
    },
    {
      "epoch": 1.091655339676448,
      "grad_norm": 1.4764983834948127,
      "learning_rate": 2.26187577247481e-06,
      "loss": 0.3091,
      "step": 20480
    },
    {
      "epoch": 1.0921883745102743,
      "grad_norm": 1.6021245669401387,
      "learning_rate": 2.261116414627613e-06,
      "loss": 0.3219,
      "step": 20490
    },
    {
      "epoch": 1.0927214093441007,
      "grad_norm": 1.6533767859742208,
      "learning_rate": 2.2603567939985263e-06,
      "loss": 0.2912,
      "step": 20500
    },
    {
      "epoch": 1.093254444177927,
      "grad_norm": 1.7035475743547959,
      "learning_rate": 2.259596910849814e-06,
      "loss": 0.3035,
      "step": 20510
    },
    {
      "epoch": 1.0937874790117534,
      "grad_norm": 1.6508501878264963,
      "learning_rate": 2.2588367654438323e-06,
      "loss": 0.2991,
      "step": 20520
    },
    {
      "epoch": 1.0943205138455798,
      "grad_norm": 1.5718945916959757,
      "learning_rate": 2.2580763580430285e-06,
      "loss": 0.3087,
      "step": 20530
    },
    {
      "epoch": 1.0948535486794062,
      "grad_norm": 1.6251940722322025,
      "learning_rate": 2.2573156889099384e-06,
      "loss": 0.3073,
      "step": 20540
    },
    {
      "epoch": 1.0953865835132326,
      "grad_norm": 1.653377268999388,
      "learning_rate": 2.2565547583071914e-06,
      "loss": 0.3101,
      "step": 20550
    },
    {
      "epoch": 1.095919618347059,
      "grad_norm": 1.75462182597595,
      "learning_rate": 2.2557935664975033e-06,
      "loss": 0.307,
      "step": 20560
    },
    {
      "epoch": 1.0964526531808854,
      "grad_norm": 1.5769222679718922,
      "learning_rate": 2.2550321137436834e-06,
      "loss": 0.2991,
      "step": 20570
    },
    {
      "epoch": 1.0969856880147117,
      "grad_norm": 1.5746337490387257,
      "learning_rate": 2.254270400308629e-06,
      "loss": 0.3212,
      "step": 20580
    },
    {
      "epoch": 1.0975187228485381,
      "grad_norm": 1.6480521242495094,
      "learning_rate": 2.253508426455329e-06,
      "loss": 0.3067,
      "step": 20590
    },
    {
      "epoch": 1.0980517576823645,
      "grad_norm": 1.7058446812952348,
      "learning_rate": 2.25274619244686e-06,
      "loss": 0.3131,
      "step": 20600
    },
    {
      "epoch": 1.098584792516191,
      "grad_norm": 1.7460479668951798,
      "learning_rate": 2.25198369854639e-06,
      "loss": 0.3067,
      "step": 20610
    },
    {
      "epoch": 1.0991178273500173,
      "grad_norm": 1.8632265859307715,
      "learning_rate": 2.251220945017178e-06,
      "loss": 0.3141,
      "step": 20620
    },
    {
      "epoch": 1.0996508621838437,
      "grad_norm": 1.8075011489080604,
      "learning_rate": 2.2504579321225696e-06,
      "loss": 0.3167,
      "step": 20630
    },
    {
      "epoch": 1.10018389701767,
      "grad_norm": 1.7099966301645766,
      "learning_rate": 2.2496946601260016e-06,
      "loss": 0.2998,
      "step": 20640
    },
    {
      "epoch": 1.1007169318514964,
      "grad_norm": 1.7027089587326318,
      "learning_rate": 2.2489311292910006e-06,
      "loss": 0.3033,
      "step": 20650
    },
    {
      "epoch": 1.1012499666853228,
      "grad_norm": 1.5919803688603003,
      "learning_rate": 2.2481673398811816e-06,
      "loss": 0.3038,
      "step": 20660
    },
    {
      "epoch": 1.1017830015191492,
      "grad_norm": 1.6024281162470557,
      "learning_rate": 2.24740329216025e-06,
      "loss": 0.3034,
      "step": 20670
    },
    {
      "epoch": 1.1023160363529756,
      "grad_norm": 1.6126467574543146,
      "learning_rate": 2.2466389863919996e-06,
      "loss": 0.3205,
      "step": 20680
    },
    {
      "epoch": 1.102849071186802,
      "grad_norm": 1.7827413541414465,
      "learning_rate": 2.2458744228403127e-06,
      "loss": 0.317,
      "step": 20690
    },
    {
      "epoch": 1.1033821060206284,
      "grad_norm": 1.7144444470739588,
      "learning_rate": 2.245109601769162e-06,
      "loss": 0.3227,
      "step": 20700
    },
    {
      "epoch": 1.1039151408544547,
      "grad_norm": 1.8014751200955856,
      "learning_rate": 2.2443445234426083e-06,
      "loss": 0.303,
      "step": 20710
    },
    {
      "epoch": 1.1044481756882811,
      "grad_norm": 1.5914755244129393,
      "learning_rate": 2.2435791881248017e-06,
      "loss": 0.3178,
      "step": 20720
    },
    {
      "epoch": 1.1049812105221077,
      "grad_norm": 1.5137246111318892,
      "learning_rate": 2.242813596079981e-06,
      "loss": 0.3102,
      "step": 20730
    },
    {
      "epoch": 1.1055142453559341,
      "grad_norm": 1.6801762479589506,
      "learning_rate": 2.2420477475724714e-06,
      "loss": 0.3113,
      "step": 20740
    },
    {
      "epoch": 1.1060472801897605,
      "grad_norm": 1.5590419708036913,
      "learning_rate": 2.241281642866691e-06,
      "loss": 0.3117,
      "step": 20750
    },
    {
      "epoch": 1.106580315023587,
      "grad_norm": 1.7898995494588794,
      "learning_rate": 2.240515282227142e-06,
      "loss": 0.3075,
      "step": 20760
    },
    {
      "epoch": 1.1071133498574133,
      "grad_norm": 1.6834613055252743,
      "learning_rate": 2.2397486659184185e-06,
      "loss": 0.303,
      "step": 20770
    },
    {
      "epoch": 1.1076463846912397,
      "grad_norm": 1.6889123149883707,
      "learning_rate": 2.2389817942052004e-06,
      "loss": 0.3089,
      "step": 20780
    },
    {
      "epoch": 1.108179419525066,
      "grad_norm": 1.7293717410097444,
      "learning_rate": 2.238214667352257e-06,
      "loss": 0.3155,
      "step": 20790
    },
    {
      "epoch": 1.1087124543588924,
      "grad_norm": 1.8093789039301587,
      "learning_rate": 2.237447285624445e-06,
      "loss": 0.2983,
      "step": 20800
    },
    {
      "epoch": 1.1092454891927188,
      "grad_norm": 1.5501234210844848,
      "learning_rate": 2.2366796492867095e-06,
      "loss": 0.3053,
      "step": 20810
    },
    {
      "epoch": 1.1097785240265452,
      "grad_norm": 1.5685323574540726,
      "learning_rate": 2.2359117586040837e-06,
      "loss": 0.3109,
      "step": 20820
    },
    {
      "epoch": 1.1103115588603716,
      "grad_norm": 1.7710444058403545,
      "learning_rate": 2.2351436138416883e-06,
      "loss": 0.3161,
      "step": 20830
    },
    {
      "epoch": 1.110844593694198,
      "grad_norm": 1.6265566997745051,
      "learning_rate": 2.2343752152647314e-06,
      "loss": 0.298,
      "step": 20840
    },
    {
      "epoch": 1.1113776285280244,
      "grad_norm": 1.7343040660344997,
      "learning_rate": 2.2336065631385097e-06,
      "loss": 0.3221,
      "step": 20850
    },
    {
      "epoch": 1.1119106633618507,
      "grad_norm": 1.470688582654047,
      "learning_rate": 2.2328376577284066e-06,
      "loss": 0.3083,
      "step": 20860
    },
    {
      "epoch": 1.1124436981956771,
      "grad_norm": 1.4296680219315556,
      "learning_rate": 2.232068499299893e-06,
      "loss": 0.3165,
      "step": 20870
    },
    {
      "epoch": 1.1129767330295035,
      "grad_norm": 1.676832282763617,
      "learning_rate": 2.2312990881185277e-06,
      "loss": 0.2973,
      "step": 20880
    },
    {
      "epoch": 1.11350976786333,
      "grad_norm": 1.774768679716971,
      "learning_rate": 2.2305294244499563e-06,
      "loss": 0.3058,
      "step": 20890
    },
    {
      "epoch": 1.1140428026971563,
      "grad_norm": 1.7799279652979498,
      "learning_rate": 2.229759508559912e-06,
      "loss": 0.3076,
      "step": 20900
    },
    {
      "epoch": 1.1145758375309827,
      "grad_norm": 1.5867822428300116,
      "learning_rate": 2.228989340714214e-06,
      "loss": 0.3138,
      "step": 20910
    },
    {
      "epoch": 1.115108872364809,
      "grad_norm": 1.5487436829033223,
      "learning_rate": 2.2282189211787697e-06,
      "loss": 0.2968,
      "step": 20920
    },
    {
      "epoch": 1.1156419071986354,
      "grad_norm": 1.6873803592953087,
      "learning_rate": 2.2274482502195733e-06,
      "loss": 0.3136,
      "step": 20930
    },
    {
      "epoch": 1.1161749420324618,
      "grad_norm": 1.559447683810378,
      "learning_rate": 2.2266773281027054e-06,
      "loss": 0.3095,
      "step": 20940
    },
    {
      "epoch": 1.1167079768662882,
      "grad_norm": 1.775421677185727,
      "learning_rate": 2.225906155094333e-06,
      "loss": 0.318,
      "step": 20950
    },
    {
      "epoch": 1.1172410117001146,
      "grad_norm": 1.7386104982302553,
      "learning_rate": 2.22513473146071e-06,
      "loss": 0.3059,
      "step": 20960
    },
    {
      "epoch": 1.117774046533941,
      "grad_norm": 1.7534474388667118,
      "learning_rate": 2.2243630574681775e-06,
      "loss": 0.2964,
      "step": 20970
    },
    {
      "epoch": 1.1183070813677674,
      "grad_norm": 1.7102947030492215,
      "learning_rate": 2.2235911333831627e-06,
      "loss": 0.3003,
      "step": 20980
    },
    {
      "epoch": 1.1188401162015937,
      "grad_norm": 1.609490185760629,
      "learning_rate": 2.222818959472178e-06,
      "loss": 0.3153,
      "step": 20990
    },
    {
      "epoch": 1.1193731510354201,
      "grad_norm": 1.5880032628420742,
      "learning_rate": 2.222046536001823e-06,
      "loss": 0.3091,
      "step": 21000
    },
    {
      "epoch": 1.1199061858692465,
      "grad_norm": 1.6401802227458995,
      "learning_rate": 2.2212738632387845e-06,
      "loss": 0.3094,
      "step": 21010
    },
    {
      "epoch": 1.120439220703073,
      "grad_norm": 1.8675617159676652,
      "learning_rate": 2.220500941449834e-06,
      "loss": 0.3127,
      "step": 21020
    },
    {
      "epoch": 1.1209722555368993,
      "grad_norm": 1.7019986096359507,
      "learning_rate": 2.2197277709018286e-06,
      "loss": 0.3084,
      "step": 21030
    },
    {
      "epoch": 1.1215052903707257,
      "grad_norm": 1.80160179780082,
      "learning_rate": 2.2189543518617124e-06,
      "loss": 0.3103,
      "step": 21040
    },
    {
      "epoch": 1.122038325204552,
      "grad_norm": 1.491361860680009,
      "learning_rate": 2.218180684596515e-06,
      "loss": 0.3039,
      "step": 21050
    },
    {
      "epoch": 1.1225713600383784,
      "grad_norm": 1.742841298640953,
      "learning_rate": 2.217406769373351e-06,
      "loss": 0.299,
      "step": 21060
    },
    {
      "epoch": 1.1231043948722048,
      "grad_norm": 1.7785023317108535,
      "learning_rate": 2.216632606459422e-06,
      "loss": 0.3137,
      "step": 21070
    },
    {
      "epoch": 1.1236374297060312,
      "grad_norm": 1.5783916705654453,
      "learning_rate": 2.215858196122014e-06,
      "loss": 0.3125,
      "step": 21080
    },
    {
      "epoch": 1.1241704645398576,
      "grad_norm": 1.577594442437047,
      "learning_rate": 2.215083538628499e-06,
      "loss": 0.3166,
      "step": 21090
    },
    {
      "epoch": 1.124703499373684,
      "grad_norm": 1.7045582673880306,
      "learning_rate": 2.214308634246333e-06,
      "loss": 0.3101,
      "step": 21100
    },
    {
      "epoch": 1.1252365342075104,
      "grad_norm": 1.6790255485034211,
      "learning_rate": 2.21353348324306e-06,
      "loss": 0.3064,
      "step": 21110
    },
    {
      "epoch": 1.1257695690413367,
      "grad_norm": 1.6277292643713521,
      "learning_rate": 2.2127580858863057e-06,
      "loss": 0.3139,
      "step": 21120
    },
    {
      "epoch": 1.1263026038751631,
      "grad_norm": 1.6873977006034813,
      "learning_rate": 2.2119824424437836e-06,
      "loss": 0.3153,
      "step": 21130
    },
    {
      "epoch": 1.1268356387089897,
      "grad_norm": 1.728680549544708,
      "learning_rate": 2.211206553183291e-06,
      "loss": 0.3086,
      "step": 21140
    },
    {
      "epoch": 1.1273686735428161,
      "grad_norm": 1.632470745754458,
      "learning_rate": 2.2104304183727102e-06,
      "loss": 0.3076,
      "step": 21150
    },
    {
      "epoch": 1.1279017083766425,
      "grad_norm": 1.7190508575469043,
      "learning_rate": 2.2096540382800085e-06,
      "loss": 0.3162,
      "step": 21160
    },
    {
      "epoch": 1.1284347432104689,
      "grad_norm": 1.8658650127720948,
      "learning_rate": 2.2088774131732376e-06,
      "loss": 0.3135,
      "step": 21170
    },
    {
      "epoch": 1.1289677780442953,
      "grad_norm": 1.5545140142065954,
      "learning_rate": 2.2081005433205334e-06,
      "loss": 0.3149,
      "step": 21180
    },
    {
      "epoch": 1.1295008128781217,
      "grad_norm": 1.6608493215939488,
      "learning_rate": 2.207323428990118e-06,
      "loss": 0.3155,
      "step": 21190
    },
    {
      "epoch": 1.130033847711948,
      "grad_norm": 1.70239514678641,
      "learning_rate": 2.2065460704502954e-06,
      "loss": 0.315,
      "step": 21200
    },
    {
      "epoch": 1.1305668825457744,
      "grad_norm": 1.6496785874268263,
      "learning_rate": 2.2057684679694557e-06,
      "loss": 0.3025,
      "step": 21210
    },
    {
      "epoch": 1.1310999173796008,
      "grad_norm": 1.7550833561227488,
      "learning_rate": 2.2049906218160724e-06,
      "loss": 0.3075,
      "step": 21220
    },
    {
      "epoch": 1.1316329522134272,
      "grad_norm": 1.64706352637559,
      "learning_rate": 2.2042125322587048e-06,
      "loss": 0.3103,
      "step": 21230
    },
    {
      "epoch": 1.1321659870472536,
      "grad_norm": 1.687132715132537,
      "learning_rate": 2.2034341995659938e-06,
      "loss": 0.3158,
      "step": 21240
    },
    {
      "epoch": 1.13269902188108,
      "grad_norm": 1.6543614390022436,
      "learning_rate": 2.202655624006665e-06,
      "loss": 0.2985,
      "step": 21250
    },
    {
      "epoch": 1.1332320567149063,
      "grad_norm": 1.69337061810349,
      "learning_rate": 2.2018768058495292e-06,
      "loss": 0.3066,
      "step": 21260
    },
    {
      "epoch": 1.1337650915487327,
      "grad_norm": 1.5222096149297102,
      "learning_rate": 2.2010977453634794e-06,
      "loss": 0.3193,
      "step": 21270
    },
    {
      "epoch": 1.1342981263825591,
      "grad_norm": 1.622855621916122,
      "learning_rate": 2.2003184428174932e-06,
      "loss": 0.3034,
      "step": 21280
    },
    {
      "epoch": 1.1348311612163855,
      "grad_norm": 1.6956987624602966,
      "learning_rate": 2.1995388984806308e-06,
      "loss": 0.296,
      "step": 21290
    },
    {
      "epoch": 1.1353641960502119,
      "grad_norm": 1.6649515239504777,
      "learning_rate": 2.1987591126220368e-06,
      "loss": 0.3148,
      "step": 21300
    },
    {
      "epoch": 1.1358972308840383,
      "grad_norm": 1.8845503856479293,
      "learning_rate": 2.1979790855109394e-06,
      "loss": 0.3154,
      "step": 21310
    },
    {
      "epoch": 1.1364302657178647,
      "grad_norm": 1.6307121195488599,
      "learning_rate": 2.1971988174166485e-06,
      "loss": 0.3063,
      "step": 21320
    },
    {
      "epoch": 1.136963300551691,
      "grad_norm": 1.6593939526643333,
      "learning_rate": 2.196418308608559e-06,
      "loss": 0.31,
      "step": 21330
    },
    {
      "epoch": 1.1374963353855174,
      "grad_norm": 1.497105604345187,
      "learning_rate": 2.195637559356149e-06,
      "loss": 0.3192,
      "step": 21340
    },
    {
      "epoch": 1.1380293702193438,
      "grad_norm": 1.558046444444482,
      "learning_rate": 2.194856569928977e-06,
      "loss": 0.2963,
      "step": 21350
    },
    {
      "epoch": 1.1385624050531702,
      "grad_norm": 1.7640136875409804,
      "learning_rate": 2.194075340596687e-06,
      "loss": 0.3052,
      "step": 21360
    },
    {
      "epoch": 1.1390954398869966,
      "grad_norm": 1.5952577900721743,
      "learning_rate": 2.1932938716290055e-06,
      "loss": 0.3113,
      "step": 21370
    },
    {
      "epoch": 1.139628474720823,
      "grad_norm": 1.695727652741668,
      "learning_rate": 2.1925121632957406e-06,
      "loss": 0.3027,
      "step": 21380
    },
    {
      "epoch": 1.1401615095546493,
      "grad_norm": 1.5878130950883684,
      "learning_rate": 2.1917302158667838e-06,
      "loss": 0.3124,
      "step": 21390
    },
    {
      "epoch": 1.1406945443884757,
      "grad_norm": 1.7535662725272476,
      "learning_rate": 2.1909480296121097e-06,
      "loss": 0.3134,
      "step": 21400
    },
    {
      "epoch": 1.1412275792223021,
      "grad_norm": 1.7125827345330777,
      "learning_rate": 2.1901656048017742e-06,
      "loss": 0.3116,
      "step": 21410
    },
    {
      "epoch": 1.1417606140561285,
      "grad_norm": 1.480422649066469,
      "learning_rate": 2.189382941705916e-06,
      "loss": 0.3097,
      "step": 21420
    },
    {
      "epoch": 1.1422936488899549,
      "grad_norm": 1.8264689307430122,
      "learning_rate": 2.1886000405947567e-06,
      "loss": 0.3134,
      "step": 21430
    },
    {
      "epoch": 1.1428266837237813,
      "grad_norm": 1.641378088819736,
      "learning_rate": 2.187816901738599e-06,
      "loss": 0.3097,
      "step": 21440
    },
    {
      "epoch": 1.1433597185576077,
      "grad_norm": 1.6014079320825023,
      "learning_rate": 2.1870335254078285e-06,
      "loss": 0.3103,
      "step": 21450
    },
    {
      "epoch": 1.143892753391434,
      "grad_norm": 1.8008018324362969,
      "learning_rate": 2.1862499118729126e-06,
      "loss": 0.3124,
      "step": 21460
    },
    {
      "epoch": 1.1444257882252606,
      "grad_norm": 1.71813725982525,
      "learning_rate": 2.1854660614044003e-06,
      "loss": 0.3024,
      "step": 21470
    },
    {
      "epoch": 1.144958823059087,
      "grad_norm": 1.7574964653927196,
      "learning_rate": 2.1846819742729226e-06,
      "loss": 0.3092,
      "step": 21480
    },
    {
      "epoch": 1.1454918578929134,
      "grad_norm": 1.7483329992033536,
      "learning_rate": 2.1838976507491923e-06,
      "loss": 0.3068,
      "step": 21490
    },
    {
      "epoch": 1.1460248927267398,
      "grad_norm": 1.5211470719893452,
      "learning_rate": 2.183113091104004e-06,
      "loss": 0.3101,
      "step": 21500
    },
    {
      "epoch": 1.1465579275605662,
      "grad_norm": 1.7043341464293227,
      "learning_rate": 2.182328295608233e-06,
      "loss": 0.3126,
      "step": 21510
    },
    {
      "epoch": 1.1470909623943926,
      "grad_norm": 1.60302367740354,
      "learning_rate": 2.181543264532837e-06,
      "loss": 0.3122,
      "step": 21520
    },
    {
      "epoch": 1.147623997228219,
      "grad_norm": 1.6727570095138387,
      "learning_rate": 2.1807579981488544e-06,
      "loss": 0.3089,
      "step": 21530
    },
    {
      "epoch": 1.1481570320620453,
      "grad_norm": 1.5566991037908373,
      "learning_rate": 2.1799724967274057e-06,
      "loss": 0.3086,
      "step": 21540
    },
    {
      "epoch": 1.1486900668958717,
      "grad_norm": 1.5842030034937153,
      "learning_rate": 2.179186760539691e-06,
      "loss": 0.3075,
      "step": 21550
    },
    {
      "epoch": 1.149223101729698,
      "grad_norm": 1.5494798850814615,
      "learning_rate": 2.178400789856993e-06,
      "loss": 0.3087,
      "step": 21560
    },
    {
      "epoch": 1.1497561365635245,
      "grad_norm": 1.5190822090674974,
      "learning_rate": 2.1776145849506747e-06,
      "loss": 0.3242,
      "step": 21570
    },
    {
      "epoch": 1.1502891713973509,
      "grad_norm": 1.7427207164534948,
      "learning_rate": 2.17682814609218e-06,
      "loss": 0.3087,
      "step": 21580
    },
    {
      "epoch": 1.1508222062311773,
      "grad_norm": 1.6870435612018546,
      "learning_rate": 2.176041473553033e-06,
      "loss": 0.3117,
      "step": 21590
    },
    {
      "epoch": 1.1513552410650036,
      "grad_norm": 1.6056465538593587,
      "learning_rate": 2.1752545676048393e-06,
      "loss": 0.3053,
      "step": 21600
    },
    {
      "epoch": 1.15188827589883,
      "grad_norm": 1.5915232972592372,
      "learning_rate": 2.174467428519285e-06,
      "loss": 0.2988,
      "step": 21610
    },
    {
      "epoch": 1.1524213107326564,
      "grad_norm": 1.8515770996618846,
      "learning_rate": 2.1736800565681373e-06,
      "loss": 0.3082,
      "step": 21620
    },
    {
      "epoch": 1.1529543455664828,
      "grad_norm": 1.604787640476992,
      "learning_rate": 2.172892452023242e-06,
      "loss": 0.316,
      "step": 21630
    },
    {
      "epoch": 1.1534873804003092,
      "grad_norm": 1.6983919866739992,
      "learning_rate": 2.1721046151565264e-06,
      "loss": 0.303,
      "step": 21640
    },
    {
      "epoch": 1.1540204152341356,
      "grad_norm": 1.4715984036168932,
      "learning_rate": 2.171316546239998e-06,
      "loss": 0.3044,
      "step": 21650
    },
    {
      "epoch": 1.154553450067962,
      "grad_norm": 1.553740217520958,
      "learning_rate": 2.1705282455457443e-06,
      "loss": 0.2974,
      "step": 21660
    },
    {
      "epoch": 1.1550864849017883,
      "grad_norm": 1.7916951176843285,
      "learning_rate": 2.169739713345933e-06,
      "loss": 0.3097,
      "step": 21670
    },
    {
      "epoch": 1.1556195197356147,
      "grad_norm": 1.9372575436666049,
      "learning_rate": 2.168950949912811e-06,
      "loss": 0.301,
      "step": 21680
    },
    {
      "epoch": 1.156152554569441,
      "grad_norm": 1.7182848690242787,
      "learning_rate": 2.1681619555187056e-06,
      "loss": 0.3183,
      "step": 21690
    },
    {
      "epoch": 1.1566855894032675,
      "grad_norm": 1.639511494870053,
      "learning_rate": 2.1673727304360248e-06,
      "loss": 0.3087,
      "step": 21700
    },
    {
      "epoch": 1.1572186242370939,
      "grad_norm": 1.8034293067921714,
      "learning_rate": 2.1665832749372545e-06,
      "loss": 0.2964,
      "step": 21710
    },
    {
      "epoch": 1.1577516590709203,
      "grad_norm": 1.5385832132752124,
      "learning_rate": 2.16579358929496e-06,
      "loss": 0.31,
      "step": 21720
    },
    {
      "epoch": 1.1582846939047466,
      "grad_norm": 1.7144540520139608,
      "learning_rate": 2.1650036737817887e-06,
      "loss": 0.2997,
      "step": 21730
    },
    {
      "epoch": 1.158817728738573,
      "grad_norm": 1.5596836257732085,
      "learning_rate": 2.1642135286704646e-06,
      "loss": 0.3057,
      "step": 21740
    },
    {
      "epoch": 1.1593507635723994,
      "grad_norm": 1.634680174118517,
      "learning_rate": 2.163423154233792e-06,
      "loss": 0.3068,
      "step": 21750
    },
    {
      "epoch": 1.1598837984062258,
      "grad_norm": 1.5657800662297066,
      "learning_rate": 2.1626325507446546e-06,
      "loss": 0.3061,
      "step": 21760
    },
    {
      "epoch": 1.1604168332400522,
      "grad_norm": 1.744269865003573,
      "learning_rate": 2.161841718476015e-06,
      "loss": 0.3111,
      "step": 21770
    },
    {
      "epoch": 1.1609498680738786,
      "grad_norm": 1.6058667266676994,
      "learning_rate": 2.1610506577009143e-06,
      "loss": 0.3144,
      "step": 21780
    },
    {
      "epoch": 1.161482902907705,
      "grad_norm": 1.6710651992799488,
      "learning_rate": 2.160259368692474e-06,
      "loss": 0.2988,
      "step": 21790
    },
    {
      "epoch": 1.1620159377415313,
      "grad_norm": 1.9227312302736368,
      "learning_rate": 2.159467851723892e-06,
      "loss": 0.3072,
      "step": 21800
    },
    {
      "epoch": 1.1625489725753577,
      "grad_norm": 1.7040728840656527,
      "learning_rate": 2.158676107068447e-06,
      "loss": 0.3066,
      "step": 21810
    },
    {
      "epoch": 1.163082007409184,
      "grad_norm": 1.6147110825363211,
      "learning_rate": 2.157884134999495e-06,
      "loss": 0.3152,
      "step": 21820
    },
    {
      "epoch": 1.1636150422430105,
      "grad_norm": 1.571570932644986,
      "learning_rate": 2.1570919357904716e-06,
      "loss": 0.3094,
      "step": 21830
    },
    {
      "epoch": 1.1641480770768369,
      "grad_norm": 1.6328097280554397,
      "learning_rate": 2.156299509714891e-06,
      "loss": 0.3095,
      "step": 21840
    },
    {
      "epoch": 1.1646811119106633,
      "grad_norm": 1.7166939795346328,
      "learning_rate": 2.1555068570463434e-06,
      "loss": 0.3124,
      "step": 21850
    },
    {
      "epoch": 1.1652141467444896,
      "grad_norm": 1.6741991831516585,
      "learning_rate": 2.1547139780585e-06,
      "loss": 0.2976,
      "step": 21860
    },
    {
      "epoch": 1.165747181578316,
      "grad_norm": 1.7286356505487748,
      "learning_rate": 2.1539208730251082e-06,
      "loss": 0.3059,
      "step": 21870
    },
    {
      "epoch": 1.1662802164121424,
      "grad_norm": 1.5976550217520216,
      "learning_rate": 2.153127542219995e-06,
      "loss": 0.3141,
      "step": 21880
    },
    {
      "epoch": 1.166813251245969,
      "grad_norm": 1.7611518622741738,
      "learning_rate": 2.152333985917064e-06,
      "loss": 0.3241,
      "step": 21890
    },
    {
      "epoch": 1.1673462860797954,
      "grad_norm": 1.8151169172675925,
      "learning_rate": 2.1515402043902973e-06,
      "loss": 0.3066,
      "step": 21900
    },
    {
      "epoch": 1.1678793209136218,
      "grad_norm": 1.6553179755714715,
      "learning_rate": 2.1507461979137556e-06,
      "loss": 0.3118,
      "step": 21910
    },
    {
      "epoch": 1.1684123557474482,
      "grad_norm": 1.5655280372293336,
      "learning_rate": 2.149951966761575e-06,
      "loss": 0.3062,
      "step": 21920
    },
    {
      "epoch": 1.1689453905812746,
      "grad_norm": 1.6644324250326497,
      "learning_rate": 2.149157511207971e-06,
      "loss": 0.2993,
      "step": 21930
    },
    {
      "epoch": 1.169478425415101,
      "grad_norm": 1.8397480389823289,
      "learning_rate": 2.1483628315272365e-06,
      "loss": 0.3029,
      "step": 21940
    },
    {
      "epoch": 1.1700114602489273,
      "grad_norm": 1.8807998797665644,
      "learning_rate": 2.1475679279937414e-06,
      "loss": 0.3122,
      "step": 21950
    },
    {
      "epoch": 1.1705444950827537,
      "grad_norm": 1.5626249997871782,
      "learning_rate": 2.1467728008819328e-06,
      "loss": 0.3077,
      "step": 21960
    },
    {
      "epoch": 1.17107752991658,
      "grad_norm": 1.8431944130382634,
      "learning_rate": 2.1459774504663344e-06,
      "loss": 0.305,
      "step": 21970
    },
    {
      "epoch": 1.1716105647504065,
      "grad_norm": 1.6110578833170766,
      "learning_rate": 2.145181877021548e-06,
      "loss": 0.3039,
      "step": 21980
    },
    {
      "epoch": 1.1721435995842329,
      "grad_norm": 1.7211334909055216,
      "learning_rate": 2.1443860808222526e-06,
      "loss": 0.3108,
      "step": 21990
    },
    {
      "epoch": 1.1726766344180592,
      "grad_norm": 1.5750700781222808,
      "learning_rate": 2.1435900621432035e-06,
      "loss": 0.3096,
      "step": 22000
    },
    {
      "epoch": 1.1732096692518856,
      "grad_norm": 1.8140573432231122,
      "learning_rate": 2.1427938212592327e-06,
      "loss": 0.3021,
      "step": 22010
    },
    {
      "epoch": 1.173742704085712,
      "grad_norm": 1.605119764072049,
      "learning_rate": 2.1419973584452487e-06,
      "loss": 0.3068,
      "step": 22020
    },
    {
      "epoch": 1.1742757389195384,
      "grad_norm": 1.5827382536387942,
      "learning_rate": 2.1412006739762374e-06,
      "loss": 0.3142,
      "step": 22030
    },
    {
      "epoch": 1.1748087737533648,
      "grad_norm": 1.7016552488188008,
      "learning_rate": 2.1404037681272608e-06,
      "loss": 0.3045,
      "step": 22040
    },
    {
      "epoch": 1.1753418085871912,
      "grad_norm": 1.5043105100648253,
      "learning_rate": 2.139606641173457e-06,
      "loss": 0.2995,
      "step": 22050
    },
    {
      "epoch": 1.1758748434210176,
      "grad_norm": 1.929533869811147,
      "learning_rate": 2.138809293390042e-06,
      "loss": 0.3108,
      "step": 22060
    },
    {
      "epoch": 1.176407878254844,
      "grad_norm": 1.7875585483390928,
      "learning_rate": 2.138011725052306e-06,
      "loss": 0.3027,
      "step": 22070
    },
    {
      "epoch": 1.1769409130886703,
      "grad_norm": 1.5814222403582903,
      "learning_rate": 2.137213936435616e-06,
      "loss": 0.2852,
      "step": 22080
    },
    {
      "epoch": 1.1774739479224967,
      "grad_norm": 1.7272130776912558,
      "learning_rate": 2.136415927815416e-06,
      "loss": 0.3078,
      "step": 22090
    },
    {
      "epoch": 1.178006982756323,
      "grad_norm": 1.7461628884736005,
      "learning_rate": 2.135617699467225e-06,
      "loss": 0.2961,
      "step": 22100
    },
    {
      "epoch": 1.1785400175901495,
      "grad_norm": 1.5605762093119409,
      "learning_rate": 2.1348192516666377e-06,
      "loss": 0.2948,
      "step": 22110
    },
    {
      "epoch": 1.1790730524239759,
      "grad_norm": 1.8333495712183872,
      "learning_rate": 2.134020584689325e-06,
      "loss": 0.2976,
      "step": 22120
    },
    {
      "epoch": 1.1796060872578022,
      "grad_norm": 1.705994538463318,
      "learning_rate": 2.1332216988110344e-06,
      "loss": 0.3076,
      "step": 22130
    },
    {
      "epoch": 1.1801391220916286,
      "grad_norm": 1.7756943606191598,
      "learning_rate": 2.132422594307587e-06,
      "loss": 0.2998,
      "step": 22140
    },
    {
      "epoch": 1.180672156925455,
      "grad_norm": 1.5844862841286445,
      "learning_rate": 2.1316232714548803e-06,
      "loss": 0.2941,
      "step": 22150
    },
    {
      "epoch": 1.1812051917592814,
      "grad_norm": 1.622051563424511,
      "learning_rate": 2.130823730528888e-06,
      "loss": 0.3082,
      "step": 22160
    },
    {
      "epoch": 1.1817382265931078,
      "grad_norm": 1.7068849154162093,
      "learning_rate": 2.130023971805658e-06,
      "loss": 0.307,
      "step": 22170
    },
    {
      "epoch": 1.1822712614269342,
      "grad_norm": 1.5949571040846469,
      "learning_rate": 2.129223995561314e-06,
      "loss": 0.3,
      "step": 22180
    },
    {
      "epoch": 1.1828042962607606,
      "grad_norm": 1.8926393243747694,
      "learning_rate": 2.128423802072055e-06,
      "loss": 0.3142,
      "step": 22190
    },
    {
      "epoch": 1.183337331094587,
      "grad_norm": 1.6853788729619184,
      "learning_rate": 2.1276233916141527e-06,
      "loss": 0.314,
      "step": 22200
    },
    {
      "epoch": 1.1838703659284133,
      "grad_norm": 1.698763361363233,
      "learning_rate": 2.1268227644639577e-06,
      "loss": 0.3031,
      "step": 22210
    },
    {
      "epoch": 1.18440340076224,
      "grad_norm": 1.8072365688515866,
      "learning_rate": 2.1260219208978926e-06,
      "loss": 0.306,
      "step": 22220
    },
    {
      "epoch": 1.1849364355960663,
      "grad_norm": 1.6223053361664546,
      "learning_rate": 2.1252208611924554e-06,
      "loss": 0.2873,
      "step": 22230
    },
    {
      "epoch": 1.1854694704298927,
      "grad_norm": 1.6422801506442333,
      "learning_rate": 2.1244195856242185e-06,
      "loss": 0.2927,
      "step": 22240
    },
    {
      "epoch": 1.186002505263719,
      "grad_norm": 1.7164118640906687,
      "learning_rate": 2.123618094469829e-06,
      "loss": 0.3104,
      "step": 22250
    },
    {
      "epoch": 1.1865355400975455,
      "grad_norm": 1.6489880561579486,
      "learning_rate": 2.122816388006009e-06,
      "loss": 0.3155,
      "step": 22260
    },
    {
      "epoch": 1.1870685749313719,
      "grad_norm": 1.7848319733890206,
      "learning_rate": 2.1220144665095546e-06,
      "loss": 0.3117,
      "step": 22270
    },
    {
      "epoch": 1.1876016097651982,
      "grad_norm": 1.7385490230216296,
      "learning_rate": 2.1212123302573347e-06,
      "loss": 0.3054,
      "step": 22280
    },
    {
      "epoch": 1.1881346445990246,
      "grad_norm": 1.5841782393072172,
      "learning_rate": 2.1204099795262954e-06,
      "loss": 0.3047,
      "step": 22290
    },
    {
      "epoch": 1.188667679432851,
      "grad_norm": 1.6262560566098443,
      "learning_rate": 2.1196074145934537e-06,
      "loss": 0.3048,
      "step": 22300
    },
    {
      "epoch": 1.1892007142666774,
      "grad_norm": 1.6558276281356448,
      "learning_rate": 2.118804635735903e-06,
      "loss": 0.3143,
      "step": 22310
    },
    {
      "epoch": 1.1897337491005038,
      "grad_norm": 1.780908636868358,
      "learning_rate": 2.118001643230809e-06,
      "loss": 0.3023,
      "step": 22320
    },
    {
      "epoch": 1.1902667839343302,
      "grad_norm": 1.5467631173805412,
      "learning_rate": 2.1171984373554108e-06,
      "loss": 0.3005,
      "step": 22330
    },
    {
      "epoch": 1.1907998187681565,
      "grad_norm": 1.5574493707268737,
      "learning_rate": 2.1163950183870234e-06,
      "loss": 0.3101,
      "step": 22340
    },
    {
      "epoch": 1.191332853601983,
      "grad_norm": 1.6038570190265888,
      "learning_rate": 2.1155913866030337e-06,
      "loss": 0.3058,
      "step": 22350
    },
    {
      "epoch": 1.1918658884358093,
      "grad_norm": 1.7535745529465345,
      "learning_rate": 2.1147875422809024e-06,
      "loss": 0.3156,
      "step": 22360
    },
    {
      "epoch": 1.1923989232696357,
      "grad_norm": 1.6832580511940625,
      "learning_rate": 2.113983485698163e-06,
      "loss": 0.3108,
      "step": 22370
    },
    {
      "epoch": 1.192931958103462,
      "grad_norm": 1.699811789318154,
      "learning_rate": 2.113179217132423e-06,
      "loss": 0.3027,
      "step": 22380
    },
    {
      "epoch": 1.1934649929372885,
      "grad_norm": 1.7739012963629168,
      "learning_rate": 2.112374736861363e-06,
      "loss": 0.3093,
      "step": 22390
    },
    {
      "epoch": 1.1939980277711149,
      "grad_norm": 1.73807525084544,
      "learning_rate": 2.1115700451627374e-06,
      "loss": 0.298,
      "step": 22400
    },
    {
      "epoch": 1.1945310626049412,
      "grad_norm": 1.6147640813956223,
      "learning_rate": 2.1107651423143718e-06,
      "loss": 0.3071,
      "step": 22410
    },
    {
      "epoch": 1.1950640974387676,
      "grad_norm": 1.6095317332785417,
      "learning_rate": 2.109960028594166e-06,
      "loss": 0.299,
      "step": 22420
    },
    {
      "epoch": 1.195597132272594,
      "grad_norm": 1.9104919232107886,
      "learning_rate": 2.1091547042800923e-06,
      "loss": 0.3037,
      "step": 22430
    },
    {
      "epoch": 1.1961301671064204,
      "grad_norm": 1.9327535929164992,
      "learning_rate": 2.108349169650196e-06,
      "loss": 0.3099,
      "step": 22440
    },
    {
      "epoch": 1.1966632019402468,
      "grad_norm": 1.872728388314714,
      "learning_rate": 2.107543424982595e-06,
      "loss": 0.3173,
      "step": 22450
    },
    {
      "epoch": 1.1971962367740732,
      "grad_norm": 1.5868814110069622,
      "learning_rate": 2.106737470555479e-06,
      "loss": 0.3119,
      "step": 22460
    },
    {
      "epoch": 1.1977292716078995,
      "grad_norm": 1.7977527909977369,
      "learning_rate": 2.1059313066471106e-06,
      "loss": 0.2959,
      "step": 22470
    },
    {
      "epoch": 1.198262306441726,
      "grad_norm": 1.6519040627369146,
      "learning_rate": 2.105124933535825e-06,
      "loss": 0.2951,
      "step": 22480
    },
    {
      "epoch": 1.1987953412755523,
      "grad_norm": 1.5675419032610383,
      "learning_rate": 2.104318351500029e-06,
      "loss": 0.3103,
      "step": 22490
    },
    {
      "epoch": 1.1993283761093787,
      "grad_norm": 1.5919344949554,
      "learning_rate": 2.103511560818202e-06,
      "loss": 0.3067,
      "step": 22500
    },
    {
      "epoch": 1.199861410943205,
      "grad_norm": 1.7545445742383547,
      "learning_rate": 2.1027045617688956e-06,
      "loss": 0.3095,
      "step": 22510
    },
    {
      "epoch": 1.2003944457770315,
      "grad_norm": 1.6591192004643252,
      "learning_rate": 2.1018973546307326e-06,
      "loss": 0.3085,
      "step": 22520
    },
    {
      "epoch": 1.2009274806108579,
      "grad_norm": 1.555190480116379,
      "learning_rate": 2.1010899396824084e-06,
      "loss": 0.3031,
      "step": 22530
    },
    {
      "epoch": 1.2014605154446842,
      "grad_norm": 1.6180993802345915,
      "learning_rate": 2.1002823172026896e-06,
      "loss": 0.2962,
      "step": 22540
    },
    {
      "epoch": 1.2019935502785106,
      "grad_norm": 1.7505641941296128,
      "learning_rate": 2.0994744874704153e-06,
      "loss": 0.3044,
      "step": 22550
    },
    {
      "epoch": 1.202526585112337,
      "grad_norm": 1.629273304302741,
      "learning_rate": 2.098666450764495e-06,
      "loss": 0.2981,
      "step": 22560
    },
    {
      "epoch": 1.2030596199461634,
      "grad_norm": 1.6435175769695696,
      "learning_rate": 2.0978582073639103e-06,
      "loss": 0.2974,
      "step": 22570
    },
    {
      "epoch": 1.2035926547799898,
      "grad_norm": 1.7556312935660925,
      "learning_rate": 2.0970497575477143e-06,
      "loss": 0.2896,
      "step": 22580
    },
    {
      "epoch": 1.2041256896138162,
      "grad_norm": 1.7036150566418415,
      "learning_rate": 2.0962411015950308e-06,
      "loss": 0.3033,
      "step": 22590
    },
    {
      "epoch": 1.2046587244476425,
      "grad_norm": 1.5520789600676863,
      "learning_rate": 2.0954322397850556e-06,
      "loss": 0.3159,
      "step": 22600
    },
    {
      "epoch": 1.205191759281469,
      "grad_norm": 1.6845509526261502,
      "learning_rate": 2.0946231723970543e-06,
      "loss": 0.3196,
      "step": 22610
    },
    {
      "epoch": 1.2057247941152953,
      "grad_norm": 1.8151009079214864,
      "learning_rate": 2.0938138997103653e-06,
      "loss": 0.3046,
      "step": 22620
    },
    {
      "epoch": 1.2062578289491217,
      "grad_norm": 1.7163462039429207,
      "learning_rate": 2.093004422004396e-06,
      "loss": 0.3019,
      "step": 22630
    },
    {
      "epoch": 1.2067908637829483,
      "grad_norm": 1.557164469511081,
      "learning_rate": 2.092194739558626e-06,
      "loss": 0.3047,
      "step": 22640
    },
    {
      "epoch": 1.2073238986167747,
      "grad_norm": 1.7043293480417827,
      "learning_rate": 2.0913848526526047e-06,
      "loss": 0.3068,
      "step": 22650
    },
    {
      "epoch": 1.207856933450601,
      "grad_norm": 1.6921223365466198,
      "learning_rate": 2.0905747615659524e-06,
      "loss": 0.307,
      "step": 22660
    },
    {
      "epoch": 1.2083899682844275,
      "grad_norm": 1.557180842419274,
      "learning_rate": 2.0897644665783604e-06,
      "loss": 0.3112,
      "step": 22670
    },
    {
      "epoch": 1.2089230031182538,
      "grad_norm": 1.850495550684652,
      "learning_rate": 2.088953967969589e-06,
      "loss": 0.3032,
      "step": 22680
    },
    {
      "epoch": 1.2094560379520802,
      "grad_norm": 1.6137623896619744,
      "learning_rate": 2.0881432660194707e-06,
      "loss": 0.2935,
      "step": 22690
    },
    {
      "epoch": 1.2099890727859066,
      "grad_norm": 1.7013619393664292,
      "learning_rate": 2.0873323610079065e-06,
      "loss": 0.2978,
      "step": 22700
    },
    {
      "epoch": 1.210522107619733,
      "grad_norm": 1.7337313325172732,
      "learning_rate": 2.0865212532148687e-06,
      "loss": 0.2957,
      "step": 22710
    },
    {
      "epoch": 1.2110551424535594,
      "grad_norm": 1.8440407641188579,
      "learning_rate": 2.0857099429203988e-06,
      "loss": 0.3004,
      "step": 22720
    },
    {
      "epoch": 1.2115881772873858,
      "grad_norm": 1.5981660767100057,
      "learning_rate": 2.0848984304046087e-06,
      "loss": 0.3085,
      "step": 22730
    },
    {
      "epoch": 1.2121212121212122,
      "grad_norm": 1.564937949736306,
      "learning_rate": 2.0840867159476802e-06,
      "loss": 0.3026,
      "step": 22740
    },
    {
      "epoch": 1.2126542469550385,
      "grad_norm": 1.6425259906821708,
      "learning_rate": 2.0832747998298638e-06,
      "loss": 0.3038,
      "step": 22750
    },
    {
      "epoch": 1.213187281788865,
      "grad_norm": 1.656477692170798,
      "learning_rate": 2.082462682331481e-06,
      "loss": 0.3008,
      "step": 22760
    },
    {
      "epoch": 1.2137203166226913,
      "grad_norm": 1.6495460353345768,
      "learning_rate": 2.0816503637329227e-06,
      "loss": 0.3081,
      "step": 22770
    },
    {
      "epoch": 1.2142533514565177,
      "grad_norm": 1.587330993510244,
      "learning_rate": 2.0808378443146483e-06,
      "loss": 0.3013,
      "step": 22780
    },
    {
      "epoch": 1.214786386290344,
      "grad_norm": 1.708540271652481,
      "learning_rate": 2.0800251243571867e-06,
      "loss": 0.3097,
      "step": 22790
    },
    {
      "epoch": 1.2153194211241705,
      "grad_norm": 1.7538441867916539,
      "learning_rate": 2.0792122041411366e-06,
      "loss": 0.296,
      "step": 22800
    },
    {
      "epoch": 1.2158524559579968,
      "grad_norm": 1.6445908707269592,
      "learning_rate": 2.0783990839471654e-06,
      "loss": 0.3063,
      "step": 22810
    },
    {
      "epoch": 1.2163854907918232,
      "grad_norm": 1.7426460312004,
      "learning_rate": 2.0775857640560103e-06,
      "loss": 0.3068,
      "step": 22820
    },
    {
      "epoch": 1.2169185256256496,
      "grad_norm": 1.8772432626326854,
      "learning_rate": 2.0767722447484767e-06,
      "loss": 0.3095,
      "step": 22830
    },
    {
      "epoch": 1.217451560459476,
      "grad_norm": 1.7378019374550453,
      "learning_rate": 2.075958526305439e-06,
      "loss": 0.2986,
      "step": 22840
    },
    {
      "epoch": 1.2179845952933024,
      "grad_norm": 1.7433443010702305,
      "learning_rate": 2.0751446090078395e-06,
      "loss": 0.3131,
      "step": 22850
    },
    {
      "epoch": 1.2185176301271288,
      "grad_norm": 2.0335376172963398,
      "learning_rate": 2.074330493136692e-06,
      "loss": 0.3109,
      "step": 22860
    },
    {
      "epoch": 1.2190506649609552,
      "grad_norm": 1.61318448002519,
      "learning_rate": 2.073516178973075e-06,
      "loss": 0.2995,
      "step": 22870
    },
    {
      "epoch": 1.2195836997947815,
      "grad_norm": 1.7159222475285165,
      "learning_rate": 2.072701666798138e-06,
      "loss": 0.3071,
      "step": 22880
    },
    {
      "epoch": 1.220116734628608,
      "grad_norm": 1.5555244480679535,
      "learning_rate": 2.0718869568930985e-06,
      "loss": 0.3101,
      "step": 22890
    },
    {
      "epoch": 1.2206497694624343,
      "grad_norm": 1.6272077642632332,
      "learning_rate": 2.071072049539242e-06,
      "loss": 0.3245,
      "step": 22900
    },
    {
      "epoch": 1.2211828042962607,
      "grad_norm": 1.6101601507559524,
      "learning_rate": 2.070256945017922e-06,
      "loss": 0.3008,
      "step": 22910
    },
    {
      "epoch": 1.221715839130087,
      "grad_norm": 1.9018779298422441,
      "learning_rate": 2.06944164361056e-06,
      "loss": 0.3041,
      "step": 22920
    },
    {
      "epoch": 1.2222488739639135,
      "grad_norm": 1.7500364475077854,
      "learning_rate": 2.068626145598646e-06,
      "loss": 0.3022,
      "step": 22930
    },
    {
      "epoch": 1.2227819087977398,
      "grad_norm": 1.7172411071749432,
      "learning_rate": 2.067810451263737e-06,
      "loss": 0.305,
      "step": 22940
    },
    {
      "epoch": 1.2233149436315662,
      "grad_norm": 1.5086814906774648,
      "learning_rate": 2.0669945608874597e-06,
      "loss": 0.3098,
      "step": 22950
    },
    {
      "epoch": 1.2238479784653926,
      "grad_norm": 1.7174971633111764,
      "learning_rate": 2.0661784747515065e-06,
      "loss": 0.3079,
      "step": 22960
    },
    {
      "epoch": 1.2243810132992192,
      "grad_norm": 1.7003702955675264,
      "learning_rate": 2.0653621931376375e-06,
      "loss": 0.2991,
      "step": 22970
    },
    {
      "epoch": 1.2249140481330456,
      "grad_norm": 1.639439207392191,
      "learning_rate": 2.064545716327681e-06,
      "loss": 0.2998,
      "step": 22980
    },
    {
      "epoch": 1.225447082966872,
      "grad_norm": 1.9022879696254273,
      "learning_rate": 2.0637290446035336e-06,
      "loss": 0.3063,
      "step": 22990
    },
    {
      "epoch": 1.2259801178006984,
      "grad_norm": 1.617982685002921,
      "learning_rate": 2.062912178247157e-06,
      "loss": 0.2968,
      "step": 23000
    },
    {
      "epoch": 1.2265131526345248,
      "grad_norm": 1.9969217312839123,
      "learning_rate": 2.062095117540581e-06,
      "loss": 0.317,
      "step": 23010
    },
    {
      "epoch": 1.2270461874683511,
      "grad_norm": 1.6920728336270645,
      "learning_rate": 2.061277862765904e-06,
      "loss": 0.2977,
      "step": 23020
    },
    {
      "epoch": 1.2275792223021775,
      "grad_norm": 1.6313947480776152,
      "learning_rate": 2.0604604142052892e-06,
      "loss": 0.3073,
      "step": 23030
    },
    {
      "epoch": 1.228112257136004,
      "grad_norm": 1.6564264788419336,
      "learning_rate": 2.0596427721409686e-06,
      "loss": 0.3039,
      "step": 23040
    },
    {
      "epoch": 1.2286452919698303,
      "grad_norm": 1.6976190444671193,
      "learning_rate": 2.058824936855239e-06,
      "loss": 0.3066,
      "step": 23050
    },
    {
      "epoch": 1.2291783268036567,
      "grad_norm": 1.6700520547031814,
      "learning_rate": 2.058006908630465e-06,
      "loss": 0.3104,
      "step": 23060
    },
    {
      "epoch": 1.229711361637483,
      "grad_norm": 1.7890008600368514,
      "learning_rate": 2.057188687749078e-06,
      "loss": 0.3142,
      "step": 23070
    },
    {
      "epoch": 1.2302443964713095,
      "grad_norm": 1.7733026675262766,
      "learning_rate": 2.056370274493576e-06,
      "loss": 0.3093,
      "step": 23080
    },
    {
      "epoch": 1.2307774313051358,
      "grad_norm": 1.8081693412874023,
      "learning_rate": 2.055551669146523e-06,
      "loss": 0.3066,
      "step": 23090
    },
    {
      "epoch": 1.2313104661389622,
      "grad_norm": 1.5979703342482794,
      "learning_rate": 2.0547328719905486e-06,
      "loss": 0.3009,
      "step": 23100
    },
    {
      "epoch": 1.2318435009727886,
      "grad_norm": 1.6355757331358172,
      "learning_rate": 2.053913883308351e-06,
      "loss": 0.3058,
      "step": 23110
    },
    {
      "epoch": 1.232376535806615,
      "grad_norm": 1.7272048493867864,
      "learning_rate": 2.053094703382692e-06,
      "loss": 0.3068,
      "step": 23120
    },
    {
      "epoch": 1.2329095706404414,
      "grad_norm": 1.5675457306442544,
      "learning_rate": 2.0522753324964004e-06,
      "loss": 0.3091,
      "step": 23130
    },
    {
      "epoch": 1.2334426054742678,
      "grad_norm": 1.6575492285086713,
      "learning_rate": 2.0514557709323707e-06,
      "loss": 0.2946,
      "step": 23140
    },
    {
      "epoch": 1.2339756403080941,
      "grad_norm": 1.655614314037684,
      "learning_rate": 2.0506360189735643e-06,
      "loss": 0.3103,
      "step": 23150
    },
    {
      "epoch": 1.2345086751419205,
      "grad_norm": 1.6184479052156073,
      "learning_rate": 2.049816076903007e-06,
      "loss": 0.3052,
      "step": 23160
    },
    {
      "epoch": 1.235041709975747,
      "grad_norm": 1.668374985332481,
      "learning_rate": 2.0489959450037915e-06,
      "loss": 0.3063,
      "step": 23170
    },
    {
      "epoch": 1.2355747448095733,
      "grad_norm": 1.759364050447776,
      "learning_rate": 2.048175623559074e-06,
      "loss": 0.3169,
      "step": 23180
    },
    {
      "epoch": 1.2361077796433997,
      "grad_norm": 1.7854334105978855,
      "learning_rate": 2.0473551128520778e-06,
      "loss": 0.3003,
      "step": 23190
    },
    {
      "epoch": 1.236640814477226,
      "grad_norm": 1.8656959072383144,
      "learning_rate": 2.0465344131660917e-06,
      "loss": 0.3144,
      "step": 23200
    },
    {
      "epoch": 1.2371738493110525,
      "grad_norm": 1.7862010636796275,
      "learning_rate": 2.045713524784469e-06,
      "loss": 0.3114,
      "step": 23210
    },
    {
      "epoch": 1.2377068841448788,
      "grad_norm": 1.7644424348514764,
      "learning_rate": 2.0448924479906282e-06,
      "loss": 0.303,
      "step": 23220
    },
    {
      "epoch": 1.2382399189787052,
      "grad_norm": 1.6219955535347945,
      "learning_rate": 2.044071183068053e-06,
      "loss": 0.3029,
      "step": 23230
    },
    {
      "epoch": 1.2387729538125316,
      "grad_norm": 1.6118746740378496,
      "learning_rate": 2.0432497303002926e-06,
      "loss": 0.3161,
      "step": 23240
    },
    {
      "epoch": 1.239305988646358,
      "grad_norm": 1.6387435018177512,
      "learning_rate": 2.04242808997096e-06,
      "loss": 0.2994,
      "step": 23250
    },
    {
      "epoch": 1.2398390234801844,
      "grad_norm": 1.8153250914283579,
      "learning_rate": 2.0416062623637334e-06,
      "loss": 0.3061,
      "step": 23260
    },
    {
      "epoch": 1.2403720583140108,
      "grad_norm": 1.7144466259165616,
      "learning_rate": 2.040784247762356e-06,
      "loss": 0.3087,
      "step": 23270
    },
    {
      "epoch": 1.2409050931478371,
      "grad_norm": 1.639913306693048,
      "learning_rate": 2.039962046450635e-06,
      "loss": 0.3058,
      "step": 23280
    },
    {
      "epoch": 1.2414381279816635,
      "grad_norm": 1.9478436724477137,
      "learning_rate": 2.039139658712444e-06,
      "loss": 0.3115,
      "step": 23290
    },
    {
      "epoch": 1.24197116281549,
      "grad_norm": 1.5123708939139904,
      "learning_rate": 2.0383170848317162e-06,
      "loss": 0.3092,
      "step": 23300
    },
    {
      "epoch": 1.2425041976493163,
      "grad_norm": 1.7801015955009891,
      "learning_rate": 2.037494325092455e-06,
      "loss": 0.299,
      "step": 23310
    },
    {
      "epoch": 1.2430372324831427,
      "grad_norm": 1.6946527927867592,
      "learning_rate": 2.0366713797787235e-06,
      "loss": 0.3081,
      "step": 23320
    },
    {
      "epoch": 1.243570267316969,
      "grad_norm": 1.5099028998630448,
      "learning_rate": 2.0358482491746515e-06,
      "loss": 0.3046,
      "step": 23330
    },
    {
      "epoch": 1.2441033021507955,
      "grad_norm": 2.1318087650046933,
      "learning_rate": 2.035024933564432e-06,
      "loss": 0.2998,
      "step": 23340
    },
    {
      "epoch": 1.2446363369846218,
      "grad_norm": 1.7193063490887475,
      "learning_rate": 2.0342014332323208e-06,
      "loss": 0.3026,
      "step": 23350
    },
    {
      "epoch": 1.2451693718184482,
      "grad_norm": 1.6864763705247356,
      "learning_rate": 2.0333777484626384e-06,
      "loss": 0.3168,
      "step": 23360
    },
    {
      "epoch": 1.2457024066522746,
      "grad_norm": 1.6974719414453956,
      "learning_rate": 2.0325538795397695e-06,
      "loss": 0.3108,
      "step": 23370
    },
    {
      "epoch": 1.246235441486101,
      "grad_norm": 1.7508322173549118,
      "learning_rate": 2.031729826748162e-06,
      "loss": 0.2997,
      "step": 23380
    },
    {
      "epoch": 1.2467684763199276,
      "grad_norm": 1.560396403119923,
      "learning_rate": 2.0309055903723265e-06,
      "loss": 0.3138,
      "step": 23390
    },
    {
      "epoch": 1.247301511153754,
      "grad_norm": 1.7609993321443171,
      "learning_rate": 2.030081170696838e-06,
      "loss": 0.3064,
      "step": 23400
    },
    {
      "epoch": 1.2478345459875804,
      "grad_norm": 1.7677053612413458,
      "learning_rate": 2.029256568006334e-06,
      "loss": 0.3049,
      "step": 23410
    },
    {
      "epoch": 1.2483675808214068,
      "grad_norm": 1.573857880434198,
      "learning_rate": 2.0284317825855156e-06,
      "loss": 0.3203,
      "step": 23420
    },
    {
      "epoch": 1.2489006156552331,
      "grad_norm": 1.6417171599729599,
      "learning_rate": 2.0276068147191473e-06,
      "loss": 0.3166,
      "step": 23430
    },
    {
      "epoch": 1.2494336504890595,
      "grad_norm": 1.9465722922208217,
      "learning_rate": 2.026781664692056e-06,
      "loss": 0.3089,
      "step": 23440
    },
    {
      "epoch": 1.249966685322886,
      "grad_norm": 1.9061177520198818,
      "learning_rate": 2.0259563327891316e-06,
      "loss": 0.2945,
      "step": 23450
    },
    {
      "epoch": 1.2504997201567123,
      "grad_norm": 1.6761650418704517,
      "learning_rate": 2.0251308192953273e-06,
      "loss": 0.3066,
      "step": 23460
    },
    {
      "epoch": 1.2510327549905387,
      "grad_norm": 1.6406278957445162,
      "learning_rate": 2.024305124495659e-06,
      "loss": 0.3,
      "step": 23470
    },
    {
      "epoch": 1.251565789824365,
      "grad_norm": 1.7390805170953534,
      "learning_rate": 2.0234792486752037e-06,
      "loss": 0.3201,
      "step": 23480
    },
    {
      "epoch": 1.2520988246581914,
      "grad_norm": 1.8175449778740163,
      "learning_rate": 2.0226531921191023e-06,
      "loss": 0.3137,
      "step": 23490
    },
    {
      "epoch": 1.2526318594920178,
      "grad_norm": 1.6807641917012166,
      "learning_rate": 2.0218269551125587e-06,
      "loss": 0.3011,
      "step": 23500
    },
    {
      "epoch": 1.2531648943258442,
      "grad_norm": 1.7763183860779088,
      "learning_rate": 2.021000537940838e-06,
      "loss": 0.3193,
      "step": 23510
    },
    {
      "epoch": 1.2536979291596706,
      "grad_norm": 1.8974129488929747,
      "learning_rate": 2.020173940889267e-06,
      "loss": 0.3019,
      "step": 23520
    },
    {
      "epoch": 1.254230963993497,
      "grad_norm": 1.7254528384907368,
      "learning_rate": 2.0193471642432355e-06,
      "loss": 0.3068,
      "step": 23530
    },
    {
      "epoch": 1.2547639988273234,
      "grad_norm": 1.9913061487305976,
      "learning_rate": 2.018520208288195e-06,
      "loss": 0.2923,
      "step": 23540
    },
    {
      "epoch": 1.2552970336611498,
      "grad_norm": 1.6330855881949171,
      "learning_rate": 2.0176930733096604e-06,
      "loss": 0.3076,
      "step": 23550
    },
    {
      "epoch": 1.2558300684949761,
      "grad_norm": 1.907718882882335,
      "learning_rate": 2.0168657595932057e-06,
      "loss": 0.3037,
      "step": 23560
    },
    {
      "epoch": 1.2563631033288025,
      "grad_norm": 1.8810064935227189,
      "learning_rate": 2.0160382674244684e-06,
      "loss": 0.301,
      "step": 23570
    },
    {
      "epoch": 1.256896138162629,
      "grad_norm": 1.7241431314280322,
      "learning_rate": 2.0152105970891473e-06,
      "loss": 0.3002,
      "step": 23580
    },
    {
      "epoch": 1.2574291729964553,
      "grad_norm": 1.7668103365406842,
      "learning_rate": 2.014382748873002e-06,
      "loss": 0.2965,
      "step": 23590
    },
    {
      "epoch": 1.2579622078302817,
      "grad_norm": 1.6528251100277758,
      "learning_rate": 2.013554723061855e-06,
      "loss": 0.2989,
      "step": 23600
    },
    {
      "epoch": 1.258495242664108,
      "grad_norm": 1.7693076532602492,
      "learning_rate": 2.0127265199415885e-06,
      "loss": 0.3077,
      "step": 23610
    },
    {
      "epoch": 1.2590282774979344,
      "grad_norm": 1.7991073891585239,
      "learning_rate": 2.0118981397981467e-06,
      "loss": 0.3158,
      "step": 23620
    },
    {
      "epoch": 1.2595613123317608,
      "grad_norm": 1.6531727663471893,
      "learning_rate": 2.0110695829175354e-06,
      "loss": 0.3084,
      "step": 23630
    },
    {
      "epoch": 1.2600943471655872,
      "grad_norm": 1.5372854325296668,
      "learning_rate": 2.0102408495858204e-06,
      "loss": 0.3178,
      "step": 23640
    },
    {
      "epoch": 1.2606273819994136,
      "grad_norm": 1.657453755504069,
      "learning_rate": 2.0094119400891293e-06,
      "loss": 0.3103,
      "step": 23650
    },
    {
      "epoch": 1.26116041683324,
      "grad_norm": 1.6390090310046577,
      "learning_rate": 2.00858285471365e-06,
      "loss": 0.3084,
      "step": 23660
    },
    {
      "epoch": 1.2616934516670664,
      "grad_norm": 1.9280536501717365,
      "learning_rate": 2.00775359374563e-06,
      "loss": 0.3019,
      "step": 23670
    },
    {
      "epoch": 1.2622264865008928,
      "grad_norm": 1.749625768797954,
      "learning_rate": 2.006924157471381e-06,
      "loss": 0.3049,
      "step": 23680
    },
    {
      "epoch": 1.2627595213347194,
      "grad_norm": 1.6734495570784924,
      "learning_rate": 2.0060945461772714e-06,
      "loss": 0.3214,
      "step": 23690
    },
    {
      "epoch": 1.2632925561685457,
      "grad_norm": 1.7941123780022543,
      "learning_rate": 2.005264760149732e-06,
      "loss": 0.2998,
      "step": 23700
    },
    {
      "epoch": 1.2638255910023721,
      "grad_norm": 1.6987616290417822,
      "learning_rate": 2.0044347996752533e-06,
      "loss": 0.3146,
      "step": 23710
    },
    {
      "epoch": 1.2643586258361985,
      "grad_norm": 1.811070881348571,
      "learning_rate": 2.0036046650403857e-06,
      "loss": 0.3075,
      "step": 23720
    },
    {
      "epoch": 1.264891660670025,
      "grad_norm": 1.6382451184178053,
      "learning_rate": 2.0027743565317413e-06,
      "loss": 0.2982,
      "step": 23730
    },
    {
      "epoch": 1.2654246955038513,
      "grad_norm": 1.7278127555851586,
      "learning_rate": 2.0019438744359903e-06,
      "loss": 0.3072,
      "step": 23740
    },
    {
      "epoch": 1.2659577303376777,
      "grad_norm": 1.6695385711788127,
      "learning_rate": 2.0011132190398635e-06,
      "loss": 0.3,
      "step": 23750
    },
    {
      "epoch": 1.266490765171504,
      "grad_norm": 1.6744063988070408,
      "learning_rate": 2.0002823906301527e-06,
      "loss": 0.3087,
      "step": 23760
    },
    {
      "epoch": 1.2670238000053304,
      "grad_norm": 1.649259089639654,
      "learning_rate": 1.9994513894937082e-06,
      "loss": 0.3111,
      "step": 23770
    },
    {
      "epoch": 1.2675568348391568,
      "grad_norm": 1.6006381710566826,
      "learning_rate": 1.998620215917439e-06,
      "loss": 0.3112,
      "step": 23780
    },
    {
      "epoch": 1.2680898696729832,
      "grad_norm": 1.6288001078473064,
      "learning_rate": 1.9977888701883164e-06,
      "loss": 0.3092,
      "step": 23790
    },
    {
      "epoch": 1.2686229045068096,
      "grad_norm": 1.6958816442340083,
      "learning_rate": 1.9969573525933686e-06,
      "loss": 0.3026,
      "step": 23800
    },
    {
      "epoch": 1.269155939340636,
      "grad_norm": 1.7264966338483279,
      "learning_rate": 1.996125663419684e-06,
      "loss": 0.3048,
      "step": 23810
    },
    {
      "epoch": 1.2696889741744624,
      "grad_norm": 1.650861921664192,
      "learning_rate": 1.9952938029544113e-06,
      "loss": 0.2967,
      "step": 23820
    },
    {
      "epoch": 1.2702220090082887,
      "grad_norm": 1.755834153464092,
      "learning_rate": 1.994461771484756e-06,
      "loss": 0.3017,
      "step": 23830
    },
    {
      "epoch": 1.2707550438421151,
      "grad_norm": 1.767874990410785,
      "learning_rate": 1.9936295692979856e-06,
      "loss": 0.2892,
      "step": 23840
    },
    {
      "epoch": 1.2712880786759415,
      "grad_norm": 1.5762801171439977,
      "learning_rate": 1.992797196681424e-06,
      "loss": 0.2948,
      "step": 23850
    },
    {
      "epoch": 1.271821113509768,
      "grad_norm": 1.7255827702454796,
      "learning_rate": 1.991964653922454e-06,
      "loss": 0.3083,
      "step": 23860
    },
    {
      "epoch": 1.2723541483435943,
      "grad_norm": 1.5257989472792648,
      "learning_rate": 1.99113194130852e-06,
      "loss": 0.2901,
      "step": 23870
    },
    {
      "epoch": 1.2728871831774207,
      "grad_norm": 1.7370431079395767,
      "learning_rate": 1.990299059127122e-06,
      "loss": 0.2999,
      "step": 23880
    },
    {
      "epoch": 1.273420218011247,
      "grad_norm": 1.5211423379019517,
      "learning_rate": 1.9894660076658198e-06,
      "loss": 0.2887,
      "step": 23890
    },
    {
      "epoch": 1.2739532528450734,
      "grad_norm": 1.7139661543978686,
      "learning_rate": 1.9886327872122317e-06,
      "loss": 0.3093,
      "step": 23900
    },
    {
      "epoch": 1.2744862876788998,
      "grad_norm": 1.6859724829742593,
      "learning_rate": 1.9877993980540335e-06,
      "loss": 0.3089,
      "step": 23910
    },
    {
      "epoch": 1.2750193225127262,
      "grad_norm": 1.7840117165235774,
      "learning_rate": 1.9869658404789602e-06,
      "loss": 0.3017,
      "step": 23920
    },
    {
      "epoch": 1.2755523573465526,
      "grad_norm": 1.8081796216229582,
      "learning_rate": 1.9861321147748045e-06,
      "loss": 0.3167,
      "step": 23930
    },
    {
      "epoch": 1.276085392180379,
      "grad_norm": 1.8944870103297562,
      "learning_rate": 1.985298221229418e-06,
      "loss": 0.303,
      "step": 23940
    },
    {
      "epoch": 1.2766184270142054,
      "grad_norm": 1.7834241070789296,
      "learning_rate": 1.9844641601307084e-06,
      "loss": 0.3085,
      "step": 23950
    },
    {
      "epoch": 1.2771514618480317,
      "grad_norm": 1.7272653383426313,
      "learning_rate": 1.9836299317666434e-06,
      "loss": 0.3108,
      "step": 23960
    },
    {
      "epoch": 1.2776844966818581,
      "grad_norm": 1.807055148793654,
      "learning_rate": 1.9827955364252463e-06,
      "loss": 0.2955,
      "step": 23970
    },
    {
      "epoch": 1.2782175315156845,
      "grad_norm": 1.581988851962794,
      "learning_rate": 1.9819609743946004e-06,
      "loss": 0.3046,
      "step": 23980
    },
    {
      "epoch": 1.278750566349511,
      "grad_norm": 1.7575255484880108,
      "learning_rate": 1.9811262459628446e-06,
      "loss": 0.3063,
      "step": 23990
    },
    {
      "epoch": 1.2792836011833373,
      "grad_norm": 1.8110477661759514,
      "learning_rate": 1.9802913514181755e-06,
      "loss": 0.312,
      "step": 24000
    },
    {
      "epoch": 1.2798166360171637,
      "grad_norm": 1.7025572078999478,
      "learning_rate": 1.979456291048848e-06,
      "loss": 0.3054,
      "step": 24010
    },
    {
      "epoch": 1.28034967085099,
      "grad_norm": 1.674077627318722,
      "learning_rate": 1.9786210651431745e-06,
      "loss": 0.3076,
      "step": 24020
    },
    {
      "epoch": 1.2808827056848164,
      "grad_norm": 1.6409421003288633,
      "learning_rate": 1.977785673989523e-06,
      "loss": 0.3029,
      "step": 24030
    },
    {
      "epoch": 1.2814157405186428,
      "grad_norm": 1.754176402845564,
      "learning_rate": 1.976950117876319e-06,
      "loss": 0.3041,
      "step": 24040
    },
    {
      "epoch": 1.2819487753524692,
      "grad_norm": 1.6992314594222562,
      "learning_rate": 1.9761143970920463e-06,
      "loss": 0.2985,
      "step": 24050
    },
    {
      "epoch": 1.2824818101862956,
      "grad_norm": 1.6114792478961888,
      "learning_rate": 1.975278511925244e-06,
      "loss": 0.3048,
      "step": 24060
    },
    {
      "epoch": 1.283014845020122,
      "grad_norm": 1.6771823817485154,
      "learning_rate": 1.974442462664509e-06,
      "loss": 0.3117,
      "step": 24070
    },
    {
      "epoch": 1.2835478798539484,
      "grad_norm": 1.5929546704872253,
      "learning_rate": 1.9736062495984937e-06,
      "loss": 0.3166,
      "step": 24080
    },
    {
      "epoch": 1.2840809146877747,
      "grad_norm": 1.6926011350693226,
      "learning_rate": 1.972769873015908e-06,
      "loss": 0.3058,
      "step": 24090
    },
    {
      "epoch": 1.2846139495216011,
      "grad_norm": 1.7662870946208142,
      "learning_rate": 1.971933333205518e-06,
      "loss": 0.3088,
      "step": 24100
    },
    {
      "epoch": 1.2851469843554275,
      "grad_norm": 1.544872415656281,
      "learning_rate": 1.9710966304561466e-06,
      "loss": 0.3002,
      "step": 24110
    },
    {
      "epoch": 1.285680019189254,
      "grad_norm": 1.6154117522996114,
      "learning_rate": 1.970259765056672e-06,
      "loss": 0.3007,
      "step": 24120
    },
    {
      "epoch": 1.2862130540230803,
      "grad_norm": 1.6703854411528758,
      "learning_rate": 1.9694227372960295e-06,
      "loss": 0.3129,
      "step": 24130
    },
    {
      "epoch": 1.2867460888569067,
      "grad_norm": 1.6759575566336262,
      "learning_rate": 1.968585547463209e-06,
      "loss": 0.3036,
      "step": 24140
    },
    {
      "epoch": 1.287279123690733,
      "grad_norm": 1.8525776571802695,
      "learning_rate": 1.967748195847259e-06,
      "loss": 0.3073,
      "step": 24150
    },
    {
      "epoch": 1.2878121585245597,
      "grad_norm": 1.6265519772383061,
      "learning_rate": 1.966910682737281e-06,
      "loss": 0.2998,
      "step": 24160
    },
    {
      "epoch": 1.288345193358386,
      "grad_norm": 1.723349511020782,
      "learning_rate": 1.966073008422434e-06,
      "loss": 0.3075,
      "step": 24170
    },
    {
      "epoch": 1.2888782281922124,
      "grad_norm": 1.6553144898350696,
      "learning_rate": 1.965235173191932e-06,
      "loss": 0.2908,
      "step": 24180
    },
    {
      "epoch": 1.2894112630260388,
      "grad_norm": 2.010016923609899,
      "learning_rate": 1.9643971773350443e-06,
      "loss": 0.2989,
      "step": 24190
    },
    {
      "epoch": 1.2899442978598652,
      "grad_norm": 1.7007418823381912,
      "learning_rate": 1.9635590211410975e-06,
      "loss": 0.2976,
      "step": 24200
    },
    {
      "epoch": 1.2904773326936916,
      "grad_norm": 1.7832007763299094,
      "learning_rate": 1.9627207048994702e-06,
      "loss": 0.3128,
      "step": 24210
    },
    {
      "epoch": 1.291010367527518,
      "grad_norm": 1.7460412252130852,
      "learning_rate": 1.9618822288995995e-06,
      "loss": 0.3026,
      "step": 24220
    },
    {
      "epoch": 1.2915434023613444,
      "grad_norm": 1.694423923360052,
      "learning_rate": 1.9610435934309763e-06,
      "loss": 0.3078,
      "step": 24230
    },
    {
      "epoch": 1.2920764371951707,
      "grad_norm": 1.632519395205416,
      "learning_rate": 1.960204798783146e-06,
      "loss": 0.3082,
      "step": 24240
    },
    {
      "epoch": 1.2926094720289971,
      "grad_norm": 1.6988954222354853,
      "learning_rate": 1.9593658452457098e-06,
      "loss": 0.2975,
      "step": 24250
    },
    {
      "epoch": 1.2931425068628235,
      "grad_norm": 1.6967655423997856,
      "learning_rate": 1.9585267331083235e-06,
      "loss": 0.3028,
      "step": 24260
    },
    {
      "epoch": 1.2936755416966499,
      "grad_norm": 1.6389139785690472,
      "learning_rate": 1.9576874626606975e-06,
      "loss": 0.3064,
      "step": 24270
    },
    {
      "epoch": 1.2942085765304763,
      "grad_norm": 1.6194908734913336,
      "learning_rate": 1.9568480341925975e-06,
      "loss": 0.2883,
      "step": 24280
    },
    {
      "epoch": 1.2947416113643027,
      "grad_norm": 1.5632776829525976,
      "learning_rate": 1.9560084479938435e-06,
      "loss": 0.3022,
      "step": 24290
    },
    {
      "epoch": 1.295274646198129,
      "grad_norm": 1.768719916978527,
      "learning_rate": 1.9551687043543087e-06,
      "loss": 0.3145,
      "step": 24300
    },
    {
      "epoch": 1.2958076810319554,
      "grad_norm": 1.801955872522533,
      "learning_rate": 1.954328803563923e-06,
      "loss": 0.3031,
      "step": 24310
    },
    {
      "epoch": 1.2963407158657818,
      "grad_norm": 1.698453690277252,
      "learning_rate": 1.953488745912667e-06,
      "loss": 0.2984,
      "step": 24320
    },
    {
      "epoch": 1.2968737506996082,
      "grad_norm": 1.698893672874514,
      "learning_rate": 1.952648531690581e-06,
      "loss": 0.306,
      "step": 24330
    },
    {
      "epoch": 1.2974067855334346,
      "grad_norm": 1.7322497236870733,
      "learning_rate": 1.9518081611877537e-06,
      "loss": 0.303,
      "step": 24340
    },
    {
      "epoch": 1.297939820367261,
      "grad_norm": 1.6072320884166629,
      "learning_rate": 1.950967634694331e-06,
      "loss": 0.2952,
      "step": 24350
    },
    {
      "epoch": 1.2984728552010874,
      "grad_norm": 1.7462103423578497,
      "learning_rate": 1.950126952500511e-06,
      "loss": 0.3014,
      "step": 24360
    },
    {
      "epoch": 1.2990058900349137,
      "grad_norm": 1.6838721781792219,
      "learning_rate": 1.9492861148965475e-06,
      "loss": 0.3005,
      "step": 24370
    },
    {
      "epoch": 1.2995389248687401,
      "grad_norm": 1.646669854976831,
      "learning_rate": 1.9484451221727463e-06,
      "loss": 0.3117,
      "step": 24380
    },
    {
      "epoch": 1.3000719597025665,
      "grad_norm": 1.7060404482923701,
      "learning_rate": 1.947603974619467e-06,
      "loss": 0.3041,
      "step": 24390
    },
    {
      "epoch": 1.3006049945363929,
      "grad_norm": 1.7827910818425539,
      "learning_rate": 1.9467626725271233e-06,
      "loss": 0.3129,
      "step": 24400
    },
    {
      "epoch": 1.3011380293702193,
      "grad_norm": 1.6018056908764948,
      "learning_rate": 1.9459212161861814e-06,
      "loss": 0.2969,
      "step": 24410
    },
    {
      "epoch": 1.3016710642040457,
      "grad_norm": 1.8344774186788915,
      "learning_rate": 1.945079605887162e-06,
      "loss": 0.3038,
      "step": 24420
    },
    {
      "epoch": 1.302204099037872,
      "grad_norm": 1.5518862170990173,
      "learning_rate": 1.944237841920637e-06,
      "loss": 0.3019,
      "step": 24430
    },
    {
      "epoch": 1.3027371338716986,
      "grad_norm": 1.5844239031975051,
      "learning_rate": 1.943395924577234e-06,
      "loss": 0.2997,
      "step": 24440
    },
    {
      "epoch": 1.303270168705525,
      "grad_norm": 1.5187460673777748,
      "learning_rate": 1.942553854147631e-06,
      "loss": 0.3157,
      "step": 24450
    },
    {
      "epoch": 1.3038032035393514,
      "grad_norm": 1.54301777331229,
      "learning_rate": 1.9417116309225604e-06,
      "loss": 0.2973,
      "step": 24460
    },
    {
      "epoch": 1.3043362383731778,
      "grad_norm": 1.6361812824608672,
      "learning_rate": 1.940869255192807e-06,
      "loss": 0.3049,
      "step": 24470
    },
    {
      "epoch": 1.3048692732070042,
      "grad_norm": 1.6280012072784735,
      "learning_rate": 1.940026727249208e-06,
      "loss": 0.3095,
      "step": 24480
    },
    {
      "epoch": 1.3054023080408306,
      "grad_norm": 1.5670169298927417,
      "learning_rate": 1.9391840473826525e-06,
      "loss": 0.3184,
      "step": 24490
    },
    {
      "epoch": 1.305935342874657,
      "grad_norm": 1.7785725883376424,
      "learning_rate": 1.9383412158840845e-06,
      "loss": 0.2975,
      "step": 24500
    },
    {
      "epoch": 1.3064683777084833,
      "grad_norm": 1.708854521225467,
      "learning_rate": 1.937498233044498e-06,
      "loss": 0.2997,
      "step": 24510
    },
    {
      "epoch": 1.3070014125423097,
      "grad_norm": 1.600293951707941,
      "learning_rate": 1.936655099154939e-06,
      "loss": 0.308,
      "step": 24520
    },
    {
      "epoch": 1.307534447376136,
      "grad_norm": 1.595263837936558,
      "learning_rate": 1.935811814506508e-06,
      "loss": 0.2872,
      "step": 24530
    },
    {
      "epoch": 1.3080674822099625,
      "grad_norm": 1.7202152799755208,
      "learning_rate": 1.934968379390356e-06,
      "loss": 0.2963,
      "step": 24540
    },
    {
      "epoch": 1.3086005170437889,
      "grad_norm": 1.6362949253673207,
      "learning_rate": 1.934124794097685e-06,
      "loss": 0.3042,
      "step": 24550
    },
    {
      "epoch": 1.3091335518776153,
      "grad_norm": 1.5888924483958626,
      "learning_rate": 1.933281058919751e-06,
      "loss": 0.2988,
      "step": 24560
    },
    {
      "epoch": 1.3096665867114416,
      "grad_norm": 1.5366126403555402,
      "learning_rate": 1.9324371741478605e-06,
      "loss": 0.3003,
      "step": 24570
    },
    {
      "epoch": 1.310199621545268,
      "grad_norm": 1.6488126300560526,
      "learning_rate": 1.9315931400733716e-06,
      "loss": 0.2939,
      "step": 24580
    },
    {
      "epoch": 1.3107326563790944,
      "grad_norm": 1.9741920031563378,
      "learning_rate": 1.9307489569876948e-06,
      "loss": 0.3078,
      "step": 24590
    },
    {
      "epoch": 1.3112656912129208,
      "grad_norm": 1.8484048168636145,
      "learning_rate": 1.929904625182291e-06,
      "loss": 0.2986,
      "step": 24600
    },
    {
      "epoch": 1.3117987260467472,
      "grad_norm": 1.852771724086025,
      "learning_rate": 1.9290601449486724e-06,
      "loss": 0.3016,
      "step": 24610
    },
    {
      "epoch": 1.3123317608805736,
      "grad_norm": 1.725630833258073,
      "learning_rate": 1.928215516578404e-06,
      "loss": 0.2998,
      "step": 24620
    },
    {
      "epoch": 1.3128647957144,
      "grad_norm": 1.7745457828630256,
      "learning_rate": 1.9273707403631007e-06,
      "loss": 0.3086,
      "step": 24630
    },
    {
      "epoch": 1.3133978305482263,
      "grad_norm": 1.8030306995194472,
      "learning_rate": 1.926525816594428e-06,
      "loss": 0.3077,
      "step": 24640
    },
    {
      "epoch": 1.3139308653820527,
      "grad_norm": 1.6079744676091676,
      "learning_rate": 1.9256807455641036e-06,
      "loss": 0.3028,
      "step": 24650
    },
    {
      "epoch": 1.314463900215879,
      "grad_norm": 1.6697650753363078,
      "learning_rate": 1.924835527563895e-06,
      "loss": 0.3054,
      "step": 24660
    },
    {
      "epoch": 1.3149969350497055,
      "grad_norm": 1.6758335021339845,
      "learning_rate": 1.923990162885621e-06,
      "loss": 0.3068,
      "step": 24670
    },
    {
      "epoch": 1.3155299698835319,
      "grad_norm": 2.053399679183637,
      "learning_rate": 1.923144651821151e-06,
      "loss": 0.3042,
      "step": 24680
    },
    {
      "epoch": 1.3160630047173583,
      "grad_norm": 1.712923362383969,
      "learning_rate": 1.9222989946624053e-06,
      "loss": 0.3068,
      "step": 24690
    },
    {
      "epoch": 1.3165960395511846,
      "grad_norm": 1.6096277019474694,
      "learning_rate": 1.9214531917013537e-06,
      "loss": 0.2952,
      "step": 24700
    },
    {
      "epoch": 1.317129074385011,
      "grad_norm": 1.7770554268339862,
      "learning_rate": 1.920607243230017e-06,
      "loss": 0.2926,
      "step": 24710
    },
    {
      "epoch": 1.3176621092188374,
      "grad_norm": 1.71902939999981,
      "learning_rate": 1.9197611495404657e-06,
      "loss": 0.2882,
      "step": 24720
    },
    {
      "epoch": 1.3181951440526638,
      "grad_norm": 1.9011281934636743,
      "learning_rate": 1.918914910924821e-06,
      "loss": 0.2987,
      "step": 24730
    },
    {
      "epoch": 1.3187281788864902,
      "grad_norm": 1.8191261316065641,
      "learning_rate": 1.9180685276752544e-06,
      "loss": 0.294,
      "step": 24740
    },
    {
      "epoch": 1.3192612137203166,
      "grad_norm": 1.672957086841675,
      "learning_rate": 1.917222000083986e-06,
      "loss": 0.3103,
      "step": 24750
    },
    {
      "epoch": 1.319794248554143,
      "grad_norm": 1.659211577025086,
      "learning_rate": 1.9163753284432878e-06,
      "loss": 0.3045,
      "step": 24760
    },
    {
      "epoch": 1.3203272833879693,
      "grad_norm": 1.6602295046304174,
      "learning_rate": 1.9155285130454794e-06,
      "loss": 0.2941,
      "step": 24770
    },
    {
      "epoch": 1.3208603182217957,
      "grad_norm": 1.6376469812095842,
      "learning_rate": 1.914681554182931e-06,
      "loss": 0.2882,
      "step": 24780
    },
    {
      "epoch": 1.321393353055622,
      "grad_norm": 1.6196913179364119,
      "learning_rate": 1.913834452148063e-06,
      "loss": 0.3026,
      "step": 24790
    },
    {
      "epoch": 1.3219263878894485,
      "grad_norm": 1.6322955649802322,
      "learning_rate": 1.912987207233344e-06,
      "loss": 0.2953,
      "step": 24800
    },
    {
      "epoch": 1.3224594227232749,
      "grad_norm": 1.6843421006635075,
      "learning_rate": 1.912139819731293e-06,
      "loss": 0.3038,
      "step": 24810
    },
    {
      "epoch": 1.3229924575571013,
      "grad_norm": 1.6885799190564523,
      "learning_rate": 1.9112922899344765e-06,
      "loss": 0.2802,
      "step": 24820
    },
    {
      "epoch": 1.3235254923909276,
      "grad_norm": 1.693764267732668,
      "learning_rate": 1.9104446181355132e-06,
      "loss": 0.3098,
      "step": 24830
    },
    {
      "epoch": 1.324058527224754,
      "grad_norm": 1.786059441085882,
      "learning_rate": 1.9095968046270675e-06,
      "loss": 0.3038,
      "step": 24840
    },
    {
      "epoch": 1.3245915620585804,
      "grad_norm": 1.4806942027984702,
      "learning_rate": 1.9087488497018554e-06,
      "loss": 0.3007,
      "step": 24850
    },
    {
      "epoch": 1.3251245968924068,
      "grad_norm": 1.8650841428094405,
      "learning_rate": 1.9079007536526385e-06,
      "loss": 0.3197,
      "step": 24860
    },
    {
      "epoch": 1.3256576317262332,
      "grad_norm": 1.698426286117286,
      "learning_rate": 1.9070525167722315e-06,
      "loss": 0.2976,
      "step": 24870
    },
    {
      "epoch": 1.3261906665600596,
      "grad_norm": 1.8221646509067384,
      "learning_rate": 1.9062041393534939e-06,
      "loss": 0.2949,
      "step": 24880
    },
    {
      "epoch": 1.326723701393886,
      "grad_norm": 1.6906448973418389,
      "learning_rate": 1.905355621689336e-06,
      "loss": 0.3022,
      "step": 24890
    },
    {
      "epoch": 1.3272567362277126,
      "grad_norm": 1.7774432607120814,
      "learning_rate": 1.9045069640727152e-06,
      "loss": 0.2932,
      "step": 24900
    },
    {
      "epoch": 1.327789771061539,
      "grad_norm": 1.6379111782146818,
      "learning_rate": 1.9036581667966376e-06,
      "loss": 0.2892,
      "step": 24910
    },
    {
      "epoch": 1.3283228058953653,
      "grad_norm": 1.9302657990680174,
      "learning_rate": 1.902809230154158e-06,
      "loss": 0.2985,
      "step": 24920
    },
    {
      "epoch": 1.3288558407291917,
      "grad_norm": 1.7767253763951694,
      "learning_rate": 1.9019601544383792e-06,
      "loss": 0.2938,
      "step": 24930
    },
    {
      "epoch": 1.329388875563018,
      "grad_norm": 1.7220449827622353,
      "learning_rate": 1.901110939942451e-06,
      "loss": 0.3018,
      "step": 24940
    },
    {
      "epoch": 1.3299219103968445,
      "grad_norm": 1.7537968927336551,
      "learning_rate": 1.9002615869595727e-06,
      "loss": 0.315,
      "step": 24950
    },
    {
      "epoch": 1.3304549452306709,
      "grad_norm": 1.6801148746324377,
      "learning_rate": 1.8994120957829902e-06,
      "loss": 0.3029,
      "step": 24960
    },
    {
      "epoch": 1.3309879800644973,
      "grad_norm": 1.5738669557039318,
      "learning_rate": 1.898562466705998e-06,
      "loss": 0.298,
      "step": 24970
    },
    {
      "epoch": 1.3315210148983236,
      "grad_norm": 1.527901098261024,
      "learning_rate": 1.897712700021937e-06,
      "loss": 0.3003,
      "step": 24980
    },
    {
      "epoch": 1.33205404973215,
      "grad_norm": 1.8627992857368125,
      "learning_rate": 1.8968627960241965e-06,
      "loss": 0.2937,
      "step": 24990
    },
    {
      "epoch": 1.3325870845659764,
      "grad_norm": 1.5427152804467288,
      "learning_rate": 1.8960127550062133e-06,
      "loss": 0.3114,
      "step": 25000
    },
    {
      "epoch": 1.3331201193998028,
      "grad_norm": 1.8199112210337107,
      "learning_rate": 1.8951625772614712e-06,
      "loss": 0.2893,
      "step": 25010
    },
    {
      "epoch": 1.3336531542336292,
      "grad_norm": 1.711954876820095,
      "learning_rate": 1.894312263083502e-06,
      "loss": 0.3002,
      "step": 25020
    },
    {
      "epoch": 1.3341861890674556,
      "grad_norm": 1.5829722363345358,
      "learning_rate": 1.8934618127658822e-06,
      "loss": 0.2878,
      "step": 25030
    },
    {
      "epoch": 1.334719223901282,
      "grad_norm": 1.6146549782155446,
      "learning_rate": 1.8926112266022383e-06,
      "loss": 0.2905,
      "step": 25040
    },
    {
      "epoch": 1.3352522587351083,
      "grad_norm": 1.7308783549678697,
      "learning_rate": 1.891760504886242e-06,
      "loss": 0.2931,
      "step": 25050
    },
    {
      "epoch": 1.3357852935689347,
      "grad_norm": 1.7538149980414772,
      "learning_rate": 1.8909096479116126e-06,
      "loss": 0.2987,
      "step": 25060
    },
    {
      "epoch": 1.336318328402761,
      "grad_norm": 1.7923175460136251,
      "learning_rate": 1.8900586559721145e-06,
      "loss": 0.3036,
      "step": 25070
    },
    {
      "epoch": 1.3368513632365875,
      "grad_norm": 1.6176174897012252,
      "learning_rate": 1.889207529361561e-06,
      "loss": 0.2941,
      "step": 25080
    },
    {
      "epoch": 1.3373843980704139,
      "grad_norm": 1.7268935325575119,
      "learning_rate": 1.8883562683738106e-06,
      "loss": 0.2987,
      "step": 25090
    },
    {
      "epoch": 1.3379174329042403,
      "grad_norm": 1.6814566738927406,
      "learning_rate": 1.8875048733027682e-06,
      "loss": 0.2935,
      "step": 25100
    },
    {
      "epoch": 1.3384504677380666,
      "grad_norm": 1.6084892383698028,
      "learning_rate": 1.8866533444423852e-06,
      "loss": 0.2952,
      "step": 25110
    },
    {
      "epoch": 1.338983502571893,
      "grad_norm": 1.5047901580876593,
      "learning_rate": 1.8858016820866589e-06,
      "loss": 0.2971,
      "step": 25120
    },
    {
      "epoch": 1.3395165374057194,
      "grad_norm": 1.6268242355973896,
      "learning_rate": 1.8849498865296328e-06,
      "loss": 0.2988,
      "step": 25130
    },
    {
      "epoch": 1.3400495722395458,
      "grad_norm": 1.7294076091299029,
      "learning_rate": 1.8840979580653976e-06,
      "loss": 0.2984,
      "step": 25140
    },
    {
      "epoch": 1.3405826070733722,
      "grad_norm": 1.7472393140295948,
      "learning_rate": 1.883245896988088e-06,
      "loss": 0.3122,
      "step": 25150
    },
    {
      "epoch": 1.3411156419071986,
      "grad_norm": 1.837804481892729,
      "learning_rate": 1.882393703591885e-06,
      "loss": 0.3048,
      "step": 25160
    },
    {
      "epoch": 1.341648676741025,
      "grad_norm": 1.7821030265278661,
      "learning_rate": 1.8815413781710162e-06,
      "loss": 0.2914,
      "step": 25170
    },
    {
      "epoch": 1.3421817115748516,
      "grad_norm": 1.6991749907385412,
      "learning_rate": 1.8806889210197539e-06,
      "loss": 0.3018,
      "step": 25180
    },
    {
      "epoch": 1.342714746408678,
      "grad_norm": 1.5966895156481757,
      "learning_rate": 1.8798363324324165e-06,
      "loss": 0.3024,
      "step": 25190
    },
    {
      "epoch": 1.3432477812425043,
      "grad_norm": 1.4917889160441473,
      "learning_rate": 1.878983612703367e-06,
      "loss": 0.2943,
      "step": 25200
    },
    {
      "epoch": 1.3437808160763307,
      "grad_norm": 1.6929870555282298,
      "learning_rate": 1.8781307621270141e-06,
      "loss": 0.2979,
      "step": 25210
    },
    {
      "epoch": 1.344313850910157,
      "grad_norm": 1.7982102288359176,
      "learning_rate": 1.877277780997812e-06,
      "loss": 0.2994,
      "step": 25220
    },
    {
      "epoch": 1.3448468857439835,
      "grad_norm": 1.6044560556604885,
      "learning_rate": 1.8764246696102599e-06,
      "loss": 0.2948,
      "step": 25230
    },
    {
      "epoch": 1.3453799205778099,
      "grad_norm": 1.7623280762646298,
      "learning_rate": 1.875571428258901e-06,
      "loss": 0.3108,
      "step": 25240
    },
    {
      "epoch": 1.3459129554116362,
      "grad_norm": 1.8527228921505159,
      "learning_rate": 1.874718057238324e-06,
      "loss": 0.3151,
      "step": 25250
    },
    {
      "epoch": 1.3464459902454626,
      "grad_norm": 1.6735918364759137,
      "learning_rate": 1.8738645568431634e-06,
      "loss": 0.3002,
      "step": 25260
    },
    {
      "epoch": 1.346979025079289,
      "grad_norm": 1.563025738268665,
      "learning_rate": 1.8730109273680973e-06,
      "loss": 0.3072,
      "step": 25270
    },
    {
      "epoch": 1.3475120599131154,
      "grad_norm": 1.6396888735271204,
      "learning_rate": 1.8721571691078476e-06,
      "loss": 0.2936,
      "step": 25280
    },
    {
      "epoch": 1.3480450947469418,
      "grad_norm": 1.6498885530892358,
      "learning_rate": 1.8713032823571826e-06,
      "loss": 0.2965,
      "step": 25290
    },
    {
      "epoch": 1.3485781295807682,
      "grad_norm": 1.7885364385884202,
      "learning_rate": 1.870449267410913e-06,
      "loss": 0.2996,
      "step": 25300
    },
    {
      "epoch": 1.3491111644145946,
      "grad_norm": 1.724548081568859,
      "learning_rate": 1.869595124563896e-06,
      "loss": 0.2999,
      "step": 25310
    },
    {
      "epoch": 1.349644199248421,
      "grad_norm": 1.653491280163915,
      "learning_rate": 1.8687408541110303e-06,
      "loss": 0.2952,
      "step": 25320
    },
    {
      "epoch": 1.3501772340822473,
      "grad_norm": 1.8524470987183679,
      "learning_rate": 1.8678864563472606e-06,
      "loss": 0.2931,
      "step": 25330
    },
    {
      "epoch": 1.3507102689160737,
      "grad_norm": 1.7025876542375056,
      "learning_rate": 1.867031931567575e-06,
      "loss": 0.2834,
      "step": 25340
    },
    {
      "epoch": 1.3512433037499,
      "grad_norm": 1.591168911517188,
      "learning_rate": 1.8661772800670054e-06,
      "loss": 0.2844,
      "step": 25350
    },
    {
      "epoch": 1.3517763385837265,
      "grad_norm": 1.5023369096934813,
      "learning_rate": 1.8653225021406272e-06,
      "loss": 0.3004,
      "step": 25360
    },
    {
      "epoch": 1.3523093734175529,
      "grad_norm": 1.7902580587599353,
      "learning_rate": 1.8644675980835603e-06,
      "loss": 0.2903,
      "step": 25370
    },
    {
      "epoch": 1.3528424082513792,
      "grad_norm": 1.7956207687859305,
      "learning_rate": 1.8636125681909669e-06,
      "loss": 0.2983,
      "step": 25380
    },
    {
      "epoch": 1.3533754430852056,
      "grad_norm": 1.8986260767342624,
      "learning_rate": 1.8627574127580535e-06,
      "loss": 0.2963,
      "step": 25390
    },
    {
      "epoch": 1.353908477919032,
      "grad_norm": 1.8732280093213718,
      "learning_rate": 1.8619021320800703e-06,
      "loss": 0.299,
      "step": 25400
    },
    {
      "epoch": 1.3544415127528584,
      "grad_norm": 1.6210959314138322,
      "learning_rate": 1.86104672645231e-06,
      "loss": 0.3052,
      "step": 25410
    },
    {
      "epoch": 1.3549745475866848,
      "grad_norm": 1.6340301583357382,
      "learning_rate": 1.860191196170108e-06,
      "loss": 0.3024,
      "step": 25420
    },
    {
      "epoch": 1.3555075824205112,
      "grad_norm": 1.6351253586406966,
      "learning_rate": 1.8593355415288437e-06,
      "loss": 0.2914,
      "step": 25430
    },
    {
      "epoch": 1.3560406172543376,
      "grad_norm": 1.6417334726401889,
      "learning_rate": 1.8584797628239394e-06,
      "loss": 0.3036,
      "step": 25440
    },
    {
      "epoch": 1.356573652088164,
      "grad_norm": 1.7706793869060462,
      "learning_rate": 1.8576238603508596e-06,
      "loss": 0.3044,
      "step": 25450
    },
    {
      "epoch": 1.3571066869219903,
      "grad_norm": 1.702811056499645,
      "learning_rate": 1.856767834405112e-06,
      "loss": 0.2985,
      "step": 25460
    },
    {
      "epoch": 1.3576397217558167,
      "grad_norm": 1.5455985609970666,
      "learning_rate": 1.855911685282247e-06,
      "loss": 0.2955,
      "step": 25470
    },
    {
      "epoch": 1.358172756589643,
      "grad_norm": 1.713875855767372,
      "learning_rate": 1.8550554132778573e-06,
      "loss": 0.2993,
      "step": 25480
    },
    {
      "epoch": 1.3587057914234695,
      "grad_norm": 1.8676661657624682,
      "learning_rate": 1.854199018687578e-06,
      "loss": 0.3039,
      "step": 25490
    },
    {
      "epoch": 1.3592388262572959,
      "grad_norm": 1.9242276623718506,
      "learning_rate": 1.8533425018070863e-06,
      "loss": 0.3137,
      "step": 25500
    },
    {
      "epoch": 1.3597718610911222,
      "grad_norm": 1.5109209459416384,
      "learning_rate": 1.8524858629321022e-06,
      "loss": 0.292,
      "step": 25510
    },
    {
      "epoch": 1.3603048959249486,
      "grad_norm": 1.7736688794176796,
      "learning_rate": 1.8516291023583874e-06,
      "loss": 0.3006,
      "step": 25520
    },
    {
      "epoch": 1.360837930758775,
      "grad_norm": 1.5974972156106002,
      "learning_rate": 1.8507722203817463e-06,
      "loss": 0.2926,
      "step": 25530
    },
    {
      "epoch": 1.3613709655926014,
      "grad_norm": 1.709460510599122,
      "learning_rate": 1.8499152172980238e-06,
      "loss": 0.3078,
      "step": 25540
    },
    {
      "epoch": 1.3619040004264278,
      "grad_norm": 1.6988453871414007,
      "learning_rate": 1.8490580934031076e-06,
      "loss": 0.2943,
      "step": 25550
    },
    {
      "epoch": 1.3624370352602542,
      "grad_norm": 1.6459710862486738,
      "learning_rate": 1.8482008489929276e-06,
      "loss": 0.2978,
      "step": 25560
    },
    {
      "epoch": 1.3629700700940806,
      "grad_norm": 1.9608193129331608,
      "learning_rate": 1.8473434843634545e-06,
      "loss": 0.3049,
      "step": 25570
    },
    {
      "epoch": 1.363503104927907,
      "grad_norm": 1.5874307082002002,
      "learning_rate": 1.8464859998107007e-06,
      "loss": 0.307,
      "step": 25580
    },
    {
      "epoch": 1.3640361397617333,
      "grad_norm": 1.6431515878646898,
      "learning_rate": 1.8456283956307193e-06,
      "loss": 0.3006,
      "step": 25590
    },
    {
      "epoch": 1.3645691745955597,
      "grad_norm": 1.7186107937097803,
      "learning_rate": 1.8447706721196062e-06,
      "loss": 0.3008,
      "step": 25600
    },
    {
      "epoch": 1.365102209429386,
      "grad_norm": 1.564724932901496,
      "learning_rate": 1.8439128295734973e-06,
      "loss": 0.2938,
      "step": 25610
    },
    {
      "epoch": 1.3656352442632125,
      "grad_norm": 1.7183545956361774,
      "learning_rate": 1.8430548682885707e-06,
      "loss": 0.2904,
      "step": 25620
    },
    {
      "epoch": 1.3661682790970389,
      "grad_norm": 1.7157679817068054,
      "learning_rate": 1.8421967885610438e-06,
      "loss": 0.2816,
      "step": 25630
    },
    {
      "epoch": 1.3667013139308652,
      "grad_norm": 1.7082403829283734,
      "learning_rate": 1.8413385906871765e-06,
      "loss": 0.2883,
      "step": 25640
    },
    {
      "epoch": 1.3672343487646919,
      "grad_norm": 1.828671899313489,
      "learning_rate": 1.840480274963269e-06,
      "loss": 0.3079,
      "step": 25650
    },
    {
      "epoch": 1.3677673835985182,
      "grad_norm": 1.6838995679223723,
      "learning_rate": 1.8396218416856618e-06,
      "loss": 0.2857,
      "step": 25660
    },
    {
      "epoch": 1.3683004184323446,
      "grad_norm": 1.5821568461220046,
      "learning_rate": 1.838763291150736e-06,
      "loss": 0.3033,
      "step": 25670
    },
    {
      "epoch": 1.368833453266171,
      "grad_norm": 1.6990131419897607,
      "learning_rate": 1.8379046236549142e-06,
      "loss": 0.2913,
      "step": 25680
    },
    {
      "epoch": 1.3693664880999974,
      "grad_norm": 1.4897284878011985,
      "learning_rate": 1.8370458394946576e-06,
      "loss": 0.296,
      "step": 25690
    },
    {
      "epoch": 1.3698995229338238,
      "grad_norm": 1.5887160957408233,
      "learning_rate": 1.83618693896647e-06,
      "loss": 0.3061,
      "step": 25700
    },
    {
      "epoch": 1.3704325577676502,
      "grad_norm": 1.7836494945522463,
      "learning_rate": 1.835327922366893e-06,
      "loss": 0.2931,
      "step": 25710
    },
    {
      "epoch": 1.3709655926014765,
      "grad_norm": 1.7793882903729614,
      "learning_rate": 1.8344687899925097e-06,
      "loss": 0.3023,
      "step": 25720
    },
    {
      "epoch": 1.371498627435303,
      "grad_norm": 1.7685174231779794,
      "learning_rate": 1.8336095421399428e-06,
      "loss": 0.2964,
      "step": 25730
    },
    {
      "epoch": 1.3720316622691293,
      "grad_norm": 1.5550968773752014,
      "learning_rate": 1.8327501791058555e-06,
      "loss": 0.3004,
      "step": 25740
    },
    {
      "epoch": 1.3725646971029557,
      "grad_norm": 1.9417001033961754,
      "learning_rate": 1.8318907011869493e-06,
      "loss": 0.2946,
      "step": 25750
    },
    {
      "epoch": 1.373097731936782,
      "grad_norm": 1.565186024570825,
      "learning_rate": 1.8310311086799662e-06,
      "loss": 0.3029,
      "step": 25760
    },
    {
      "epoch": 1.3736307667706085,
      "grad_norm": 1.4711919777433935,
      "learning_rate": 1.8301714018816884e-06,
      "loss": 0.2989,
      "step": 25770
    },
    {
      "epoch": 1.3741638016044349,
      "grad_norm": 1.716137265923244,
      "learning_rate": 1.8293115810889365e-06,
      "loss": 0.2977,
      "step": 25780
    },
    {
      "epoch": 1.3746968364382612,
      "grad_norm": 1.7048689188613573,
      "learning_rate": 1.828451646598572e-06,
      "loss": 0.3051,
      "step": 25790
    },
    {
      "epoch": 1.3752298712720876,
      "grad_norm": 1.7180382734650073,
      "learning_rate": 1.8275915987074928e-06,
      "loss": 0.297,
      "step": 25800
    },
    {
      "epoch": 1.375762906105914,
      "grad_norm": 1.8728526945565789,
      "learning_rate": 1.8267314377126391e-06,
      "loss": 0.3052,
      "step": 25810
    },
    {
      "epoch": 1.3762959409397404,
      "grad_norm": 1.6302356608129749,
      "learning_rate": 1.8258711639109882e-06,
      "loss": 0.2937,
      "step": 25820
    },
    {
      "epoch": 1.3768289757735668,
      "grad_norm": 1.5227128346833096,
      "learning_rate": 1.825010777599557e-06,
      "loss": 0.2911,
      "step": 25830
    },
    {
      "epoch": 1.3773620106073932,
      "grad_norm": 1.9115662639842952,
      "learning_rate": 1.8241502790754012e-06,
      "loss": 0.306,
      "step": 25840
    },
    {
      "epoch": 1.3778950454412195,
      "grad_norm": 1.6411381274129313,
      "learning_rate": 1.823289668635615e-06,
      "loss": 0.3059,
      "step": 25850
    },
    {
      "epoch": 1.378428080275046,
      "grad_norm": 1.8273017090723045,
      "learning_rate": 1.8224289465773316e-06,
      "loss": 0.3151,
      "step": 25860
    },
    {
      "epoch": 1.3789611151088723,
      "grad_norm": 1.5706508452865011,
      "learning_rate": 1.821568113197723e-06,
      "loss": 0.2968,
      "step": 25870
    },
    {
      "epoch": 1.3794941499426987,
      "grad_norm": 1.8457855322102203,
      "learning_rate": 1.8207071687939989e-06,
      "loss": 0.304,
      "step": 25880
    },
    {
      "epoch": 1.380027184776525,
      "grad_norm": 1.6757696964366564,
      "learning_rate": 1.819846113663407e-06,
      "loss": 0.2978,
      "step": 25890
    },
    {
      "epoch": 1.3805602196103515,
      "grad_norm": 1.615616358562519,
      "learning_rate": 1.8189849481032346e-06,
      "loss": 0.3018,
      "step": 25900
    },
    {
      "epoch": 1.3810932544441779,
      "grad_norm": 1.7307368199348938,
      "learning_rate": 1.8181236724108069e-06,
      "loss": 0.2963,
      "step": 25910
    },
    {
      "epoch": 1.3816262892780042,
      "grad_norm": 1.6240940529558976,
      "learning_rate": 1.8172622868834854e-06,
      "loss": 0.3034,
      "step": 25920
    },
    {
      "epoch": 1.3821593241118308,
      "grad_norm": 1.6558605846339667,
      "learning_rate": 1.8164007918186712e-06,
      "loss": 0.3024,
      "step": 25930
    },
    {
      "epoch": 1.3826923589456572,
      "grad_norm": 1.6375702676984762,
      "learning_rate": 1.8155391875138032e-06,
      "loss": 0.3091,
      "step": 25940
    },
    {
      "epoch": 1.3832253937794836,
      "grad_norm": 1.759002891873053,
      "learning_rate": 1.8146774742663573e-06,
      "loss": 0.2919,
      "step": 25950
    },
    {
      "epoch": 1.38375842861331,
      "grad_norm": 1.8305215960155221,
      "learning_rate": 1.8138156523738474e-06,
      "loss": 0.2926,
      "step": 25960
    },
    {
      "epoch": 1.3842914634471364,
      "grad_norm": 1.6636208365933962,
      "learning_rate": 1.8129537221338247e-06,
      "loss": 0.3015,
      "step": 25970
    },
    {
      "epoch": 1.3848244982809628,
      "grad_norm": 1.8181836112710485,
      "learning_rate": 1.8120916838438775e-06,
      "loss": 0.2938,
      "step": 25980
    },
    {
      "epoch": 1.3853575331147892,
      "grad_norm": 1.7678384597914825,
      "learning_rate": 1.8112295378016326e-06,
      "loss": 0.2881,
      "step": 25990
    },
    {
      "epoch": 1.3858905679486155,
      "grad_norm": 1.546534730954037,
      "learning_rate": 1.8103672843047533e-06,
      "loss": 0.2922,
      "step": 26000
    },
    {
      "epoch": 1.386423602782442,
      "grad_norm": 1.758140222906188,
      "learning_rate": 1.8095049236509389e-06,
      "loss": 0.293,
      "step": 26010
    },
    {
      "epoch": 1.3869566376162683,
      "grad_norm": 1.7588011087431463,
      "learning_rate": 1.8086424561379277e-06,
      "loss": 0.298,
      "step": 26020
    },
    {
      "epoch": 1.3874896724500947,
      "grad_norm": 1.683049047556465,
      "learning_rate": 1.8077798820634931e-06,
      "loss": 0.3055,
      "step": 26030
    },
    {
      "epoch": 1.388022707283921,
      "grad_norm": 1.673331396903691,
      "learning_rate": 1.806917201725447e-06,
      "loss": 0.3043,
      "step": 26040
    },
    {
      "epoch": 1.3885557421177475,
      "grad_norm": 1.8297760493562283,
      "learning_rate": 1.8060544154216366e-06,
      "loss": 0.2976,
      "step": 26050
    },
    {
      "epoch": 1.3890887769515738,
      "grad_norm": 1.8201694755914335,
      "learning_rate": 1.805191523449946e-06,
      "loss": 0.2968,
      "step": 26060
    },
    {
      "epoch": 1.3896218117854002,
      "grad_norm": 1.8655101143362887,
      "learning_rate": 1.804328526108296e-06,
      "loss": 0.3,
      "step": 26070
    },
    {
      "epoch": 1.3901548466192266,
      "grad_norm": 1.7353512284273505,
      "learning_rate": 1.8034654236946447e-06,
      "loss": 0.2967,
      "step": 26080
    },
    {
      "epoch": 1.390687881453053,
      "grad_norm": 1.55442410359212,
      "learning_rate": 1.8026022165069843e-06,
      "loss": 0.2971,
      "step": 26090
    },
    {
      "epoch": 1.3912209162868794,
      "grad_norm": 1.582479953305048,
      "learning_rate": 1.801738904843345e-06,
      "loss": 0.2967,
      "step": 26100
    },
    {
      "epoch": 1.3917539511207058,
      "grad_norm": 1.6166498323898875,
      "learning_rate": 1.800875489001792e-06,
      "loss": 0.2832,
      "step": 26110
    },
    {
      "epoch": 1.3922869859545322,
      "grad_norm": 1.7175353069476316,
      "learning_rate": 1.8000119692804274e-06,
      "loss": 0.3019,
      "step": 26120
    },
    {
      "epoch": 1.3928200207883585,
      "grad_norm": 1.6983601849676904,
      "learning_rate": 1.7991483459773888e-06,
      "loss": 0.2929,
      "step": 26130
    },
    {
      "epoch": 1.393353055622185,
      "grad_norm": 1.6477328045513475,
      "learning_rate": 1.798284619390849e-06,
      "loss": 0.2942,
      "step": 26140
    },
    {
      "epoch": 1.3938860904560113,
      "grad_norm": 1.8751332377723902,
      "learning_rate": 1.797420789819017e-06,
      "loss": 0.2973,
      "step": 26150
    },
    {
      "epoch": 1.3944191252898377,
      "grad_norm": 1.5831994320911136,
      "learning_rate": 1.7965568575601374e-06,
      "loss": 0.2845,
      "step": 26160
    },
    {
      "epoch": 1.394952160123664,
      "grad_norm": 1.6578406734879945,
      "learning_rate": 1.7956928229124902e-06,
      "loss": 0.3113,
      "step": 26170
    },
    {
      "epoch": 1.3954851949574905,
      "grad_norm": 1.5467336199552801,
      "learning_rate": 1.7948286861743908e-06,
      "loss": 0.3102,
      "step": 26180
    },
    {
      "epoch": 1.3960182297913168,
      "grad_norm": 2.019678522633885,
      "learning_rate": 1.7939644476441897e-06,
      "loss": 0.2953,
      "step": 26190
    },
    {
      "epoch": 1.3965512646251432,
      "grad_norm": 1.4828441186240586,
      "learning_rate": 1.7931001076202722e-06,
      "loss": 0.3006,
      "step": 26200
    },
    {
      "epoch": 1.3970842994589696,
      "grad_norm": 1.4708125917698387,
      "learning_rate": 1.792235666401059e-06,
      "loss": 0.2906,
      "step": 26210
    },
    {
      "epoch": 1.397617334292796,
      "grad_norm": 1.7290179357743,
      "learning_rate": 1.7913711242850065e-06,
      "loss": 0.3048,
      "step": 26220
    },
    {
      "epoch": 1.3981503691266224,
      "grad_norm": 1.6172558371873067,
      "learning_rate": 1.7905064815706043e-06,
      "loss": 0.2972,
      "step": 26230
    },
    {
      "epoch": 1.3986834039604488,
      "grad_norm": 1.558128305715451,
      "learning_rate": 1.7896417385563782e-06,
      "loss": 0.3,
      "step": 26240
    },
    {
      "epoch": 1.3992164387942752,
      "grad_norm": 1.6550457709388975,
      "learning_rate": 1.7887768955408876e-06,
      "loss": 0.2977,
      "step": 26250
    },
    {
      "epoch": 1.3997494736281015,
      "grad_norm": 1.7918889845440522,
      "learning_rate": 1.787911952822728e-06,
      "loss": 0.2961,
      "step": 26260
    },
    {
      "epoch": 1.400282508461928,
      "grad_norm": 1.4629794316944487,
      "learning_rate": 1.7870469107005264e-06,
      "loss": 0.3064,
      "step": 26270
    },
    {
      "epoch": 1.4008155432957543,
      "grad_norm": 1.8801017899915509,
      "learning_rate": 1.7861817694729472e-06,
      "loss": 0.3073,
      "step": 26280
    },
    {
      "epoch": 1.4013485781295807,
      "grad_norm": 1.6393017748426992,
      "learning_rate": 1.7853165294386877e-06,
      "loss": 0.3057,
      "step": 26290
    },
    {
      "epoch": 1.401881612963407,
      "grad_norm": 1.6372430845091208,
      "learning_rate": 1.7844511908964791e-06,
      "loss": 0.3062,
      "step": 26300
    },
    {
      "epoch": 1.4024146477972335,
      "grad_norm": 1.6935876510269459,
      "learning_rate": 1.7835857541450863e-06,
      "loss": 0.2954,
      "step": 26310
    },
    {
      "epoch": 1.4029476826310598,
      "grad_norm": 1.7678950660144939,
      "learning_rate": 1.7827202194833094e-06,
      "loss": 0.2951,
      "step": 26320
    },
    {
      "epoch": 1.4034807174648862,
      "grad_norm": 1.4629954158245355,
      "learning_rate": 1.7818545872099815e-06,
      "loss": 0.29,
      "step": 26330
    },
    {
      "epoch": 1.4040137522987126,
      "grad_norm": 2.078019597319422,
      "learning_rate": 1.7809888576239693e-06,
      "loss": 0.3012,
      "step": 26340
    },
    {
      "epoch": 1.404546787132539,
      "grad_norm": 1.7719334412054681,
      "learning_rate": 1.7801230310241734e-06,
      "loss": 0.2999,
      "step": 26350
    },
    {
      "epoch": 1.4050798219663654,
      "grad_norm": 1.7377464565240786,
      "learning_rate": 1.7792571077095275e-06,
      "loss": 0.2998,
      "step": 26360
    },
    {
      "epoch": 1.4056128568001918,
      "grad_norm": 1.6352342111484524,
      "learning_rate": 1.7783910879789991e-06,
      "loss": 0.2964,
      "step": 26370
    },
    {
      "epoch": 1.4061458916340182,
      "grad_norm": 1.697009670134385,
      "learning_rate": 1.777524972131589e-06,
      "loss": 0.2975,
      "step": 26380
    },
    {
      "epoch": 1.4066789264678445,
      "grad_norm": 1.6718912433255346,
      "learning_rate": 1.7766587604663312e-06,
      "loss": 0.293,
      "step": 26390
    },
    {
      "epoch": 1.4072119613016711,
      "grad_norm": 1.8140347491014321,
      "learning_rate": 1.7757924532822922e-06,
      "loss": 0.3056,
      "step": 26400
    },
    {
      "epoch": 1.4077449961354975,
      "grad_norm": 1.7031312617346268,
      "learning_rate": 1.774926050878572e-06,
      "loss": 0.2985,
      "step": 26410
    },
    {
      "epoch": 1.408278030969324,
      "grad_norm": 1.5502865665642145,
      "learning_rate": 1.7740595535543035e-06,
      "loss": 0.2983,
      "step": 26420
    },
    {
      "epoch": 1.4088110658031503,
      "grad_norm": 1.7930078600790027,
      "learning_rate": 1.7731929616086523e-06,
      "loss": 0.298,
      "step": 26430
    },
    {
      "epoch": 1.4093441006369767,
      "grad_norm": 1.7710400672071704,
      "learning_rate": 1.772326275340816e-06,
      "loss": 0.2904,
      "step": 26440
    },
    {
      "epoch": 1.409877135470803,
      "grad_norm": 1.623529556403256,
      "learning_rate": 1.7714594950500264e-06,
      "loss": 0.3017,
      "step": 26450
    },
    {
      "epoch": 1.4104101703046295,
      "grad_norm": 1.6345170122712465,
      "learning_rate": 1.7705926210355458e-06,
      "loss": 0.2994,
      "step": 26460
    },
    {
      "epoch": 1.4109432051384558,
      "grad_norm": 1.5084067854427958,
      "learning_rate": 1.7697256535966705e-06,
      "loss": 0.3034,
      "step": 26470
    },
    {
      "epoch": 1.4114762399722822,
      "grad_norm": 1.6569700893461776,
      "learning_rate": 1.7688585930327278e-06,
      "loss": 0.2941,
      "step": 26480
    },
    {
      "epoch": 1.4120092748061086,
      "grad_norm": 1.714763331108559,
      "learning_rate": 1.7679914396430778e-06,
      "loss": 0.3001,
      "step": 26490
    },
    {
      "epoch": 1.412542309639935,
      "grad_norm": 1.7836240709395328,
      "learning_rate": 1.767124193727113e-06,
      "loss": 0.2961,
      "step": 26500
    },
    {
      "epoch": 1.4130753444737614,
      "grad_norm": 1.7018376151859356,
      "learning_rate": 1.766256855584257e-06,
      "loss": 0.2918,
      "step": 26510
    },
    {
      "epoch": 1.4136083793075878,
      "grad_norm": 1.7181227659134677,
      "learning_rate": 1.765389425513966e-06,
      "loss": 0.3075,
      "step": 26520
    },
    {
      "epoch": 1.4141414141414141,
      "grad_norm": 1.7638829892478762,
      "learning_rate": 1.7645219038157272e-06,
      "loss": 0.2881,
      "step": 26530
    },
    {
      "epoch": 1.4146744489752405,
      "grad_norm": 1.7173535893014005,
      "learning_rate": 1.7636542907890596e-06,
      "loss": 0.2994,
      "step": 26540
    },
    {
      "epoch": 1.415207483809067,
      "grad_norm": 1.5617235448068514,
      "learning_rate": 1.7627865867335146e-06,
      "loss": 0.2937,
      "step": 26550
    },
    {
      "epoch": 1.4157405186428933,
      "grad_norm": 1.7695225608047267,
      "learning_rate": 1.7619187919486743e-06,
      "loss": 0.2961,
      "step": 26560
    },
    {
      "epoch": 1.4162735534767197,
      "grad_norm": 1.8172435753323006,
      "learning_rate": 1.7610509067341514e-06,
      "loss": 0.2989,
      "step": 26570
    },
    {
      "epoch": 1.416806588310546,
      "grad_norm": 1.6783220829773293,
      "learning_rate": 1.7601829313895917e-06,
      "loss": 0.2981,
      "step": 26580
    },
    {
      "epoch": 1.4173396231443725,
      "grad_norm": 1.6915333985115013,
      "learning_rate": 1.75931486621467e-06,
      "loss": 0.298,
      "step": 26590
    },
    {
      "epoch": 1.4178726579781988,
      "grad_norm": 1.865018561949842,
      "learning_rate": 1.7584467115090942e-06,
      "loss": 0.2988,
      "step": 26600
    },
    {
      "epoch": 1.4184056928120252,
      "grad_norm": 1.8309259886049176,
      "learning_rate": 1.7575784675726015e-06,
      "loss": 0.2965,
      "step": 26610
    },
    {
      "epoch": 1.4189387276458516,
      "grad_norm": 1.6991754119084301,
      "learning_rate": 1.7567101347049603e-06,
      "loss": 0.2918,
      "step": 26620
    },
    {
      "epoch": 1.419471762479678,
      "grad_norm": 1.678854945907068,
      "learning_rate": 1.7558417132059702e-06,
      "loss": 0.2785,
      "step": 26630
    },
    {
      "epoch": 1.4200047973135044,
      "grad_norm": 1.7536621029688282,
      "learning_rate": 1.7549732033754614e-06,
      "loss": 0.3021,
      "step": 26640
    },
    {
      "epoch": 1.4205378321473308,
      "grad_norm": 1.9258702576188684,
      "learning_rate": 1.7541046055132934e-06,
      "loss": 0.2978,
      "step": 26650
    },
    {
      "epoch": 1.4210708669811571,
      "grad_norm": 1.7228957156370122,
      "learning_rate": 1.7532359199193575e-06,
      "loss": 0.3033,
      "step": 26660
    },
    {
      "epoch": 1.4216039018149835,
      "grad_norm": 1.585862859633546,
      "learning_rate": 1.7523671468935746e-06,
      "loss": 0.2949,
      "step": 26670
    },
    {
      "epoch": 1.4221369366488101,
      "grad_norm": 1.5090748696304355,
      "learning_rate": 1.751498286735896e-06,
      "loss": 0.299,
      "step": 26680
    },
    {
      "epoch": 1.4226699714826365,
      "grad_norm": 1.7709645874493618,
      "learning_rate": 1.7506293397463036e-06,
      "loss": 0.2898,
      "step": 26690
    },
    {
      "epoch": 1.423203006316463,
      "grad_norm": 1.8129163334222032,
      "learning_rate": 1.7497603062248078e-06,
      "loss": 0.2896,
      "step": 26700
    },
    {
      "epoch": 1.4237360411502893,
      "grad_norm": 1.7912628299896205,
      "learning_rate": 1.7488911864714504e-06,
      "loss": 0.3034,
      "step": 26710
    },
    {
      "epoch": 1.4242690759841157,
      "grad_norm": 1.6228269461486757,
      "learning_rate": 1.7480219807863019e-06,
      "loss": 0.3063,
      "step": 26720
    },
    {
      "epoch": 1.424802110817942,
      "grad_norm": 1.7485948212265539,
      "learning_rate": 1.7471526894694635e-06,
      "loss": 0.304,
      "step": 26730
    },
    {
      "epoch": 1.4253351456517684,
      "grad_norm": 1.7824154268628587,
      "learning_rate": 1.7462833128210645e-06,
      "loss": 0.2987,
      "step": 26740
    },
    {
      "epoch": 1.4258681804855948,
      "grad_norm": 1.8384270223840056,
      "learning_rate": 1.7454138511412656e-06,
      "loss": 0.3085,
      "step": 26750
    },
    {
      "epoch": 1.4264012153194212,
      "grad_norm": 1.715369316429233,
      "learning_rate": 1.744544304730255e-06,
      "loss": 0.2974,
      "step": 26760
    },
    {
      "epoch": 1.4269342501532476,
      "grad_norm": 1.8797341325758168,
      "learning_rate": 1.7436746738882517e-06,
      "loss": 0.3044,
      "step": 26770
    },
    {
      "epoch": 1.427467284987074,
      "grad_norm": 1.8423086639407293,
      "learning_rate": 1.742804958915503e-06,
      "loss": 0.2932,
      "step": 26780
    },
    {
      "epoch": 1.4280003198209004,
      "grad_norm": 1.605966035729234,
      "learning_rate": 1.7419351601122844e-06,
      "loss": 0.3009,
      "step": 26790
    },
    {
      "epoch": 1.4285333546547268,
      "grad_norm": 1.6855910361001112,
      "learning_rate": 1.7410652777789017e-06,
      "loss": 0.3024,
      "step": 26800
    },
    {
      "epoch": 1.4290663894885531,
      "grad_norm": 1.6547056730254632,
      "learning_rate": 1.7401953122156902e-06,
      "loss": 0.3004,
      "step": 26810
    },
    {
      "epoch": 1.4295994243223795,
      "grad_norm": 1.5932137378981208,
      "learning_rate": 1.739325263723012e-06,
      "loss": 0.2963,
      "step": 26820
    },
    {
      "epoch": 1.430132459156206,
      "grad_norm": 1.8332058492007164,
      "learning_rate": 1.7384551326012583e-06,
      "loss": 0.2997,
      "step": 26830
    },
    {
      "epoch": 1.4306654939900323,
      "grad_norm": 1.6335589615625263,
      "learning_rate": 1.7375849191508501e-06,
      "loss": 0.2997,
      "step": 26840
    },
    {
      "epoch": 1.4311985288238587,
      "grad_norm": 1.9018798996551556,
      "learning_rate": 1.7367146236722354e-06,
      "loss": 0.2991,
      "step": 26850
    },
    {
      "epoch": 1.431731563657685,
      "grad_norm": 1.830783303735447,
      "learning_rate": 1.7358442464658916e-06,
      "loss": 0.308,
      "step": 26860
    },
    {
      "epoch": 1.4322645984915114,
      "grad_norm": 1.8585369463117338,
      "learning_rate": 1.7349737878323234e-06,
      "loss": 0.3045,
      "step": 26870
    },
    {
      "epoch": 1.4327976333253378,
      "grad_norm": 1.674828166380364,
      "learning_rate": 1.7341032480720638e-06,
      "loss": 0.3041,
      "step": 26880
    },
    {
      "epoch": 1.4333306681591642,
      "grad_norm": 1.7175904573254501,
      "learning_rate": 1.733232627485675e-06,
      "loss": 0.2971,
      "step": 26890
    },
    {
      "epoch": 1.4338637029929906,
      "grad_norm": 1.681304584144117,
      "learning_rate": 1.7323619263737447e-06,
      "loss": 0.2928,
      "step": 26900
    },
    {
      "epoch": 1.434396737826817,
      "grad_norm": 1.9145423118374085,
      "learning_rate": 1.7314911450368915e-06,
      "loss": 0.2955,
      "step": 26910
    },
    {
      "epoch": 1.4349297726606434,
      "grad_norm": 1.411654315384082,
      "learning_rate": 1.7306202837757587e-06,
      "loss": 0.3025,
      "step": 26920
    },
    {
      "epoch": 1.4354628074944698,
      "grad_norm": 1.9406175068060867,
      "learning_rate": 1.729749342891019e-06,
      "loss": 0.3054,
      "step": 26930
    },
    {
      "epoch": 1.4359958423282961,
      "grad_norm": 1.8086314299800497,
      "learning_rate": 1.728878322683372e-06,
      "loss": 0.2978,
      "step": 26940
    },
    {
      "epoch": 1.4365288771621225,
      "grad_norm": 1.5945339211070315,
      "learning_rate": 1.7280072234535455e-06,
      "loss": 0.2967,
      "step": 26950
    },
    {
      "epoch": 1.437061911995949,
      "grad_norm": 1.7239929839957984,
      "learning_rate": 1.7271360455022926e-06,
      "loss": 0.2917,
      "step": 26960
    },
    {
      "epoch": 1.4375949468297753,
      "grad_norm": 1.7233501885608034,
      "learning_rate": 1.7262647891303958e-06,
      "loss": 0.3094,
      "step": 26970
    },
    {
      "epoch": 1.4381279816636017,
      "grad_norm": 1.9588769534508925,
      "learning_rate": 1.7253934546386637e-06,
      "loss": 0.3015,
      "step": 26980
    },
    {
      "epoch": 1.438661016497428,
      "grad_norm": 1.6982989302274034,
      "learning_rate": 1.7245220423279319e-06,
      "loss": 0.2858,
      "step": 26990
    },
    {
      "epoch": 1.4391940513312544,
      "grad_norm": 1.6486048024521598,
      "learning_rate": 1.7236505524990622e-06,
      "loss": 0.29,
      "step": 27000
    },
    {
      "epoch": 1.4397270861650808,
      "grad_norm": 1.6755161868593285,
      "learning_rate": 1.7227789854529447e-06,
      "loss": 0.3044,
      "step": 27010
    },
    {
      "epoch": 1.4402601209989072,
      "grad_norm": 1.718201720778143,
      "learning_rate": 1.7219073414904948e-06,
      "loss": 0.313,
      "step": 27020
    },
    {
      "epoch": 1.4407931558327336,
      "grad_norm": 1.737733157286925,
      "learning_rate": 1.7210356209126555e-06,
      "loss": 0.2938,
      "step": 27030
    },
    {
      "epoch": 1.44132619066656,
      "grad_norm": 1.9469823233023167,
      "learning_rate": 1.7201638240203952e-06,
      "loss": 0.2948,
      "step": 27040
    },
    {
      "epoch": 1.4418592255003864,
      "grad_norm": 1.6523027607812257,
      "learning_rate": 1.719291951114709e-06,
      "loss": 0.293,
      "step": 27050
    },
    {
      "epoch": 1.4423922603342128,
      "grad_norm": 1.9657408010892838,
      "learning_rate": 1.718420002496619e-06,
      "loss": 0.2869,
      "step": 27060
    },
    {
      "epoch": 1.4429252951680391,
      "grad_norm": 1.7535294375076989,
      "learning_rate": 1.7175479784671726e-06,
      "loss": 0.2973,
      "step": 27070
    },
    {
      "epoch": 1.4434583300018655,
      "grad_norm": 1.567368036356119,
      "learning_rate": 1.7166758793274435e-06,
      "loss": 0.2899,
      "step": 27080
    },
    {
      "epoch": 1.443991364835692,
      "grad_norm": 1.79695931807065,
      "learning_rate": 1.7158037053785313e-06,
      "loss": 0.3046,
      "step": 27090
    },
    {
      "epoch": 1.4445243996695183,
      "grad_norm": 1.5860986291818169,
      "learning_rate": 1.7149314569215609e-06,
      "loss": 0.2907,
      "step": 27100
    },
    {
      "epoch": 1.4450574345033447,
      "grad_norm": 1.833465580118085,
      "learning_rate": 1.7140591342576844e-06,
      "loss": 0.2952,
      "step": 27110
    },
    {
      "epoch": 1.445590469337171,
      "grad_norm": 1.583922563925103,
      "learning_rate": 1.7131867376880786e-06,
      "loss": 0.2915,
      "step": 27120
    },
    {
      "epoch": 1.4461235041709974,
      "grad_norm": 1.6714845641154914,
      "learning_rate": 1.7123142675139445e-06,
      "loss": 0.2991,
      "step": 27130
    },
    {
      "epoch": 1.4466565390048238,
      "grad_norm": 1.7227014574038273,
      "learning_rate": 1.7114417240365111e-06,
      "loss": 0.2837,
      "step": 27140
    },
    {
      "epoch": 1.4471895738386504,
      "grad_norm": 1.662503588245676,
      "learning_rate": 1.7105691075570311e-06,
      "loss": 0.3012,
      "step": 27150
    },
    {
      "epoch": 1.4477226086724768,
      "grad_norm": 1.7030517097273221,
      "learning_rate": 1.709696418376783e-06,
      "loss": 0.3009,
      "step": 27160
    },
    {
      "epoch": 1.4482556435063032,
      "grad_norm": 1.5937351659627172,
      "learning_rate": 1.7088236567970695e-06,
      "loss": 0.2923,
      "step": 27170
    },
    {
      "epoch": 1.4487886783401296,
      "grad_norm": 1.7278318252163622,
      "learning_rate": 1.7079508231192188e-06,
      "loss": 0.3085,
      "step": 27180
    },
    {
      "epoch": 1.449321713173956,
      "grad_norm": 1.6368988557855801,
      "learning_rate": 1.707077917644585e-06,
      "loss": 0.2912,
      "step": 27190
    },
    {
      "epoch": 1.4498547480077824,
      "grad_norm": 1.6791057003365149,
      "learning_rate": 1.7062049406745461e-06,
      "loss": 0.2871,
      "step": 27200
    },
    {
      "epoch": 1.4503877828416087,
      "grad_norm": 1.7093953675074673,
      "learning_rate": 1.7053318925105039e-06,
      "loss": 0.3,
      "step": 27210
    },
    {
      "epoch": 1.4509208176754351,
      "grad_norm": 1.8096692528204446,
      "learning_rate": 1.704458773453886e-06,
      "loss": 0.2955,
      "step": 27220
    },
    {
      "epoch": 1.4514538525092615,
      "grad_norm": 1.6433602154068048,
      "learning_rate": 1.7035855838061443e-06,
      "loss": 0.2983,
      "step": 27230
    },
    {
      "epoch": 1.451986887343088,
      "grad_norm": 1.9044622947576912,
      "learning_rate": 1.7027123238687548e-06,
      "loss": 0.288,
      "step": 27240
    },
    {
      "epoch": 1.4525199221769143,
      "grad_norm": 1.847103434506775,
      "learning_rate": 1.7018389939432185e-06,
      "loss": 0.2966,
      "step": 27250
    },
    {
      "epoch": 1.4530529570107407,
      "grad_norm": 1.7134411365194073,
      "learning_rate": 1.700965594331059e-06,
      "loss": 0.2986,
      "step": 27260
    },
    {
      "epoch": 1.453585991844567,
      "grad_norm": 1.6134127006346162,
      "learning_rate": 1.7000921253338256e-06,
      "loss": 0.2946,
      "step": 27270
    },
    {
      "epoch": 1.4541190266783934,
      "grad_norm": 1.5666513665906094,
      "learning_rate": 1.6992185872530904e-06,
      "loss": 0.2997,
      "step": 27280
    },
    {
      "epoch": 1.4546520615122198,
      "grad_norm": 1.610372338951449,
      "learning_rate": 1.6983449803904504e-06,
      "loss": 0.2971,
      "step": 27290
    },
    {
      "epoch": 1.4551850963460462,
      "grad_norm": 1.7723047540000008,
      "learning_rate": 1.6974713050475254e-06,
      "loss": 0.2987,
      "step": 27300
    },
    {
      "epoch": 1.4557181311798726,
      "grad_norm": 1.778586198140521,
      "learning_rate": 1.696597561525959e-06,
      "loss": 0.2853,
      "step": 27310
    },
    {
      "epoch": 1.456251166013699,
      "grad_norm": 1.6191848208021444,
      "learning_rate": 1.6957237501274193e-06,
      "loss": 0.2973,
      "step": 27320
    },
    {
      "epoch": 1.4567842008475254,
      "grad_norm": 1.556652193442352,
      "learning_rate": 1.6948498711535963e-06,
      "loss": 0.296,
      "step": 27330
    },
    {
      "epoch": 1.4573172356813517,
      "grad_norm": 1.802907753062817,
      "learning_rate": 1.6939759249062045e-06,
      "loss": 0.3092,
      "step": 27340
    },
    {
      "epoch": 1.4578502705151781,
      "grad_norm": 1.6794447937674568,
      "learning_rate": 1.6931019116869808e-06,
      "loss": 0.3113,
      "step": 27350
    },
    {
      "epoch": 1.4583833053490045,
      "grad_norm": 1.7767525767084973,
      "learning_rate": 1.6922278317976864e-06,
      "loss": 0.306,
      "step": 27360
    },
    {
      "epoch": 1.458916340182831,
      "grad_norm": 1.8303255899968982,
      "learning_rate": 1.6913536855401045e-06,
      "loss": 0.2956,
      "step": 27370
    },
    {
      "epoch": 1.4594493750166573,
      "grad_norm": 1.8539109764589967,
      "learning_rate": 1.6904794732160412e-06,
      "loss": 0.2883,
      "step": 27380
    },
    {
      "epoch": 1.4599824098504837,
      "grad_norm": 1.575562634112439,
      "learning_rate": 1.6896051951273258e-06,
      "loss": 0.3006,
      "step": 27390
    },
    {
      "epoch": 1.46051544468431,
      "grad_norm": 1.6053287187694665,
      "learning_rate": 1.6887308515758107e-06,
      "loss": 0.3018,
      "step": 27400
    },
    {
      "epoch": 1.4610484795181364,
      "grad_norm": 1.676656958360302,
      "learning_rate": 1.6878564428633698e-06,
      "loss": 0.3038,
      "step": 27410
    },
    {
      "epoch": 1.4615815143519628,
      "grad_norm": 1.764302001674795,
      "learning_rate": 1.6869819692919005e-06,
      "loss": 0.2885,
      "step": 27420
    },
    {
      "epoch": 1.4621145491857894,
      "grad_norm": 1.6274000913965454,
      "learning_rate": 1.6861074311633221e-06,
      "loss": 0.2887,
      "step": 27430
    },
    {
      "epoch": 1.4626475840196158,
      "grad_norm": 1.8137987646484843,
      "learning_rate": 1.6852328287795761e-06,
      "loss": 0.3024,
      "step": 27440
    },
    {
      "epoch": 1.4631806188534422,
      "grad_norm": 1.5585894595314522,
      "learning_rate": 1.6843581624426272e-06,
      "loss": 0.3044,
      "step": 27450
    },
    {
      "epoch": 1.4637136536872686,
      "grad_norm": 1.792163610404331,
      "learning_rate": 1.68348343245446e-06,
      "loss": 0.3081,
      "step": 27460
    },
    {
      "epoch": 1.464246688521095,
      "grad_norm": 1.8482431515830744,
      "learning_rate": 1.682608639117084e-06,
      "loss": 0.2996,
      "step": 27470
    },
    {
      "epoch": 1.4647797233549213,
      "grad_norm": 1.801132634260648,
      "learning_rate": 1.6817337827325275e-06,
      "loss": 0.2961,
      "step": 27480
    },
    {
      "epoch": 1.4653127581887477,
      "grad_norm": 1.6830322923094274,
      "learning_rate": 1.6808588636028427e-06,
      "loss": 0.3069,
      "step": 27490
    },
    {
      "epoch": 1.4658457930225741,
      "grad_norm": 1.648516389852871,
      "learning_rate": 1.6799838820301031e-06,
      "loss": 0.2934,
      "step": 27500
    },
    {
      "epoch": 1.4663788278564005,
      "grad_norm": 1.6178595526168695,
      "learning_rate": 1.6791088383164037e-06,
      "loss": 0.294,
      "step": 27510
    },
    {
      "epoch": 1.4669118626902269,
      "grad_norm": 1.632930528278504,
      "learning_rate": 1.6782337327638599e-06,
      "loss": 0.3016,
      "step": 27520
    },
    {
      "epoch": 1.4674448975240533,
      "grad_norm": 1.7733791927326759,
      "learning_rate": 1.67735856567461e-06,
      "loss": 0.3028,
      "step": 27530
    },
    {
      "epoch": 1.4679779323578797,
      "grad_norm": 1.7264020767448875,
      "learning_rate": 1.6764833373508124e-06,
      "loss": 0.297,
      "step": 27540
    },
    {
      "epoch": 1.468510967191706,
      "grad_norm": 1.69479142766262,
      "learning_rate": 1.6756080480946477e-06,
      "loss": 0.3065,
      "step": 27550
    },
    {
      "epoch": 1.4690440020255324,
      "grad_norm": 1.750597489582251,
      "learning_rate": 1.6747326982083164e-06,
      "loss": 0.294,
      "step": 27560
    },
    {
      "epoch": 1.4695770368593588,
      "grad_norm": 1.8259750607950043,
      "learning_rate": 1.6738572879940405e-06,
      "loss": 0.2925,
      "step": 27570
    },
    {
      "epoch": 1.4701100716931852,
      "grad_norm": 1.6500194638415577,
      "learning_rate": 1.6729818177540627e-06,
      "loss": 0.2994,
      "step": 27580
    },
    {
      "epoch": 1.4706431065270116,
      "grad_norm": 1.8644354107399674,
      "learning_rate": 1.6721062877906473e-06,
      "loss": 0.2986,
      "step": 27590
    },
    {
      "epoch": 1.471176141360838,
      "grad_norm": 1.586703376417515,
      "learning_rate": 1.6712306984060777e-06,
      "loss": 0.2902,
      "step": 27600
    },
    {
      "epoch": 1.4717091761946643,
      "grad_norm": 1.8229037829596009,
      "learning_rate": 1.6703550499026583e-06,
      "loss": 0.2982,
      "step": 27610
    },
    {
      "epoch": 1.4722422110284907,
      "grad_norm": 1.6711875682475372,
      "learning_rate": 1.669479342582715e-06,
      "loss": 0.2914,
      "step": 27620
    },
    {
      "epoch": 1.4727752458623171,
      "grad_norm": 1.717026421882349,
      "learning_rate": 1.6686035767485925e-06,
      "loss": 0.2928,
      "step": 27630
    },
    {
      "epoch": 1.4733082806961435,
      "grad_norm": 1.8111832276628441,
      "learning_rate": 1.6677277527026568e-06,
      "loss": 0.3046,
      "step": 27640
    },
    {
      "epoch": 1.4738413155299699,
      "grad_norm": 1.7503214352699468,
      "learning_rate": 1.6668518707472933e-06,
      "loss": 0.2994,
      "step": 27650
    },
    {
      "epoch": 1.4743743503637963,
      "grad_norm": 1.586385859752052,
      "learning_rate": 1.6659759311849082e-06,
      "loss": 0.2986,
      "step": 27660
    },
    {
      "epoch": 1.4749073851976227,
      "grad_norm": 1.540224465059517,
      "learning_rate": 1.6650999343179261e-06,
      "loss": 0.2935,
      "step": 27670
    },
    {
      "epoch": 1.475440420031449,
      "grad_norm": 1.4959973826274766,
      "learning_rate": 1.664223880448793e-06,
      "loss": 0.2857,
      "step": 27680
    },
    {
      "epoch": 1.4759734548652754,
      "grad_norm": 1.6336195860243121,
      "learning_rate": 1.663347769879974e-06,
      "loss": 0.2956,
      "step": 27690
    },
    {
      "epoch": 1.4765064896991018,
      "grad_norm": 1.6383625263474346,
      "learning_rate": 1.662471602913953e-06,
      "loss": 0.2841,
      "step": 27700
    },
    {
      "epoch": 1.4770395245329282,
      "grad_norm": 1.663860284596052,
      "learning_rate": 1.661595379853235e-06,
      "loss": 0.3022,
      "step": 27710
    },
    {
      "epoch": 1.4775725593667546,
      "grad_norm": 1.7020235515232844,
      "learning_rate": 1.660719101000343e-06,
      "loss": 0.2992,
      "step": 27720
    },
    {
      "epoch": 1.478105594200581,
      "grad_norm": 1.824488715931212,
      "learning_rate": 1.6598427666578194e-06,
      "loss": 0.2964,
      "step": 27730
    },
    {
      "epoch": 1.4786386290344073,
      "grad_norm": 1.6517601663000931,
      "learning_rate": 1.6589663771282268e-06,
      "loss": 0.3024,
      "step": 27740
    },
    {
      "epoch": 1.4791716638682337,
      "grad_norm": 1.7121110341342616,
      "learning_rate": 1.658089932714145e-06,
      "loss": 0.2942,
      "step": 27750
    },
    {
      "epoch": 1.4797046987020601,
      "grad_norm": 1.6880610449541016,
      "learning_rate": 1.6572134337181752e-06,
      "loss": 0.3151,
      "step": 27760
    },
    {
      "epoch": 1.4802377335358865,
      "grad_norm": 1.6285883850939522,
      "learning_rate": 1.656336880442935e-06,
      "loss": 0.2935,
      "step": 27770
    },
    {
      "epoch": 1.4807707683697129,
      "grad_norm": 1.8933376190946465,
      "learning_rate": 1.6554602731910623e-06,
      "loss": 0.2953,
      "step": 27780
    },
    {
      "epoch": 1.4813038032035393,
      "grad_norm": 1.731183515260805,
      "learning_rate": 1.654583612265213e-06,
      "loss": 0.3004,
      "step": 27790
    },
    {
      "epoch": 1.4818368380373657,
      "grad_norm": 1.6965312285850074,
      "learning_rate": 1.6537068979680626e-06,
      "loss": 0.2892,
      "step": 27800
    },
    {
      "epoch": 1.482369872871192,
      "grad_norm": 1.9242253509759262,
      "learning_rate": 1.6528301306023024e-06,
      "loss": 0.2856,
      "step": 27810
    },
    {
      "epoch": 1.4829029077050184,
      "grad_norm": 1.7044853439965404,
      "learning_rate": 1.6519533104706452e-06,
      "loss": 0.3032,
      "step": 27820
    },
    {
      "epoch": 1.4834359425388448,
      "grad_norm": 1.5883266991698182,
      "learning_rate": 1.65107643787582e-06,
      "loss": 0.2932,
      "step": 27830
    },
    {
      "epoch": 1.4839689773726712,
      "grad_norm": 1.5867537930341205,
      "learning_rate": 1.6501995131205746e-06,
      "loss": 0.2955,
      "step": 27840
    },
    {
      "epoch": 1.4845020122064976,
      "grad_norm": 1.6901333763132695,
      "learning_rate": 1.6493225365076754e-06,
      "loss": 0.2983,
      "step": 27850
    },
    {
      "epoch": 1.485035047040324,
      "grad_norm": 1.7558129590322686,
      "learning_rate": 1.6484455083399049e-06,
      "loss": 0.2939,
      "step": 27860
    },
    {
      "epoch": 1.4855680818741503,
      "grad_norm": 1.8427958295314535,
      "learning_rate": 1.6475684289200645e-06,
      "loss": 0.3053,
      "step": 27870
    },
    {
      "epoch": 1.4861011167079767,
      "grad_norm": 1.690141242985493,
      "learning_rate": 1.6466912985509738e-06,
      "loss": 0.2931,
      "step": 27880
    },
    {
      "epoch": 1.4866341515418033,
      "grad_norm": 1.7201623676231372,
      "learning_rate": 1.6458141175354704e-06,
      "loss": 0.3032,
      "step": 27890
    },
    {
      "epoch": 1.4871671863756297,
      "grad_norm": 1.7328747967864193,
      "learning_rate": 1.6449368861764068e-06,
      "loss": 0.3024,
      "step": 27900
    },
    {
      "epoch": 1.487700221209456,
      "grad_norm": 1.8989262757327343,
      "learning_rate": 1.6440596047766552e-06,
      "loss": 0.2925,
      "step": 27910
    },
    {
      "epoch": 1.4882332560432825,
      "grad_norm": 1.676818091566833,
      "learning_rate": 1.6431822736391045e-06,
      "loss": 0.2898,
      "step": 27920
    },
    {
      "epoch": 1.4887662908771089,
      "grad_norm": 1.7035824751704662,
      "learning_rate": 1.6423048930666609e-06,
      "loss": 0.298,
      "step": 27930
    },
    {
      "epoch": 1.4892993257109353,
      "grad_norm": 1.5962473302819875,
      "learning_rate": 1.641427463362247e-06,
      "loss": 0.3001,
      "step": 27940
    },
    {
      "epoch": 1.4898323605447616,
      "grad_norm": 1.8504533027552956,
      "learning_rate": 1.6405499848288036e-06,
      "loss": 0.3003,
      "step": 27950
    },
    {
      "epoch": 1.490365395378588,
      "grad_norm": 1.6907730220777866,
      "learning_rate": 1.6396724577692868e-06,
      "loss": 0.2927,
      "step": 27960
    },
    {
      "epoch": 1.4908984302124144,
      "grad_norm": 1.8161335413322999,
      "learning_rate": 1.638794882486671e-06,
      "loss": 0.3034,
      "step": 27970
    },
    {
      "epoch": 1.4914314650462408,
      "grad_norm": 1.80666646244297,
      "learning_rate": 1.6379172592839456e-06,
      "loss": 0.294,
      "step": 27980
    },
    {
      "epoch": 1.4919644998800672,
      "grad_norm": 1.826714047419686,
      "learning_rate": 1.637039588464118e-06,
      "loss": 0.2972,
      "step": 27990
    },
    {
      "epoch": 1.4924975347138936,
      "grad_norm": 1.722561708487875,
      "learning_rate": 1.6361618703302114e-06,
      "loss": 0.2856,
      "step": 28000
    },
    {
      "epoch": 1.49303056954772,
      "grad_norm": 1.5969410968034268,
      "learning_rate": 1.6352841051852655e-06,
      "loss": 0.2905,
      "step": 28010
    },
    {
      "epoch": 1.4935636043815463,
      "grad_norm": 1.6096847253544357,
      "learning_rate": 1.6344062933323362e-06,
      "loss": 0.2946,
      "step": 28020
    },
    {
      "epoch": 1.4940966392153727,
      "grad_norm": 1.7731542890231147,
      "learning_rate": 1.6335284350744953e-06,
      "loss": 0.3001,
      "step": 28030
    },
    {
      "epoch": 1.494629674049199,
      "grad_norm": 1.7508154811463157,
      "learning_rate": 1.6326505307148308e-06,
      "loss": 0.3041,
      "step": 28040
    },
    {
      "epoch": 1.4951627088830255,
      "grad_norm": 1.8031010237971825,
      "learning_rate": 1.6317725805564465e-06,
      "loss": 0.2939,
      "step": 28050
    },
    {
      "epoch": 1.4956957437168519,
      "grad_norm": 1.6997027672173084,
      "learning_rate": 1.630894584902463e-06,
      "loss": 0.3024,
      "step": 28060
    },
    {
      "epoch": 1.4962287785506783,
      "grad_norm": 1.6270487738608646,
      "learning_rate": 1.6300165440560144e-06,
      "loss": 0.289,
      "step": 28070
    },
    {
      "epoch": 1.4967618133845046,
      "grad_norm": 1.845574235405915,
      "learning_rate": 1.6291384583202525e-06,
      "loss": 0.2887,
      "step": 28080
    },
    {
      "epoch": 1.497294848218331,
      "grad_norm": 1.73738188908622,
      "learning_rate": 1.6282603279983439e-06,
      "loss": 0.2868,
      "step": 28090
    },
    {
      "epoch": 1.4978278830521574,
      "grad_norm": 1.7358893721831805,
      "learning_rate": 1.6273821533934703e-06,
      "loss": 0.2935,
      "step": 28100
    },
    {
      "epoch": 1.4983609178859838,
      "grad_norm": 1.67699992309884,
      "learning_rate": 1.626503934808829e-06,
      "loss": 0.2999,
      "step": 28110
    },
    {
      "epoch": 1.4988939527198102,
      "grad_norm": 1.8370640778598912,
      "learning_rate": 1.6256256725476325e-06,
      "loss": 0.2843,
      "step": 28120
    },
    {
      "epoch": 1.4994269875536366,
      "grad_norm": 1.6861679442667195,
      "learning_rate": 1.6247473669131085e-06,
      "loss": 0.3015,
      "step": 28130
    },
    {
      "epoch": 1.499960022387463,
      "grad_norm": 1.642119874643243,
      "learning_rate": 1.6238690182084987e-06,
      "loss": 0.2943,
      "step": 28140
    },
    {
      "epoch": 1.5004930572212896,
      "grad_norm": 1.6569100336113909,
      "learning_rate": 1.622990626737061e-06,
      "loss": 0.2965,
      "step": 28150
    },
    {
      "epoch": 1.501026092055116,
      "grad_norm": 1.9649148361340172,
      "learning_rate": 1.6221121928020672e-06,
      "loss": 0.3024,
      "step": 28160
    },
    {
      "epoch": 1.5015591268889423,
      "grad_norm": 1.6958069942772516,
      "learning_rate": 1.6212337167068045e-06,
      "loss": 0.2929,
      "step": 28170
    },
    {
      "epoch": 1.5020921617227687,
      "grad_norm": 1.8097213385035291,
      "learning_rate": 1.6203551987545737e-06,
      "loss": 0.3053,
      "step": 28180
    },
    {
      "epoch": 1.502625196556595,
      "grad_norm": 1.6659347999034089,
      "learning_rate": 1.6194766392486912e-06,
      "loss": 0.2852,
      "step": 28190
    },
    {
      "epoch": 1.5031582313904215,
      "grad_norm": 1.8238563165414425,
      "learning_rate": 1.6185980384924865e-06,
      "loss": 0.2994,
      "step": 28200
    },
    {
      "epoch": 1.5036912662242479,
      "grad_norm": 1.658849686670842,
      "learning_rate": 1.617719396789304e-06,
      "loss": 0.294,
      "step": 28210
    },
    {
      "epoch": 1.5042243010580743,
      "grad_norm": 1.7539900164457038,
      "learning_rate": 1.6168407144425025e-06,
      "loss": 0.2919,
      "step": 28220
    },
    {
      "epoch": 1.5047573358919006,
      "grad_norm": 1.6412991696555816,
      "learning_rate": 1.6159619917554542e-06,
      "loss": 0.2899,
      "step": 28230
    },
    {
      "epoch": 1.505290370725727,
      "grad_norm": 1.518043561074825,
      "learning_rate": 1.6150832290315463e-06,
      "loss": 0.292,
      "step": 28240
    },
    {
      "epoch": 1.5058234055595534,
      "grad_norm": 1.8154556112911517,
      "learning_rate": 1.6142044265741783e-06,
      "loss": 0.2893,
      "step": 28250
    },
    {
      "epoch": 1.5063564403933798,
      "grad_norm": 1.705924089473256,
      "learning_rate": 1.6133255846867643e-06,
      "loss": 0.2889,
      "step": 28260
    },
    {
      "epoch": 1.5068894752272062,
      "grad_norm": 1.6869212430216896,
      "learning_rate": 1.6124467036727321e-06,
      "loss": 0.2905,
      "step": 28270
    },
    {
      "epoch": 1.5074225100610326,
      "grad_norm": 1.853956929348773,
      "learning_rate": 1.611567783835523e-06,
      "loss": 0.3092,
      "step": 28280
    },
    {
      "epoch": 1.507955544894859,
      "grad_norm": 1.720981687695877,
      "learning_rate": 1.6106888254785909e-06,
      "loss": 0.3004,
      "step": 28290
    },
    {
      "epoch": 1.5084885797286853,
      "grad_norm": 1.7978555077404486,
      "learning_rate": 1.6098098289054042e-06,
      "loss": 0.3025,
      "step": 28300
    },
    {
      "epoch": 1.5090216145625117,
      "grad_norm": 1.7210103903336027,
      "learning_rate": 1.6089307944194434e-06,
      "loss": 0.2989,
      "step": 28310
    },
    {
      "epoch": 1.509554649396338,
      "grad_norm": 1.6837185360746634,
      "learning_rate": 1.6080517223242035e-06,
      "loss": 0.3026,
      "step": 28320
    },
    {
      "epoch": 1.5100876842301645,
      "grad_norm": 1.6144801646589821,
      "learning_rate": 1.6071726129231905e-06,
      "loss": 0.3024,
      "step": 28330
    },
    {
      "epoch": 1.5106207190639909,
      "grad_norm": 1.9599970383905356,
      "learning_rate": 1.6062934665199251e-06,
      "loss": 0.3055,
      "step": 28340
    },
    {
      "epoch": 1.5111537538978173,
      "grad_norm": 1.6931084525565177,
      "learning_rate": 1.6054142834179396e-06,
      "loss": 0.2926,
      "step": 28350
    },
    {
      "epoch": 1.5116867887316436,
      "grad_norm": 1.7486058775920885,
      "learning_rate": 1.6045350639207802e-06,
      "loss": 0.2951,
      "step": 28360
    },
    {
      "epoch": 1.51221982356547,
      "grad_norm": 1.6882984582026879,
      "learning_rate": 1.6036558083320037e-06,
      "loss": 0.3052,
      "step": 28370
    },
    {
      "epoch": 1.5127528583992964,
      "grad_norm": 1.6172156545956373,
      "learning_rate": 1.602776516955181e-06,
      "loss": 0.2999,
      "step": 28380
    },
    {
      "epoch": 1.5132858932331228,
      "grad_norm": 1.700471594039228,
      "learning_rate": 1.6018971900938953e-06,
      "loss": 0.2992,
      "step": 28390
    },
    {
      "epoch": 1.5138189280669492,
      "grad_norm": 1.8247550784338393,
      "learning_rate": 1.6010178280517415e-06,
      "loss": 0.2938,
      "step": 28400
    },
    {
      "epoch": 1.5143519629007756,
      "grad_norm": 1.6243837729790183,
      "learning_rate": 1.6001384311323268e-06,
      "loss": 0.3021,
      "step": 28410
    },
    {
      "epoch": 1.514884997734602,
      "grad_norm": 1.6028482402628443,
      "learning_rate": 1.5992589996392703e-06,
      "loss": 0.2888,
      "step": 28420
    },
    {
      "epoch": 1.5154180325684283,
      "grad_norm": 1.747798202328879,
      "learning_rate": 1.5983795338762029e-06,
      "loss": 0.3014,
      "step": 28430
    },
    {
      "epoch": 1.5159510674022547,
      "grad_norm": 1.614565188496978,
      "learning_rate": 1.597500034146768e-06,
      "loss": 0.2971,
      "step": 28440
    },
    {
      "epoch": 1.516484102236081,
      "grad_norm": 1.7266266657771903,
      "learning_rate": 1.5966205007546207e-06,
      "loss": 0.2926,
      "step": 28450
    },
    {
      "epoch": 1.5170171370699075,
      "grad_norm": 1.634752127596423,
      "learning_rate": 1.5957409340034263e-06,
      "loss": 0.2882,
      "step": 28460
    },
    {
      "epoch": 1.5175501719037339,
      "grad_norm": 1.6965604001420744,
      "learning_rate": 1.5948613341968633e-06,
      "loss": 0.2951,
      "step": 28470
    },
    {
      "epoch": 1.5180832067375603,
      "grad_norm": 1.8262756714258117,
      "learning_rate": 1.5939817016386212e-06,
      "loss": 0.2943,
      "step": 28480
    },
    {
      "epoch": 1.5186162415713866,
      "grad_norm": 1.7446263180707948,
      "learning_rate": 1.5931020366324005e-06,
      "loss": 0.2941,
      "step": 28490
    },
    {
      "epoch": 1.519149276405213,
      "grad_norm": 1.7628376517561695,
      "learning_rate": 1.5922223394819126e-06,
      "loss": 0.2942,
      "step": 28500
    },
    {
      "epoch": 1.5196823112390394,
      "grad_norm": 1.7531577696458098,
      "learning_rate": 1.5913426104908807e-06,
      "loss": 0.2922,
      "step": 28510
    },
    {
      "epoch": 1.5202153460728658,
      "grad_norm": 1.5861094510233353,
      "learning_rate": 1.5904628499630384e-06,
      "loss": 0.2964,
      "step": 28520
    },
    {
      "epoch": 1.5207483809066922,
      "grad_norm": 1.7111374605667955,
      "learning_rate": 1.5895830582021312e-06,
      "loss": 0.293,
      "step": 28530
    },
    {
      "epoch": 1.5212814157405186,
      "grad_norm": 1.703220797847656,
      "learning_rate": 1.5887032355119136e-06,
      "loss": 0.3101,
      "step": 28540
    },
    {
      "epoch": 1.521814450574345,
      "grad_norm": 2.3977667351995158,
      "learning_rate": 1.5878233821961527e-06,
      "loss": 0.3036,
      "step": 28550
    },
    {
      "epoch": 1.5223474854081713,
      "grad_norm": 1.8617842680729169,
      "learning_rate": 1.5869434985586248e-06,
      "loss": 0.2969,
      "step": 28560
    },
    {
      "epoch": 1.5228805202419977,
      "grad_norm": 1.7435109249642375,
      "learning_rate": 1.5860635849031177e-06,
      "loss": 0.2927,
      "step": 28570
    },
    {
      "epoch": 1.523413555075824,
      "grad_norm": 1.6339853635474115,
      "learning_rate": 1.5851836415334289e-06,
      "loss": 0.294,
      "step": 28580
    },
    {
      "epoch": 1.5239465899096505,
      "grad_norm": 1.8421012470615414,
      "learning_rate": 1.584303668753366e-06,
      "loss": 0.2981,
      "step": 28590
    },
    {
      "epoch": 1.5244796247434769,
      "grad_norm": 1.8311849255284487,
      "learning_rate": 1.5834236668667468e-06,
      "loss": 0.3023,
      "step": 28600
    },
    {
      "epoch": 1.5250126595773033,
      "grad_norm": 1.5577061448746736,
      "learning_rate": 1.5825436361774e-06,
      "loss": 0.2933,
      "step": 28610
    },
    {
      "epoch": 1.5255456944111296,
      "grad_norm": 1.6624329481275517,
      "learning_rate": 1.5816635769891636e-06,
      "loss": 0.2964,
      "step": 28620
    },
    {
      "epoch": 1.526078729244956,
      "grad_norm": 1.636190525351784,
      "learning_rate": 1.5807834896058853e-06,
      "loss": 0.286,
      "step": 28630
    },
    {
      "epoch": 1.5266117640787824,
      "grad_norm": 1.783624078408808,
      "learning_rate": 1.5799033743314229e-06,
      "loss": 0.2941,
      "step": 28640
    },
    {
      "epoch": 1.5271447989126088,
      "grad_norm": 1.7325584803786656,
      "learning_rate": 1.5790232314696433e-06,
      "loss": 0.2953,
      "step": 28650
    },
    {
      "epoch": 1.5276778337464352,
      "grad_norm": 1.9083094539837078,
      "learning_rate": 1.5781430613244236e-06,
      "loss": 0.2933,
      "step": 28660
    },
    {
      "epoch": 1.5282108685802616,
      "grad_norm": 2.0451081571029452,
      "learning_rate": 1.57726286419965e-06,
      "loss": 0.2974,
      "step": 28670
    },
    {
      "epoch": 1.528743903414088,
      "grad_norm": 1.8343846593639332,
      "learning_rate": 1.5763826403992178e-06,
      "loss": 0.2961,
      "step": 28680
    },
    {
      "epoch": 1.5292769382479146,
      "grad_norm": 1.9549566614955023,
      "learning_rate": 1.5755023902270317e-06,
      "loss": 0.298,
      "step": 28690
    },
    {
      "epoch": 1.529809973081741,
      "grad_norm": 1.8536852563050172,
      "learning_rate": 1.574622113987006e-06,
      "loss": 0.2997,
      "step": 28700
    },
    {
      "epoch": 1.5303430079155673,
      "grad_norm": 1.6592000043536876,
      "learning_rate": 1.573741811983063e-06,
      "loss": 0.294,
      "step": 28710
    },
    {
      "epoch": 1.5308760427493937,
      "grad_norm": 1.8881292882742458,
      "learning_rate": 1.5728614845191347e-06,
      "loss": 0.2919,
      "step": 28720
    },
    {
      "epoch": 1.53140907758322,
      "grad_norm": 1.825312457054052,
      "learning_rate": 1.5719811318991617e-06,
      "loss": 0.2942,
      "step": 28730
    },
    {
      "epoch": 1.5319421124170465,
      "grad_norm": 1.7571945083263387,
      "learning_rate": 1.5711007544270927e-06,
      "loss": 0.3029,
      "step": 28740
    },
    {
      "epoch": 1.5324751472508729,
      "grad_norm": 1.8683441116480533,
      "learning_rate": 1.5702203524068864e-06,
      "loss": 0.2928,
      "step": 28750
    },
    {
      "epoch": 1.5330081820846992,
      "grad_norm": 1.5973701567812721,
      "learning_rate": 1.569339926142508e-06,
      "loss": 0.286,
      "step": 28760
    },
    {
      "epoch": 1.5335412169185256,
      "grad_norm": 1.7716311390117534,
      "learning_rate": 1.5684594759379325e-06,
      "loss": 0.303,
      "step": 28770
    },
    {
      "epoch": 1.534074251752352,
      "grad_norm": 1.5439971205746117,
      "learning_rate": 1.5675790020971431e-06,
      "loss": 0.2857,
      "step": 28780
    },
    {
      "epoch": 1.5346072865861784,
      "grad_norm": 2.201419785810831,
      "learning_rate": 1.5666985049241312e-06,
      "loss": 0.3007,
      "step": 28790
    },
    {
      "epoch": 1.5351403214200048,
      "grad_norm": 1.7183377729934377,
      "learning_rate": 1.5658179847228947e-06,
      "loss": 0.2898,
      "step": 28800
    },
    {
      "epoch": 1.5356733562538312,
      "grad_norm": 1.802440538999993,
      "learning_rate": 1.5649374417974417e-06,
      "loss": 0.2894,
      "step": 28810
    },
    {
      "epoch": 1.5362063910876576,
      "grad_norm": 1.703431075434067,
      "learning_rate": 1.5640568764517867e-06,
      "loss": 0.296,
      "step": 28820
    },
    {
      "epoch": 1.536739425921484,
      "grad_norm": 1.885849139426709,
      "learning_rate": 1.563176288989952e-06,
      "loss": 0.2799,
      "step": 28830
    },
    {
      "epoch": 1.5372724607553103,
      "grad_norm": 1.634048682347014,
      "learning_rate": 1.562295679715969e-06,
      "loss": 0.2935,
      "step": 28840
    },
    {
      "epoch": 1.5378054955891367,
      "grad_norm": 1.7473911943796552,
      "learning_rate": 1.5614150489338739e-06,
      "loss": 0.2933,
      "step": 28850
    },
    {
      "epoch": 1.538338530422963,
      "grad_norm": 1.7636035666036844,
      "learning_rate": 1.5605343969477132e-06,
      "loss": 0.2972,
      "step": 28860
    },
    {
      "epoch": 1.5388715652567897,
      "grad_norm": 1.8471786793401228,
      "learning_rate": 1.559653724061539e-06,
      "loss": 0.2924,
      "step": 28870
    },
    {
      "epoch": 1.539404600090616,
      "grad_norm": 1.6199480807082696,
      "learning_rate": 1.5587730305794114e-06,
      "loss": 0.2892,
      "step": 28880
    },
    {
      "epoch": 1.5399376349244425,
      "grad_norm": 1.5483058660371363,
      "learning_rate": 1.557892316805397e-06,
      "loss": 0.303,
      "step": 28890
    },
    {
      "epoch": 1.5404706697582689,
      "grad_norm": 1.653927269709408,
      "learning_rate": 1.5570115830435693e-06,
      "loss": 0.303,
      "step": 28900
    },
    {
      "epoch": 1.5410037045920952,
      "grad_norm": 1.8433909449753805,
      "learning_rate": 1.55613082959801e-06,
      "loss": 0.2882,
      "step": 28910
    },
    {
      "epoch": 1.5415367394259216,
      "grad_norm": 1.8081350242458483,
      "learning_rate": 1.5552500567728064e-06,
      "loss": 0.2954,
      "step": 28920
    },
    {
      "epoch": 1.542069774259748,
      "grad_norm": 1.7542713104999963,
      "learning_rate": 1.554369264872052e-06,
      "loss": 0.2934,
      "step": 28930
    },
    {
      "epoch": 1.5426028090935744,
      "grad_norm": 1.7446725348283751,
      "learning_rate": 1.5534884541998486e-06,
      "loss": 0.2768,
      "step": 28940
    },
    {
      "epoch": 1.5431358439274008,
      "grad_norm": 1.6791827727591018,
      "learning_rate": 1.5526076250603028e-06,
      "loss": 0.2927,
      "step": 28950
    },
    {
      "epoch": 1.5436688787612272,
      "grad_norm": 1.8438925899239147,
      "learning_rate": 1.551726777757529e-06,
      "loss": 0.2893,
      "step": 28960
    },
    {
      "epoch": 1.5442019135950535,
      "grad_norm": 1.8057102543856545,
      "learning_rate": 1.550845912595647e-06,
      "loss": 0.2971,
      "step": 28970
    },
    {
      "epoch": 1.54473494842888,
      "grad_norm": 1.7197293993214577,
      "learning_rate": 1.549965029878783e-06,
      "loss": 0.2866,
      "step": 28980
    },
    {
      "epoch": 1.5452679832627063,
      "grad_norm": 1.7656653900122072,
      "learning_rate": 1.549084129911069e-06,
      "loss": 0.2928,
      "step": 28990
    },
    {
      "epoch": 1.5458010180965327,
      "grad_norm": 1.713177521103493,
      "learning_rate": 1.5482032129966427e-06,
      "loss": 0.2833,
      "step": 29000
    },
    {
      "epoch": 1.546334052930359,
      "grad_norm": 1.7665636184546882,
      "learning_rate": 1.5473222794396497e-06,
      "loss": 0.2927,
      "step": 29010
    },
    {
      "epoch": 1.5468670877641855,
      "grad_norm": 1.8510141714175767,
      "learning_rate": 1.5464413295442382e-06,
      "loss": 0.2875,
      "step": 29020
    },
    {
      "epoch": 1.5474001225980119,
      "grad_norm": 2.120257671777005,
      "learning_rate": 1.5455603636145643e-06,
      "loss": 0.3013,
      "step": 29030
    },
    {
      "epoch": 1.5479331574318382,
      "grad_norm": 1.7090098778980756,
      "learning_rate": 1.544679381954789e-06,
      "loss": 0.2904,
      "step": 29040
    },
    {
      "epoch": 1.5484661922656646,
      "grad_norm": 1.7900042413336352,
      "learning_rate": 1.5437983848690785e-06,
      "loss": 0.2869,
      "step": 29050
    },
    {
      "epoch": 1.548999227099491,
      "grad_norm": 1.7323622840467536,
      "learning_rate": 1.5429173726616043e-06,
      "loss": 0.2949,
      "step": 29060
    },
    {
      "epoch": 1.5495322619333174,
      "grad_norm": 1.8085541732086141,
      "learning_rate": 1.5420363456365438e-06,
      "loss": 0.3003,
      "step": 29070
    },
    {
      "epoch": 1.5500652967671438,
      "grad_norm": 1.7063765707716472,
      "learning_rate": 1.5411553040980786e-06,
      "loss": 0.3023,
      "step": 29080
    },
    {
      "epoch": 1.5505983316009702,
      "grad_norm": 1.7301599684775033,
      "learning_rate": 1.5402742483503962e-06,
      "loss": 0.2932,
      "step": 29090
    },
    {
      "epoch": 1.5511313664347965,
      "grad_norm": 1.7365182226562361,
      "learning_rate": 1.5393931786976884e-06,
      "loss": 0.2758,
      "step": 29100
    },
    {
      "epoch": 1.551664401268623,
      "grad_norm": 1.8591936086396716,
      "learning_rate": 1.5385120954441518e-06,
      "loss": 0.2883,
      "step": 29110
    },
    {
      "epoch": 1.5521974361024493,
      "grad_norm": 1.5892261429339873,
      "learning_rate": 1.5376309988939877e-06,
      "loss": 0.3022,
      "step": 29120
    },
    {
      "epoch": 1.5527304709362757,
      "grad_norm": 1.8519464360031765,
      "learning_rate": 1.5367498893514023e-06,
      "loss": 0.294,
      "step": 29130
    },
    {
      "epoch": 1.553263505770102,
      "grad_norm": 1.8891603306431821,
      "learning_rate": 1.535868767120607e-06,
      "loss": 0.2923,
      "step": 29140
    },
    {
      "epoch": 1.5537965406039285,
      "grad_norm": 1.6263498224727384,
      "learning_rate": 1.5349876325058154e-06,
      "loss": 0.2942,
      "step": 29150
    },
    {
      "epoch": 1.5543295754377549,
      "grad_norm": 1.6935832837348501,
      "learning_rate": 1.534106485811247e-06,
      "loss": 0.2972,
      "step": 29160
    },
    {
      "epoch": 1.5548626102715812,
      "grad_norm": 1.6496853466921766,
      "learning_rate": 1.5332253273411256e-06,
      "loss": 0.3004,
      "step": 29170
    },
    {
      "epoch": 1.5553956451054076,
      "grad_norm": 1.7745318232485863,
      "learning_rate": 1.532344157399679e-06,
      "loss": 0.2968,
      "step": 29180
    },
    {
      "epoch": 1.555928679939234,
      "grad_norm": 1.7223290085284912,
      "learning_rate": 1.5314629762911375e-06,
      "loss": 0.2898,
      "step": 29190
    },
    {
      "epoch": 1.5564617147730604,
      "grad_norm": 1.79787418568104,
      "learning_rate": 1.530581784319737e-06,
      "loss": 0.3009,
      "step": 29200
    },
    {
      "epoch": 1.5569947496068868,
      "grad_norm": 1.738331588677691,
      "learning_rate": 1.5297005817897162e-06,
      "loss": 0.2969,
      "step": 29210
    },
    {
      "epoch": 1.5575277844407132,
      "grad_norm": 1.8017796862562803,
      "learning_rate": 1.5288193690053182e-06,
      "loss": 0.2853,
      "step": 29220
    },
    {
      "epoch": 1.5580608192745395,
      "grad_norm": 1.824425767854045,
      "learning_rate": 1.5279381462707887e-06,
      "loss": 0.3009,
      "step": 29230
    },
    {
      "epoch": 1.558593854108366,
      "grad_norm": 1.6177833474108034,
      "learning_rate": 1.5270569138903774e-06,
      "loss": 0.295,
      "step": 29240
    },
    {
      "epoch": 1.5591268889421923,
      "grad_norm": 1.7554613683017242,
      "learning_rate": 1.5261756721683374e-06,
      "loss": 0.3021,
      "step": 29250
    },
    {
      "epoch": 1.5596599237760187,
      "grad_norm": 1.8089123772227371,
      "learning_rate": 1.525294421408925e-06,
      "loss": 0.3006,
      "step": 29260
    },
    {
      "epoch": 1.560192958609845,
      "grad_norm": 1.7861263387117634,
      "learning_rate": 1.5244131619163992e-06,
      "loss": 0.2841,
      "step": 29270
    },
    {
      "epoch": 1.5607259934436715,
      "grad_norm": 1.820631455076131,
      "learning_rate": 1.5235318939950222e-06,
      "loss": 0.2973,
      "step": 29280
    },
    {
      "epoch": 1.5612590282774979,
      "grad_norm": 1.7481690762500246,
      "learning_rate": 1.5226506179490595e-06,
      "loss": 0.2942,
      "step": 29290
    },
    {
      "epoch": 1.5617920631113242,
      "grad_norm": 1.6575625506593998,
      "learning_rate": 1.5217693340827792e-06,
      "loss": 0.2967,
      "step": 29300
    },
    {
      "epoch": 1.5623250979451506,
      "grad_norm": 1.8455027689733514,
      "learning_rate": 1.520888042700452e-06,
      "loss": 0.2959,
      "step": 29310
    },
    {
      "epoch": 1.562858132778977,
      "grad_norm": 1.6597897673256523,
      "learning_rate": 1.5200067441063505e-06,
      "loss": 0.2847,
      "step": 29320
    },
    {
      "epoch": 1.5633911676128034,
      "grad_norm": 1.7121060363437959,
      "learning_rate": 1.5191254386047516e-06,
      "loss": 0.2838,
      "step": 29330
    },
    {
      "epoch": 1.5639242024466298,
      "grad_norm": 1.5863357753350653,
      "learning_rate": 1.5182441264999327e-06,
      "loss": 0.2968,
      "step": 29340
    },
    {
      "epoch": 1.5644572372804562,
      "grad_norm": 1.8229288180980128,
      "learning_rate": 1.5173628080961747e-06,
      "loss": 0.2928,
      "step": 29350
    },
    {
      "epoch": 1.5649902721142825,
      "grad_norm": 1.5331678552178638,
      "learning_rate": 1.5164814836977604e-06,
      "loss": 0.2962,
      "step": 29360
    },
    {
      "epoch": 1.565523306948109,
      "grad_norm": 1.6484295238174518,
      "learning_rate": 1.5156001536089735e-06,
      "loss": 0.2908,
      "step": 29370
    },
    {
      "epoch": 1.5660563417819353,
      "grad_norm": 1.6315555651845868,
      "learning_rate": 1.514718818134102e-06,
      "loss": 0.2921,
      "step": 29380
    },
    {
      "epoch": 1.5665893766157617,
      "grad_norm": 1.653521081680679,
      "learning_rate": 1.513837477577434e-06,
      "loss": 0.2989,
      "step": 29390
    },
    {
      "epoch": 1.567122411449588,
      "grad_norm": 1.6794389480599035,
      "learning_rate": 1.512956132243259e-06,
      "loss": 0.2931,
      "step": 29400
    },
    {
      "epoch": 1.5676554462834145,
      "grad_norm": 1.7506217509942266,
      "learning_rate": 1.51207478243587e-06,
      "loss": 0.2951,
      "step": 29410
    },
    {
      "epoch": 1.5681884811172409,
      "grad_norm": 1.7427523537299987,
      "learning_rate": 1.5111934284595598e-06,
      "loss": 0.2929,
      "step": 29420
    },
    {
      "epoch": 1.5687215159510672,
      "grad_norm": 1.8252874255787626,
      "learning_rate": 1.5103120706186239e-06,
      "loss": 0.2843,
      "step": 29430
    },
    {
      "epoch": 1.5692545507848938,
      "grad_norm": 1.8166711691739543,
      "learning_rate": 1.5094307092173578e-06,
      "loss": 0.2892,
      "step": 29440
    },
    {
      "epoch": 1.5697875856187202,
      "grad_norm": 1.7704441845382535,
      "learning_rate": 1.5085493445600593e-06,
      "loss": 0.2982,
      "step": 29450
    },
    {
      "epoch": 1.5703206204525466,
      "grad_norm": 1.7339631843224352,
      "learning_rate": 1.5076679769510277e-06,
      "loss": 0.2917,
      "step": 29460
    },
    {
      "epoch": 1.570853655286373,
      "grad_norm": 1.7169978904517043,
      "learning_rate": 1.5067866066945622e-06,
      "loss": 0.2877,
      "step": 29470
    },
    {
      "epoch": 1.5713866901201994,
      "grad_norm": 1.727066373523391,
      "learning_rate": 1.5059052340949632e-06,
      "loss": 0.285,
      "step": 29480
    },
    {
      "epoch": 1.5719197249540258,
      "grad_norm": 1.7622201639321324,
      "learning_rate": 1.5050238594565317e-06,
      "loss": 0.2824,
      "step": 29490
    },
    {
      "epoch": 1.5724527597878522,
      "grad_norm": 1.8964643669719574,
      "learning_rate": 1.5041424830835704e-06,
      "loss": 0.2923,
      "step": 29500
    },
    {
      "epoch": 1.5729857946216785,
      "grad_norm": 1.6733681966839968,
      "learning_rate": 1.5032611052803816e-06,
      "loss": 0.2998,
      "step": 29510
    },
    {
      "epoch": 1.573518829455505,
      "grad_norm": 2.0127908832695067,
      "learning_rate": 1.5023797263512683e-06,
      "loss": 0.3069,
      "step": 29520
    },
    {
      "epoch": 1.5740518642893313,
      "grad_norm": 1.8383242398927957,
      "learning_rate": 1.5014983466005348e-06,
      "loss": 0.2868,
      "step": 29530
    },
    {
      "epoch": 1.5745848991231577,
      "grad_norm": 1.7218919525028973,
      "learning_rate": 1.5006169663324842e-06,
      "loss": 0.2977,
      "step": 29540
    },
    {
      "epoch": 1.575117933956984,
      "grad_norm": 1.7953636226015546,
      "learning_rate": 1.4997355858514206e-06,
      "loss": 0.2883,
      "step": 29550
    },
    {
      "epoch": 1.5756509687908105,
      "grad_norm": 1.8905724320980837,
      "learning_rate": 1.4988542054616486e-06,
      "loss": 0.3007,
      "step": 29560
    },
    {
      "epoch": 1.5761840036246368,
      "grad_norm": 1.6717849790344281,
      "learning_rate": 1.4979728254674717e-06,
      "loss": 0.2913,
      "step": 29570
    },
    {
      "epoch": 1.5767170384584632,
      "grad_norm": 1.6829301510296975,
      "learning_rate": 1.4970914461731938e-06,
      "loss": 0.3011,
      "step": 29580
    },
    {
      "epoch": 1.5772500732922896,
      "grad_norm": 1.7106737360506645,
      "learning_rate": 1.496210067883119e-06,
      "loss": 0.2851,
      "step": 29590
    },
    {
      "epoch": 1.577783108126116,
      "grad_norm": 1.7771988372987726,
      "learning_rate": 1.4953286909015505e-06,
      "loss": 0.2849,
      "step": 29600
    },
    {
      "epoch": 1.5783161429599424,
      "grad_norm": 1.7076412774843983,
      "learning_rate": 1.4944473155327908e-06,
      "loss": 0.285,
      "step": 29610
    },
    {
      "epoch": 1.578849177793769,
      "grad_norm": 2.0100980705442923,
      "learning_rate": 1.4935659420811425e-06,
      "loss": 0.2976,
      "step": 29620
    },
    {
      "epoch": 1.5793822126275954,
      "grad_norm": 2.0364791644781692,
      "learning_rate": 1.4926845708509071e-06,
      "loss": 0.2921,
      "step": 29630
    },
    {
      "epoch": 1.5799152474614218,
      "grad_norm": 1.528965050552814,
      "learning_rate": 1.491803202146386e-06,
      "loss": 0.2806,
      "step": 29640
    },
    {
      "epoch": 1.5804482822952481,
      "grad_norm": 1.6917418998977327,
      "learning_rate": 1.4909218362718783e-06,
      "loss": 0.2887,
      "step": 29650
    },
    {
      "epoch": 1.5809813171290745,
      "grad_norm": 1.5949104978843631,
      "learning_rate": 1.4900404735316837e-06,
      "loss": 0.2887,
      "step": 29660
    },
    {
      "epoch": 1.581514351962901,
      "grad_norm": 1.6591268495298312,
      "learning_rate": 1.4891591142301e-06,
      "loss": 0.2979,
      "step": 29670
    },
    {
      "epoch": 1.5820473867967273,
      "grad_norm": 1.7864903686969482,
      "learning_rate": 1.488277758671424e-06,
      "loss": 0.2959,
      "step": 29680
    },
    {
      "epoch": 1.5825804216305537,
      "grad_norm": 1.6149836453851365,
      "learning_rate": 1.4873964071599515e-06,
      "loss": 0.2909,
      "step": 29690
    },
    {
      "epoch": 1.58311345646438,
      "grad_norm": 1.7203432392741604,
      "learning_rate": 1.4865150599999757e-06,
      "loss": 0.2926,
      "step": 29700
    },
    {
      "epoch": 1.5836464912982064,
      "grad_norm": 1.9222255085932465,
      "learning_rate": 1.4856337174957898e-06,
      "loss": 0.3035,
      "step": 29710
    },
    {
      "epoch": 1.5841795261320328,
      "grad_norm": 1.9505240563639825,
      "learning_rate": 1.4847523799516846e-06,
      "loss": 0.2912,
      "step": 29720
    },
    {
      "epoch": 1.5847125609658592,
      "grad_norm": 1.495431000600028,
      "learning_rate": 1.4838710476719498e-06,
      "loss": 0.2878,
      "step": 29730
    },
    {
      "epoch": 1.5852455957996856,
      "grad_norm": 1.6225851485779565,
      "learning_rate": 1.482989720960872e-06,
      "loss": 0.2875,
      "step": 29740
    },
    {
      "epoch": 1.585778630633512,
      "grad_norm": 1.8520363950078764,
      "learning_rate": 1.4821084001227373e-06,
      "loss": 0.2948,
      "step": 29750
    },
    {
      "epoch": 1.5863116654673384,
      "grad_norm": 1.9114972356068047,
      "learning_rate": 1.481227085461829e-06,
      "loss": 0.298,
      "step": 29760
    },
    {
      "epoch": 1.5868447003011648,
      "grad_norm": 1.681523278171821,
      "learning_rate": 1.4803457772824287e-06,
      "loss": 0.2878,
      "step": 29770
    },
    {
      "epoch": 1.5873777351349911,
      "grad_norm": 2.10014393489051,
      "learning_rate": 1.4794644758888154e-06,
      "loss": 0.2852,
      "step": 29780
    },
    {
      "epoch": 1.5879107699688175,
      "grad_norm": 1.6029038473812447,
      "learning_rate": 1.4785831815852652e-06,
      "loss": 0.2994,
      "step": 29790
    },
    {
      "epoch": 1.588443804802644,
      "grad_norm": 1.7279237101284677,
      "learning_rate": 1.4777018946760534e-06,
      "loss": 0.2922,
      "step": 29800
    },
    {
      "epoch": 1.5889768396364703,
      "grad_norm": 1.7693786663101572,
      "learning_rate": 1.4768206154654517e-06,
      "loss": 0.2856,
      "step": 29810
    },
    {
      "epoch": 1.5895098744702967,
      "grad_norm": 1.7807937146128132,
      "learning_rate": 1.4759393442577287e-06,
      "loss": 0.2823,
      "step": 29820
    },
    {
      "epoch": 1.590042909304123,
      "grad_norm": 1.6651240447297195,
      "learning_rate": 1.4750580813571512e-06,
      "loss": 0.2872,
      "step": 29830
    },
    {
      "epoch": 1.5905759441379494,
      "grad_norm": 1.5479437443507142,
      "learning_rate": 1.4741768270679817e-06,
      "loss": 0.2909,
      "step": 29840
    },
    {
      "epoch": 1.5911089789717758,
      "grad_norm": 1.9486996128425538,
      "learning_rate": 1.4732955816944818e-06,
      "loss": 0.3067,
      "step": 29850
    },
    {
      "epoch": 1.5916420138056022,
      "grad_norm": 1.776230465946072,
      "learning_rate": 1.4724143455409086e-06,
      "loss": 0.2936,
      "step": 29860
    },
    {
      "epoch": 1.5921750486394286,
      "grad_norm": 1.8519639812750501,
      "learning_rate": 1.4715331189115164e-06,
      "loss": 0.2908,
      "step": 29870
    },
    {
      "epoch": 1.592708083473255,
      "grad_norm": 1.7729475522446254,
      "learning_rate": 1.470651902110556e-06,
      "loss": 0.2894,
      "step": 29880
    },
    {
      "epoch": 1.5932411183070814,
      "grad_norm": 1.7436151462487275,
      "learning_rate": 1.4697706954422749e-06,
      "loss": 0.2984,
      "step": 29890
    },
    {
      "epoch": 1.5937741531409078,
      "grad_norm": 1.807041226681752,
      "learning_rate": 1.4688894992109173e-06,
      "loss": 0.2845,
      "step": 29900
    },
    {
      "epoch": 1.5943071879747341,
      "grad_norm": 1.7369816471603763,
      "learning_rate": 1.4680083137207238e-06,
      "loss": 0.295,
      "step": 29910
    },
    {
      "epoch": 1.5948402228085605,
      "grad_norm": 1.7320307158069488,
      "learning_rate": 1.4671271392759311e-06,
      "loss": 0.2839,
      "step": 29920
    },
    {
      "epoch": 1.595373257642387,
      "grad_norm": 1.7203403014891072,
      "learning_rate": 1.466245976180772e-06,
      "loss": 0.289,
      "step": 29930
    },
    {
      "epoch": 1.5959062924762133,
      "grad_norm": 2.131879893106814,
      "learning_rate": 1.4653648247394755e-06,
      "loss": 0.297,
      "step": 29940
    },
    {
      "epoch": 1.5964393273100397,
      "grad_norm": 1.6403213281206186,
      "learning_rate": 1.4644836852562668e-06,
      "loss": 0.2937,
      "step": 29950
    },
    {
      "epoch": 1.596972362143866,
      "grad_norm": 1.9166361004830939,
      "learning_rate": 1.4636025580353668e-06,
      "loss": 0.2994,
      "step": 29960
    },
    {
      "epoch": 1.5975053969776924,
      "grad_norm": 1.7065153110500177,
      "learning_rate": 1.462721443380992e-06,
      "loss": 0.294,
      "step": 29970
    },
    {
      "epoch": 1.5980384318115188,
      "grad_norm": 1.6068123360904143,
      "learning_rate": 1.4618403415973546e-06,
      "loss": 0.2989,
      "step": 29980
    },
    {
      "epoch": 1.5985714666453452,
      "grad_norm": 1.9757782848674599,
      "learning_rate": 1.4609592529886624e-06,
      "loss": 0.2797,
      "step": 29990
    },
    {
      "epoch": 1.5991045014791716,
      "grad_norm": 1.9114273826012875,
      "learning_rate": 1.4600781778591194e-06,
      "loss": 0.2906,
      "step": 30000
    },
    {
      "epoch": 1.599637536312998,
      "grad_norm": 1.6977416149440836,
      "learning_rate": 1.4591971165129234e-06,
      "loss": 0.2916,
      "step": 30010
    },
    {
      "epoch": 1.6001705711468244,
      "grad_norm": 1.8617417990393528,
      "learning_rate": 1.4583160692542685e-06,
      "loss": 0.286,
      "step": 30020
    },
    {
      "epoch": 1.6007036059806508,
      "grad_norm": 1.7390391425669267,
      "learning_rate": 1.457435036387344e-06,
      "loss": 0.2905,
      "step": 30030
    },
    {
      "epoch": 1.6012366408144771,
      "grad_norm": 1.7204244586294035,
      "learning_rate": 1.456554018216334e-06,
      "loss": 0.2915,
      "step": 30040
    },
    {
      "epoch": 1.6017696756483035,
      "grad_norm": 1.495972777362757,
      "learning_rate": 1.455673015045417e-06,
      "loss": 0.2886,
      "step": 30050
    },
    {
      "epoch": 1.60230271048213,
      "grad_norm": 1.6698287745784772,
      "learning_rate": 1.4547920271787674e-06,
      "loss": 0.2919,
      "step": 30060
    },
    {
      "epoch": 1.6028357453159563,
      "grad_norm": 1.6563096457504,
      "learning_rate": 1.4539110549205531e-06,
      "loss": 0.2862,
      "step": 30070
    },
    {
      "epoch": 1.6033687801497827,
      "grad_norm": 1.5696180412578158,
      "learning_rate": 1.4530300985749376e-06,
      "loss": 0.2903,
      "step": 30080
    },
    {
      "epoch": 1.603901814983609,
      "grad_norm": 1.9328779705136152,
      "learning_rate": 1.452149158446079e-06,
      "loss": 0.2886,
      "step": 30090
    },
    {
      "epoch": 1.6044348498174354,
      "grad_norm": 1.6968063245285405,
      "learning_rate": 1.4512682348381285e-06,
      "loss": 0.3034,
      "step": 30100
    },
    {
      "epoch": 1.6049678846512618,
      "grad_norm": 1.7284877634340405,
      "learning_rate": 1.4503873280552326e-06,
      "loss": 0.2878,
      "step": 30110
    },
    {
      "epoch": 1.6055009194850882,
      "grad_norm": 1.5766855279671077,
      "learning_rate": 1.4495064384015323e-06,
      "loss": 0.2866,
      "step": 30120
    },
    {
      "epoch": 1.6060339543189146,
      "grad_norm": 1.6629848182261564,
      "learning_rate": 1.4486255661811621e-06,
      "loss": 0.2846,
      "step": 30130
    },
    {
      "epoch": 1.606566989152741,
      "grad_norm": 1.752335694125959,
      "learning_rate": 1.4477447116982503e-06,
      "loss": 0.3058,
      "step": 30140
    },
    {
      "epoch": 1.6071000239865674,
      "grad_norm": 1.8416926587540483,
      "learning_rate": 1.4468638752569193e-06,
      "loss": 0.2791,
      "step": 30150
    },
    {
      "epoch": 1.6076330588203938,
      "grad_norm": 1.7743146963666445,
      "learning_rate": 1.4459830571612857e-06,
      "loss": 0.291,
      "step": 30160
    },
    {
      "epoch": 1.6081660936542201,
      "grad_norm": 1.7775585093653938,
      "learning_rate": 1.4451022577154597e-06,
      "loss": 0.2964,
      "step": 30170
    },
    {
      "epoch": 1.6086991284880467,
      "grad_norm": 1.8841816144061794,
      "learning_rate": 1.4442214772235438e-06,
      "loss": 0.2923,
      "step": 30180
    },
    {
      "epoch": 1.6092321633218731,
      "grad_norm": 1.7808132885140648,
      "learning_rate": 1.4433407159896356e-06,
      "loss": 0.2928,
      "step": 30190
    },
    {
      "epoch": 1.6097651981556995,
      "grad_norm": 1.5396400528966006,
      "learning_rate": 1.4424599743178252e-06,
      "loss": 0.29,
      "step": 30200
    },
    {
      "epoch": 1.610298232989526,
      "grad_norm": 1.6457019661135046,
      "learning_rate": 1.4415792525121966e-06,
      "loss": 0.2983,
      "step": 30210
    },
    {
      "epoch": 1.6108312678233523,
      "grad_norm": 1.6855099835813079,
      "learning_rate": 1.4406985508768257e-06,
      "loss": 0.2874,
      "step": 30220
    },
    {
      "epoch": 1.6113643026571787,
      "grad_norm": 1.6990597726711465,
      "learning_rate": 1.4398178697157825e-06,
      "loss": 0.2919,
      "step": 30230
    },
    {
      "epoch": 1.611897337491005,
      "grad_norm": 1.6592408992278302,
      "learning_rate": 1.4389372093331295e-06,
      "loss": 0.2803,
      "step": 30240
    },
    {
      "epoch": 1.6124303723248314,
      "grad_norm": 1.6141288403459748,
      "learning_rate": 1.4380565700329225e-06,
      "loss": 0.2932,
      "step": 30250
    },
    {
      "epoch": 1.6129634071586578,
      "grad_norm": 1.6825199751823052,
      "learning_rate": 1.4371759521192097e-06,
      "loss": 0.2842,
      "step": 30260
    },
    {
      "epoch": 1.6134964419924842,
      "grad_norm": 1.6805441093015927,
      "learning_rate": 1.4362953558960316e-06,
      "loss": 0.2855,
      "step": 30270
    },
    {
      "epoch": 1.6140294768263106,
      "grad_norm": 1.6479847989463403,
      "learning_rate": 1.4354147816674215e-06,
      "loss": 0.2842,
      "step": 30280
    },
    {
      "epoch": 1.614562511660137,
      "grad_norm": 1.8092148696833215,
      "learning_rate": 1.4345342297374052e-06,
      "loss": 0.2986,
      "step": 30290
    },
    {
      "epoch": 1.6150955464939634,
      "grad_norm": 1.7517838444573417,
      "learning_rate": 1.4336537004100011e-06,
      "loss": 0.2855,
      "step": 30300
    },
    {
      "epoch": 1.6156285813277897,
      "grad_norm": 1.7383655206374091,
      "learning_rate": 1.4327731939892188e-06,
      "loss": 0.2991,
      "step": 30310
    },
    {
      "epoch": 1.6161616161616161,
      "grad_norm": 1.785947456975646,
      "learning_rate": 1.431892710779061e-06,
      "loss": 0.2939,
      "step": 30320
    },
    {
      "epoch": 1.6166946509954425,
      "grad_norm": 1.6288688315096507,
      "learning_rate": 1.4310122510835215e-06,
      "loss": 0.2971,
      "step": 30330
    },
    {
      "epoch": 1.617227685829269,
      "grad_norm": 1.760875376836615,
      "learning_rate": 1.430131815206587e-06,
      "loss": 0.2947,
      "step": 30340
    },
    {
      "epoch": 1.6177607206630953,
      "grad_norm": 1.9411375850550032,
      "learning_rate": 1.4292514034522354e-06,
      "loss": 0.2834,
      "step": 30350
    },
    {
      "epoch": 1.6182937554969217,
      "grad_norm": 1.707061858586393,
      "learning_rate": 1.428371016124436e-06,
      "loss": 0.2898,
      "step": 30360
    },
    {
      "epoch": 1.6188267903307483,
      "grad_norm": 1.7593713579007588,
      "learning_rate": 1.4274906535271503e-06,
      "loss": 0.303,
      "step": 30370
    },
    {
      "epoch": 1.6193598251645747,
      "grad_norm": 1.7196003736923966,
      "learning_rate": 1.426610315964331e-06,
      "loss": 0.2907,
      "step": 30380
    },
    {
      "epoch": 1.619892859998401,
      "grad_norm": 1.7452496295663822,
      "learning_rate": 1.4257300037399212e-06,
      "loss": 0.282,
      "step": 30390
    },
    {
      "epoch": 1.6204258948322274,
      "grad_norm": 1.604164679326339,
      "learning_rate": 1.4248497171578574e-06,
      "loss": 0.2966,
      "step": 30400
    },
    {
      "epoch": 1.6209589296660538,
      "grad_norm": 1.7681589780914448,
      "learning_rate": 1.4239694565220652e-06,
      "loss": 0.2856,
      "step": 30410
    },
    {
      "epoch": 1.6214919644998802,
      "grad_norm": 1.6882324712829686,
      "learning_rate": 1.4230892221364628e-06,
      "loss": 0.2956,
      "step": 30420
    },
    {
      "epoch": 1.6220249993337066,
      "grad_norm": 2.1158217342634247,
      "learning_rate": 1.4222090143049577e-06,
      "loss": 0.2983,
      "step": 30430
    },
    {
      "epoch": 1.622558034167533,
      "grad_norm": 1.643460807442831,
      "learning_rate": 1.4213288333314494e-06,
      "loss": 0.2922,
      "step": 30440
    },
    {
      "epoch": 1.6230910690013594,
      "grad_norm": 1.6784694368518194,
      "learning_rate": 1.420448679519828e-06,
      "loss": 0.2833,
      "step": 30450
    },
    {
      "epoch": 1.6236241038351857,
      "grad_norm": 1.7100141632747143,
      "learning_rate": 1.419568553173974e-06,
      "loss": 0.3038,
      "step": 30460
    },
    {
      "epoch": 1.6241571386690121,
      "grad_norm": 1.7312092930819818,
      "learning_rate": 1.4186884545977587e-06,
      "loss": 0.3029,
      "step": 30470
    },
    {
      "epoch": 1.6246901735028385,
      "grad_norm": 1.6933592248217049,
      "learning_rate": 1.4178083840950432e-06,
      "loss": 0.2846,
      "step": 30480
    },
    {
      "epoch": 1.625223208336665,
      "grad_norm": 1.743081006999338,
      "learning_rate": 1.4169283419696794e-06,
      "loss": 0.2906,
      "step": 30490
    },
    {
      "epoch": 1.6257562431704913,
      "grad_norm": 1.585804316310054,
      "learning_rate": 1.4160483285255092e-06,
      "loss": 0.2947,
      "step": 30500
    },
    {
      "epoch": 1.6262892780043177,
      "grad_norm": 1.6939153616740914,
      "learning_rate": 1.4151683440663654e-06,
      "loss": 0.2928,
      "step": 30510
    },
    {
      "epoch": 1.626822312838144,
      "grad_norm": 1.666220650877759,
      "learning_rate": 1.4142883888960691e-06,
      "loss": 0.2931,
      "step": 30520
    },
    {
      "epoch": 1.6273553476719704,
      "grad_norm": 1.5856672062829056,
      "learning_rate": 1.413408463318433e-06,
      "loss": 0.2851,
      "step": 30530
    },
    {
      "epoch": 1.6278883825057968,
      "grad_norm": 1.718183107222889,
      "learning_rate": 1.4125285676372582e-06,
      "loss": 0.2993,
      "step": 30540
    },
    {
      "epoch": 1.6284214173396232,
      "grad_norm": 1.71088841435637,
      "learning_rate": 1.411648702156337e-06,
      "loss": 0.2933,
      "step": 30550
    },
    {
      "epoch": 1.6289544521734496,
      "grad_norm": 1.575896504819221,
      "learning_rate": 1.4107688671794496e-06,
      "loss": 0.2894,
      "step": 30560
    },
    {
      "epoch": 1.629487487007276,
      "grad_norm": 1.5914106948402416,
      "learning_rate": 1.409889063010367e-06,
      "loss": 0.289,
      "step": 30570
    },
    {
      "epoch": 1.6300205218411024,
      "grad_norm": 1.8033248791093652,
      "learning_rate": 1.4090092899528486e-06,
      "loss": 0.2989,
      "step": 30580
    },
    {
      "epoch": 1.6305535566749287,
      "grad_norm": 1.896503632842607,
      "learning_rate": 1.4081295483106442e-06,
      "loss": 0.2821,
      "step": 30590
    },
    {
      "epoch": 1.6310865915087551,
      "grad_norm": 1.869582302899287,
      "learning_rate": 1.4072498383874913e-06,
      "loss": 0.2977,
      "step": 30600
    },
    {
      "epoch": 1.6316196263425815,
      "grad_norm": 1.8903593892500128,
      "learning_rate": 1.4063701604871174e-06,
      "loss": 0.3051,
      "step": 30610
    },
    {
      "epoch": 1.632152661176408,
      "grad_norm": 1.6453743564447636,
      "learning_rate": 1.4054905149132392e-06,
      "loss": 0.2974,
      "step": 30620
    },
    {
      "epoch": 1.6326856960102343,
      "grad_norm": 1.6357825558152437,
      "learning_rate": 1.404610901969561e-06,
      "loss": 0.283,
      "step": 30630
    },
    {
      "epoch": 1.6332187308440607,
      "grad_norm": 1.7210731554212018,
      "learning_rate": 1.4037313219597777e-06,
      "loss": 0.3019,
      "step": 30640
    },
    {
      "epoch": 1.633751765677887,
      "grad_norm": 1.8132888963016724,
      "learning_rate": 1.4028517751875706e-06,
      "loss": 0.2814,
      "step": 30650
    },
    {
      "epoch": 1.6342848005117134,
      "grad_norm": 1.8213050459750213,
      "learning_rate": 1.4019722619566114e-06,
      "loss": 0.2867,
      "step": 30660
    },
    {
      "epoch": 1.6348178353455398,
      "grad_norm": 1.7681456431689744,
      "learning_rate": 1.4010927825705594e-06,
      "loss": 0.2812,
      "step": 30670
    },
    {
      "epoch": 1.6353508701793662,
      "grad_norm": 2.060029572899713,
      "learning_rate": 1.4002133373330624e-06,
      "loss": 0.2927,
      "step": 30680
    },
    {
      "epoch": 1.6358839050131926,
      "grad_norm": 1.824451681296403,
      "learning_rate": 1.3993339265477558e-06,
      "loss": 0.287,
      "step": 30690
    },
    {
      "epoch": 1.636416939847019,
      "grad_norm": 1.7516508437153122,
      "learning_rate": 1.3984545505182638e-06,
      "loss": 0.294,
      "step": 30700
    },
    {
      "epoch": 1.6369499746808454,
      "grad_norm": 1.5737967599368465,
      "learning_rate": 1.3975752095481988e-06,
      "loss": 0.2763,
      "step": 30710
    },
    {
      "epoch": 1.6374830095146717,
      "grad_norm": 1.7734480332693277,
      "learning_rate": 1.3966959039411607e-06,
      "loss": 0.2903,
      "step": 30720
    },
    {
      "epoch": 1.6380160443484981,
      "grad_norm": 1.6660175896003442,
      "learning_rate": 1.3958166340007367e-06,
      "loss": 0.2836,
      "step": 30730
    },
    {
      "epoch": 1.6385490791823245,
      "grad_norm": 1.606722986016552,
      "learning_rate": 1.3949374000305026e-06,
      "loss": 0.2832,
      "step": 30740
    },
    {
      "epoch": 1.639082114016151,
      "grad_norm": 1.70063760810336,
      "learning_rate": 1.394058202334021e-06,
      "loss": 0.2826,
      "step": 30750
    },
    {
      "epoch": 1.6396151488499773,
      "grad_norm": 1.9001934845585937,
      "learning_rate": 1.3931790412148433e-06,
      "loss": 0.3007,
      "step": 30760
    },
    {
      "epoch": 1.6401481836838037,
      "grad_norm": 1.8850635456783504,
      "learning_rate": 1.392299916976506e-06,
      "loss": 0.292,
      "step": 30770
    },
    {
      "epoch": 1.64068121851763,
      "grad_norm": 1.961167911384526,
      "learning_rate": 1.3914208299225353e-06,
      "loss": 0.2722,
      "step": 30780
    },
    {
      "epoch": 1.6412142533514564,
      "grad_norm": 1.814877337325891,
      "learning_rate": 1.3905417803564427e-06,
      "loss": 0.2861,
      "step": 30790
    },
    {
      "epoch": 1.6417472881852828,
      "grad_norm": 1.5430378872135873,
      "learning_rate": 1.3896627685817279e-06,
      "loss": 0.2818,
      "step": 30800
    },
    {
      "epoch": 1.6422803230191092,
      "grad_norm": 1.6713813683934147,
      "learning_rate": 1.3887837949018775e-06,
      "loss": 0.2814,
      "step": 30810
    },
    {
      "epoch": 1.6428133578529356,
      "grad_norm": 1.6676642066215461,
      "learning_rate": 1.3879048596203636e-06,
      "loss": 0.2826,
      "step": 30820
    },
    {
      "epoch": 1.643346392686762,
      "grad_norm": 1.7736379747019733,
      "learning_rate": 1.3870259630406469e-06,
      "loss": 0.3006,
      "step": 30830
    },
    {
      "epoch": 1.6438794275205884,
      "grad_norm": 1.7487955680603482,
      "learning_rate": 1.3861471054661735e-06,
      "loss": 0.2904,
      "step": 30840
    },
    {
      "epoch": 1.6444124623544147,
      "grad_norm": 1.7673180412405818,
      "learning_rate": 1.3852682872003768e-06,
      "loss": 0.2917,
      "step": 30850
    },
    {
      "epoch": 1.6449454971882411,
      "grad_norm": 1.743238318718656,
      "learning_rate": 1.3843895085466758e-06,
      "loss": 0.2942,
      "step": 30860
    },
    {
      "epoch": 1.6454785320220675,
      "grad_norm": 1.8441954661311348,
      "learning_rate": 1.3835107698084766e-06,
      "loss": 0.2906,
      "step": 30870
    },
    {
      "epoch": 1.646011566855894,
      "grad_norm": 1.7166551298838426,
      "learning_rate": 1.3826320712891708e-06,
      "loss": 0.282,
      "step": 30880
    },
    {
      "epoch": 1.6465446016897203,
      "grad_norm": 1.5554577935018792,
      "learning_rate": 1.3817534132921374e-06,
      "loss": 0.2919,
      "step": 30890
    },
    {
      "epoch": 1.6470776365235467,
      "grad_norm": 1.57727885412223,
      "learning_rate": 1.3808747961207393e-06,
      "loss": 0.2878,
      "step": 30900
    },
    {
      "epoch": 1.647610671357373,
      "grad_norm": 1.678339484696579,
      "learning_rate": 1.3799962200783274e-06,
      "loss": 0.2792,
      "step": 30910
    },
    {
      "epoch": 1.6481437061911994,
      "grad_norm": 1.851238629627464,
      "learning_rate": 1.379117685468237e-06,
      "loss": 0.287,
      "step": 30920
    },
    {
      "epoch": 1.648676741025026,
      "grad_norm": 1.960066308299213,
      "learning_rate": 1.3782391925937903e-06,
      "loss": 0.2893,
      "step": 30930
    },
    {
      "epoch": 1.6492097758588524,
      "grad_norm": 1.6934858548087313,
      "learning_rate": 1.3773607417582937e-06,
      "loss": 0.2922,
      "step": 30940
    },
    {
      "epoch": 1.6497428106926788,
      "grad_norm": 1.7847875685908474,
      "learning_rate": 1.37648233326504e-06,
      "loss": 0.2884,
      "step": 30950
    },
    {
      "epoch": 1.6502758455265052,
      "grad_norm": 1.9306341100900701,
      "learning_rate": 1.3756039674173074e-06,
      "loss": 0.2907,
      "step": 30960
    },
    {
      "epoch": 1.6508088803603316,
      "grad_norm": 1.7903053432401599,
      "learning_rate": 1.3747256445183592e-06,
      "loss": 0.2893,
      "step": 30970
    },
    {
      "epoch": 1.651341915194158,
      "grad_norm": 1.876027980060498,
      "learning_rate": 1.3738473648714433e-06,
      "loss": 0.2912,
      "step": 30980
    },
    {
      "epoch": 1.6518749500279843,
      "grad_norm": 1.8155863850131362,
      "learning_rate": 1.3729691287797936e-06,
      "loss": 0.2893,
      "step": 30990
    },
    {
      "epoch": 1.6524079848618107,
      "grad_norm": 1.6603977771056089,
      "learning_rate": 1.3720909365466284e-06,
      "loss": 0.2925,
      "step": 31000
    },
    {
      "epoch": 1.6529410196956371,
      "grad_norm": 1.8083439913306736,
      "learning_rate": 1.3712127884751509e-06,
      "loss": 0.2799,
      "step": 31010
    },
    {
      "epoch": 1.6534740545294635,
      "grad_norm": 1.7179088723258629,
      "learning_rate": 1.3703346848685498e-06,
      "loss": 0.2916,
      "step": 31020
    },
    {
      "epoch": 1.6540070893632899,
      "grad_norm": 1.6414321465099093,
      "learning_rate": 1.369456626029997e-06,
      "loss": 0.2867,
      "step": 31030
    },
    {
      "epoch": 1.6545401241971163,
      "grad_norm": 1.683825838768604,
      "learning_rate": 1.36857861226265e-06,
      "loss": 0.2902,
      "step": 31040
    },
    {
      "epoch": 1.6550731590309427,
      "grad_norm": 1.7446146880372986,
      "learning_rate": 1.3677006438696508e-06,
      "loss": 0.2986,
      "step": 31050
    },
    {
      "epoch": 1.655606193864769,
      "grad_norm": 1.856553643919025,
      "learning_rate": 1.3668227211541252e-06,
      "loss": 0.2802,
      "step": 31060
    },
    {
      "epoch": 1.6561392286985954,
      "grad_norm": 1.7194379533897897,
      "learning_rate": 1.3659448444191834e-06,
      "loss": 0.2915,
      "step": 31070
    },
    {
      "epoch": 1.6566722635324218,
      "grad_norm": 1.749573299054026,
      "learning_rate": 1.3650670139679197e-06,
      "loss": 0.2932,
      "step": 31080
    },
    {
      "epoch": 1.6572052983662482,
      "grad_norm": 1.659394213860439,
      "learning_rate": 1.3641892301034129e-06,
      "loss": 0.2978,
      "step": 31090
    },
    {
      "epoch": 1.6577383332000746,
      "grad_norm": 1.751725710891011,
      "learning_rate": 1.363311493128725e-06,
      "loss": 0.291,
      "step": 31100
    },
    {
      "epoch": 1.658271368033901,
      "grad_norm": 1.6966003718667395,
      "learning_rate": 1.3624338033469023e-06,
      "loss": 0.2884,
      "step": 31110
    },
    {
      "epoch": 1.6588044028677276,
      "grad_norm": 1.5954907987729654,
      "learning_rate": 1.3615561610609744e-06,
      "loss": 0.289,
      "step": 31120
    },
    {
      "epoch": 1.659337437701554,
      "grad_norm": 1.8955414236679382,
      "learning_rate": 1.3606785665739547e-06,
      "loss": 0.2888,
      "step": 31130
    },
    {
      "epoch": 1.6598704725353803,
      "grad_norm": 1.635158551506039,
      "learning_rate": 1.3598010201888405e-06,
      "loss": 0.277,
      "step": 31140
    },
    {
      "epoch": 1.6604035073692067,
      "grad_norm": 1.814875722318185,
      "learning_rate": 1.358923522208612e-06,
      "loss": 0.2952,
      "step": 31150
    },
    {
      "epoch": 1.660936542203033,
      "grad_norm": 1.8490175131236,
      "learning_rate": 1.3580460729362325e-06,
      "loss": 0.2791,
      "step": 31160
    },
    {
      "epoch": 1.6614695770368595,
      "grad_norm": 1.5805777785080926,
      "learning_rate": 1.357168672674649e-06,
      "loss": 0.283,
      "step": 31170
    },
    {
      "epoch": 1.6620026118706859,
      "grad_norm": 1.68892409205369,
      "learning_rate": 1.3562913217267922e-06,
      "loss": 0.2956,
      "step": 31180
    },
    {
      "epoch": 1.6625356467045123,
      "grad_norm": 1.9158083003418402,
      "learning_rate": 1.355414020395573e-06,
      "loss": 0.2874,
      "step": 31190
    },
    {
      "epoch": 1.6630686815383386,
      "grad_norm": 1.795163319654632,
      "learning_rate": 1.3545367689838892e-06,
      "loss": 0.297,
      "step": 31200
    },
    {
      "epoch": 1.663601716372165,
      "grad_norm": 1.7260731918489889,
      "learning_rate": 1.3536595677946177e-06,
      "loss": 0.2897,
      "step": 31210
    },
    {
      "epoch": 1.6641347512059914,
      "grad_norm": 1.9107183609417713,
      "learning_rate": 1.3527824171306204e-06,
      "loss": 0.2958,
      "step": 31220
    },
    {
      "epoch": 1.6646677860398178,
      "grad_norm": 1.6750206083512973,
      "learning_rate": 1.3519053172947408e-06,
      "loss": 0.2836,
      "step": 31230
    },
    {
      "epoch": 1.6652008208736442,
      "grad_norm": 1.7052345419021837,
      "learning_rate": 1.3510282685898053e-06,
      "loss": 0.2822,
      "step": 31240
    },
    {
      "epoch": 1.6657338557074706,
      "grad_norm": 1.6297672522784827,
      "learning_rate": 1.3501512713186224e-06,
      "loss": 0.2798,
      "step": 31250
    },
    {
      "epoch": 1.666266890541297,
      "grad_norm": 1.669543425779207,
      "learning_rate": 1.3492743257839824e-06,
      "loss": 0.295,
      "step": 31260
    },
    {
      "epoch": 1.6667999253751233,
      "grad_norm": 1.6369560608739928,
      "learning_rate": 1.3483974322886585e-06,
      "loss": 0.2911,
      "step": 31270
    },
    {
      "epoch": 1.6673329602089497,
      "grad_norm": 1.7891980025993794,
      "learning_rate": 1.3475205911354057e-06,
      "loss": 0.2854,
      "step": 31280
    },
    {
      "epoch": 1.667865995042776,
      "grad_norm": 1.6609783141643637,
      "learning_rate": 1.3466438026269607e-06,
      "loss": 0.286,
      "step": 31290
    },
    {
      "epoch": 1.6683990298766025,
      "grad_norm": 1.9536200326989912,
      "learning_rate": 1.345767067066042e-06,
      "loss": 0.2876,
      "step": 31300
    },
    {
      "epoch": 1.6689320647104289,
      "grad_norm": 1.5598354111095247,
      "learning_rate": 1.34489038475535e-06,
      "loss": 0.2796,
      "step": 31310
    },
    {
      "epoch": 1.6694650995442553,
      "grad_norm": 1.7762987162645032,
      "learning_rate": 1.344013755997567e-06,
      "loss": 0.2863,
      "step": 31320
    },
    {
      "epoch": 1.6699981343780816,
      "grad_norm": 3.349481124235958,
      "learning_rate": 1.3431371810953566e-06,
      "loss": 0.2949,
      "step": 31330
    },
    {
      "epoch": 1.670531169211908,
      "grad_norm": 1.58855393309782,
      "learning_rate": 1.3422606603513629e-06,
      "loss": 0.2883,
      "step": 31340
    },
    {
      "epoch": 1.6710642040457344,
      "grad_norm": 1.6516568219929992,
      "learning_rate": 1.3413841940682129e-06,
      "loss": 0.284,
      "step": 31350
    },
    {
      "epoch": 1.6715972388795608,
      "grad_norm": 1.5289002355984553,
      "learning_rate": 1.3405077825485137e-06,
      "loss": 0.2929,
      "step": 31360
    },
    {
      "epoch": 1.6721302737133872,
      "grad_norm": 1.5399826412917716,
      "learning_rate": 1.3396314260948543e-06,
      "loss": 0.2883,
      "step": 31370
    },
    {
      "epoch": 1.6726633085472136,
      "grad_norm": 1.7998503249067999,
      "learning_rate": 1.3387551250098033e-06,
      "loss": 0.2947,
      "step": 31380
    },
    {
      "epoch": 1.67319634338104,
      "grad_norm": 1.8359330897805242,
      "learning_rate": 1.3378788795959117e-06,
      "loss": 0.283,
      "step": 31390
    },
    {
      "epoch": 1.6737293782148663,
      "grad_norm": 1.6775115709540163,
      "learning_rate": 1.3370026901557104e-06,
      "loss": 0.2902,
      "step": 31400
    },
    {
      "epoch": 1.6742624130486927,
      "grad_norm": 1.8343363698692996,
      "learning_rate": 1.3361265569917112e-06,
      "loss": 0.2921,
      "step": 31410
    },
    {
      "epoch": 1.674795447882519,
      "grad_norm": 1.9977825013841777,
      "learning_rate": 1.3352504804064071e-06,
      "loss": 0.2852,
      "step": 31420
    },
    {
      "epoch": 1.6753284827163455,
      "grad_norm": 1.8634209358258391,
      "learning_rate": 1.3343744607022701e-06,
      "loss": 0.2849,
      "step": 31430
    },
    {
      "epoch": 1.6758615175501719,
      "grad_norm": 1.821551382582791,
      "learning_rate": 1.3334984981817539e-06,
      "loss": 0.2852,
      "step": 31440
    },
    {
      "epoch": 1.6763945523839983,
      "grad_norm": 1.6794818871386097,
      "learning_rate": 1.3326225931472917e-06,
      "loss": 0.277,
      "step": 31450
    },
    {
      "epoch": 1.6769275872178246,
      "grad_norm": 1.633606121551204,
      "learning_rate": 1.3317467459012978e-06,
      "loss": 0.2951,
      "step": 31460
    },
    {
      "epoch": 1.677460622051651,
      "grad_norm": 2.1307577204739307,
      "learning_rate": 1.330870956746165e-06,
      "loss": 0.2914,
      "step": 31470
    },
    {
      "epoch": 1.6779936568854774,
      "grad_norm": 1.7030936589122425,
      "learning_rate": 1.3299952259842674e-06,
      "loss": 0.2889,
      "step": 31480
    },
    {
      "epoch": 1.6785266917193038,
      "grad_norm": 1.7935922091801022,
      "learning_rate": 1.3291195539179584e-06,
      "loss": 0.2909,
      "step": 31490
    },
    {
      "epoch": 1.6790597265531302,
      "grad_norm": 1.7550348987162039,
      "learning_rate": 1.3282439408495715e-06,
      "loss": 0.3001,
      "step": 31500
    },
    {
      "epoch": 1.6795927613869566,
      "grad_norm": 1.824332956287139,
      "learning_rate": 1.327368387081419e-06,
      "loss": 0.301,
      "step": 31510
    },
    {
      "epoch": 1.680125796220783,
      "grad_norm": 1.9263867992383745,
      "learning_rate": 1.326492892915793e-06,
      "loss": 0.2858,
      "step": 31520
    },
    {
      "epoch": 1.6806588310546093,
      "grad_norm": 1.638530444452028,
      "learning_rate": 1.3256174586549662e-06,
      "loss": 0.2782,
      "step": 31530
    },
    {
      "epoch": 1.6811918658884357,
      "grad_norm": 1.858423045256905,
      "learning_rate": 1.3247420846011894e-06,
      "loss": 0.2813,
      "step": 31540
    },
    {
      "epoch": 1.681724900722262,
      "grad_norm": 1.7383105626396633,
      "learning_rate": 1.3238667710566928e-06,
      "loss": 0.2863,
      "step": 31550
    },
    {
      "epoch": 1.6822579355560885,
      "grad_norm": 1.8716211776981713,
      "learning_rate": 1.3229915183236852e-06,
      "loss": 0.2939,
      "step": 31560
    },
    {
      "epoch": 1.6827909703899149,
      "grad_norm": 1.6548848631003752,
      "learning_rate": 1.3221163267043561e-06,
      "loss": 0.2922,
      "step": 31570
    },
    {
      "epoch": 1.6833240052237413,
      "grad_norm": 1.8097577187118805,
      "learning_rate": 1.321241196500872e-06,
      "loss": 0.2768,
      "step": 31580
    },
    {
      "epoch": 1.6838570400575676,
      "grad_norm": 1.835159675615701,
      "learning_rate": 1.3203661280153798e-06,
      "loss": 0.3014,
      "step": 31590
    },
    {
      "epoch": 1.684390074891394,
      "grad_norm": 1.6512864314305946,
      "learning_rate": 1.3194911215500038e-06,
      "loss": 0.2828,
      "step": 31600
    },
    {
      "epoch": 1.6849231097252204,
      "grad_norm": 1.782270675904756,
      "learning_rate": 1.318616177406847e-06,
      "loss": 0.2856,
      "step": 31610
    },
    {
      "epoch": 1.6854561445590468,
      "grad_norm": 1.8577533867222902,
      "learning_rate": 1.317741295887992e-06,
      "loss": 0.2795,
      "step": 31620
    },
    {
      "epoch": 1.6859891793928732,
      "grad_norm": 1.8659766822133308,
      "learning_rate": 1.316866477295499e-06,
      "loss": 0.2879,
      "step": 31630
    },
    {
      "epoch": 1.6865222142266996,
      "grad_norm": 1.967231372853565,
      "learning_rate": 1.3159917219314058e-06,
      "loss": 0.2973,
      "step": 31640
    },
    {
      "epoch": 1.687055249060526,
      "grad_norm": 1.9068235135729854,
      "learning_rate": 1.3151170300977299e-06,
      "loss": 0.2868,
      "step": 31650
    },
    {
      "epoch": 1.6875882838943523,
      "grad_norm": 1.759374970615166,
      "learning_rate": 1.3142424020964657e-06,
      "loss": 0.2784,
      "step": 31660
    },
    {
      "epoch": 1.6881213187281787,
      "grad_norm": 1.8294860719491626,
      "learning_rate": 1.3133678382295864e-06,
      "loss": 0.2925,
      "step": 31670
    },
    {
      "epoch": 1.6886543535620053,
      "grad_norm": 1.7031323954040583,
      "learning_rate": 1.3124933387990417e-06,
      "loss": 0.283,
      "step": 31680
    },
    {
      "epoch": 1.6891873883958317,
      "grad_norm": 1.5014912640479998,
      "learning_rate": 1.3116189041067607e-06,
      "loss": 0.2911,
      "step": 31690
    },
    {
      "epoch": 1.689720423229658,
      "grad_norm": 1.5626481458831571,
      "learning_rate": 1.310744534454649e-06,
      "loss": 0.2956,
      "step": 31700
    },
    {
      "epoch": 1.6902534580634845,
      "grad_norm": 1.742988721027641,
      "learning_rate": 1.3098702301445904e-06,
      "loss": 0.2838,
      "step": 31710
    },
    {
      "epoch": 1.6907864928973109,
      "grad_norm": 1.7449672312982958,
      "learning_rate": 1.3089959914784453e-06,
      "loss": 0.2818,
      "step": 31720
    },
    {
      "epoch": 1.6913195277311373,
      "grad_norm": 1.8748681739051887,
      "learning_rate": 1.3081218187580527e-06,
      "loss": 0.2907,
      "step": 31730
    },
    {
      "epoch": 1.6918525625649636,
      "grad_norm": 1.9013414254794463,
      "learning_rate": 1.3072477122852278e-06,
      "loss": 0.2877,
      "step": 31740
    },
    {
      "epoch": 1.69238559739879,
      "grad_norm": 1.668232613451,
      "learning_rate": 1.3063736723617634e-06,
      "loss": 0.2908,
      "step": 31750
    },
    {
      "epoch": 1.6929186322326164,
      "grad_norm": 1.6469132163245612,
      "learning_rate": 1.3054996992894294e-06,
      "loss": 0.2788,
      "step": 31760
    },
    {
      "epoch": 1.6934516670664428,
      "grad_norm": 1.7962405199138152,
      "learning_rate": 1.3046257933699718e-06,
      "loss": 0.291,
      "step": 31770
    },
    {
      "epoch": 1.6939847019002692,
      "grad_norm": 1.8084771430798992,
      "learning_rate": 1.3037519549051141e-06,
      "loss": 0.2899,
      "step": 31780
    },
    {
      "epoch": 1.6945177367340956,
      "grad_norm": 1.7644452209017247,
      "learning_rate": 1.302878184196557e-06,
      "loss": 0.2907,
      "step": 31790
    },
    {
      "epoch": 1.695050771567922,
      "grad_norm": 1.9754336026117922,
      "learning_rate": 1.3020044815459771e-06,
      "loss": 0.291,
      "step": 31800
    },
    {
      "epoch": 1.6955838064017483,
      "grad_norm": 1.664595344877802,
      "learning_rate": 1.3011308472550273e-06,
      "loss": 0.2794,
      "step": 31810
    },
    {
      "epoch": 1.6961168412355747,
      "grad_norm": 1.757012749517835,
      "learning_rate": 1.3002572816253371e-06,
      "loss": 0.283,
      "step": 31820
    },
    {
      "epoch": 1.696649876069401,
      "grad_norm": 1.587702243368332,
      "learning_rate": 1.299383784958513e-06,
      "loss": 0.2921,
      "step": 31830
    },
    {
      "epoch": 1.6971829109032275,
      "grad_norm": 1.6658819359421824,
      "learning_rate": 1.298510357556137e-06,
      "loss": 0.2856,
      "step": 31840
    },
    {
      "epoch": 1.6977159457370539,
      "grad_norm": 1.8413089384085608,
      "learning_rate": 1.297636999719767e-06,
      "loss": 0.2955,
      "step": 31850
    },
    {
      "epoch": 1.6982489805708805,
      "grad_norm": 1.645899438988138,
      "learning_rate": 1.2967637117509375e-06,
      "loss": 0.2864,
      "step": 31860
    },
    {
      "epoch": 1.6987820154047069,
      "grad_norm": 1.639350930993552,
      "learning_rate": 1.2958904939511582e-06,
      "loss": 0.2814,
      "step": 31870
    },
    {
      "epoch": 1.6993150502385332,
      "grad_norm": 1.6914028177368043,
      "learning_rate": 1.2950173466219154e-06,
      "loss": 0.2814,
      "step": 31880
    },
    {
      "epoch": 1.6998480850723596,
      "grad_norm": 1.6859154122951039,
      "learning_rate": 1.2941442700646704e-06,
      "loss": 0.2984,
      "step": 31890
    },
    {
      "epoch": 1.700381119906186,
      "grad_norm": 1.69675818981569,
      "learning_rate": 1.29327126458086e-06,
      "loss": 0.2856,
      "step": 31900
    },
    {
      "epoch": 1.7009141547400124,
      "grad_norm": 1.837627928283781,
      "learning_rate": 1.292398330471897e-06,
      "loss": 0.2946,
      "step": 31910
    },
    {
      "epoch": 1.7014471895738388,
      "grad_norm": 1.6825116669360909,
      "learning_rate": 1.2915254680391699e-06,
      "loss": 0.2902,
      "step": 31920
    },
    {
      "epoch": 1.7019802244076652,
      "grad_norm": 1.6862398600509785,
      "learning_rate": 1.2906526775840408e-06,
      "loss": 0.2888,
      "step": 31930
    },
    {
      "epoch": 1.7025132592414916,
      "grad_norm": 1.605175864135079,
      "learning_rate": 1.2897799594078484e-06,
      "loss": 0.2875,
      "step": 31940
    },
    {
      "epoch": 1.703046294075318,
      "grad_norm": 1.800408623747545,
      "learning_rate": 1.2889073138119063e-06,
      "loss": 0.2882,
      "step": 31950
    },
    {
      "epoch": 1.7035793289091443,
      "grad_norm": 1.9003792829068324,
      "learning_rate": 1.2880347410975023e-06,
      "loss": 0.2909,
      "step": 31960
    },
    {
      "epoch": 1.7041123637429707,
      "grad_norm": 1.5733354035455709,
      "learning_rate": 1.2871622415659002e-06,
      "loss": 0.2962,
      "step": 31970
    },
    {
      "epoch": 1.704645398576797,
      "grad_norm": 1.8486221273412649,
      "learning_rate": 1.2862898155183372e-06,
      "loss": 0.2872,
      "step": 31980
    },
    {
      "epoch": 1.7051784334106235,
      "grad_norm": 1.8013937278620276,
      "learning_rate": 1.2854174632560262e-06,
      "loss": 0.2906,
      "step": 31990
    },
    {
      "epoch": 1.7057114682444499,
      "grad_norm": 1.7335084828565217,
      "learning_rate": 1.284545185080154e-06,
      "loss": 0.2903,
      "step": 32000
    },
    {
      "epoch": 1.7062445030782762,
      "grad_norm": 1.8179861791064678,
      "learning_rate": 1.2836729812918823e-06,
      "loss": 0.2841,
      "step": 32010
    },
    {
      "epoch": 1.7067775379121026,
      "grad_norm": 1.7012698782629172,
      "learning_rate": 1.2828008521923463e-06,
      "loss": 0.2807,
      "step": 32020
    },
    {
      "epoch": 1.707310572745929,
      "grad_norm": 1.8686016778614292,
      "learning_rate": 1.2819287980826565e-06,
      "loss": 0.2907,
      "step": 32030
    },
    {
      "epoch": 1.7078436075797554,
      "grad_norm": 1.814373854463964,
      "learning_rate": 1.2810568192638966e-06,
      "loss": 0.2855,
      "step": 32040
    },
    {
      "epoch": 1.7083766424135818,
      "grad_norm": 1.731405453747505,
      "learning_rate": 1.2801849160371252e-06,
      "loss": 0.2878,
      "step": 32050
    },
    {
      "epoch": 1.7089096772474082,
      "grad_norm": 1.918965569057865,
      "learning_rate": 1.2793130887033736e-06,
      "loss": 0.2881,
      "step": 32060
    },
    {
      "epoch": 1.7094427120812345,
      "grad_norm": 1.8066009656542463,
      "learning_rate": 1.278441337563648e-06,
      "loss": 0.2963,
      "step": 32070
    },
    {
      "epoch": 1.709975746915061,
      "grad_norm": 1.6391056949774767,
      "learning_rate": 1.2775696629189275e-06,
      "loss": 0.2933,
      "step": 32080
    },
    {
      "epoch": 1.7105087817488873,
      "grad_norm": 1.724589478012094,
      "learning_rate": 1.276698065070166e-06,
      "loss": 0.2929,
      "step": 32090
    },
    {
      "epoch": 1.7110418165827137,
      "grad_norm": 1.6124000033160877,
      "learning_rate": 1.275826544318289e-06,
      "loss": 0.295,
      "step": 32100
    },
    {
      "epoch": 1.71157485141654,
      "grad_norm": 1.9262350031671895,
      "learning_rate": 1.274955100964197e-06,
      "loss": 0.299,
      "step": 32110
    },
    {
      "epoch": 1.7121078862503665,
      "grad_norm": 1.7295431598476851,
      "learning_rate": 1.274083735308763e-06,
      "loss": 0.2824,
      "step": 32120
    },
    {
      "epoch": 1.7126409210841929,
      "grad_norm": 1.6797257611595653,
      "learning_rate": 1.2732124476528336e-06,
      "loss": 0.2848,
      "step": 32130
    },
    {
      "epoch": 1.7131739559180192,
      "grad_norm": 1.6373776177580501,
      "learning_rate": 1.272341238297228e-06,
      "loss": 0.3034,
      "step": 32140
    },
    {
      "epoch": 1.7137069907518456,
      "grad_norm": 1.6542781596549008,
      "learning_rate": 1.2714701075427388e-06,
      "loss": 0.2946,
      "step": 32150
    },
    {
      "epoch": 1.714240025585672,
      "grad_norm": 1.7592181423552475,
      "learning_rate": 1.270599055690131e-06,
      "loss": 0.296,
      "step": 32160
    },
    {
      "epoch": 1.7147730604194984,
      "grad_norm": 1.7796172949126436,
      "learning_rate": 1.269728083040143e-06,
      "loss": 0.2888,
      "step": 32170
    },
    {
      "epoch": 1.7153060952533248,
      "grad_norm": 1.6094262967291313,
      "learning_rate": 1.2688571898934856e-06,
      "loss": 0.2897,
      "step": 32180
    },
    {
      "epoch": 1.7158391300871512,
      "grad_norm": 1.7634994229426055,
      "learning_rate": 1.2679863765508413e-06,
      "loss": 0.2922,
      "step": 32190
    },
    {
      "epoch": 1.7163721649209775,
      "grad_norm": 1.675866725691634,
      "learning_rate": 1.2671156433128662e-06,
      "loss": 0.2947,
      "step": 32200
    },
    {
      "epoch": 1.716905199754804,
      "grad_norm": 1.8231333892473982,
      "learning_rate": 1.2662449904801885e-06,
      "loss": 0.2806,
      "step": 32210
    },
    {
      "epoch": 1.7174382345886303,
      "grad_norm": 1.7315291727014117,
      "learning_rate": 1.2653744183534088e-06,
      "loss": 0.2882,
      "step": 32220
    },
    {
      "epoch": 1.7179712694224567,
      "grad_norm": 1.8892073254805635,
      "learning_rate": 1.2645039272330985e-06,
      "loss": 0.2819,
      "step": 32230
    },
    {
      "epoch": 1.718504304256283,
      "grad_norm": 1.9271801104470458,
      "learning_rate": 1.2636335174198024e-06,
      "loss": 0.2792,
      "step": 32240
    },
    {
      "epoch": 1.7190373390901095,
      "grad_norm": 1.8985297799323286,
      "learning_rate": 1.2627631892140374e-06,
      "loss": 0.2899,
      "step": 32250
    },
    {
      "epoch": 1.7195703739239359,
      "grad_norm": 1.8352453750448383,
      "learning_rate": 1.2618929429162914e-06,
      "loss": 0.2833,
      "step": 32260
    },
    {
      "epoch": 1.7201034087577622,
      "grad_norm": 1.7603140151922858,
      "learning_rate": 1.2610227788270238e-06,
      "loss": 0.2893,
      "step": 32270
    },
    {
      "epoch": 1.7206364435915886,
      "grad_norm": 2.066031521938059,
      "learning_rate": 1.2601526972466669e-06,
      "loss": 0.2893,
      "step": 32280
    },
    {
      "epoch": 1.721169478425415,
      "grad_norm": 1.7923899227892706,
      "learning_rate": 1.259282698475623e-06,
      "loss": 0.2979,
      "step": 32290
    },
    {
      "epoch": 1.7217025132592414,
      "grad_norm": 1.8145893131581892,
      "learning_rate": 1.258412782814267e-06,
      "loss": 0.288,
      "step": 32300
    },
    {
      "epoch": 1.7222355480930678,
      "grad_norm": 1.7860490298144265,
      "learning_rate": 1.257542950562945e-06,
      "loss": 0.288,
      "step": 32310
    },
    {
      "epoch": 1.7227685829268942,
      "grad_norm": 1.7058416146793243,
      "learning_rate": 1.2566732020219733e-06,
      "loss": 0.2731,
      "step": 32320
    },
    {
      "epoch": 1.7233016177607205,
      "grad_norm": 1.6392637297748773,
      "learning_rate": 1.25580353749164e-06,
      "loss": 0.2794,
      "step": 32330
    },
    {
      "epoch": 1.723834652594547,
      "grad_norm": 1.7619685791656245,
      "learning_rate": 1.2549339572722045e-06,
      "loss": 0.298,
      "step": 32340
    },
    {
      "epoch": 1.7243676874283733,
      "grad_norm": 1.6928993014860996,
      "learning_rate": 1.254064461663897e-06,
      "loss": 0.282,
      "step": 32350
    },
    {
      "epoch": 1.7249007222621997,
      "grad_norm": 1.8634610303765131,
      "learning_rate": 1.2531950509669173e-06,
      "loss": 0.2871,
      "step": 32360
    },
    {
      "epoch": 1.725433757096026,
      "grad_norm": 1.8198287653540872,
      "learning_rate": 1.2523257254814373e-06,
      "loss": 0.2772,
      "step": 32370
    },
    {
      "epoch": 1.7259667919298525,
      "grad_norm": 1.7712246352972696,
      "learning_rate": 1.251456485507599e-06,
      "loss": 0.2972,
      "step": 32380
    },
    {
      "epoch": 1.7264998267636789,
      "grad_norm": 1.6928742444614127,
      "learning_rate": 1.2505873313455152e-06,
      "loss": 0.2861,
      "step": 32390
    },
    {
      "epoch": 1.7270328615975052,
      "grad_norm": 1.5743304135552794,
      "learning_rate": 1.249718263295268e-06,
      "loss": 0.2782,
      "step": 32400
    },
    {
      "epoch": 1.7275658964313316,
      "grad_norm": 1.8000940569408972,
      "learning_rate": 1.248849281656911e-06,
      "loss": 0.2937,
      "step": 32410
    },
    {
      "epoch": 1.7280989312651582,
      "grad_norm": 2.1364880791695002,
      "learning_rate": 1.2479803867304674e-06,
      "loss": 0.2927,
      "step": 32420
    },
    {
      "epoch": 1.7286319660989846,
      "grad_norm": 1.7289884596703256,
      "learning_rate": 1.2471115788159307e-06,
      "loss": 0.298,
      "step": 32430
    },
    {
      "epoch": 1.729165000932811,
      "grad_norm": 1.7885437705839957,
      "learning_rate": 1.246242858213264e-06,
      "loss": 0.2805,
      "step": 32440
    },
    {
      "epoch": 1.7296980357666374,
      "grad_norm": 1.7213749843065442,
      "learning_rate": 1.2453742252224001e-06,
      "loss": 0.2828,
      "step": 32450
    },
    {
      "epoch": 1.7302310706004638,
      "grad_norm": 1.734110349766069,
      "learning_rate": 1.2445056801432426e-06,
      "loss": 0.2817,
      "step": 32460
    },
    {
      "epoch": 1.7307641054342902,
      "grad_norm": 1.793158908262709,
      "learning_rate": 1.2436372232756633e-06,
      "loss": 0.2904,
      "step": 32470
    },
    {
      "epoch": 1.7312971402681165,
      "grad_norm": 1.9125650433337338,
      "learning_rate": 1.2427688549195054e-06,
      "loss": 0.2831,
      "step": 32480
    },
    {
      "epoch": 1.731830175101943,
      "grad_norm": 1.587736332768039,
      "learning_rate": 1.2419005753745791e-06,
      "loss": 0.29,
      "step": 32490
    },
    {
      "epoch": 1.7323632099357693,
      "grad_norm": 1.6885322881543168,
      "learning_rate": 1.241032384940666e-06,
      "loss": 0.2938,
      "step": 32500
    },
    {
      "epoch": 1.7328962447695957,
      "grad_norm": 1.684896672377855,
      "learning_rate": 1.2401642839175164e-06,
      "loss": 0.2824,
      "step": 32510
    },
    {
      "epoch": 1.733429279603422,
      "grad_norm": 1.826670321336252,
      "learning_rate": 1.2392962726048492e-06,
      "loss": 0.2906,
      "step": 32520
    },
    {
      "epoch": 1.7339623144372485,
      "grad_norm": 1.6999067401066261,
      "learning_rate": 1.2384283513023525e-06,
      "loss": 0.2933,
      "step": 32530
    },
    {
      "epoch": 1.7344953492710748,
      "grad_norm": 1.8878656399525928,
      "learning_rate": 1.2375605203096843e-06,
      "loss": 0.2812,
      "step": 32540
    },
    {
      "epoch": 1.7350283841049012,
      "grad_norm": 1.8051267375967035,
      "learning_rate": 1.2366927799264693e-06,
      "loss": 0.2806,
      "step": 32550
    },
    {
      "epoch": 1.7355614189387276,
      "grad_norm": 2.028139662727251,
      "learning_rate": 1.2358251304523029e-06,
      "loss": 0.2907,
      "step": 32560
    },
    {
      "epoch": 1.736094453772554,
      "grad_norm": 1.76113236862697,
      "learning_rate": 1.2349575721867486e-06,
      "loss": 0.278,
      "step": 32570
    },
    {
      "epoch": 1.7366274886063804,
      "grad_norm": 1.8044038208104018,
      "learning_rate": 1.234090105429338e-06,
      "loss": 0.2932,
      "step": 32580
    },
    {
      "epoch": 1.7371605234402068,
      "grad_norm": 1.901315843664375,
      "learning_rate": 1.2332227304795716e-06,
      "loss": 0.2946,
      "step": 32590
    },
    {
      "epoch": 1.7376935582740332,
      "grad_norm": 1.6246469206639678,
      "learning_rate": 1.2323554476369176e-06,
      "loss": 0.2848,
      "step": 32600
    },
    {
      "epoch": 1.7382265931078598,
      "grad_norm": 1.8612589241364976,
      "learning_rate": 1.231488257200813e-06,
      "loss": 0.2808,
      "step": 32610
    },
    {
      "epoch": 1.7387596279416861,
      "grad_norm": 1.9816092252933535,
      "learning_rate": 1.2306211594706627e-06,
      "loss": 0.2848,
      "step": 32620
    },
    {
      "epoch": 1.7392926627755125,
      "grad_norm": 1.6454266590077804,
      "learning_rate": 1.2297541547458392e-06,
      "loss": 0.2943,
      "step": 32630
    },
    {
      "epoch": 1.739825697609339,
      "grad_norm": 1.6934677428973497,
      "learning_rate": 1.2288872433256833e-06,
      "loss": 0.2865,
      "step": 32640
    },
    {
      "epoch": 1.7403587324431653,
      "grad_norm": 1.8349130663313056,
      "learning_rate": 1.2280204255095037e-06,
      "loss": 0.2968,
      "step": 32650
    },
    {
      "epoch": 1.7408917672769917,
      "grad_norm": 1.8082006055395339,
      "learning_rate": 1.2271537015965771e-06,
      "loss": 0.2922,
      "step": 32660
    },
    {
      "epoch": 1.741424802110818,
      "grad_norm": 1.6132799450246456,
      "learning_rate": 1.226287071886146e-06,
      "loss": 0.2926,
      "step": 32670
    },
    {
      "epoch": 1.7419578369446445,
      "grad_norm": 1.8330073109081042,
      "learning_rate": 1.2254205366774225e-06,
      "loss": 0.2963,
      "step": 32680
    },
    {
      "epoch": 1.7424908717784708,
      "grad_norm": 1.7490262019324814,
      "learning_rate": 1.2245540962695852e-06,
      "loss": 0.2817,
      "step": 32690
    },
    {
      "epoch": 1.7430239066122972,
      "grad_norm": 1.5694743555498218,
      "learning_rate": 1.2236877509617797e-06,
      "loss": 0.3006,
      "step": 32700
    },
    {
      "epoch": 1.7435569414461236,
      "grad_norm": 1.6365517300644163,
      "learning_rate": 1.22282150105312e-06,
      "loss": 0.2815,
      "step": 32710
    },
    {
      "epoch": 1.74408997627995,
      "grad_norm": 1.6707725670788345,
      "learning_rate": 1.2219553468426846e-06,
      "loss": 0.2902,
      "step": 32720
    },
    {
      "epoch": 1.7446230111137764,
      "grad_norm": 1.6665716145520126,
      "learning_rate": 1.2210892886295216e-06,
      "loss": 0.276,
      "step": 32730
    },
    {
      "epoch": 1.7451560459476028,
      "grad_norm": 1.8149634728887676,
      "learning_rate": 1.220223326712645e-06,
      "loss": 0.289,
      "step": 32740
    },
    {
      "epoch": 1.7456890807814291,
      "grad_norm": 1.7024464590510855,
      "learning_rate": 1.2193574613910354e-06,
      "loss": 0.2795,
      "step": 32750
    },
    {
      "epoch": 1.7462221156152555,
      "grad_norm": 1.6677739205639304,
      "learning_rate": 1.2184916929636395e-06,
      "loss": 0.2943,
      "step": 32760
    },
    {
      "epoch": 1.746755150449082,
      "grad_norm": 1.8810025155021723,
      "learning_rate": 1.2176260217293719e-06,
      "loss": 0.2756,
      "step": 32770
    },
    {
      "epoch": 1.7472881852829083,
      "grad_norm": 1.6084077896840048,
      "learning_rate": 1.2167604479871126e-06,
      "loss": 0.2778,
      "step": 32780
    },
    {
      "epoch": 1.7478212201167347,
      "grad_norm": 1.8202668096359373,
      "learning_rate": 1.2158949720357088e-06,
      "loss": 0.2844,
      "step": 32790
    },
    {
      "epoch": 1.748354254950561,
      "grad_norm": 1.727782645554656,
      "learning_rate": 1.2150295941739725e-06,
      "loss": 0.2904,
      "step": 32800
    },
    {
      "epoch": 1.7488872897843875,
      "grad_norm": 1.5626873948146105,
      "learning_rate": 1.2141643147006835e-06,
      "loss": 0.2816,
      "step": 32810
    },
    {
      "epoch": 1.7494203246182138,
      "grad_norm": 1.7828010861627483,
      "learning_rate": 1.2132991339145864e-06,
      "loss": 0.2951,
      "step": 32820
    },
    {
      "epoch": 1.7499533594520402,
      "grad_norm": 1.765954718394173,
      "learning_rate": 1.2124340521143927e-06,
      "loss": 0.2842,
      "step": 32830
    },
    {
      "epoch": 1.7504863942858666,
      "grad_norm": 2.133384100820481,
      "learning_rate": 1.2115690695987787e-06,
      "loss": 0.2981,
      "step": 32840
    },
    {
      "epoch": 1.751019429119693,
      "grad_norm": 1.7693334962578569,
      "learning_rate": 1.2107041866663871e-06,
      "loss": 0.286,
      "step": 32850
    },
    {
      "epoch": 1.7515524639535194,
      "grad_norm": 1.9185758906996053,
      "learning_rate": 1.2098394036158261e-06,
      "loss": 0.2856,
      "step": 32860
    },
    {
      "epoch": 1.7520854987873458,
      "grad_norm": 1.631677298504431,
      "learning_rate": 1.2089747207456698e-06,
      "loss": 0.2783,
      "step": 32870
    },
    {
      "epoch": 1.7526185336211721,
      "grad_norm": 2.002703240882398,
      "learning_rate": 1.2081101383544565e-06,
      "loss": 0.2818,
      "step": 32880
    },
    {
      "epoch": 1.7531515684549985,
      "grad_norm": 1.857754950565365,
      "learning_rate": 1.207245656740691e-06,
      "loss": 0.2934,
      "step": 32890
    },
    {
      "epoch": 1.753684603288825,
      "grad_norm": 1.7517615486818996,
      "learning_rate": 1.2063812762028426e-06,
      "loss": 0.2905,
      "step": 32900
    },
    {
      "epoch": 1.7542176381226513,
      "grad_norm": 1.6296068835059556,
      "learning_rate": 1.2055169970393467e-06,
      "loss": 0.2906,
      "step": 32910
    },
    {
      "epoch": 1.7547506729564777,
      "grad_norm": 1.7247746637727355,
      "learning_rate": 1.2046528195486024e-06,
      "loss": 0.282,
      "step": 32920
    },
    {
      "epoch": 1.755283707790304,
      "grad_norm": 1.8232212179343361,
      "learning_rate": 1.2037887440289747e-06,
      "loss": 0.2812,
      "step": 32930
    },
    {
      "epoch": 1.7558167426241305,
      "grad_norm": 1.8554989663379087,
      "learning_rate": 1.2029247707787924e-06,
      "loss": 0.2996,
      "step": 32940
    },
    {
      "epoch": 1.7563497774579568,
      "grad_norm": 1.822530092997045,
      "learning_rate": 1.2020609000963501e-06,
      "loss": 0.2725,
      "step": 32950
    },
    {
      "epoch": 1.7568828122917832,
      "grad_norm": 1.629573216162499,
      "learning_rate": 1.2011971322799067e-06,
      "loss": 0.2914,
      "step": 32960
    },
    {
      "epoch": 1.7574158471256096,
      "grad_norm": 1.809886682543272,
      "learning_rate": 1.2003334676276846e-06,
      "loss": 0.2972,
      "step": 32970
    },
    {
      "epoch": 1.757948881959436,
      "grad_norm": 1.6532586884485587,
      "learning_rate": 1.1994699064378718e-06,
      "loss": 0.2944,
      "step": 32980
    },
    {
      "epoch": 1.7584819167932624,
      "grad_norm": 1.7716347872072205,
      "learning_rate": 1.1986064490086199e-06,
      "loss": 0.2795,
      "step": 32990
    },
    {
      "epoch": 1.7590149516270888,
      "grad_norm": 1.8549978661483972,
      "learning_rate": 1.1977430956380455e-06,
      "loss": 0.289,
      "step": 33000
    },
    {
      "epoch": 1.7595479864609151,
      "grad_norm": 1.7326000225639766,
      "learning_rate": 1.1968798466242278e-06,
      "loss": 0.2854,
      "step": 33010
    },
    {
      "epoch": 1.7600810212947415,
      "grad_norm": 1.754159471927188,
      "learning_rate": 1.1960167022652112e-06,
      "loss": 0.2822,
      "step": 33020
    },
    {
      "epoch": 1.760614056128568,
      "grad_norm": 1.7283279761900754,
      "learning_rate": 1.1951536628590037e-06,
      "loss": 0.2852,
      "step": 33030
    },
    {
      "epoch": 1.7611470909623943,
      "grad_norm": 1.793610205395446,
      "learning_rate": 1.1942907287035773e-06,
      "loss": 0.2814,
      "step": 33040
    },
    {
      "epoch": 1.7616801257962207,
      "grad_norm": 1.7109075087930035,
      "learning_rate": 1.1934279000968665e-06,
      "loss": 0.2923,
      "step": 33050
    },
    {
      "epoch": 1.762213160630047,
      "grad_norm": 1.7921948066413613,
      "learning_rate": 1.1925651773367706e-06,
      "loss": 0.2844,
      "step": 33060
    },
    {
      "epoch": 1.7627461954638735,
      "grad_norm": 1.5363061497054005,
      "learning_rate": 1.1917025607211521e-06,
      "loss": 0.2674,
      "step": 33070
    },
    {
      "epoch": 1.7632792302976998,
      "grad_norm": 1.709148251541826,
      "learning_rate": 1.1908400505478365e-06,
      "loss": 0.2842,
      "step": 33080
    },
    {
      "epoch": 1.7638122651315262,
      "grad_norm": 1.6682557427616687,
      "learning_rate": 1.1899776471146133e-06,
      "loss": 0.2956,
      "step": 33090
    },
    {
      "epoch": 1.7643452999653526,
      "grad_norm": 1.686295591873106,
      "learning_rate": 1.1891153507192334e-06,
      "loss": 0.2814,
      "step": 33100
    },
    {
      "epoch": 1.764878334799179,
      "grad_norm": 1.621018205567231,
      "learning_rate": 1.1882531616594128e-06,
      "loss": 0.286,
      "step": 33110
    },
    {
      "epoch": 1.7654113696330054,
      "grad_norm": 1.6571672657074825,
      "learning_rate": 1.1873910802328296e-06,
      "loss": 0.2867,
      "step": 33120
    },
    {
      "epoch": 1.7659444044668318,
      "grad_norm": 1.7193043288993533,
      "learning_rate": 1.1865291067371246e-06,
      "loss": 0.2853,
      "step": 33130
    },
    {
      "epoch": 1.7664774393006581,
      "grad_norm": 2.0076084089130024,
      "learning_rate": 1.185667241469901e-06,
      "loss": 0.2958,
      "step": 33140
    },
    {
      "epoch": 1.7670104741344845,
      "grad_norm": 1.793770422364409,
      "learning_rate": 1.1848054847287258e-06,
      "loss": 0.2698,
      "step": 33150
    },
    {
      "epoch": 1.767543508968311,
      "grad_norm": 1.7129722386046085,
      "learning_rate": 1.1839438368111272e-06,
      "loss": 0.278,
      "step": 33160
    },
    {
      "epoch": 1.7680765438021375,
      "grad_norm": 1.803384510958122,
      "learning_rate": 1.1830822980145972e-06,
      "loss": 0.2928,
      "step": 33170
    },
    {
      "epoch": 1.768609578635964,
      "grad_norm": 1.7457379163257314,
      "learning_rate": 1.1822208686365884e-06,
      "loss": 0.2894,
      "step": 33180
    },
    {
      "epoch": 1.7691426134697903,
      "grad_norm": 1.7378909312847994,
      "learning_rate": 1.1813595489745174e-06,
      "loss": 0.2818,
      "step": 33190
    },
    {
      "epoch": 1.7696756483036167,
      "grad_norm": 1.742084780582119,
      "learning_rate": 1.1804983393257616e-06,
      "loss": 0.2819,
      "step": 33200
    },
    {
      "epoch": 1.770208683137443,
      "grad_norm": 1.8024718189393962,
      "learning_rate": 1.1796372399876616e-06,
      "loss": 0.2881,
      "step": 33210
    },
    {
      "epoch": 1.7707417179712694,
      "grad_norm": 1.5911692022405042,
      "learning_rate": 1.1787762512575181e-06,
      "loss": 0.2916,
      "step": 33220
    },
    {
      "epoch": 1.7712747528050958,
      "grad_norm": 1.8971458335615976,
      "learning_rate": 1.1779153734325958e-06,
      "loss": 0.2769,
      "step": 33230
    },
    {
      "epoch": 1.7718077876389222,
      "grad_norm": 1.928163282189597,
      "learning_rate": 1.1770546068101195e-06,
      "loss": 0.2938,
      "step": 33240
    },
    {
      "epoch": 1.7723408224727486,
      "grad_norm": 1.7823246537926254,
      "learning_rate": 1.1761939516872764e-06,
      "loss": 0.2814,
      "step": 33250
    },
    {
      "epoch": 1.772873857306575,
      "grad_norm": 1.8134904587564507,
      "learning_rate": 1.1753334083612154e-06,
      "loss": 0.2835,
      "step": 33260
    },
    {
      "epoch": 1.7734068921404014,
      "grad_norm": 1.6642545826318993,
      "learning_rate": 1.1744729771290451e-06,
      "loss": 0.2826,
      "step": 33270
    },
    {
      "epoch": 1.7739399269742278,
      "grad_norm": 1.656731292080873,
      "learning_rate": 1.173612658287838e-06,
      "loss": 0.284,
      "step": 33280
    },
    {
      "epoch": 1.7744729618080541,
      "grad_norm": 1.6587836047364595,
      "learning_rate": 1.1727524521346256e-06,
      "loss": 0.2855,
      "step": 33290
    },
    {
      "epoch": 1.7750059966418805,
      "grad_norm": 1.6321025009912309,
      "learning_rate": 1.171892358966402e-06,
      "loss": 0.2818,
      "step": 33300
    },
    {
      "epoch": 1.775539031475707,
      "grad_norm": 1.800968689436437,
      "learning_rate": 1.1710323790801214e-06,
      "loss": 0.2844,
      "step": 33310
    },
    {
      "epoch": 1.7760720663095333,
      "grad_norm": 1.6390910368016185,
      "learning_rate": 1.1701725127726987e-06,
      "loss": 0.2857,
      "step": 33320
    },
    {
      "epoch": 1.7766051011433597,
      "grad_norm": 1.6677627476701735,
      "learning_rate": 1.1693127603410107e-06,
      "loss": 0.2793,
      "step": 33330
    },
    {
      "epoch": 1.777138135977186,
      "grad_norm": 1.577193594826864,
      "learning_rate": 1.1684531220818942e-06,
      "loss": 0.2936,
      "step": 33340
    },
    {
      "epoch": 1.7776711708110124,
      "grad_norm": 1.6814733061446683,
      "learning_rate": 1.167593598292146e-06,
      "loss": 0.2781,
      "step": 33350
    },
    {
      "epoch": 1.778204205644839,
      "grad_norm": 1.8138118523789826,
      "learning_rate": 1.1667341892685243e-06,
      "loss": 0.266,
      "step": 33360
    },
    {
      "epoch": 1.7787372404786654,
      "grad_norm": 1.755745646007814,
      "learning_rate": 1.1658748953077474e-06,
      "loss": 0.2897,
      "step": 33370
    },
    {
      "epoch": 1.7792702753124918,
      "grad_norm": 1.7647426318655384,
      "learning_rate": 1.1650157167064938e-06,
      "loss": 0.2828,
      "step": 33380
    },
    {
      "epoch": 1.7798033101463182,
      "grad_norm": 1.7501861552037314,
      "learning_rate": 1.1641566537614021e-06,
      "loss": 0.2822,
      "step": 33390
    },
    {
      "epoch": 1.7803363449801446,
      "grad_norm": 1.6381073635539256,
      "learning_rate": 1.163297706769071e-06,
      "loss": 0.2924,
      "step": 33400
    },
    {
      "epoch": 1.780869379813971,
      "grad_norm": 1.6528912100178759,
      "learning_rate": 1.1624388760260597e-06,
      "loss": 0.2879,
      "step": 33410
    },
    {
      "epoch": 1.7814024146477974,
      "grad_norm": 1.9693691431225955,
      "learning_rate": 1.161580161828886e-06,
      "loss": 0.2907,
      "step": 33420
    },
    {
      "epoch": 1.7819354494816237,
      "grad_norm": 1.6775167126991481,
      "learning_rate": 1.1607215644740293e-06,
      "loss": 0.2912,
      "step": 33430
    },
    {
      "epoch": 1.7824684843154501,
      "grad_norm": 1.8501231068500794,
      "learning_rate": 1.1598630842579266e-06,
      "loss": 0.2862,
      "step": 33440
    },
    {
      "epoch": 1.7830015191492765,
      "grad_norm": 1.7929903654136852,
      "learning_rate": 1.1590047214769758e-06,
      "loss": 0.2715,
      "step": 33450
    },
    {
      "epoch": 1.783534553983103,
      "grad_norm": 1.9417642928228735,
      "learning_rate": 1.1581464764275342e-06,
      "loss": 0.2804,
      "step": 33460
    },
    {
      "epoch": 1.7840675888169293,
      "grad_norm": 1.887765723109769,
      "learning_rate": 1.157288349405918e-06,
      "loss": 0.2811,
      "step": 33470
    },
    {
      "epoch": 1.7846006236507557,
      "grad_norm": 1.8122568285710194,
      "learning_rate": 1.1564303407084027e-06,
      "loss": 0.2861,
      "step": 33480
    },
    {
      "epoch": 1.785133658484582,
      "grad_norm": 1.753965790461032,
      "learning_rate": 1.1555724506312232e-06,
      "loss": 0.289,
      "step": 33490
    },
    {
      "epoch": 1.7856666933184084,
      "grad_norm": 1.8806664581253916,
      "learning_rate": 1.1547146794705729e-06,
      "loss": 0.2937,
      "step": 33500
    },
    {
      "epoch": 1.7861997281522348,
      "grad_norm": 1.7289241688389556,
      "learning_rate": 1.1538570275226055e-06,
      "loss": 0.2911,
      "step": 33510
    },
    {
      "epoch": 1.7867327629860612,
      "grad_norm": 1.8719795433769644,
      "learning_rate": 1.1529994950834317e-06,
      "loss": 0.2948,
      "step": 33520
    },
    {
      "epoch": 1.7872657978198876,
      "grad_norm": 1.645858881282186,
      "learning_rate": 1.1521420824491222e-06,
      "loss": 0.289,
      "step": 33530
    },
    {
      "epoch": 1.787798832653714,
      "grad_norm": 1.805836445270633,
      "learning_rate": 1.151284789915706e-06,
      "loss": 0.2873,
      "step": 33540
    },
    {
      "epoch": 1.7883318674875404,
      "grad_norm": 1.498351679995341,
      "learning_rate": 1.150427617779171e-06,
      "loss": 0.281,
      "step": 33550
    },
    {
      "epoch": 1.7888649023213667,
      "grad_norm": 1.8921876971501468,
      "learning_rate": 1.1495705663354622e-06,
      "loss": 0.293,
      "step": 33560
    },
    {
      "epoch": 1.7893979371551931,
      "grad_norm": 1.7940753021292293,
      "learning_rate": 1.1487136358804845e-06,
      "loss": 0.2942,
      "step": 33570
    },
    {
      "epoch": 1.7899309719890195,
      "grad_norm": 1.6198606871866887,
      "learning_rate": 1.1478568267101003e-06,
      "loss": 0.2802,
      "step": 33580
    },
    {
      "epoch": 1.790464006822846,
      "grad_norm": 1.8640272317564233,
      "learning_rate": 1.1470001391201304e-06,
      "loss": 0.2839,
      "step": 33590
    },
    {
      "epoch": 1.7909970416566723,
      "grad_norm": 1.6737504891714863,
      "learning_rate": 1.146143573406353e-06,
      "loss": 0.2833,
      "step": 33600
    },
    {
      "epoch": 1.7915300764904987,
      "grad_norm": 1.8129577594218758,
      "learning_rate": 1.145287129864505e-06,
      "loss": 0.2781,
      "step": 33610
    },
    {
      "epoch": 1.792063111324325,
      "grad_norm": 1.7690860864153717,
      "learning_rate": 1.1444308087902804e-06,
      "loss": 0.2854,
      "step": 33620
    },
    {
      "epoch": 1.7925961461581514,
      "grad_norm": 1.7611740146371033,
      "learning_rate": 1.143574610479332e-06,
      "loss": 0.2856,
      "step": 33630
    },
    {
      "epoch": 1.7931291809919778,
      "grad_norm": 1.5250532183756842,
      "learning_rate": 1.1427185352272694e-06,
      "loss": 0.2821,
      "step": 33640
    },
    {
      "epoch": 1.7936622158258042,
      "grad_norm": 1.8295559105368613,
      "learning_rate": 1.1418625833296592e-06,
      "loss": 0.2834,
      "step": 33650
    },
    {
      "epoch": 1.7941952506596306,
      "grad_norm": 1.737860592041329,
      "learning_rate": 1.1410067550820265e-06,
      "loss": 0.2772,
      "step": 33660
    },
    {
      "epoch": 1.794728285493457,
      "grad_norm": 1.910099360063565,
      "learning_rate": 1.140151050779853e-06,
      "loss": 0.2859,
      "step": 33670
    },
    {
      "epoch": 1.7952613203272834,
      "grad_norm": 1.6248629726418629,
      "learning_rate": 1.1392954707185786e-06,
      "loss": 0.2844,
      "step": 33680
    },
    {
      "epoch": 1.7957943551611097,
      "grad_norm": 1.700431966603427,
      "learning_rate": 1.1384400151935985e-06,
      "loss": 0.2804,
      "step": 33690
    },
    {
      "epoch": 1.7963273899949361,
      "grad_norm": 1.688983542272381,
      "learning_rate": 1.1375846845002666e-06,
      "loss": 0.2926,
      "step": 33700
    },
    {
      "epoch": 1.7968604248287625,
      "grad_norm": 1.753162322925547,
      "learning_rate": 1.1367294789338928e-06,
      "loss": 0.283,
      "step": 33710
    },
    {
      "epoch": 1.797393459662589,
      "grad_norm": 1.7652441781406532,
      "learning_rate": 1.1358743987897448e-06,
      "loss": 0.2873,
      "step": 33720
    },
    {
      "epoch": 1.7979264944964153,
      "grad_norm": 1.8861411903139922,
      "learning_rate": 1.135019444363045e-06,
      "loss": 0.2776,
      "step": 33730
    },
    {
      "epoch": 1.7984595293302417,
      "grad_norm": 1.690489967507977,
      "learning_rate": 1.1341646159489744e-06,
      "loss": 0.2832,
      "step": 33740
    },
    {
      "epoch": 1.798992564164068,
      "grad_norm": 1.6167954041258796,
      "learning_rate": 1.13330991384267e-06,
      "loss": 0.2778,
      "step": 33750
    },
    {
      "epoch": 1.7995255989978944,
      "grad_norm": 1.756470122286585,
      "learning_rate": 1.1324553383392246e-06,
      "loss": 0.2917,
      "step": 33760
    },
    {
      "epoch": 1.8000586338317208,
      "grad_norm": 1.6593793448073022,
      "learning_rate": 1.1316008897336873e-06,
      "loss": 0.2832,
      "step": 33770
    },
    {
      "epoch": 1.8005916686655472,
      "grad_norm": 1.7523875609619814,
      "learning_rate": 1.1307465683210644e-06,
      "loss": 0.2753,
      "step": 33780
    },
    {
      "epoch": 1.8011247034993736,
      "grad_norm": 1.8807585832714437,
      "learning_rate": 1.1298923743963172e-06,
      "loss": 0.2763,
      "step": 33790
    },
    {
      "epoch": 1.8016577383332,
      "grad_norm": 2.133302933958573,
      "learning_rate": 1.1290383082543633e-06,
      "loss": 0.2954,
      "step": 33800
    },
    {
      "epoch": 1.8021907731670264,
      "grad_norm": 1.8325941685987879,
      "learning_rate": 1.1281843701900768e-06,
      "loss": 0.2897,
      "step": 33810
    },
    {
      "epoch": 1.8027238080008527,
      "grad_norm": 1.7055468079395337,
      "learning_rate": 1.1273305604982865e-06,
      "loss": 0.287,
      "step": 33820
    },
    {
      "epoch": 1.8032568428346791,
      "grad_norm": 1.982174723381969,
      "learning_rate": 1.1264768794737777e-06,
      "loss": 0.291,
      "step": 33830
    },
    {
      "epoch": 1.8037898776685055,
      "grad_norm": 1.6439751284410773,
      "learning_rate": 1.1256233274112907e-06,
      "loss": 0.2845,
      "step": 33840
    },
    {
      "epoch": 1.804322912502332,
      "grad_norm": 1.830414371209443,
      "learning_rate": 1.1247699046055222e-06,
      "loss": 0.285,
      "step": 33850
    },
    {
      "epoch": 1.8048559473361583,
      "grad_norm": 1.7439309798165994,
      "learning_rate": 1.123916611351123e-06,
      "loss": 0.2757,
      "step": 33860
    },
    {
      "epoch": 1.8053889821699847,
      "grad_norm": 1.712434726007058,
      "learning_rate": 1.1230634479427004e-06,
      "loss": 0.2798,
      "step": 33870
    },
    {
      "epoch": 1.805922017003811,
      "grad_norm": 1.7080465587597295,
      "learning_rate": 1.1222104146748156e-06,
      "loss": 0.2914,
      "step": 33880
    },
    {
      "epoch": 1.8064550518376374,
      "grad_norm": 1.7349800287185455,
      "learning_rate": 1.1213575118419866e-06,
      "loss": 0.2941,
      "step": 33890
    },
    {
      "epoch": 1.8069880866714638,
      "grad_norm": 1.788350132423697,
      "learning_rate": 1.1205047397386841e-06,
      "loss": 0.2821,
      "step": 33900
    },
    {
      "epoch": 1.8075211215052902,
      "grad_norm": 1.672050369300618,
      "learning_rate": 1.1196520986593359e-06,
      "loss": 0.2848,
      "step": 33910
    },
    {
      "epoch": 1.8080541563391168,
      "grad_norm": 1.8399765950580336,
      "learning_rate": 1.1187995888983227e-06,
      "loss": 0.2818,
      "step": 33920
    },
    {
      "epoch": 1.8085871911729432,
      "grad_norm": 1.8086788054055678,
      "learning_rate": 1.1179472107499813e-06,
      "loss": 0.2932,
      "step": 33930
    },
    {
      "epoch": 1.8091202260067696,
      "grad_norm": 1.8289005131436555,
      "learning_rate": 1.1170949645086022e-06,
      "loss": 0.2847,
      "step": 33940
    },
    {
      "epoch": 1.809653260840596,
      "grad_norm": 1.5490717063639103,
      "learning_rate": 1.1162428504684313e-06,
      "loss": 0.2889,
      "step": 33950
    },
    {
      "epoch": 1.8101862956744224,
      "grad_norm": 1.7283938741145624,
      "learning_rate": 1.115390868923667e-06,
      "loss": 0.2943,
      "step": 33960
    },
    {
      "epoch": 1.8107193305082487,
      "grad_norm": 1.8194921712267031,
      "learning_rate": 1.114539020168464e-06,
      "loss": 0.277,
      "step": 33970
    },
    {
      "epoch": 1.8112523653420751,
      "grad_norm": 1.639176107193273,
      "learning_rate": 1.1136873044969302e-06,
      "loss": 0.2804,
      "step": 33980
    },
    {
      "epoch": 1.8117854001759015,
      "grad_norm": 1.7458436628548724,
      "learning_rate": 1.1128357222031276e-06,
      "loss": 0.2746,
      "step": 33990
    },
    {
      "epoch": 1.812318435009728,
      "grad_norm": 1.8650303802442367,
      "learning_rate": 1.111984273581072e-06,
      "loss": 0.2818,
      "step": 34000
    },
    {
      "epoch": 1.8128514698435543,
      "grad_norm": 1.655443839449947,
      "learning_rate": 1.1111329589247336e-06,
      "loss": 0.2763,
      "step": 34010
    },
    {
      "epoch": 1.8133845046773807,
      "grad_norm": 1.8102412312389937,
      "learning_rate": 1.1102817785280354e-06,
      "loss": 0.2889,
      "step": 34020
    },
    {
      "epoch": 1.813917539511207,
      "grad_norm": 1.6134138131586813,
      "learning_rate": 1.1094307326848554e-06,
      "loss": 0.2886,
      "step": 34030
    },
    {
      "epoch": 1.8144505743450334,
      "grad_norm": 1.7450590474867633,
      "learning_rate": 1.108579821689024e-06,
      "loss": 0.2822,
      "step": 34040
    },
    {
      "epoch": 1.8149836091788598,
      "grad_norm": 1.5897483201065454,
      "learning_rate": 1.1077290458343256e-06,
      "loss": 0.2761,
      "step": 34050
    },
    {
      "epoch": 1.8155166440126862,
      "grad_norm": 1.883617289604175,
      "learning_rate": 1.1068784054144973e-06,
      "loss": 0.2829,
      "step": 34060
    },
    {
      "epoch": 1.8160496788465126,
      "grad_norm": 1.7311891204055823,
      "learning_rate": 1.1060279007232305e-06,
      "loss": 0.2821,
      "step": 34070
    },
    {
      "epoch": 1.816582713680339,
      "grad_norm": 1.828293791159772,
      "learning_rate": 1.105177532054169e-06,
      "loss": 0.2871,
      "step": 34080
    },
    {
      "epoch": 1.8171157485141654,
      "grad_norm": 1.7575674008176536,
      "learning_rate": 1.1043272997009096e-06,
      "loss": 0.2849,
      "step": 34090
    },
    {
      "epoch": 1.8176487833479917,
      "grad_norm": 1.9130918964340493,
      "learning_rate": 1.103477203957002e-06,
      "loss": 0.2975,
      "step": 34100
    },
    {
      "epoch": 1.8181818181818183,
      "grad_norm": 1.5610682780039224,
      "learning_rate": 1.1026272451159496e-06,
      "loss": 0.288,
      "step": 34110
    },
    {
      "epoch": 1.8187148530156447,
      "grad_norm": 1.8156298165923574,
      "learning_rate": 1.1017774234712077e-06,
      "loss": 0.2685,
      "step": 34120
    },
    {
      "epoch": 1.8192478878494711,
      "grad_norm": 1.8981220587794834,
      "learning_rate": 1.1009277393161841e-06,
      "loss": 0.276,
      "step": 34130
    },
    {
      "epoch": 1.8197809226832975,
      "grad_norm": 1.7014268583428251,
      "learning_rate": 1.1000781929442394e-06,
      "loss": 0.2872,
      "step": 34140
    },
    {
      "epoch": 1.8203139575171239,
      "grad_norm": 1.6707214951958549,
      "learning_rate": 1.099228784648687e-06,
      "loss": 0.2851,
      "step": 34150
    },
    {
      "epoch": 1.8208469923509503,
      "grad_norm": 1.8265949184513688,
      "learning_rate": 1.0983795147227926e-06,
      "loss": 0.2912,
      "step": 34160
    },
    {
      "epoch": 1.8213800271847767,
      "grad_norm": 1.8979884812981793,
      "learning_rate": 1.0975303834597734e-06,
      "loss": 0.2817,
      "step": 34170
    },
    {
      "epoch": 1.821913062018603,
      "grad_norm": 1.713077673795763,
      "learning_rate": 1.096681391152799e-06,
      "loss": 0.2878,
      "step": 34180
    },
    {
      "epoch": 1.8224460968524294,
      "grad_norm": 1.9329467397832927,
      "learning_rate": 1.0958325380949914e-06,
      "loss": 0.2773,
      "step": 34190
    },
    {
      "epoch": 1.8229791316862558,
      "grad_norm": 1.751671910749437,
      "learning_rate": 1.0949838245794244e-06,
      "loss": 0.2778,
      "step": 34200
    },
    {
      "epoch": 1.8235121665200822,
      "grad_norm": 1.8642896118551555,
      "learning_rate": 1.0941352508991238e-06,
      "loss": 0.2868,
      "step": 34210
    },
    {
      "epoch": 1.8240452013539086,
      "grad_norm": 1.784714945417271,
      "learning_rate": 1.0932868173470662e-06,
      "loss": 0.289,
      "step": 34220
    },
    {
      "epoch": 1.824578236187735,
      "grad_norm": 1.9068453575818782,
      "learning_rate": 1.0924385242161805e-06,
      "loss": 0.2803,
      "step": 34230
    },
    {
      "epoch": 1.8251112710215613,
      "grad_norm": 1.7116479484949956,
      "learning_rate": 1.0915903717993477e-06,
      "loss": 0.274,
      "step": 34240
    },
    {
      "epoch": 1.8256443058553877,
      "grad_norm": 1.748251781989164,
      "learning_rate": 1.0907423603893996e-06,
      "loss": 0.2838,
      "step": 34250
    },
    {
      "epoch": 1.8261773406892141,
      "grad_norm": 1.7088192878892445,
      "learning_rate": 1.0898944902791186e-06,
      "loss": 0.2911,
      "step": 34260
    },
    {
      "epoch": 1.8267103755230405,
      "grad_norm": 1.6764555278790207,
      "learning_rate": 1.0890467617612395e-06,
      "loss": 0.2838,
      "step": 34270
    },
    {
      "epoch": 1.8272434103568669,
      "grad_norm": 1.7424091503380401,
      "learning_rate": 1.0881991751284478e-06,
      "loss": 0.2844,
      "step": 34280
    },
    {
      "epoch": 1.8277764451906933,
      "grad_norm": 1.7797215963768982,
      "learning_rate": 1.0873517306733797e-06,
      "loss": 0.2878,
      "step": 34290
    },
    {
      "epoch": 1.8283094800245197,
      "grad_norm": 1.5453777935549813,
      "learning_rate": 1.086504428688623e-06,
      "loss": 0.28,
      "step": 34300
    },
    {
      "epoch": 1.828842514858346,
      "grad_norm": 1.9896540890870467,
      "learning_rate": 1.0856572694667153e-06,
      "loss": 0.2876,
      "step": 34310
    },
    {
      "epoch": 1.8293755496921724,
      "grad_norm": 1.7179419355397545,
      "learning_rate": 1.0848102533001461e-06,
      "loss": 0.2935,
      "step": 34320
    },
    {
      "epoch": 1.8299085845259988,
      "grad_norm": 2.067669891429877,
      "learning_rate": 1.0839633804813547e-06,
      "loss": 0.2888,
      "step": 34330
    },
    {
      "epoch": 1.8304416193598252,
      "grad_norm": 2.226691334150769,
      "learning_rate": 1.083116651302731e-06,
      "loss": 0.2868,
      "step": 34340
    },
    {
      "epoch": 1.8309746541936516,
      "grad_norm": 1.8149770027364398,
      "learning_rate": 1.0822700660566153e-06,
      "loss": 0.2796,
      "step": 34350
    },
    {
      "epoch": 1.831507689027478,
      "grad_norm": 1.704603511554391,
      "learning_rate": 1.0814236250352986e-06,
      "loss": 0.2797,
      "step": 34360
    },
    {
      "epoch": 1.8320407238613043,
      "grad_norm": 1.6429964179576118,
      "learning_rate": 1.0805773285310217e-06,
      "loss": 0.2861,
      "step": 34370
    },
    {
      "epoch": 1.8325737586951307,
      "grad_norm": 1.7717173469199823,
      "learning_rate": 1.0797311768359758e-06,
      "loss": 0.2814,
      "step": 34380
    },
    {
      "epoch": 1.8331067935289571,
      "grad_norm": 1.8919850655287325,
      "learning_rate": 1.078885170242302e-06,
      "loss": 0.2811,
      "step": 34390
    },
    {
      "epoch": 1.8336398283627835,
      "grad_norm": 1.9150816559010473,
      "learning_rate": 1.0780393090420906e-06,
      "loss": 0.2742,
      "step": 34400
    },
    {
      "epoch": 1.8341728631966099,
      "grad_norm": 1.753970404084367,
      "learning_rate": 1.077193593527383e-06,
      "loss": 0.2802,
      "step": 34410
    },
    {
      "epoch": 1.8347058980304363,
      "grad_norm": 1.6730326269014233,
      "learning_rate": 1.0763480239901699e-06,
      "loss": 0.2765,
      "step": 34420
    },
    {
      "epoch": 1.8352389328642627,
      "grad_norm": 1.7477151992531457,
      "learning_rate": 1.0755026007223904e-06,
      "loss": 0.2766,
      "step": 34430
    },
    {
      "epoch": 1.835771967698089,
      "grad_norm": 1.6148231936669728,
      "learning_rate": 1.0746573240159346e-06,
      "loss": 0.2807,
      "step": 34440
    },
    {
      "epoch": 1.8363050025319154,
      "grad_norm": 1.5447582855547874,
      "learning_rate": 1.0738121941626415e-06,
      "loss": 0.2898,
      "step": 34450
    },
    {
      "epoch": 1.8368380373657418,
      "grad_norm": 1.7959913831018646,
      "learning_rate": 1.0729672114542995e-06,
      "loss": 0.2838,
      "step": 34460
    },
    {
      "epoch": 1.8373710721995682,
      "grad_norm": 1.8160129417948139,
      "learning_rate": 1.0721223761826454e-06,
      "loss": 0.289,
      "step": 34470
    },
    {
      "epoch": 1.8379041070333946,
      "grad_norm": 1.710630021315159,
      "learning_rate": 1.0712776886393663e-06,
      "loss": 0.285,
      "step": 34480
    },
    {
      "epoch": 1.838437141867221,
      "grad_norm": 1.6643078191367287,
      "learning_rate": 1.0704331491160973e-06,
      "loss": 0.2842,
      "step": 34490
    },
    {
      "epoch": 1.8389701767010473,
      "grad_norm": 1.6908870434165215,
      "learning_rate": 1.0695887579044235e-06,
      "loss": 0.2787,
      "step": 34500
    },
    {
      "epoch": 1.8395032115348737,
      "grad_norm": 1.6267838092012359,
      "learning_rate": 1.0687445152958773e-06,
      "loss": 0.2777,
      "step": 34510
    },
    {
      "epoch": 1.8400362463687001,
      "grad_norm": 1.7506964768494824,
      "learning_rate": 1.0679004215819412e-06,
      "loss": 0.2964,
      "step": 34520
    },
    {
      "epoch": 1.8405692812025265,
      "grad_norm": 1.6474621816685022,
      "learning_rate": 1.0670564770540455e-06,
      "loss": 0.2707,
      "step": 34530
    },
    {
      "epoch": 1.8411023160363529,
      "grad_norm": 1.8086883973577987,
      "learning_rate": 1.0662126820035693e-06,
      "loss": 0.2758,
      "step": 34540
    },
    {
      "epoch": 1.8416353508701793,
      "grad_norm": 1.7431654890755537,
      "learning_rate": 1.06536903672184e-06,
      "loss": 0.2677,
      "step": 34550
    },
    {
      "epoch": 1.8421683857040057,
      "grad_norm": 1.6709738595213355,
      "learning_rate": 1.0645255415001327e-06,
      "loss": 0.2759,
      "step": 34560
    },
    {
      "epoch": 1.842701420537832,
      "grad_norm": 1.6165678326395383,
      "learning_rate": 1.0636821966296719e-06,
      "loss": 0.2807,
      "step": 34570
    },
    {
      "epoch": 1.8432344553716584,
      "grad_norm": 1.7028473422401444,
      "learning_rate": 1.0628390024016296e-06,
      "loss": 0.2833,
      "step": 34580
    },
    {
      "epoch": 1.8437674902054848,
      "grad_norm": 1.7405113975136388,
      "learning_rate": 1.0619959591071258e-06,
      "loss": 0.2833,
      "step": 34590
    },
    {
      "epoch": 1.8443005250393112,
      "grad_norm": 1.6600466254275668,
      "learning_rate": 1.0611530670372275e-06,
      "loss": 0.2871,
      "step": 34600
    },
    {
      "epoch": 1.8448335598731376,
      "grad_norm": 1.7079471415002405,
      "learning_rate": 1.0603103264829513e-06,
      "loss": 0.2863,
      "step": 34610
    },
    {
      "epoch": 1.845366594706964,
      "grad_norm": 1.6562824961931186,
      "learning_rate": 1.0594677377352601e-06,
      "loss": 0.2756,
      "step": 34620
    },
    {
      "epoch": 1.8458996295407903,
      "grad_norm": 1.7123599775145584,
      "learning_rate": 1.058625301085065e-06,
      "loss": 0.2892,
      "step": 34630
    },
    {
      "epoch": 1.8464326643746167,
      "grad_norm": 1.7734656830606317,
      "learning_rate": 1.057783016823224e-06,
      "loss": 0.2838,
      "step": 34640
    },
    {
      "epoch": 1.8469656992084431,
      "grad_norm": 1.695548261231936,
      "learning_rate": 1.0569408852405433e-06,
      "loss": 0.2957,
      "step": 34650
    },
    {
      "epoch": 1.8474987340422695,
      "grad_norm": 1.7494164440516087,
      "learning_rate": 1.0560989066277755e-06,
      "loss": 0.2774,
      "step": 34660
    },
    {
      "epoch": 1.848031768876096,
      "grad_norm": 1.7330502578030142,
      "learning_rate": 1.0552570812756215e-06,
      "loss": 0.2857,
      "step": 34670
    },
    {
      "epoch": 1.8485648037099225,
      "grad_norm": 1.6587549978959704,
      "learning_rate": 1.0544154094747284e-06,
      "loss": 0.2773,
      "step": 34680
    },
    {
      "epoch": 1.8490978385437489,
      "grad_norm": 1.849411392429873,
      "learning_rate": 1.05357389151569e-06,
      "loss": 0.2823,
      "step": 34690
    },
    {
      "epoch": 1.8496308733775753,
      "grad_norm": 1.8212631169038112,
      "learning_rate": 1.052732527689048e-06,
      "loss": 0.2868,
      "step": 34700
    },
    {
      "epoch": 1.8501639082114016,
      "grad_norm": 1.7483996194095333,
      "learning_rate": 1.0518913182852905e-06,
      "loss": 0.2808,
      "step": 34710
    },
    {
      "epoch": 1.850696943045228,
      "grad_norm": 1.7317317091779638,
      "learning_rate": 1.0510502635948518e-06,
      "loss": 0.2935,
      "step": 34720
    },
    {
      "epoch": 1.8512299778790544,
      "grad_norm": 1.8719800453928523,
      "learning_rate": 1.0502093639081132e-06,
      "loss": 0.2918,
      "step": 34730
    },
    {
      "epoch": 1.8517630127128808,
      "grad_norm": 1.7022234984337459,
      "learning_rate": 1.0493686195154024e-06,
      "loss": 0.2842,
      "step": 34740
    },
    {
      "epoch": 1.8522960475467072,
      "grad_norm": 1.6309342788612091,
      "learning_rate": 1.048528030706994e-06,
      "loss": 0.2837,
      "step": 34750
    },
    {
      "epoch": 1.8528290823805336,
      "grad_norm": 1.8445263935654417,
      "learning_rate": 1.0476875977731083e-06,
      "loss": 0.2789,
      "step": 34760
    },
    {
      "epoch": 1.85336211721436,
      "grad_norm": 1.8379393888632174,
      "learning_rate": 1.0468473210039113e-06,
      "loss": 0.2778,
      "step": 34770
    },
    {
      "epoch": 1.8538951520481863,
      "grad_norm": 1.8667607983656693,
      "learning_rate": 1.0460072006895163e-06,
      "loss": 0.2933,
      "step": 34780
    },
    {
      "epoch": 1.8544281868820127,
      "grad_norm": 1.8299120005426075,
      "learning_rate": 1.0451672371199816e-06,
      "loss": 0.2867,
      "step": 34790
    },
    {
      "epoch": 1.854961221715839,
      "grad_norm": 1.8132107372428896,
      "learning_rate": 1.0443274305853121e-06,
      "loss": 0.2787,
      "step": 34800
    },
    {
      "epoch": 1.8554942565496655,
      "grad_norm": 1.75534628956572,
      "learning_rate": 1.0434877813754576e-06,
      "loss": 0.2881,
      "step": 34810
    },
    {
      "epoch": 1.8560272913834919,
      "grad_norm": 1.6882702107320098,
      "learning_rate": 1.0426482897803142e-06,
      "loss": 0.278,
      "step": 34820
    },
    {
      "epoch": 1.8565603262173183,
      "grad_norm": 1.9191956479523045,
      "learning_rate": 1.0418089560897239e-06,
      "loss": 0.2926,
      "step": 34830
    },
    {
      "epoch": 1.8570933610511446,
      "grad_norm": 1.971987032346656,
      "learning_rate": 1.0409697805934738e-06,
      "loss": 0.2784,
      "step": 34840
    },
    {
      "epoch": 1.8576263958849712,
      "grad_norm": 1.7343757430387683,
      "learning_rate": 1.0401307635812957e-06,
      "loss": 0.2871,
      "step": 34850
    },
    {
      "epoch": 1.8581594307187976,
      "grad_norm": 1.9720783163694182,
      "learning_rate": 1.0392919053428676e-06,
      "loss": 0.2743,
      "step": 34860
    },
    {
      "epoch": 1.858692465552624,
      "grad_norm": 1.7974747096027834,
      "learning_rate": 1.0384532061678129e-06,
      "loss": 0.2763,
      "step": 34870
    },
    {
      "epoch": 1.8592255003864504,
      "grad_norm": 1.5831927083400277,
      "learning_rate": 1.0376146663456992e-06,
      "loss": 0.2832,
      "step": 34880
    },
    {
      "epoch": 1.8597585352202768,
      "grad_norm": 1.9479818234794424,
      "learning_rate": 1.0367762861660393e-06,
      "loss": 0.2881,
      "step": 34890
    },
    {
      "epoch": 1.8602915700541032,
      "grad_norm": 1.6627760289046543,
      "learning_rate": 1.0359380659182913e-06,
      "loss": 0.2744,
      "step": 34900
    },
    {
      "epoch": 1.8608246048879296,
      "grad_norm": 1.8344299218133981,
      "learning_rate": 1.0351000058918579e-06,
      "loss": 0.287,
      "step": 34910
    },
    {
      "epoch": 1.861357639721756,
      "grad_norm": 1.5524926230550007,
      "learning_rate": 1.0342621063760864e-06,
      "loss": 0.2938,
      "step": 34920
    },
    {
      "epoch": 1.8618906745555823,
      "grad_norm": 1.6587993644360186,
      "learning_rate": 1.0334243676602687e-06,
      "loss": 0.2831,
      "step": 34930
    },
    {
      "epoch": 1.8624237093894087,
      "grad_norm": 1.6894005460719788,
      "learning_rate": 1.0325867900336408e-06,
      "loss": 0.2974,
      "step": 34940
    },
    {
      "epoch": 1.862956744223235,
      "grad_norm": 1.8227726132270095,
      "learning_rate": 1.0317493737853838e-06,
      "loss": 0.2779,
      "step": 34950
    },
    {
      "epoch": 1.8634897790570615,
      "grad_norm": 1.7289818601710043,
      "learning_rate": 1.030912119204623e-06,
      "loss": 0.2804,
      "step": 34960
    },
    {
      "epoch": 1.8640228138908879,
      "grad_norm": 1.7836242972182799,
      "learning_rate": 1.0300750265804276e-06,
      "loss": 0.2772,
      "step": 34970
    },
    {
      "epoch": 1.8645558487247142,
      "grad_norm": 1.682765619651718,
      "learning_rate": 1.0292380962018102e-06,
      "loss": 0.2807,
      "step": 34980
    },
    {
      "epoch": 1.8650888835585406,
      "grad_norm": 2.0450127028175333,
      "learning_rate": 1.0284013283577288e-06,
      "loss": 0.2943,
      "step": 34990
    },
    {
      "epoch": 1.865621918392367,
      "grad_norm": 1.730411000942452,
      "learning_rate": 1.0275647233370843e-06,
      "loss": 0.2748,
      "step": 35000
    },
    {
      "epoch": 1.8661549532261934,
      "grad_norm": 1.6277389975419827,
      "learning_rate": 1.0267282814287222e-06,
      "loss": 0.2832,
      "step": 35010
    },
    {
      "epoch": 1.8666879880600198,
      "grad_norm": 1.5334358552233316,
      "learning_rate": 1.0258920029214304e-06,
      "loss": 0.2815,
      "step": 35020
    },
    {
      "epoch": 1.8672210228938462,
      "grad_norm": 1.6268896579282706,
      "learning_rate": 1.0250558881039415e-06,
      "loss": 0.2786,
      "step": 35030
    },
    {
      "epoch": 1.8677540577276726,
      "grad_norm": 1.6822030193127806,
      "learning_rate": 1.024219937264931e-06,
      "loss": 0.2838,
      "step": 35040
    },
    {
      "epoch": 1.868287092561499,
      "grad_norm": 1.72656067068977,
      "learning_rate": 1.0233841506930188e-06,
      "loss": 0.2873,
      "step": 35050
    },
    {
      "epoch": 1.8688201273953253,
      "grad_norm": 1.6681560326786644,
      "learning_rate": 1.0225485286767662e-06,
      "loss": 0.2802,
      "step": 35060
    },
    {
      "epoch": 1.8693531622291517,
      "grad_norm": 1.7187599855179903,
      "learning_rate": 1.0217130715046794e-06,
      "loss": 0.2855,
      "step": 35070
    },
    {
      "epoch": 1.869886197062978,
      "grad_norm": 1.8269575746769546,
      "learning_rate": 1.020877779465207e-06,
      "loss": 0.2882,
      "step": 35080
    },
    {
      "epoch": 1.8704192318968045,
      "grad_norm": 1.6038886577188158,
      "learning_rate": 1.0200426528467408e-06,
      "loss": 0.2841,
      "step": 35090
    },
    {
      "epoch": 1.8709522667306309,
      "grad_norm": 1.816057630472255,
      "learning_rate": 1.0192076919376146e-06,
      "loss": 0.2821,
      "step": 35100
    },
    {
      "epoch": 1.8714853015644572,
      "grad_norm": 1.7406501251151498,
      "learning_rate": 1.0183728970261065e-06,
      "loss": 0.2797,
      "step": 35110
    },
    {
      "epoch": 1.8720183363982836,
      "grad_norm": 1.8775122599692728,
      "learning_rate": 1.0175382684004364e-06,
      "loss": 0.2885,
      "step": 35120
    },
    {
      "epoch": 1.87255137123211,
      "grad_norm": 1.6496897557508083,
      "learning_rate": 1.0167038063487669e-06,
      "loss": 0.2841,
      "step": 35130
    },
    {
      "epoch": 1.8730844060659364,
      "grad_norm": 1.7212205169023977,
      "learning_rate": 1.015869511159203e-06,
      "loss": 0.2877,
      "step": 35140
    },
    {
      "epoch": 1.8736174408997628,
      "grad_norm": 1.7308538260272222,
      "learning_rate": 1.015035383119792e-06,
      "loss": 0.2882,
      "step": 35150
    },
    {
      "epoch": 1.8741504757335892,
      "grad_norm": 1.9290899388149554,
      "learning_rate": 1.0142014225185238e-06,
      "loss": 0.2866,
      "step": 35160
    },
    {
      "epoch": 1.8746835105674156,
      "grad_norm": 1.7682914896377402,
      "learning_rate": 1.01336762964333e-06,
      "loss": 0.2845,
      "step": 35170
    },
    {
      "epoch": 1.875216545401242,
      "grad_norm": 1.7473730247270731,
      "learning_rate": 1.012534004782086e-06,
      "loss": 0.2799,
      "step": 35180
    },
    {
      "epoch": 1.8757495802350683,
      "grad_norm": 1.9216826681234516,
      "learning_rate": 1.0117005482226065e-06,
      "loss": 0.2781,
      "step": 35190
    },
    {
      "epoch": 1.8762826150688947,
      "grad_norm": 1.9470854303714606,
      "learning_rate": 1.0108672602526496e-06,
      "loss": 0.2769,
      "step": 35200
    },
    {
      "epoch": 1.876815649902721,
      "grad_norm": 1.7026592356819465,
      "learning_rate": 1.0100341411599153e-06,
      "loss": 0.2781,
      "step": 35210
    },
    {
      "epoch": 1.8773486847365475,
      "grad_norm": 1.7986917689102309,
      "learning_rate": 1.0092011912320455e-06,
      "loss": 0.2895,
      "step": 35220
    },
    {
      "epoch": 1.8778817195703739,
      "grad_norm": 1.6321121618712109,
      "learning_rate": 1.0083684107566223e-06,
      "loss": 0.2842,
      "step": 35230
    },
    {
      "epoch": 1.8784147544042002,
      "grad_norm": 1.8371304734996443,
      "learning_rate": 1.0075358000211705e-06,
      "loss": 0.2858,
      "step": 35240
    },
    {
      "epoch": 1.8789477892380266,
      "grad_norm": 1.883124366511591,
      "learning_rate": 1.006703359313157e-06,
      "loss": 0.2828,
      "step": 35250
    },
    {
      "epoch": 1.879480824071853,
      "grad_norm": 1.769211253565573,
      "learning_rate": 1.0058710889199874e-06,
      "loss": 0.2846,
      "step": 35260
    },
    {
      "epoch": 1.8800138589056794,
      "grad_norm": 1.8463400326176658,
      "learning_rate": 1.005038989129011e-06,
      "loss": 0.2831,
      "step": 35270
    },
    {
      "epoch": 1.8805468937395058,
      "grad_norm": 1.761129299368819,
      "learning_rate": 1.0042070602275173e-06,
      "loss": 0.2766,
      "step": 35280
    },
    {
      "epoch": 1.8810799285733322,
      "grad_norm": 1.659425918482653,
      "learning_rate": 1.0033753025027365e-06,
      "loss": 0.2895,
      "step": 35290
    },
    {
      "epoch": 1.8816129634071586,
      "grad_norm": 1.6361794089062556,
      "learning_rate": 1.00254371624184e-06,
      "loss": 0.2785,
      "step": 35300
    },
    {
      "epoch": 1.882145998240985,
      "grad_norm": 1.7094848670515799,
      "learning_rate": 1.0017123017319401e-06,
      "loss": 0.2798,
      "step": 35310
    },
    {
      "epoch": 1.8826790330748113,
      "grad_norm": 1.8248155417864,
      "learning_rate": 1.0008810592600904e-06,
      "loss": 0.2794,
      "step": 35320
    },
    {
      "epoch": 1.8832120679086377,
      "grad_norm": 1.736341576073759,
      "learning_rate": 1.0000499891132832e-06,
      "loss": 0.2838,
      "step": 35330
    },
    {
      "epoch": 1.883745102742464,
      "grad_norm": 1.754804169792267,
      "learning_rate": 9.992190915784532e-07,
      "loss": 0.2874,
      "step": 35340
    },
    {
      "epoch": 1.8842781375762905,
      "grad_norm": 1.764348330078149,
      "learning_rate": 9.983883669424747e-07,
      "loss": 0.2816,
      "step": 35350
    },
    {
      "epoch": 1.8848111724101169,
      "grad_norm": 1.8070731732141163,
      "learning_rate": 9.975578154921623e-07,
      "loss": 0.2687,
      "step": 35360
    },
    {
      "epoch": 1.8853442072439432,
      "grad_norm": 1.7172715655997957,
      "learning_rate": 9.967274375142714e-07,
      "loss": 0.2767,
      "step": 35370
    },
    {
      "epoch": 1.8858772420777696,
      "grad_norm": 1.5972635910144992,
      "learning_rate": 9.958972332954966e-07,
      "loss": 0.2806,
      "step": 35380
    },
    {
      "epoch": 1.886410276911596,
      "grad_norm": 1.838781091571979,
      "learning_rate": 9.950672031224729e-07,
      "loss": 0.289,
      "step": 35390
    },
    {
      "epoch": 1.8869433117454224,
      "grad_norm": 1.7094539989863005,
      "learning_rate": 9.942373472817752e-07,
      "loss": 0.2739,
      "step": 35400
    },
    {
      "epoch": 1.887476346579249,
      "grad_norm": 1.6199132374367613,
      "learning_rate": 9.934076660599186e-07,
      "loss": 0.2754,
      "step": 35410
    },
    {
      "epoch": 1.8880093814130754,
      "grad_norm": 1.7708410627416784,
      "learning_rate": 9.925781597433575e-07,
      "loss": 0.2703,
      "step": 35420
    },
    {
      "epoch": 1.8885424162469018,
      "grad_norm": 1.8058806282938,
      "learning_rate": 9.917488286184857e-07,
      "loss": 0.2822,
      "step": 35430
    },
    {
      "epoch": 1.8890754510807282,
      "grad_norm": 1.7438978329239643,
      "learning_rate": 9.90919672971637e-07,
      "loss": 0.2729,
      "step": 35440
    },
    {
      "epoch": 1.8896084859145545,
      "grad_norm": 1.989782730664509,
      "learning_rate": 9.900906930890846e-07,
      "loss": 0.3002,
      "step": 35450
    },
    {
      "epoch": 1.890141520748381,
      "grad_norm": 1.837710642569694,
      "learning_rate": 9.892618892570405e-07,
      "loss": 0.2879,
      "step": 35460
    },
    {
      "epoch": 1.8906745555822073,
      "grad_norm": 1.7368568619897187,
      "learning_rate": 9.884332617616562e-07,
      "loss": 0.2701,
      "step": 35470
    },
    {
      "epoch": 1.8912075904160337,
      "grad_norm": 1.7628832750362922,
      "learning_rate": 9.876048108890227e-07,
      "loss": 0.2759,
      "step": 35480
    },
    {
      "epoch": 1.89174062524986,
      "grad_norm": 1.7675087536218128,
      "learning_rate": 9.867765369251697e-07,
      "loss": 0.2763,
      "step": 35490
    },
    {
      "epoch": 1.8922736600836865,
      "grad_norm": 1.6785749400781123,
      "learning_rate": 9.85948440156065e-07,
      "loss": 0.2763,
      "step": 35500
    },
    {
      "epoch": 1.8928066949175129,
      "grad_norm": 1.9082838220747536,
      "learning_rate": 9.851205208676171e-07,
      "loss": 0.2887,
      "step": 35510
    },
    {
      "epoch": 1.8933397297513392,
      "grad_norm": 1.7062207467478618,
      "learning_rate": 9.842927793456716e-07,
      "loss": 0.2838,
      "step": 35520
    },
    {
      "epoch": 1.8938727645851656,
      "grad_norm": 1.7874699287668812,
      "learning_rate": 9.83465215876013e-07,
      "loss": 0.28,
      "step": 35530
    },
    {
      "epoch": 1.894405799418992,
      "grad_norm": 1.7579368596241673,
      "learning_rate": 9.82637830744366e-07,
      "loss": 0.2881,
      "step": 35540
    },
    {
      "epoch": 1.8949388342528184,
      "grad_norm": 1.7013266187431024,
      "learning_rate": 9.818106242363904e-07,
      "loss": 0.2804,
      "step": 35550
    },
    {
      "epoch": 1.8954718690866448,
      "grad_norm": 1.5813589158444141,
      "learning_rate": 9.80983596637687e-07,
      "loss": 0.2704,
      "step": 35560
    },
    {
      "epoch": 1.8960049039204712,
      "grad_norm": 1.7628655745151582,
      "learning_rate": 9.801567482337948e-07,
      "loss": 0.2858,
      "step": 35570
    },
    {
      "epoch": 1.8965379387542975,
      "grad_norm": 1.7395075223067322,
      "learning_rate": 9.793300793101896e-07,
      "loss": 0.2836,
      "step": 35580
    },
    {
      "epoch": 1.897070973588124,
      "grad_norm": 1.9104675550605692,
      "learning_rate": 9.785035901522857e-07,
      "loss": 0.2833,
      "step": 35590
    },
    {
      "epoch": 1.8976040084219505,
      "grad_norm": 1.6850286592323753,
      "learning_rate": 9.776772810454357e-07,
      "loss": 0.2769,
      "step": 35600
    },
    {
      "epoch": 1.898137043255777,
      "grad_norm": 1.7539249995871227,
      "learning_rate": 9.768511522749295e-07,
      "loss": 0.2661,
      "step": 35610
    },
    {
      "epoch": 1.8986700780896033,
      "grad_norm": 1.7587347179509758,
      "learning_rate": 9.760252041259962e-07,
      "loss": 0.2868,
      "step": 35620
    },
    {
      "epoch": 1.8992031129234297,
      "grad_norm": 1.907491760471663,
      "learning_rate": 9.751994368837998e-07,
      "loss": 0.2707,
      "step": 35630
    },
    {
      "epoch": 1.899736147757256,
      "grad_norm": 1.6173770880120335,
      "learning_rate": 9.743738508334442e-07,
      "loss": 0.281,
      "step": 35640
    },
    {
      "epoch": 1.9002691825910825,
      "grad_norm": 1.6612008469887718,
      "learning_rate": 9.735484462599701e-07,
      "loss": 0.2817,
      "step": 35650
    },
    {
      "epoch": 1.9008022174249088,
      "grad_norm": 1.7500269182005797,
      "learning_rate": 9.727232234483556e-07,
      "loss": 0.284,
      "step": 35660
    },
    {
      "epoch": 1.9013352522587352,
      "grad_norm": 1.7985921034784609,
      "learning_rate": 9.718981826835153e-07,
      "loss": 0.2765,
      "step": 35670
    },
    {
      "epoch": 1.9018682870925616,
      "grad_norm": 1.736399626177598,
      "learning_rate": 9.710733242503017e-07,
      "loss": 0.276,
      "step": 35680
    },
    {
      "epoch": 1.902401321926388,
      "grad_norm": 1.836772750385422,
      "learning_rate": 9.702486484335042e-07,
      "loss": 0.2931,
      "step": 35690
    },
    {
      "epoch": 1.9029343567602144,
      "grad_norm": 2.0091555482897743,
      "learning_rate": 9.69424155517849e-07,
      "loss": 0.2757,
      "step": 35700
    },
    {
      "epoch": 1.9034673915940408,
      "grad_norm": 1.6643677560844068,
      "learning_rate": 9.685998457880001e-07,
      "loss": 0.2842,
      "step": 35710
    },
    {
      "epoch": 1.9040004264278672,
      "grad_norm": 1.9242956813229226,
      "learning_rate": 9.677757195285562e-07,
      "loss": 0.2769,
      "step": 35720
    },
    {
      "epoch": 1.9045334612616935,
      "grad_norm": 1.9400606344595024,
      "learning_rate": 9.669517770240544e-07,
      "loss": 0.2839,
      "step": 35730
    },
    {
      "epoch": 1.90506649609552,
      "grad_norm": 1.7269053299340145,
      "learning_rate": 9.661280185589678e-07,
      "loss": 0.2813,
      "step": 35740
    },
    {
      "epoch": 1.9055995309293463,
      "grad_norm": 1.6869901644100265,
      "learning_rate": 9.653044444177066e-07,
      "loss": 0.2841,
      "step": 35750
    },
    {
      "epoch": 1.9061325657631727,
      "grad_norm": 1.7378547146706524,
      "learning_rate": 9.644810548846155e-07,
      "loss": 0.2845,
      "step": 35760
    },
    {
      "epoch": 1.906665600596999,
      "grad_norm": 1.7514458267258566,
      "learning_rate": 9.636578502439776e-07,
      "loss": 0.2892,
      "step": 35770
    },
    {
      "epoch": 1.9071986354308255,
      "grad_norm": 1.8273835957891131,
      "learning_rate": 9.62834830780011e-07,
      "loss": 0.279,
      "step": 35780
    },
    {
      "epoch": 1.9077316702646518,
      "grad_norm": 1.9273408099327018,
      "learning_rate": 9.620119967768706e-07,
      "loss": 0.2745,
      "step": 35790
    },
    {
      "epoch": 1.9082647050984782,
      "grad_norm": 1.6892019370881484,
      "learning_rate": 9.61189348518646e-07,
      "loss": 0.2812,
      "step": 35800
    },
    {
      "epoch": 1.9087977399323046,
      "grad_norm": 1.6635835006417588,
      "learning_rate": 9.60366886289364e-07,
      "loss": 0.2831,
      "step": 35810
    },
    {
      "epoch": 1.909330774766131,
      "grad_norm": 1.8973614749300904,
      "learning_rate": 9.595446103729864e-07,
      "loss": 0.2833,
      "step": 35820
    },
    {
      "epoch": 1.9098638095999574,
      "grad_norm": 1.7818512917547769,
      "learning_rate": 9.587225210534114e-07,
      "loss": 0.2884,
      "step": 35830
    },
    {
      "epoch": 1.9103968444337838,
      "grad_norm": 1.7489147029989127,
      "learning_rate": 9.579006186144717e-07,
      "loss": 0.286,
      "step": 35840
    },
    {
      "epoch": 1.9109298792676102,
      "grad_norm": 1.8007346747307635,
      "learning_rate": 9.570789033399367e-07,
      "loss": 0.2929,
      "step": 35850
    },
    {
      "epoch": 1.9114629141014365,
      "grad_norm": 1.9401577857368855,
      "learning_rate": 9.562573755135098e-07,
      "loss": 0.2844,
      "step": 35860
    },
    {
      "epoch": 1.911995948935263,
      "grad_norm": 1.5981758023953478,
      "learning_rate": 9.55436035418831e-07,
      "loss": 0.2814,
      "step": 35870
    },
    {
      "epoch": 1.9125289837690893,
      "grad_norm": 2.017519936804557,
      "learning_rate": 9.54614883339475e-07,
      "loss": 0.279,
      "step": 35880
    },
    {
      "epoch": 1.9130620186029157,
      "grad_norm": 1.809624859786025,
      "learning_rate": 9.537939195589513e-07,
      "loss": 0.2904,
      "step": 35890
    },
    {
      "epoch": 1.913595053436742,
      "grad_norm": 1.7320534903736615,
      "learning_rate": 9.529731443607042e-07,
      "loss": 0.271,
      "step": 35900
    },
    {
      "epoch": 1.9141280882705685,
      "grad_norm": 1.7357395668854863,
      "learning_rate": 9.521525580281139e-07,
      "loss": 0.2805,
      "step": 35910
    },
    {
      "epoch": 1.9146611231043948,
      "grad_norm": 1.722736631229961,
      "learning_rate": 9.513321608444949e-07,
      "loss": 0.2724,
      "step": 35920
    },
    {
      "epoch": 1.9151941579382212,
      "grad_norm": 1.9241266918096638,
      "learning_rate": 9.505119530930956e-07,
      "loss": 0.2839,
      "step": 35930
    },
    {
      "epoch": 1.9157271927720476,
      "grad_norm": 1.6270102749515718,
      "learning_rate": 9.496919350571001e-07,
      "loss": 0.3006,
      "step": 35940
    },
    {
      "epoch": 1.916260227605874,
      "grad_norm": 1.7199252096332123,
      "learning_rate": 9.488721070196264e-07,
      "loss": 0.283,
      "step": 35950
    },
    {
      "epoch": 1.9167932624397004,
      "grad_norm": 1.807800038786866,
      "learning_rate": 9.480524692637277e-07,
      "loss": 0.2845,
      "step": 35960
    },
    {
      "epoch": 1.9173262972735268,
      "grad_norm": 1.6166340248636812,
      "learning_rate": 9.472330220723898e-07,
      "loss": 0.2762,
      "step": 35970
    },
    {
      "epoch": 1.9178593321073532,
      "grad_norm": 1.9497393724588654,
      "learning_rate": 9.464137657285344e-07,
      "loss": 0.2756,
      "step": 35980
    },
    {
      "epoch": 1.9183923669411795,
      "grad_norm": 1.6361964040944617,
      "learning_rate": 9.455947005150166e-07,
      "loss": 0.2904,
      "step": 35990
    },
    {
      "epoch": 1.918925401775006,
      "grad_norm": 2.1849960188785404,
      "learning_rate": 9.447758267146262e-07,
      "loss": 0.2831,
      "step": 36000
    },
    {
      "epoch": 1.9194584366088323,
      "grad_norm": 1.6556210488547158,
      "learning_rate": 9.439571446100852e-07,
      "loss": 0.2801,
      "step": 36010
    },
    {
      "epoch": 1.9199914714426587,
      "grad_norm": 1.6712818507703093,
      "learning_rate": 9.431386544840513e-07,
      "loss": 0.2802,
      "step": 36020
    },
    {
      "epoch": 1.920524506276485,
      "grad_norm": 1.804559281641379,
      "learning_rate": 9.423203566191149e-07,
      "loss": 0.2804,
      "step": 36030
    },
    {
      "epoch": 1.9210575411103115,
      "grad_norm": 1.8679026065027806,
      "learning_rate": 9.415022512978004e-07,
      "loss": 0.2689,
      "step": 36040
    },
    {
      "epoch": 1.9215905759441378,
      "grad_norm": 1.7565237113543857,
      "learning_rate": 9.406843388025654e-07,
      "loss": 0.2824,
      "step": 36050
    },
    {
      "epoch": 1.9221236107779642,
      "grad_norm": 1.7958007320276206,
      "learning_rate": 9.398666194158013e-07,
      "loss": 0.2978,
      "step": 36060
    },
    {
      "epoch": 1.9226566456117906,
      "grad_norm": 1.6720638039570859,
      "learning_rate": 9.390490934198325e-07,
      "loss": 0.2824,
      "step": 36070
    },
    {
      "epoch": 1.923189680445617,
      "grad_norm": 1.8667543386615508,
      "learning_rate": 9.382317610969168e-07,
      "loss": 0.2929,
      "step": 36080
    },
    {
      "epoch": 1.9237227152794434,
      "grad_norm": 1.6668783405022032,
      "learning_rate": 9.374146227292455e-07,
      "loss": 0.2741,
      "step": 36090
    },
    {
      "epoch": 1.9242557501132698,
      "grad_norm": 1.900786039580737,
      "learning_rate": 9.365976785989421e-07,
      "loss": 0.2899,
      "step": 36100
    },
    {
      "epoch": 1.9247887849470962,
      "grad_norm": 1.8803647811781352,
      "learning_rate": 9.357809289880633e-07,
      "loss": 0.279,
      "step": 36110
    },
    {
      "epoch": 1.9253218197809225,
      "grad_norm": 1.8016212559393892,
      "learning_rate": 9.349643741785993e-07,
      "loss": 0.2819,
      "step": 36120
    },
    {
      "epoch": 1.925854854614749,
      "grad_norm": 1.720277545602245,
      "learning_rate": 9.341480144524727e-07,
      "loss": 0.2791,
      "step": 36130
    },
    {
      "epoch": 1.9263878894485753,
      "grad_norm": 1.823832720672372,
      "learning_rate": 9.333318500915376e-07,
      "loss": 0.2793,
      "step": 36140
    },
    {
      "epoch": 1.9269209242824017,
      "grad_norm": 1.7062857173624597,
      "learning_rate": 9.325158813775829e-07,
      "loss": 0.2698,
      "step": 36150
    },
    {
      "epoch": 1.9274539591162283,
      "grad_norm": 1.8998411596559452,
      "learning_rate": 9.317001085923278e-07,
      "loss": 0.2883,
      "step": 36160
    },
    {
      "epoch": 1.9279869939500547,
      "grad_norm": 1.8745664970895064,
      "learning_rate": 9.308845320174256e-07,
      "loss": 0.2861,
      "step": 36170
    },
    {
      "epoch": 1.928520028783881,
      "grad_norm": 1.800642889415994,
      "learning_rate": 9.300691519344602e-07,
      "loss": 0.2718,
      "step": 36180
    },
    {
      "epoch": 1.9290530636177075,
      "grad_norm": 1.7658027943527481,
      "learning_rate": 9.292539686249488e-07,
      "loss": 0.2825,
      "step": 36190
    },
    {
      "epoch": 1.9295860984515338,
      "grad_norm": 1.7326904870360502,
      "learning_rate": 9.284389823703404e-07,
      "loss": 0.2799,
      "step": 36200
    },
    {
      "epoch": 1.9301191332853602,
      "grad_norm": 1.7183900991266234,
      "learning_rate": 9.276241934520162e-07,
      "loss": 0.2776,
      "step": 36210
    },
    {
      "epoch": 1.9306521681191866,
      "grad_norm": 1.541043032916627,
      "learning_rate": 9.268096021512883e-07,
      "loss": 0.2738,
      "step": 36220
    },
    {
      "epoch": 1.931185202953013,
      "grad_norm": 1.4504211308843837,
      "learning_rate": 9.259952087494017e-07,
      "loss": 0.2741,
      "step": 36230
    },
    {
      "epoch": 1.9317182377868394,
      "grad_norm": 1.678681819627505,
      "learning_rate": 9.251810135275324e-07,
      "loss": 0.2685,
      "step": 36240
    },
    {
      "epoch": 1.9322512726206658,
      "grad_norm": 1.886181101109873,
      "learning_rate": 9.243670167667886e-07,
      "loss": 0.2823,
      "step": 36250
    },
    {
      "epoch": 1.9327843074544921,
      "grad_norm": 1.725188478274979,
      "learning_rate": 9.235532187482096e-07,
      "loss": 0.2883,
      "step": 36260
    },
    {
      "epoch": 1.9333173422883185,
      "grad_norm": 1.6693605414564963,
      "learning_rate": 9.227396197527652e-07,
      "loss": 0.2916,
      "step": 36270
    },
    {
      "epoch": 1.933850377122145,
      "grad_norm": 1.6576565386454711,
      "learning_rate": 9.219262200613581e-07,
      "loss": 0.2781,
      "step": 36280
    },
    {
      "epoch": 1.9343834119559713,
      "grad_norm": 1.9826939873446265,
      "learning_rate": 9.211130199548211e-07,
      "loss": 0.2841,
      "step": 36290
    },
    {
      "epoch": 1.9349164467897977,
      "grad_norm": 1.77360846179048,
      "learning_rate": 9.20300019713919e-07,
      "loss": 0.2772,
      "step": 36300
    },
    {
      "epoch": 1.935449481623624,
      "grad_norm": 1.8131102941049029,
      "learning_rate": 9.194872196193462e-07,
      "loss": 0.2748,
      "step": 36310
    },
    {
      "epoch": 1.9359825164574505,
      "grad_norm": 1.7480401783546256,
      "learning_rate": 9.186746199517294e-07,
      "loss": 0.2697,
      "step": 36320
    },
    {
      "epoch": 1.9365155512912768,
      "grad_norm": 1.9302671000366172,
      "learning_rate": 9.178622209916252e-07,
      "loss": 0.2842,
      "step": 36330
    },
    {
      "epoch": 1.9370485861251032,
      "grad_norm": 1.7671406445149795,
      "learning_rate": 9.170500230195219e-07,
      "loss": 0.2787,
      "step": 36340
    },
    {
      "epoch": 1.9375816209589298,
      "grad_norm": 1.8658607378820755,
      "learning_rate": 9.162380263158366e-07,
      "loss": 0.2864,
      "step": 36350
    },
    {
      "epoch": 1.9381146557927562,
      "grad_norm": 1.7325176679568046,
      "learning_rate": 9.154262311609188e-07,
      "loss": 0.2871,
      "step": 36360
    },
    {
      "epoch": 1.9386476906265826,
      "grad_norm": 1.7106999116737476,
      "learning_rate": 9.146146378350475e-07,
      "loss": 0.2843,
      "step": 36370
    },
    {
      "epoch": 1.939180725460409,
      "grad_norm": 1.927384135470018,
      "learning_rate": 9.138032466184324e-07,
      "loss": 0.2744,
      "step": 36380
    },
    {
      "epoch": 1.9397137602942354,
      "grad_norm": 1.681289791941231,
      "learning_rate": 9.129920577912127e-07,
      "loss": 0.2758,
      "step": 36390
    },
    {
      "epoch": 1.9402467951280618,
      "grad_norm": 1.654442238896739,
      "learning_rate": 9.121810716334583e-07,
      "loss": 0.2783,
      "step": 36400
    },
    {
      "epoch": 1.9407798299618881,
      "grad_norm": 1.7866227727992725,
      "learning_rate": 9.113702884251694e-07,
      "loss": 0.2637,
      "step": 36410
    },
    {
      "epoch": 1.9413128647957145,
      "grad_norm": 1.8568541751206897,
      "learning_rate": 9.105597084462756e-07,
      "loss": 0.2913,
      "step": 36420
    },
    {
      "epoch": 1.941845899629541,
      "grad_norm": 1.768017774977535,
      "learning_rate": 9.097493319766368e-07,
      "loss": 0.2854,
      "step": 36430
    },
    {
      "epoch": 1.9423789344633673,
      "grad_norm": 1.6884887208650003,
      "learning_rate": 9.089391592960417e-07,
      "loss": 0.2745,
      "step": 36440
    },
    {
      "epoch": 1.9429119692971937,
      "grad_norm": 1.7164951614461499,
      "learning_rate": 9.081291906842094e-07,
      "loss": 0.2759,
      "step": 36450
    },
    {
      "epoch": 1.94344500413102,
      "grad_norm": 1.7916389355127469,
      "learning_rate": 9.073194264207891e-07,
      "loss": 0.2854,
      "step": 36460
    },
    {
      "epoch": 1.9439780389648464,
      "grad_norm": 1.6384992054908771,
      "learning_rate": 9.065098667853582e-07,
      "loss": 0.2843,
      "step": 36470
    },
    {
      "epoch": 1.9445110737986728,
      "grad_norm": 1.608897986049546,
      "learning_rate": 9.05700512057424e-07,
      "loss": 0.2783,
      "step": 36480
    },
    {
      "epoch": 1.9450441086324992,
      "grad_norm": 1.7253520937710027,
      "learning_rate": 9.048913625164232e-07,
      "loss": 0.2827,
      "step": 36490
    },
    {
      "epoch": 1.9455771434663256,
      "grad_norm": 1.7691357094672795,
      "learning_rate": 9.040824184417215e-07,
      "loss": 0.2724,
      "step": 36500
    },
    {
      "epoch": 1.946110178300152,
      "grad_norm": 1.7388039852268735,
      "learning_rate": 9.032736801126141e-07,
      "loss": 0.2601,
      "step": 36510
    },
    {
      "epoch": 1.9466432131339784,
      "grad_norm": 1.657700283062224,
      "learning_rate": 9.024651478083242e-07,
      "loss": 0.2749,
      "step": 36520
    },
    {
      "epoch": 1.9471762479678048,
      "grad_norm": 1.8108259534309625,
      "learning_rate": 9.016568218080045e-07,
      "loss": 0.2852,
      "step": 36530
    },
    {
      "epoch": 1.9477092828016311,
      "grad_norm": 1.6513703960768635,
      "learning_rate": 9.008487023907365e-07,
      "loss": 0.28,
      "step": 36540
    },
    {
      "epoch": 1.9482423176354575,
      "grad_norm": 1.7844110344044464,
      "learning_rate": 9.000407898355308e-07,
      "loss": 0.2741,
      "step": 36550
    },
    {
      "epoch": 1.948775352469284,
      "grad_norm": 1.7524276951123188,
      "learning_rate": 8.992330844213251e-07,
      "loss": 0.2784,
      "step": 36560
    },
    {
      "epoch": 1.9493083873031103,
      "grad_norm": 1.76462533452567,
      "learning_rate": 8.98425586426987e-07,
      "loss": 0.2823,
      "step": 36570
    },
    {
      "epoch": 1.9498414221369367,
      "grad_norm": 1.9774632514881714,
      "learning_rate": 8.976182961313121e-07,
      "loss": 0.2886,
      "step": 36580
    },
    {
      "epoch": 1.950374456970763,
      "grad_norm": 1.5485616779649323,
      "learning_rate": 8.96811213813024e-07,
      "loss": 0.2733,
      "step": 36590
    },
    {
      "epoch": 1.9509074918045894,
      "grad_norm": 1.5921995713690051,
      "learning_rate": 8.960043397507753e-07,
      "loss": 0.2919,
      "step": 36600
    },
    {
      "epoch": 1.9514405266384158,
      "grad_norm": 1.851217827265196,
      "learning_rate": 8.951976742231459e-07,
      "loss": 0.2842,
      "step": 36610
    },
    {
      "epoch": 1.9519735614722422,
      "grad_norm": 1.9444409314252824,
      "learning_rate": 8.943912175086433e-07,
      "loss": 0.2862,
      "step": 36620
    },
    {
      "epoch": 1.9525065963060686,
      "grad_norm": 2.004441433151391,
      "learning_rate": 8.935849698857035e-07,
      "loss": 0.2788,
      "step": 36630
    },
    {
      "epoch": 1.953039631139895,
      "grad_norm": 1.8091019234763108,
      "learning_rate": 8.927789316326909e-07,
      "loss": 0.2766,
      "step": 36640
    },
    {
      "epoch": 1.9535726659737214,
      "grad_norm": 1.9518692871103265,
      "learning_rate": 8.919731030278968e-07,
      "loss": 0.2882,
      "step": 36650
    },
    {
      "epoch": 1.9541057008075478,
      "grad_norm": 1.7699123288702627,
      "learning_rate": 8.911674843495407e-07,
      "loss": 0.2786,
      "step": 36660
    },
    {
      "epoch": 1.9546387356413741,
      "grad_norm": 1.8185358547971173,
      "learning_rate": 8.903620758757686e-07,
      "loss": 0.2776,
      "step": 36670
    },
    {
      "epoch": 1.9551717704752005,
      "grad_norm": 2.0750874455067447,
      "learning_rate": 8.89556877884655e-07,
      "loss": 0.283,
      "step": 36680
    },
    {
      "epoch": 1.955704805309027,
      "grad_norm": 1.823021754208274,
      "learning_rate": 8.887518906542013e-07,
      "loss": 0.2871,
      "step": 36690
    },
    {
      "epoch": 1.9562378401428533,
      "grad_norm": 1.6258087870478337,
      "learning_rate": 8.879471144623363e-07,
      "loss": 0.2835,
      "step": 36700
    },
    {
      "epoch": 1.9567708749766797,
      "grad_norm": 1.625523077347558,
      "learning_rate": 8.871425495869156e-07,
      "loss": 0.2947,
      "step": 36710
    },
    {
      "epoch": 1.957303909810506,
      "grad_norm": 1.6825054503091643,
      "learning_rate": 8.86338196305722e-07,
      "loss": 0.2818,
      "step": 36720
    },
    {
      "epoch": 1.9578369446443324,
      "grad_norm": 1.7363485089066115,
      "learning_rate": 8.855340548964654e-07,
      "loss": 0.2868,
      "step": 36730
    },
    {
      "epoch": 1.9583699794781588,
      "grad_norm": 1.5282812420374434,
      "learning_rate": 8.847301256367827e-07,
      "loss": 0.2876,
      "step": 36740
    },
    {
      "epoch": 1.9589030143119852,
      "grad_norm": 1.8513297522198706,
      "learning_rate": 8.839264088042367e-07,
      "loss": 0.2868,
      "step": 36750
    },
    {
      "epoch": 1.9594360491458116,
      "grad_norm": 1.671874064013073,
      "learning_rate": 8.831229046763176e-07,
      "loss": 0.2885,
      "step": 36760
    },
    {
      "epoch": 1.959969083979638,
      "grad_norm": 1.8192626633607425,
      "learning_rate": 8.823196135304422e-07,
      "loss": 0.2779,
      "step": 36770
    },
    {
      "epoch": 1.9605021188134644,
      "grad_norm": 1.8149186109508246,
      "learning_rate": 8.815165356439539e-07,
      "loss": 0.2795,
      "step": 36780
    },
    {
      "epoch": 1.9610351536472908,
      "grad_norm": 1.7354926258018355,
      "learning_rate": 8.807136712941215e-07,
      "loss": 0.2784,
      "step": 36790
    },
    {
      "epoch": 1.9615681884811171,
      "grad_norm": 2.0301835861118507,
      "learning_rate": 8.799110207581411e-07,
      "loss": 0.2801,
      "step": 36800
    },
    {
      "epoch": 1.9621012233149435,
      "grad_norm": 1.8347235547990526,
      "learning_rate": 8.791085843131343e-07,
      "loss": 0.2763,
      "step": 36810
    },
    {
      "epoch": 1.96263425814877,
      "grad_norm": 1.655869620574944,
      "learning_rate": 8.783063622361493e-07,
      "loss": 0.2848,
      "step": 36820
    },
    {
      "epoch": 1.9631672929825963,
      "grad_norm": 1.7364512551448577,
      "learning_rate": 8.775043548041602e-07,
      "loss": 0.2734,
      "step": 36830
    },
    {
      "epoch": 1.9637003278164227,
      "grad_norm": 1.673780555752466,
      "learning_rate": 8.767025622940666e-07,
      "loss": 0.2897,
      "step": 36840
    },
    {
      "epoch": 1.964233362650249,
      "grad_norm": 1.9027837111580532,
      "learning_rate": 8.759009849826942e-07,
      "loss": 0.2823,
      "step": 36850
    },
    {
      "epoch": 1.9647663974840754,
      "grad_norm": 1.89275733224278,
      "learning_rate": 8.750996231467949e-07,
      "loss": 0.2908,
      "step": 36860
    },
    {
      "epoch": 1.9652994323179018,
      "grad_norm": 1.833761362428127,
      "learning_rate": 8.742984770630446e-07,
      "loss": 0.2816,
      "step": 36870
    },
    {
      "epoch": 1.9658324671517282,
      "grad_norm": 1.7172146271872315,
      "learning_rate": 8.734975470080465e-07,
      "loss": 0.2774,
      "step": 36880
    },
    {
      "epoch": 1.9663655019855546,
      "grad_norm": 1.5862669354506262,
      "learning_rate": 8.726968332583283e-07,
      "loss": 0.2832,
      "step": 36890
    },
    {
      "epoch": 1.966898536819381,
      "grad_norm": 1.6243517938754737,
      "learning_rate": 8.718963360903433e-07,
      "loss": 0.2818,
      "step": 36900
    },
    {
      "epoch": 1.9674315716532076,
      "grad_norm": 1.9561150905738685,
      "learning_rate": 8.710960557804697e-07,
      "loss": 0.2735,
      "step": 36910
    },
    {
      "epoch": 1.967964606487034,
      "grad_norm": 1.6272471835469826,
      "learning_rate": 8.702959926050112e-07,
      "loss": 0.2849,
      "step": 36920
    },
    {
      "epoch": 1.9684976413208604,
      "grad_norm": 1.6198491840703404,
      "learning_rate": 8.694961468401973e-07,
      "loss": 0.2793,
      "step": 36930
    },
    {
      "epoch": 1.9690306761546867,
      "grad_norm": 1.6823854341099054,
      "learning_rate": 8.6869651876218e-07,
      "loss": 0.2738,
      "step": 36940
    },
    {
      "epoch": 1.9695637109885131,
      "grad_norm": 1.9735162096961552,
      "learning_rate": 8.678971086470382e-07,
      "loss": 0.2855,
      "step": 36950
    },
    {
      "epoch": 1.9700967458223395,
      "grad_norm": 1.7759434252078614,
      "learning_rate": 8.67097916770775e-07,
      "loss": 0.2725,
      "step": 36960
    },
    {
      "epoch": 1.970629780656166,
      "grad_norm": 2.036361773071089,
      "learning_rate": 8.662989434093186e-07,
      "loss": 0.282,
      "step": 36970
    },
    {
      "epoch": 1.9711628154899923,
      "grad_norm": 2.280083996627828,
      "learning_rate": 8.655001888385208e-07,
      "loss": 0.2796,
      "step": 36980
    },
    {
      "epoch": 1.9716958503238187,
      "grad_norm": 1.8414921076702122,
      "learning_rate": 8.647016533341586e-07,
      "loss": 0.2798,
      "step": 36990
    },
    {
      "epoch": 1.972228885157645,
      "grad_norm": 1.995236770540033,
      "learning_rate": 8.639033371719332e-07,
      "loss": 0.2878,
      "step": 37000
    },
    {
      "epoch": 1.9727619199914714,
      "grad_norm": 1.7373901738460922,
      "learning_rate": 8.631052406274703e-07,
      "loss": 0.2754,
      "step": 37010
    },
    {
      "epoch": 1.9732949548252978,
      "grad_norm": 1.7654714181223115,
      "learning_rate": 8.623073639763194e-07,
      "loss": 0.2846,
      "step": 37020
    },
    {
      "epoch": 1.9738279896591242,
      "grad_norm": 1.97590645281513,
      "learning_rate": 8.615097074939534e-07,
      "loss": 0.2772,
      "step": 37030
    },
    {
      "epoch": 1.9743610244929506,
      "grad_norm": 1.7405176687167827,
      "learning_rate": 8.607122714557703e-07,
      "loss": 0.2753,
      "step": 37040
    },
    {
      "epoch": 1.974894059326777,
      "grad_norm": 1.8325113812080918,
      "learning_rate": 8.599150561370922e-07,
      "loss": 0.2819,
      "step": 37050
    },
    {
      "epoch": 1.9754270941606034,
      "grad_norm": 1.8010149742739963,
      "learning_rate": 8.59118061813164e-07,
      "loss": 0.2865,
      "step": 37060
    },
    {
      "epoch": 1.9759601289944297,
      "grad_norm": 1.8316197266917615,
      "learning_rate": 8.583212887591546e-07,
      "loss": 0.2732,
      "step": 37070
    },
    {
      "epoch": 1.9764931638282561,
      "grad_norm": 1.7149023060261717,
      "learning_rate": 8.575247372501571e-07,
      "loss": 0.285,
      "step": 37080
    },
    {
      "epoch": 1.9770261986620825,
      "grad_norm": 1.692541911361734,
      "learning_rate": 8.567284075611875e-07,
      "loss": 0.2724,
      "step": 37090
    },
    {
      "epoch": 1.9775592334959091,
      "grad_norm": 1.776816830406351,
      "learning_rate": 8.559322999671857e-07,
      "loss": 0.2783,
      "step": 37100
    },
    {
      "epoch": 1.9780922683297355,
      "grad_norm": 1.783780368753211,
      "learning_rate": 8.551364147430137e-07,
      "loss": 0.2901,
      "step": 37110
    },
    {
      "epoch": 1.9786253031635619,
      "grad_norm": 1.8756166103429739,
      "learning_rate": 8.543407521634581e-07,
      "loss": 0.2716,
      "step": 37120
    },
    {
      "epoch": 1.9791583379973883,
      "grad_norm": 1.5514200373776226,
      "learning_rate": 8.535453125032282e-07,
      "loss": 0.2783,
      "step": 37130
    },
    {
      "epoch": 1.9796913728312147,
      "grad_norm": 1.7423006707033284,
      "learning_rate": 8.527500960369563e-07,
      "loss": 0.2791,
      "step": 37140
    },
    {
      "epoch": 1.980224407665041,
      "grad_norm": 1.602861294093616,
      "learning_rate": 8.519551030391974e-07,
      "loss": 0.2744,
      "step": 37150
    },
    {
      "epoch": 1.9807574424988674,
      "grad_norm": 1.8920195961306705,
      "learning_rate": 8.511603337844299e-07,
      "loss": 0.2756,
      "step": 37160
    },
    {
      "epoch": 1.9812904773326938,
      "grad_norm": 1.5996132291653768,
      "learning_rate": 8.503657885470544e-07,
      "loss": 0.287,
      "step": 37170
    },
    {
      "epoch": 1.9818235121665202,
      "grad_norm": 1.7369523858256208,
      "learning_rate": 8.495714676013944e-07,
      "loss": 0.2782,
      "step": 37180
    },
    {
      "epoch": 1.9823565470003466,
      "grad_norm": 1.6885591506893616,
      "learning_rate": 8.487773712216966e-07,
      "loss": 0.2846,
      "step": 37190
    },
    {
      "epoch": 1.982889581834173,
      "grad_norm": 2.0194738365073075,
      "learning_rate": 8.479834996821282e-07,
      "loss": 0.2756,
      "step": 37200
    },
    {
      "epoch": 1.9834226166679993,
      "grad_norm": 1.8315859366223484,
      "learning_rate": 8.471898532567808e-07,
      "loss": 0.2729,
      "step": 37210
    },
    {
      "epoch": 1.9839556515018257,
      "grad_norm": 1.807615548568915,
      "learning_rate": 8.463964322196675e-07,
      "loss": 0.2797,
      "step": 37220
    },
    {
      "epoch": 1.9844886863356521,
      "grad_norm": 1.7267358480592316,
      "learning_rate": 8.456032368447237e-07,
      "loss": 0.2865,
      "step": 37230
    },
    {
      "epoch": 1.9850217211694785,
      "grad_norm": 1.5601398187036628,
      "learning_rate": 8.448102674058066e-07,
      "loss": 0.2688,
      "step": 37240
    },
    {
      "epoch": 1.9855547560033049,
      "grad_norm": 1.7277658703288845,
      "learning_rate": 8.44017524176696e-07,
      "loss": 0.2801,
      "step": 37250
    },
    {
      "epoch": 1.9860877908371313,
      "grad_norm": 1.533304161271189,
      "learning_rate": 8.432250074310929e-07,
      "loss": 0.2774,
      "step": 37260
    },
    {
      "epoch": 1.9866208256709577,
      "grad_norm": 1.6985046797786474,
      "learning_rate": 8.42432717442621e-07,
      "loss": 0.2837,
      "step": 37270
    },
    {
      "epoch": 1.987153860504784,
      "grad_norm": 2.0512378175590915,
      "learning_rate": 8.41640654484824e-07,
      "loss": 0.2854,
      "step": 37280
    },
    {
      "epoch": 1.9876868953386104,
      "grad_norm": 1.8761857339871766,
      "learning_rate": 8.408488188311693e-07,
      "loss": 0.2808,
      "step": 37290
    },
    {
      "epoch": 1.9882199301724368,
      "grad_norm": 1.7663108450591758,
      "learning_rate": 8.400572107550441e-07,
      "loss": 0.286,
      "step": 37300
    },
    {
      "epoch": 1.9887529650062632,
      "grad_norm": 1.6449848420513955,
      "learning_rate": 8.392658305297585e-07,
      "loss": 0.2827,
      "step": 37310
    },
    {
      "epoch": 1.9892859998400896,
      "grad_norm": 1.9266757012563016,
      "learning_rate": 8.384746784285431e-07,
      "loss": 0.2798,
      "step": 37320
    },
    {
      "epoch": 1.989819034673916,
      "grad_norm": 1.833410485729512,
      "learning_rate": 8.376837547245496e-07,
      "loss": 0.2931,
      "step": 37330
    },
    {
      "epoch": 1.9903520695077423,
      "grad_norm": 1.8753429970104938,
      "learning_rate": 8.368930596908514e-07,
      "loss": 0.2822,
      "step": 37340
    },
    {
      "epoch": 1.9908851043415687,
      "grad_norm": 1.7995855127131901,
      "learning_rate": 8.361025936004429e-07,
      "loss": 0.2824,
      "step": 37350
    },
    {
      "epoch": 1.9914181391753951,
      "grad_norm": 1.773127869758638,
      "learning_rate": 8.353123567262385e-07,
      "loss": 0.2715,
      "step": 37360
    },
    {
      "epoch": 1.9919511740092215,
      "grad_norm": 1.7889893350347559,
      "learning_rate": 8.345223493410742e-07,
      "loss": 0.276,
      "step": 37370
    },
    {
      "epoch": 1.9924842088430479,
      "grad_norm": 1.9878695322428117,
      "learning_rate": 8.337325717177072e-07,
      "loss": 0.2824,
      "step": 37380
    },
    {
      "epoch": 1.9930172436768743,
      "grad_norm": 1.6672478185804749,
      "learning_rate": 8.329430241288149e-07,
      "loss": 0.2859,
      "step": 37390
    },
    {
      "epoch": 1.9935502785107007,
      "grad_norm": 1.7051316752772439,
      "learning_rate": 8.32153706846995e-07,
      "loss": 0.2723,
      "step": 37400
    },
    {
      "epoch": 1.994083313344527,
      "grad_norm": 1.703657986558443,
      "learning_rate": 8.313646201447661e-07,
      "loss": 0.2807,
      "step": 37410
    },
    {
      "epoch": 1.9946163481783534,
      "grad_norm": 1.756571639964288,
      "learning_rate": 8.305757642945673e-07,
      "loss": 0.2897,
      "step": 37420
    },
    {
      "epoch": 1.9951493830121798,
      "grad_norm": 1.597870923328568,
      "learning_rate": 8.297871395687573e-07,
      "loss": 0.2757,
      "step": 37430
    },
    {
      "epoch": 1.9956824178460062,
      "grad_norm": 1.6485891201137066,
      "learning_rate": 8.289987462396164e-07,
      "loss": 0.2712,
      "step": 37440
    },
    {
      "epoch": 1.9962154526798326,
      "grad_norm": 1.8649702377234794,
      "learning_rate": 8.28210584579343e-07,
      "loss": 0.277,
      "step": 37450
    },
    {
      "epoch": 1.996748487513659,
      "grad_norm": 1.7786809382199258,
      "learning_rate": 8.27422654860057e-07,
      "loss": 0.277,
      "step": 37460
    },
    {
      "epoch": 1.9972815223474853,
      "grad_norm": 1.5475122501493337,
      "learning_rate": 8.266349573537975e-07,
      "loss": 0.278,
      "step": 37470
    },
    {
      "epoch": 1.9978145571813117,
      "grad_norm": 1.7395069185334262,
      "learning_rate": 8.258474923325245e-07,
      "loss": 0.2761,
      "step": 37480
    },
    {
      "epoch": 1.9983475920151381,
      "grad_norm": 1.8495904680787085,
      "learning_rate": 8.250602600681162e-07,
      "loss": 0.2897,
      "step": 37490
    },
    {
      "epoch": 1.9988806268489645,
      "grad_norm": 1.8715093356523964,
      "learning_rate": 8.242732608323717e-07,
      "loss": 0.2706,
      "step": 37500
    },
    {
      "epoch": 1.9994136616827909,
      "grad_norm": 1.730719376962117,
      "learning_rate": 8.23486494897009e-07,
      "loss": 0.2706,
      "step": 37510
    },
    {
      "epoch": 1.9999466965166173,
      "grad_norm": 1.748461421570759,
      "learning_rate": 8.226999625336663e-07,
      "loss": 0.2828,
      "step": 37520
    },
    {
      "epoch": 2.0004797313504437,
      "grad_norm": 1.5736294248881135,
      "learning_rate": 8.219136640138994e-07,
      "loss": 0.2298,
      "step": 37530
    },
    {
      "epoch": 2.00101276618427,
      "grad_norm": 1.7738230858131003,
      "learning_rate": 8.21127599609185e-07,
      "loss": 0.2413,
      "step": 37540
    },
    {
      "epoch": 2.0015458010180964,
      "grad_norm": 1.6640783817606957,
      "learning_rate": 8.203417695909185e-07,
      "loss": 0.2321,
      "step": 37550
    },
    {
      "epoch": 2.002078835851923,
      "grad_norm": 1.8111774501017874,
      "learning_rate": 8.195561742304147e-07,
      "loss": 0.2433,
      "step": 37560
    },
    {
      "epoch": 2.002611870685749,
      "grad_norm": 1.7014912795958672,
      "learning_rate": 8.187708137989066e-07,
      "loss": 0.2274,
      "step": 37570
    },
    {
      "epoch": 2.0031449055195756,
      "grad_norm": 1.6795478384000704,
      "learning_rate": 8.179856885675467e-07,
      "loss": 0.2395,
      "step": 37580
    },
    {
      "epoch": 2.003677940353402,
      "grad_norm": 1.6398399366440055,
      "learning_rate": 8.172007988074063e-07,
      "loss": 0.2311,
      "step": 37590
    },
    {
      "epoch": 2.0042109751872283,
      "grad_norm": 1.9108655627967637,
      "learning_rate": 8.164161447894752e-07,
      "loss": 0.2383,
      "step": 37600
    },
    {
      "epoch": 2.0047440100210547,
      "grad_norm": 1.959416804192182,
      "learning_rate": 8.156317267846623e-07,
      "loss": 0.2423,
      "step": 37610
    },
    {
      "epoch": 2.005277044854881,
      "grad_norm": 1.8418762869642187,
      "learning_rate": 8.148475450637933e-07,
      "loss": 0.2323,
      "step": 37620
    },
    {
      "epoch": 2.0058100796887075,
      "grad_norm": 1.7920108726808361,
      "learning_rate": 8.140635998976147e-07,
      "loss": 0.2413,
      "step": 37630
    },
    {
      "epoch": 2.006343114522534,
      "grad_norm": 1.871491738048841,
      "learning_rate": 8.132798915567895e-07,
      "loss": 0.2464,
      "step": 37640
    },
    {
      "epoch": 2.0068761493563603,
      "grad_norm": 1.8370651097602901,
      "learning_rate": 8.124964203119001e-07,
      "loss": 0.2371,
      "step": 37650
    },
    {
      "epoch": 2.0074091841901867,
      "grad_norm": 1.8895932681415564,
      "learning_rate": 8.117131864334464e-07,
      "loss": 0.2308,
      "step": 37660
    },
    {
      "epoch": 2.007942219024013,
      "grad_norm": 1.7556453087755124,
      "learning_rate": 8.109301901918466e-07,
      "loss": 0.2408,
      "step": 37670
    },
    {
      "epoch": 2.0084752538578394,
      "grad_norm": 1.849030597001621,
      "learning_rate": 8.101474318574368e-07,
      "loss": 0.2296,
      "step": 37680
    },
    {
      "epoch": 2.009008288691666,
      "grad_norm": 1.7692942312307667,
      "learning_rate": 8.093649117004716e-07,
      "loss": 0.2293,
      "step": 37690
    },
    {
      "epoch": 2.009541323525492,
      "grad_norm": 1.6097243173013291,
      "learning_rate": 8.085826299911214e-07,
      "loss": 0.2321,
      "step": 37700
    },
    {
      "epoch": 2.0100743583593186,
      "grad_norm": 1.9413411037236705,
      "learning_rate": 8.07800586999476e-07,
      "loss": 0.2357,
      "step": 37710
    },
    {
      "epoch": 2.010607393193145,
      "grad_norm": 1.8425236300592505,
      "learning_rate": 8.07018782995543e-07,
      "loss": 0.2438,
      "step": 37720
    },
    {
      "epoch": 2.0111404280269713,
      "grad_norm": 1.8096509442824977,
      "learning_rate": 8.062372182492461e-07,
      "loss": 0.2369,
      "step": 37730
    },
    {
      "epoch": 2.0116734628607977,
      "grad_norm": 1.805073782553207,
      "learning_rate": 8.054558930304279e-07,
      "loss": 0.2401,
      "step": 37740
    },
    {
      "epoch": 2.0122064976946246,
      "grad_norm": 2.0141296583955945,
      "learning_rate": 8.046748076088468e-07,
      "loss": 0.2336,
      "step": 37750
    },
    {
      "epoch": 2.012739532528451,
      "grad_norm": 1.7393311856686322,
      "learning_rate": 8.038939622541796e-07,
      "loss": 0.2365,
      "step": 37760
    },
    {
      "epoch": 2.0132725673622773,
      "grad_norm": 1.9116227455668509,
      "learning_rate": 8.031133572360197e-07,
      "loss": 0.2323,
      "step": 37770
    },
    {
      "epoch": 2.0138056021961037,
      "grad_norm": 1.8061397289574974,
      "learning_rate": 8.023329928238782e-07,
      "loss": 0.2311,
      "step": 37780
    },
    {
      "epoch": 2.01433863702993,
      "grad_norm": 1.7869109509532108,
      "learning_rate": 8.015528692871815e-07,
      "loss": 0.2371,
      "step": 37790
    },
    {
      "epoch": 2.0148716718637565,
      "grad_norm": 1.804094302500777,
      "learning_rate": 8.00772986895274e-07,
      "loss": 0.2333,
      "step": 37800
    },
    {
      "epoch": 2.015404706697583,
      "grad_norm": 1.9678191665846236,
      "learning_rate": 7.999933459174171e-07,
      "loss": 0.243,
      "step": 37810
    },
    {
      "epoch": 2.0159377415314093,
      "grad_norm": 1.6924137848581187,
      "learning_rate": 7.992139466227883e-07,
      "loss": 0.2313,
      "step": 37820
    },
    {
      "epoch": 2.0164707763652356,
      "grad_norm": 1.7948700307484415,
      "learning_rate": 7.984347892804819e-07,
      "loss": 0.2372,
      "step": 37830
    },
    {
      "epoch": 2.017003811199062,
      "grad_norm": 1.833744535386794,
      "learning_rate": 7.976558741595084e-07,
      "loss": 0.2432,
      "step": 37840
    },
    {
      "epoch": 2.0175368460328884,
      "grad_norm": 1.690808793338307,
      "learning_rate": 7.96877201528795e-07,
      "loss": 0.2442,
      "step": 37850
    },
    {
      "epoch": 2.018069880866715,
      "grad_norm": 1.8516431262820199,
      "learning_rate": 7.960987716571856e-07,
      "loss": 0.2357,
      "step": 37860
    },
    {
      "epoch": 2.018602915700541,
      "grad_norm": 1.6257392258732206,
      "learning_rate": 7.953205848134385e-07,
      "loss": 0.2396,
      "step": 37870
    },
    {
      "epoch": 2.0191359505343676,
      "grad_norm": 1.8577470034892662,
      "learning_rate": 7.945426412662299e-07,
      "loss": 0.2312,
      "step": 37880
    },
    {
      "epoch": 2.019668985368194,
      "grad_norm": 1.6841858566426315,
      "learning_rate": 7.937649412841514e-07,
      "loss": 0.2297,
      "step": 37890
    },
    {
      "epoch": 2.0202020202020203,
      "grad_norm": 1.903325187428609,
      "learning_rate": 7.929874851357104e-07,
      "loss": 0.2291,
      "step": 37900
    },
    {
      "epoch": 2.0207350550358467,
      "grad_norm": 1.8887622305980312,
      "learning_rate": 7.922102730893305e-07,
      "loss": 0.2387,
      "step": 37910
    },
    {
      "epoch": 2.021268089869673,
      "grad_norm": 1.77595963306003,
      "learning_rate": 7.914333054133503e-07,
      "loss": 0.2374,
      "step": 37920
    },
    {
      "epoch": 2.0218011247034995,
      "grad_norm": 1.97168107403706,
      "learning_rate": 7.906565823760251e-07,
      "loss": 0.239,
      "step": 37930
    },
    {
      "epoch": 2.022334159537326,
      "grad_norm": 1.701640934208281,
      "learning_rate": 7.898801042455243e-07,
      "loss": 0.2309,
      "step": 37940
    },
    {
      "epoch": 2.0228671943711523,
      "grad_norm": 1.706134416451444,
      "learning_rate": 7.89103871289935e-07,
      "loss": 0.2336,
      "step": 37950
    },
    {
      "epoch": 2.0234002292049786,
      "grad_norm": 1.7698411122501323,
      "learning_rate": 7.883278837772565e-07,
      "loss": 0.2319,
      "step": 37960
    },
    {
      "epoch": 2.023933264038805,
      "grad_norm": 1.736574804524401,
      "learning_rate": 7.875521419754062e-07,
      "loss": 0.2309,
      "step": 37970
    },
    {
      "epoch": 2.0244662988726314,
      "grad_norm": 1.7170654920574158,
      "learning_rate": 7.867766461522147e-07,
      "loss": 0.2372,
      "step": 37980
    },
    {
      "epoch": 2.024999333706458,
      "grad_norm": 2.1667226980332845,
      "learning_rate": 7.860013965754288e-07,
      "loss": 0.2389,
      "step": 37990
    },
    {
      "epoch": 2.025532368540284,
      "grad_norm": 1.8467594469420534,
      "learning_rate": 7.852263935127101e-07,
      "loss": 0.2337,
      "step": 38000
    },
    {
      "epoch": 2.0260654033741106,
      "grad_norm": 1.8233912648698773,
      "learning_rate": 7.84451637231635e-07,
      "loss": 0.2376,
      "step": 38010
    },
    {
      "epoch": 2.026598438207937,
      "grad_norm": 1.7944039914015908,
      "learning_rate": 7.836771279996946e-07,
      "loss": 0.2251,
      "step": 38020
    },
    {
      "epoch": 2.0271314730417633,
      "grad_norm": 1.6819159320036245,
      "learning_rate": 7.829028660842948e-07,
      "loss": 0.2331,
      "step": 38030
    },
    {
      "epoch": 2.0276645078755897,
      "grad_norm": 1.845999510494668,
      "learning_rate": 7.821288517527561e-07,
      "loss": 0.2324,
      "step": 38040
    },
    {
      "epoch": 2.028197542709416,
      "grad_norm": 1.7144658488912408,
      "learning_rate": 7.813550852723142e-07,
      "loss": 0.2337,
      "step": 38050
    },
    {
      "epoch": 2.0287305775432425,
      "grad_norm": 1.9575816872204819,
      "learning_rate": 7.80581566910117e-07,
      "loss": 0.2325,
      "step": 38060
    },
    {
      "epoch": 2.029263612377069,
      "grad_norm": 1.8772335533313278,
      "learning_rate": 7.798082969332292e-07,
      "loss": 0.2305,
      "step": 38070
    },
    {
      "epoch": 2.0297966472108953,
      "grad_norm": 1.8970683408147426,
      "learning_rate": 7.790352756086287e-07,
      "loss": 0.2321,
      "step": 38080
    },
    {
      "epoch": 2.0303296820447216,
      "grad_norm": 1.7921100108658055,
      "learning_rate": 7.782625032032077e-07,
      "loss": 0.2437,
      "step": 38090
    },
    {
      "epoch": 2.030862716878548,
      "grad_norm": 1.8013459628885917,
      "learning_rate": 7.774899799837724e-07,
      "loss": 0.2423,
      "step": 38100
    },
    {
      "epoch": 2.0313957517123744,
      "grad_norm": 1.791031280264957,
      "learning_rate": 7.76717706217043e-07,
      "loss": 0.2315,
      "step": 38110
    },
    {
      "epoch": 2.031928786546201,
      "grad_norm": 1.7816306638218051,
      "learning_rate": 7.759456821696536e-07,
      "loss": 0.2337,
      "step": 38120
    },
    {
      "epoch": 2.032461821380027,
      "grad_norm": 1.7259380841922725,
      "learning_rate": 7.751739081081518e-07,
      "loss": 0.2492,
      "step": 38130
    },
    {
      "epoch": 2.0329948562138536,
      "grad_norm": 1.6647546932772428,
      "learning_rate": 7.744023842989999e-07,
      "loss": 0.2402,
      "step": 38140
    },
    {
      "epoch": 2.03352789104768,
      "grad_norm": 1.964455360587893,
      "learning_rate": 7.736311110085721e-07,
      "loss": 0.2276,
      "step": 38150
    },
    {
      "epoch": 2.0340609258815063,
      "grad_norm": 1.9405695317132412,
      "learning_rate": 7.728600885031571e-07,
      "loss": 0.2455,
      "step": 38160
    },
    {
      "epoch": 2.0345939607153327,
      "grad_norm": 1.7219815442406532,
      "learning_rate": 7.720893170489574e-07,
      "loss": 0.2346,
      "step": 38170
    },
    {
      "epoch": 2.035126995549159,
      "grad_norm": 1.9284142510366096,
      "learning_rate": 7.713187969120882e-07,
      "loss": 0.2369,
      "step": 38180
    },
    {
      "epoch": 2.0356600303829855,
      "grad_norm": 2.131341757885997,
      "learning_rate": 7.70548528358578e-07,
      "loss": 0.236,
      "step": 38190
    },
    {
      "epoch": 2.036193065216812,
      "grad_norm": 1.9120895052394005,
      "learning_rate": 7.69778511654369e-07,
      "loss": 0.2447,
      "step": 38200
    },
    {
      "epoch": 2.0367261000506383,
      "grad_norm": 1.7948223586326344,
      "learning_rate": 7.690087470653153e-07,
      "loss": 0.2447,
      "step": 38210
    },
    {
      "epoch": 2.0372591348844646,
      "grad_norm": 1.8312476841409095,
      "learning_rate": 7.682392348571856e-07,
      "loss": 0.2356,
      "step": 38220
    },
    {
      "epoch": 2.037792169718291,
      "grad_norm": 1.9070237984493101,
      "learning_rate": 7.674699752956595e-07,
      "loss": 0.2385,
      "step": 38230
    },
    {
      "epoch": 2.0383252045521174,
      "grad_norm": 1.7664194188598732,
      "learning_rate": 7.667009686463306e-07,
      "loss": 0.2325,
      "step": 38240
    },
    {
      "epoch": 2.038858239385944,
      "grad_norm": 1.8501820426134736,
      "learning_rate": 7.659322151747054e-07,
      "loss": 0.2242,
      "step": 38250
    },
    {
      "epoch": 2.03939127421977,
      "grad_norm": 1.795331927286408,
      "learning_rate": 7.651637151462021e-07,
      "loss": 0.2461,
      "step": 38260
    },
    {
      "epoch": 2.0399243090535966,
      "grad_norm": 1.7959332250295559,
      "learning_rate": 7.643954688261523e-07,
      "loss": 0.2314,
      "step": 38270
    },
    {
      "epoch": 2.040457343887423,
      "grad_norm": 1.86266814866272,
      "learning_rate": 7.636274764797991e-07,
      "loss": 0.2469,
      "step": 38280
    },
    {
      "epoch": 2.0409903787212493,
      "grad_norm": 1.6733629886556638,
      "learning_rate": 7.628597383722985e-07,
      "loss": 0.2312,
      "step": 38290
    },
    {
      "epoch": 2.0415234135550757,
      "grad_norm": 1.7225322918660038,
      "learning_rate": 7.620922547687194e-07,
      "loss": 0.2435,
      "step": 38300
    },
    {
      "epoch": 2.042056448388902,
      "grad_norm": 1.8322505245736525,
      "learning_rate": 7.613250259340405e-07,
      "loss": 0.2309,
      "step": 38310
    },
    {
      "epoch": 2.0425894832227285,
      "grad_norm": 1.8050068586235133,
      "learning_rate": 7.605580521331549e-07,
      "loss": 0.2404,
      "step": 38320
    },
    {
      "epoch": 2.043122518056555,
      "grad_norm": 1.6833845783763375,
      "learning_rate": 7.597913336308668e-07,
      "loss": 0.2381,
      "step": 38330
    },
    {
      "epoch": 2.0436555528903813,
      "grad_norm": 1.8728951511052994,
      "learning_rate": 7.590248706918922e-07,
      "loss": 0.2353,
      "step": 38340
    },
    {
      "epoch": 2.0441885877242076,
      "grad_norm": 1.9330922143102474,
      "learning_rate": 7.582586635808589e-07,
      "loss": 0.241,
      "step": 38350
    },
    {
      "epoch": 2.044721622558034,
      "grad_norm": 1.811551623639987,
      "learning_rate": 7.574927125623062e-07,
      "loss": 0.2428,
      "step": 38360
    },
    {
      "epoch": 2.0452546573918604,
      "grad_norm": 1.779744683025959,
      "learning_rate": 7.567270179006854e-07,
      "loss": 0.239,
      "step": 38370
    },
    {
      "epoch": 2.045787692225687,
      "grad_norm": 1.7592124202232444,
      "learning_rate": 7.559615798603593e-07,
      "loss": 0.2287,
      "step": 38380
    },
    {
      "epoch": 2.046320727059513,
      "grad_norm": 1.8592685928692703,
      "learning_rate": 7.551963987056021e-07,
      "loss": 0.238,
      "step": 38390
    },
    {
      "epoch": 2.0468537618933396,
      "grad_norm": 1.8665641630841088,
      "learning_rate": 7.544314747005981e-07,
      "loss": 0.235,
      "step": 38400
    },
    {
      "epoch": 2.047386796727166,
      "grad_norm": 1.7878447557567456,
      "learning_rate": 7.536668081094439e-07,
      "loss": 0.24,
      "step": 38410
    },
    {
      "epoch": 2.0479198315609923,
      "grad_norm": 2.003101277001904,
      "learning_rate": 7.529023991961478e-07,
      "loss": 0.2305,
      "step": 38420
    },
    {
      "epoch": 2.0484528663948187,
      "grad_norm": 1.6137277206347598,
      "learning_rate": 7.521382482246283e-07,
      "loss": 0.2416,
      "step": 38430
    },
    {
      "epoch": 2.048985901228645,
      "grad_norm": 1.78262850336962,
      "learning_rate": 7.513743554587147e-07,
      "loss": 0.2343,
      "step": 38440
    },
    {
      "epoch": 2.0495189360624715,
      "grad_norm": 1.9049626532469903,
      "learning_rate": 7.506107211621475e-07,
      "loss": 0.2339,
      "step": 38450
    },
    {
      "epoch": 2.050051970896298,
      "grad_norm": 1.9196749422535442,
      "learning_rate": 7.498473455985782e-07,
      "loss": 0.2348,
      "step": 38460
    },
    {
      "epoch": 2.0505850057301243,
      "grad_norm": 1.7903853520777744,
      "learning_rate": 7.490842290315689e-07,
      "loss": 0.2376,
      "step": 38470
    },
    {
      "epoch": 2.0511180405639506,
      "grad_norm": 1.965974895304583,
      "learning_rate": 7.483213717245913e-07,
      "loss": 0.2361,
      "step": 38480
    },
    {
      "epoch": 2.0516510753977775,
      "grad_norm": 1.867664765365325,
      "learning_rate": 7.475587739410287e-07,
      "loss": 0.2423,
      "step": 38490
    },
    {
      "epoch": 2.052184110231604,
      "grad_norm": 1.8511950794225667,
      "learning_rate": 7.467964359441743e-07,
      "loss": 0.2458,
      "step": 38500
    },
    {
      "epoch": 2.0527171450654302,
      "grad_norm": 1.6968488847088037,
      "learning_rate": 7.460343579972319e-07,
      "loss": 0.2438,
      "step": 38510
    },
    {
      "epoch": 2.0532501798992566,
      "grad_norm": 1.8525495613524092,
      "learning_rate": 7.452725403633155e-07,
      "loss": 0.2352,
      "step": 38520
    },
    {
      "epoch": 2.053783214733083,
      "grad_norm": 2.050268247072766,
      "learning_rate": 7.445109833054489e-07,
      "loss": 0.2387,
      "step": 38530
    },
    {
      "epoch": 2.0543162495669094,
      "grad_norm": 1.5822781906709664,
      "learning_rate": 7.437496870865662e-07,
      "loss": 0.2407,
      "step": 38540
    },
    {
      "epoch": 2.0548492844007358,
      "grad_norm": 1.9408502637350467,
      "learning_rate": 7.429886519695113e-07,
      "loss": 0.2339,
      "step": 38550
    },
    {
      "epoch": 2.055382319234562,
      "grad_norm": 1.7503408901008903,
      "learning_rate": 7.422278782170389e-07,
      "loss": 0.2379,
      "step": 38560
    },
    {
      "epoch": 2.0559153540683885,
      "grad_norm": 1.871415401413392,
      "learning_rate": 7.414673660918111e-07,
      "loss": 0.2338,
      "step": 38570
    },
    {
      "epoch": 2.056448388902215,
      "grad_norm": 1.9291874266424698,
      "learning_rate": 7.407071158564017e-07,
      "loss": 0.2366,
      "step": 38580
    },
    {
      "epoch": 2.0569814237360413,
      "grad_norm": 1.7114149170732393,
      "learning_rate": 7.399471277732936e-07,
      "loss": 0.2391,
      "step": 38590
    },
    {
      "epoch": 2.0575144585698677,
      "grad_norm": 1.9876124650258065,
      "learning_rate": 7.391874021048793e-07,
      "loss": 0.2285,
      "step": 38600
    },
    {
      "epoch": 2.058047493403694,
      "grad_norm": 1.990972677377257,
      "learning_rate": 7.384279391134604e-07,
      "loss": 0.2454,
      "step": 38610
    },
    {
      "epoch": 2.0585805282375205,
      "grad_norm": 1.7928884933095712,
      "learning_rate": 7.376687390612482e-07,
      "loss": 0.2303,
      "step": 38620
    },
    {
      "epoch": 2.059113563071347,
      "grad_norm": 1.8264531507931925,
      "learning_rate": 7.369098022103624e-07,
      "loss": 0.2283,
      "step": 38630
    },
    {
      "epoch": 2.0596465979051732,
      "grad_norm": 1.7198716199600403,
      "learning_rate": 7.361511288228332e-07,
      "loss": 0.2407,
      "step": 38640
    },
    {
      "epoch": 2.0601796327389996,
      "grad_norm": 1.696736359376205,
      "learning_rate": 7.353927191605979e-07,
      "loss": 0.2253,
      "step": 38650
    },
    {
      "epoch": 2.060712667572826,
      "grad_norm": 1.8064393807227752,
      "learning_rate": 7.346345734855048e-07,
      "loss": 0.2336,
      "step": 38660
    },
    {
      "epoch": 2.0612457024066524,
      "grad_norm": 1.8270088564409517,
      "learning_rate": 7.338766920593093e-07,
      "loss": 0.2327,
      "step": 38670
    },
    {
      "epoch": 2.0617787372404788,
      "grad_norm": 1.710927842359247,
      "learning_rate": 7.331190751436772e-07,
      "loss": 0.2321,
      "step": 38680
    },
    {
      "epoch": 2.062311772074305,
      "grad_norm": 2.0278901653609522,
      "learning_rate": 7.323617230001815e-07,
      "loss": 0.2318,
      "step": 38690
    },
    {
      "epoch": 2.0628448069081315,
      "grad_norm": 1.7364311814129412,
      "learning_rate": 7.31604635890305e-07,
      "loss": 0.2412,
      "step": 38700
    },
    {
      "epoch": 2.063377841741958,
      "grad_norm": 1.8373109926965532,
      "learning_rate": 7.308478140754381e-07,
      "loss": 0.2343,
      "step": 38710
    },
    {
      "epoch": 2.0639108765757843,
      "grad_norm": 1.8208564095524293,
      "learning_rate": 7.300912578168801e-07,
      "loss": 0.2329,
      "step": 38720
    },
    {
      "epoch": 2.0644439114096107,
      "grad_norm": 1.8890569820637582,
      "learning_rate": 7.293349673758388e-07,
      "loss": 0.2384,
      "step": 38730
    },
    {
      "epoch": 2.064976946243437,
      "grad_norm": 1.6832413770974157,
      "learning_rate": 7.285789430134291e-07,
      "loss": 0.2311,
      "step": 38740
    },
    {
      "epoch": 2.0655099810772635,
      "grad_norm": 1.8752553035569006,
      "learning_rate": 7.278231849906751e-07,
      "loss": 0.2371,
      "step": 38750
    },
    {
      "epoch": 2.06604301591109,
      "grad_norm": 1.9815479545309946,
      "learning_rate": 7.270676935685087e-07,
      "loss": 0.2347,
      "step": 38760
    },
    {
      "epoch": 2.0665760507449162,
      "grad_norm": 1.6786789066714622,
      "learning_rate": 7.263124690077698e-07,
      "loss": 0.2331,
      "step": 38770
    },
    {
      "epoch": 2.0671090855787426,
      "grad_norm": 1.982227124213622,
      "learning_rate": 7.25557511569206e-07,
      "loss": 0.2314,
      "step": 38780
    },
    {
      "epoch": 2.067642120412569,
      "grad_norm": 1.6374566453130897,
      "learning_rate": 7.24802821513473e-07,
      "loss": 0.2331,
      "step": 38790
    },
    {
      "epoch": 2.0681751552463954,
      "grad_norm": 1.7015005455337817,
      "learning_rate": 7.240483991011336e-07,
      "loss": 0.239,
      "step": 38800
    },
    {
      "epoch": 2.0687081900802218,
      "grad_norm": 1.976667169465564,
      "learning_rate": 7.232942445926591e-07,
      "loss": 0.2346,
      "step": 38810
    },
    {
      "epoch": 2.069241224914048,
      "grad_norm": 1.874765663110959,
      "learning_rate": 7.225403582484269e-07,
      "loss": 0.2387,
      "step": 38820
    },
    {
      "epoch": 2.0697742597478745,
      "grad_norm": 1.898323939078004,
      "learning_rate": 7.21786740328723e-07,
      "loss": 0.2407,
      "step": 38830
    },
    {
      "epoch": 2.070307294581701,
      "grad_norm": 1.9122054892991258,
      "learning_rate": 7.210333910937404e-07,
      "loss": 0.2412,
      "step": 38840
    },
    {
      "epoch": 2.0708403294155273,
      "grad_norm": 1.6813385214142111,
      "learning_rate": 7.202803108035792e-07,
      "loss": 0.2412,
      "step": 38850
    },
    {
      "epoch": 2.0713733642493537,
      "grad_norm": 1.7089315268905128,
      "learning_rate": 7.195274997182471e-07,
      "loss": 0.2339,
      "step": 38860
    },
    {
      "epoch": 2.07190639908318,
      "grad_norm": 1.8119356272440994,
      "learning_rate": 7.187749580976582e-07,
      "loss": 0.2397,
      "step": 38870
    },
    {
      "epoch": 2.0724394339170065,
      "grad_norm": 1.756445736679096,
      "learning_rate": 7.180226862016338e-07,
      "loss": 0.2329,
      "step": 38880
    },
    {
      "epoch": 2.072972468750833,
      "grad_norm": 1.7488611644124183,
      "learning_rate": 7.172706842899025e-07,
      "loss": 0.2323,
      "step": 38890
    },
    {
      "epoch": 2.0735055035846592,
      "grad_norm": 1.9013774811670907,
      "learning_rate": 7.165189526220996e-07,
      "loss": 0.2296,
      "step": 38900
    },
    {
      "epoch": 2.0740385384184856,
      "grad_norm": 1.7687502529146126,
      "learning_rate": 7.157674914577656e-07,
      "loss": 0.2364,
      "step": 38910
    },
    {
      "epoch": 2.074571573252312,
      "grad_norm": 1.914945552584631,
      "learning_rate": 7.1501630105635e-07,
      "loss": 0.2297,
      "step": 38920
    },
    {
      "epoch": 2.0751046080861384,
      "grad_norm": 1.7242640875557604,
      "learning_rate": 7.142653816772069e-07,
      "loss": 0.2312,
      "step": 38930
    },
    {
      "epoch": 2.0756376429199648,
      "grad_norm": 1.9920917647403975,
      "learning_rate": 7.135147335795981e-07,
      "loss": 0.2366,
      "step": 38940
    },
    {
      "epoch": 2.076170677753791,
      "grad_norm": 2.0064009548889574,
      "learning_rate": 7.127643570226908e-07,
      "loss": 0.2363,
      "step": 38950
    },
    {
      "epoch": 2.0767037125876175,
      "grad_norm": 2.010494348160478,
      "learning_rate": 7.120142522655593e-07,
      "loss": 0.2374,
      "step": 38960
    },
    {
      "epoch": 2.077236747421444,
      "grad_norm": 1.9675213853990199,
      "learning_rate": 7.112644195671834e-07,
      "loss": 0.238,
      "step": 38970
    },
    {
      "epoch": 2.0777697822552703,
      "grad_norm": 1.773585968076842,
      "learning_rate": 7.105148591864497e-07,
      "loss": 0.2344,
      "step": 38980
    },
    {
      "epoch": 2.0783028170890967,
      "grad_norm": 1.7435585385293508,
      "learning_rate": 7.097655713821492e-07,
      "loss": 0.2263,
      "step": 38990
    },
    {
      "epoch": 2.078835851922923,
      "grad_norm": 1.903864032389861,
      "learning_rate": 7.090165564129805e-07,
      "loss": 0.2351,
      "step": 39000
    },
    {
      "epoch": 2.0793688867567495,
      "grad_norm": 1.7218540313410287,
      "learning_rate": 7.082678145375472e-07,
      "loss": 0.2317,
      "step": 39010
    },
    {
      "epoch": 2.079901921590576,
      "grad_norm": 1.887312086373205,
      "learning_rate": 7.07519346014359e-07,
      "loss": 0.2364,
      "step": 39020
    },
    {
      "epoch": 2.0804349564244022,
      "grad_norm": 2.161946878157882,
      "learning_rate": 7.06771151101831e-07,
      "loss": 0.2406,
      "step": 39030
    },
    {
      "epoch": 2.0809679912582286,
      "grad_norm": 1.952869762054716,
      "learning_rate": 7.060232300582835e-07,
      "loss": 0.2406,
      "step": 39040
    },
    {
      "epoch": 2.081501026092055,
      "grad_norm": 1.7228380440241955,
      "learning_rate": 7.052755831419428e-07,
      "loss": 0.2173,
      "step": 39050
    },
    {
      "epoch": 2.0820340609258814,
      "grad_norm": 1.7325922473151485,
      "learning_rate": 7.045282106109404e-07,
      "loss": 0.2373,
      "step": 39060
    },
    {
      "epoch": 2.0825670957597078,
      "grad_norm": 1.7394020756605846,
      "learning_rate": 7.037811127233134e-07,
      "loss": 0.2373,
      "step": 39070
    },
    {
      "epoch": 2.083100130593534,
      "grad_norm": 1.718151430492054,
      "learning_rate": 7.030342897370028e-07,
      "loss": 0.2447,
      "step": 39080
    },
    {
      "epoch": 2.0836331654273605,
      "grad_norm": 2.020804137280149,
      "learning_rate": 7.022877419098558e-07,
      "loss": 0.2305,
      "step": 39090
    },
    {
      "epoch": 2.084166200261187,
      "grad_norm": 1.8341247962630196,
      "learning_rate": 7.015414694996247e-07,
      "loss": 0.2306,
      "step": 39100
    },
    {
      "epoch": 2.0846992350950133,
      "grad_norm": 1.770468997248963,
      "learning_rate": 7.00795472763966e-07,
      "loss": 0.2268,
      "step": 39110
    },
    {
      "epoch": 2.0852322699288397,
      "grad_norm": 1.775501627589431,
      "learning_rate": 7.000497519604418e-07,
      "loss": 0.237,
      "step": 39120
    },
    {
      "epoch": 2.085765304762666,
      "grad_norm": 1.7182050160947544,
      "learning_rate": 6.993043073465181e-07,
      "loss": 0.2338,
      "step": 39130
    },
    {
      "epoch": 2.0862983395964925,
      "grad_norm": 2.0157672548255756,
      "learning_rate": 6.985591391795663e-07,
      "loss": 0.2262,
      "step": 39140
    },
    {
      "epoch": 2.086831374430319,
      "grad_norm": 1.7719185933431802,
      "learning_rate": 6.978142477168624e-07,
      "loss": 0.2347,
      "step": 39150
    },
    {
      "epoch": 2.0873644092641452,
      "grad_norm": 1.8986993285238174,
      "learning_rate": 6.970696332155855e-07,
      "loss": 0.2311,
      "step": 39160
    },
    {
      "epoch": 2.0878974440979716,
      "grad_norm": 1.8527327230404613,
      "learning_rate": 6.963252959328205e-07,
      "loss": 0.2354,
      "step": 39170
    },
    {
      "epoch": 2.088430478931798,
      "grad_norm": 1.754908281459408,
      "learning_rate": 6.955812361255563e-07,
      "loss": 0.2317,
      "step": 39180
    },
    {
      "epoch": 2.0889635137656244,
      "grad_norm": 1.784064436989912,
      "learning_rate": 6.948374540506856e-07,
      "loss": 0.2313,
      "step": 39190
    },
    {
      "epoch": 2.0894965485994508,
      "grad_norm": 1.8567877732530396,
      "learning_rate": 6.940939499650058e-07,
      "loss": 0.2377,
      "step": 39200
    },
    {
      "epoch": 2.090029583433277,
      "grad_norm": 1.7152402859454083,
      "learning_rate": 6.933507241252179e-07,
      "loss": 0.2363,
      "step": 39210
    },
    {
      "epoch": 2.0905626182671035,
      "grad_norm": 1.9249967281226645,
      "learning_rate": 6.926077767879267e-07,
      "loss": 0.2361,
      "step": 39220
    },
    {
      "epoch": 2.09109565310093,
      "grad_norm": 1.80540934872832,
      "learning_rate": 6.918651082096411e-07,
      "loss": 0.2308,
      "step": 39230
    },
    {
      "epoch": 2.0916286879347563,
      "grad_norm": 1.787963323345176,
      "learning_rate": 6.911227186467745e-07,
      "loss": 0.2335,
      "step": 39240
    },
    {
      "epoch": 2.0921617227685827,
      "grad_norm": 1.8827757615057787,
      "learning_rate": 6.903806083556418e-07,
      "loss": 0.2293,
      "step": 39250
    },
    {
      "epoch": 2.0926947576024095,
      "grad_norm": 2.063808696381716,
      "learning_rate": 6.896387775924637e-07,
      "loss": 0.2326,
      "step": 39260
    },
    {
      "epoch": 2.093227792436236,
      "grad_norm": 1.6968406597366485,
      "learning_rate": 6.888972266133634e-07,
      "loss": 0.2269,
      "step": 39270
    },
    {
      "epoch": 2.0937608272700623,
      "grad_norm": 1.9151628842902149,
      "learning_rate": 6.881559556743676e-07,
      "loss": 0.2318,
      "step": 39280
    },
    {
      "epoch": 2.0942938621038887,
      "grad_norm": 2.0935356230176065,
      "learning_rate": 6.874149650314067e-07,
      "loss": 0.2414,
      "step": 39290
    },
    {
      "epoch": 2.094826896937715,
      "grad_norm": 1.7526412945450405,
      "learning_rate": 6.866742549403135e-07,
      "loss": 0.2294,
      "step": 39300
    },
    {
      "epoch": 2.0953599317715415,
      "grad_norm": 2.1888986022680723,
      "learning_rate": 6.859338256568248e-07,
      "loss": 0.2325,
      "step": 39310
    },
    {
      "epoch": 2.095892966605368,
      "grad_norm": 2.0482125792443093,
      "learning_rate": 6.851936774365805e-07,
      "loss": 0.2349,
      "step": 39320
    },
    {
      "epoch": 2.096426001439194,
      "grad_norm": 1.7835832046778732,
      "learning_rate": 6.84453810535122e-07,
      "loss": 0.2338,
      "step": 39330
    },
    {
      "epoch": 2.0969590362730206,
      "grad_norm": 1.7853669186881789,
      "learning_rate": 6.837142252078956e-07,
      "loss": 0.2413,
      "step": 39340
    },
    {
      "epoch": 2.097492071106847,
      "grad_norm": 1.6889132930047615,
      "learning_rate": 6.829749217102484e-07,
      "loss": 0.2245,
      "step": 39350
    },
    {
      "epoch": 2.0980251059406734,
      "grad_norm": 1.8077248761450484,
      "learning_rate": 6.822359002974317e-07,
      "loss": 0.2399,
      "step": 39360
    },
    {
      "epoch": 2.0985581407744998,
      "grad_norm": 1.826672107967957,
      "learning_rate": 6.81497161224599e-07,
      "loss": 0.2297,
      "step": 39370
    },
    {
      "epoch": 2.099091175608326,
      "grad_norm": 1.658928802150467,
      "learning_rate": 6.807587047468062e-07,
      "loss": 0.2322,
      "step": 39380
    },
    {
      "epoch": 2.0996242104421525,
      "grad_norm": 1.8493166044316842,
      "learning_rate": 6.800205311190115e-07,
      "loss": 0.2325,
      "step": 39390
    },
    {
      "epoch": 2.100157245275979,
      "grad_norm": 1.7411284344819427,
      "learning_rate": 6.792826405960756e-07,
      "loss": 0.2369,
      "step": 39400
    },
    {
      "epoch": 2.1006902801098053,
      "grad_norm": 1.785820068285124,
      "learning_rate": 6.785450334327618e-07,
      "loss": 0.2401,
      "step": 39410
    },
    {
      "epoch": 2.1012233149436317,
      "grad_norm": 1.7916893581829731,
      "learning_rate": 6.778077098837354e-07,
      "loss": 0.2282,
      "step": 39420
    },
    {
      "epoch": 2.101756349777458,
      "grad_norm": 1.7425392132653401,
      "learning_rate": 6.770706702035626e-07,
      "loss": 0.2387,
      "step": 39430
    },
    {
      "epoch": 2.1022893846112845,
      "grad_norm": 1.892042111603105,
      "learning_rate": 6.763339146467133e-07,
      "loss": 0.2286,
      "step": 39440
    },
    {
      "epoch": 2.102822419445111,
      "grad_norm": 1.930893107842331,
      "learning_rate": 6.755974434675585e-07,
      "loss": 0.2335,
      "step": 39450
    },
    {
      "epoch": 2.103355454278937,
      "grad_norm": 1.897360206669216,
      "learning_rate": 6.74861256920371e-07,
      "loss": 0.2344,
      "step": 39460
    },
    {
      "epoch": 2.1038884891127636,
      "grad_norm": 1.7337369083503635,
      "learning_rate": 6.741253552593255e-07,
      "loss": 0.224,
      "step": 39470
    },
    {
      "epoch": 2.10442152394659,
      "grad_norm": 1.6695623874313024,
      "learning_rate": 6.733897387384985e-07,
      "loss": 0.2239,
      "step": 39480
    },
    {
      "epoch": 2.1049545587804164,
      "grad_norm": 1.6197288698777523,
      "learning_rate": 6.726544076118675e-07,
      "loss": 0.2347,
      "step": 39490
    },
    {
      "epoch": 2.1054875936142428,
      "grad_norm": 1.8015298501101642,
      "learning_rate": 6.71919362133312e-07,
      "loss": 0.2323,
      "step": 39500
    },
    {
      "epoch": 2.106020628448069,
      "grad_norm": 1.8076937875318893,
      "learning_rate": 6.711846025566131e-07,
      "loss": 0.2412,
      "step": 39510
    },
    {
      "epoch": 2.1065536632818955,
      "grad_norm": 1.6994128341693702,
      "learning_rate": 6.704501291354519e-07,
      "loss": 0.2324,
      "step": 39520
    },
    {
      "epoch": 2.107086698115722,
      "grad_norm": 1.8953112550396436,
      "learning_rate": 6.697159421234119e-07,
      "loss": 0.2322,
      "step": 39530
    },
    {
      "epoch": 2.1076197329495483,
      "grad_norm": 1.7324111977798289,
      "learning_rate": 6.689820417739776e-07,
      "loss": 0.2363,
      "step": 39540
    },
    {
      "epoch": 2.1081527677833747,
      "grad_norm": 1.7673454143031564,
      "learning_rate": 6.68248428340534e-07,
      "loss": 0.2358,
      "step": 39550
    },
    {
      "epoch": 2.108685802617201,
      "grad_norm": 1.7579946248665175,
      "learning_rate": 6.675151020763675e-07,
      "loss": 0.232,
      "step": 39560
    },
    {
      "epoch": 2.1092188374510275,
      "grad_norm": 1.8733235410375826,
      "learning_rate": 6.667820632346654e-07,
      "loss": 0.2215,
      "step": 39570
    },
    {
      "epoch": 2.109751872284854,
      "grad_norm": 1.9351318740912227,
      "learning_rate": 6.660493120685153e-07,
      "loss": 0.23,
      "step": 39580
    },
    {
      "epoch": 2.11028490711868,
      "grad_norm": 1.7621694950452877,
      "learning_rate": 6.653168488309065e-07,
      "loss": 0.2415,
      "step": 39590
    },
    {
      "epoch": 2.1108179419525066,
      "grad_norm": 1.733313493110517,
      "learning_rate": 6.645846737747269e-07,
      "loss": 0.2387,
      "step": 39600
    },
    {
      "epoch": 2.111350976786333,
      "grad_norm": 1.9524768422621142,
      "learning_rate": 6.638527871527668e-07,
      "loss": 0.237,
      "step": 39610
    },
    {
      "epoch": 2.1118840116201594,
      "grad_norm": 1.6983086570374055,
      "learning_rate": 6.63121189217716e-07,
      "loss": 0.2421,
      "step": 39620
    },
    {
      "epoch": 2.1124170464539858,
      "grad_norm": 1.852234657445964,
      "learning_rate": 6.623898802221653e-07,
      "loss": 0.2321,
      "step": 39630
    },
    {
      "epoch": 2.112950081287812,
      "grad_norm": 1.8625536359595851,
      "learning_rate": 6.61658860418605e-07,
      "loss": 0.2387,
      "step": 39640
    },
    {
      "epoch": 2.1134831161216385,
      "grad_norm": 1.6573335182186442,
      "learning_rate": 6.60928130059426e-07,
      "loss": 0.2258,
      "step": 39650
    },
    {
      "epoch": 2.114016150955465,
      "grad_norm": 1.6754698562016315,
      "learning_rate": 6.60197689396919e-07,
      "loss": 0.2468,
      "step": 39660
    },
    {
      "epoch": 2.1145491857892913,
      "grad_norm": 1.7886504126309044,
      "learning_rate": 6.594675386832752e-07,
      "loss": 0.2375,
      "step": 39670
    },
    {
      "epoch": 2.1150822206231177,
      "grad_norm": 1.644425466862984,
      "learning_rate": 6.587376781705855e-07,
      "loss": 0.2383,
      "step": 39680
    },
    {
      "epoch": 2.115615255456944,
      "grad_norm": 1.7138040935780128,
      "learning_rate": 6.580081081108395e-07,
      "loss": 0.2316,
      "step": 39690
    },
    {
      "epoch": 2.1161482902907705,
      "grad_norm": 1.8848922940027197,
      "learning_rate": 6.572788287559279e-07,
      "loss": 0.2352,
      "step": 39700
    },
    {
      "epoch": 2.116681325124597,
      "grad_norm": 1.8558868944049343,
      "learning_rate": 6.565498403576406e-07,
      "loss": 0.2322,
      "step": 39710
    },
    {
      "epoch": 2.117214359958423,
      "grad_norm": 1.9993093461834224,
      "learning_rate": 6.558211431676672e-07,
      "loss": 0.2413,
      "step": 39720
    },
    {
      "epoch": 2.1177473947922496,
      "grad_norm": 1.9946649051848502,
      "learning_rate": 6.550927374375964e-07,
      "loss": 0.2315,
      "step": 39730
    },
    {
      "epoch": 2.118280429626076,
      "grad_norm": 2.109337538082504,
      "learning_rate": 6.543646234189164e-07,
      "loss": 0.2308,
      "step": 39740
    },
    {
      "epoch": 2.1188134644599024,
      "grad_norm": 1.8380403322089125,
      "learning_rate": 6.536368013630148e-07,
      "loss": 0.2349,
      "step": 39750
    },
    {
      "epoch": 2.1193464992937288,
      "grad_norm": 1.9644611345038752,
      "learning_rate": 6.529092715211788e-07,
      "loss": 0.2278,
      "step": 39760
    },
    {
      "epoch": 2.119879534127555,
      "grad_norm": 1.9569142981011818,
      "learning_rate": 6.521820341445934e-07,
      "loss": 0.2415,
      "step": 39770
    },
    {
      "epoch": 2.1204125689613815,
      "grad_norm": 1.8575414784827466,
      "learning_rate": 6.514550894843439e-07,
      "loss": 0.2295,
      "step": 39780
    },
    {
      "epoch": 2.120945603795208,
      "grad_norm": 1.8877040316927678,
      "learning_rate": 6.50728437791414e-07,
      "loss": 0.2403,
      "step": 39790
    },
    {
      "epoch": 2.1214786386290343,
      "grad_norm": 1.6609374969056916,
      "learning_rate": 6.500020793166864e-07,
      "loss": 0.2335,
      "step": 39800
    },
    {
      "epoch": 2.1220116734628607,
      "grad_norm": 1.8754099560377642,
      "learning_rate": 6.492760143109424e-07,
      "loss": 0.2267,
      "step": 39810
    },
    {
      "epoch": 2.122544708296687,
      "grad_norm": 1.8831123672323284,
      "learning_rate": 6.485502430248623e-07,
      "loss": 0.2437,
      "step": 39820
    },
    {
      "epoch": 2.1230777431305135,
      "grad_norm": 1.9246693516294675,
      "learning_rate": 6.478247657090246e-07,
      "loss": 0.2346,
      "step": 39830
    },
    {
      "epoch": 2.12361077796434,
      "grad_norm": 1.9532938944496068,
      "learning_rate": 6.470995826139068e-07,
      "loss": 0.2431,
      "step": 39840
    },
    {
      "epoch": 2.124143812798166,
      "grad_norm": 2.015474153171191,
      "learning_rate": 6.463746939898849e-07,
      "loss": 0.2342,
      "step": 39850
    },
    {
      "epoch": 2.1246768476319926,
      "grad_norm": 1.8705196777189506,
      "learning_rate": 6.456501000872315e-07,
      "loss": 0.2316,
      "step": 39860
    },
    {
      "epoch": 2.125209882465819,
      "grad_norm": 1.908588922888013,
      "learning_rate": 6.449258011561196e-07,
      "loss": 0.2326,
      "step": 39870
    },
    {
      "epoch": 2.1257429172996454,
      "grad_norm": 1.8966147010270864,
      "learning_rate": 6.442017974466194e-07,
      "loss": 0.2322,
      "step": 39880
    },
    {
      "epoch": 2.1262759521334718,
      "grad_norm": 1.772780422289616,
      "learning_rate": 6.434780892086995e-07,
      "loss": 0.2336,
      "step": 39890
    },
    {
      "epoch": 2.126808986967298,
      "grad_norm": 1.7161525186765545,
      "learning_rate": 6.42754676692226e-07,
      "loss": 0.2383,
      "step": 39900
    },
    {
      "epoch": 2.1273420218011245,
      "grad_norm": 2.0559582308105133,
      "learning_rate": 6.420315601469636e-07,
      "loss": 0.2454,
      "step": 39910
    },
    {
      "epoch": 2.127875056634951,
      "grad_norm": 1.700316637880466,
      "learning_rate": 6.41308739822574e-07,
      "loss": 0.2426,
      "step": 39920
    },
    {
      "epoch": 2.1284080914687773,
      "grad_norm": 1.8919667007719585,
      "learning_rate": 6.40586215968618e-07,
      "loss": 0.235,
      "step": 39930
    },
    {
      "epoch": 2.1289411263026037,
      "grad_norm": 1.9327042429255488,
      "learning_rate": 6.398639888345515e-07,
      "loss": 0.2327,
      "step": 39940
    },
    {
      "epoch": 2.12947416113643,
      "grad_norm": 1.9661138371919133,
      "learning_rate": 6.391420586697306e-07,
      "loss": 0.2341,
      "step": 39950
    },
    {
      "epoch": 2.1300071959702565,
      "grad_norm": 1.9013372332012386,
      "learning_rate": 6.384204257234075e-07,
      "loss": 0.2371,
      "step": 39960
    },
    {
      "epoch": 2.1305402308040833,
      "grad_norm": 1.9757025184523238,
      "learning_rate": 6.376990902447321e-07,
      "loss": 0.2279,
      "step": 39970
    },
    {
      "epoch": 2.1310732656379097,
      "grad_norm": 1.8255977715193223,
      "learning_rate": 6.369780524827517e-07,
      "loss": 0.2333,
      "step": 39980
    },
    {
      "epoch": 2.131606300471736,
      "grad_norm": 2.207393599915795,
      "learning_rate": 6.362573126864106e-07,
      "loss": 0.2375,
      "step": 39990
    },
    {
      "epoch": 2.1321393353055624,
      "grad_norm": 1.6367167486604572,
      "learning_rate": 6.355368711045504e-07,
      "loss": 0.2297,
      "step": 40000
    },
    {
      "epoch": 2.132672370139389,
      "grad_norm": 1.8348815244864012,
      "learning_rate": 6.348167279859096e-07,
      "loss": 0.2343,
      "step": 40010
    },
    {
      "epoch": 2.133205404973215,
      "grad_norm": 1.851807801989843,
      "learning_rate": 6.340968835791241e-07,
      "loss": 0.2442,
      "step": 40020
    },
    {
      "epoch": 2.1337384398070416,
      "grad_norm": 1.736533017645107,
      "learning_rate": 6.333773381327256e-07,
      "loss": 0.2408,
      "step": 40030
    },
    {
      "epoch": 2.134271474640868,
      "grad_norm": 1.680825940291369,
      "learning_rate": 6.326580918951434e-07,
      "loss": 0.2449,
      "step": 40040
    },
    {
      "epoch": 2.1348045094746944,
      "grad_norm": 1.7474711548824875,
      "learning_rate": 6.319391451147038e-07,
      "loss": 0.2363,
      "step": 40050
    },
    {
      "epoch": 2.1353375443085207,
      "grad_norm": 1.6615034788368437,
      "learning_rate": 6.312204980396287e-07,
      "loss": 0.2373,
      "step": 40060
    },
    {
      "epoch": 2.135870579142347,
      "grad_norm": 1.944319080986197,
      "learning_rate": 6.305021509180375e-07,
      "loss": 0.2341,
      "step": 40070
    },
    {
      "epoch": 2.1364036139761735,
      "grad_norm": 1.7130983414424126,
      "learning_rate": 6.297841039979454e-07,
      "loss": 0.2464,
      "step": 40080
    },
    {
      "epoch": 2.13693664881,
      "grad_norm": 1.827286043198682,
      "learning_rate": 6.290663575272645e-07,
      "loss": 0.2315,
      "step": 40090
    },
    {
      "epoch": 2.1374696836438263,
      "grad_norm": 1.908884073171793,
      "learning_rate": 6.283489117538031e-07,
      "loss": 0.2381,
      "step": 40100
    },
    {
      "epoch": 2.1380027184776527,
      "grad_norm": 1.8864710073623185,
      "learning_rate": 6.276317669252645e-07,
      "loss": 0.2413,
      "step": 40110
    },
    {
      "epoch": 2.138535753311479,
      "grad_norm": 1.7561293850111093,
      "learning_rate": 6.269149232892495e-07,
      "loss": 0.2461,
      "step": 40120
    },
    {
      "epoch": 2.1390687881453054,
      "grad_norm": 1.8141178218641116,
      "learning_rate": 6.261983810932543e-07,
      "loss": 0.2332,
      "step": 40130
    },
    {
      "epoch": 2.139601822979132,
      "grad_norm": 1.7544498510912303,
      "learning_rate": 6.254821405846715e-07,
      "loss": 0.234,
      "step": 40140
    },
    {
      "epoch": 2.140134857812958,
      "grad_norm": 1.947168833596968,
      "learning_rate": 6.24766202010789e-07,
      "loss": 0.2308,
      "step": 40150
    },
    {
      "epoch": 2.1406678926467846,
      "grad_norm": 2.233100781585162,
      "learning_rate": 6.240505656187905e-07,
      "loss": 0.2314,
      "step": 40160
    },
    {
      "epoch": 2.141200927480611,
      "grad_norm": 1.8733176243835672,
      "learning_rate": 6.233352316557558e-07,
      "loss": 0.2354,
      "step": 40170
    },
    {
      "epoch": 2.1417339623144374,
      "grad_norm": 1.8502519911588187,
      "learning_rate": 6.226202003686601e-07,
      "loss": 0.2355,
      "step": 40180
    },
    {
      "epoch": 2.1422669971482637,
      "grad_norm": 1.7848900256793836,
      "learning_rate": 6.21905472004374e-07,
      "loss": 0.237,
      "step": 40190
    },
    {
      "epoch": 2.14280003198209,
      "grad_norm": 1.8951635801878093,
      "learning_rate": 6.211910468096631e-07,
      "loss": 0.2362,
      "step": 40200
    },
    {
      "epoch": 2.1433330668159165,
      "grad_norm": 1.7006800035725336,
      "learning_rate": 6.204769250311889e-07,
      "loss": 0.2244,
      "step": 40210
    },
    {
      "epoch": 2.143866101649743,
      "grad_norm": 1.7212870778421754,
      "learning_rate": 6.197631069155081e-07,
      "loss": 0.2429,
      "step": 40220
    },
    {
      "epoch": 2.1443991364835693,
      "grad_norm": 1.7735006585455204,
      "learning_rate": 6.190495927090725e-07,
      "loss": 0.231,
      "step": 40230
    },
    {
      "epoch": 2.1449321713173957,
      "grad_norm": 1.8210038930716173,
      "learning_rate": 6.183363826582287e-07,
      "loss": 0.245,
      "step": 40240
    },
    {
      "epoch": 2.145465206151222,
      "grad_norm": 1.9251286014134974,
      "learning_rate": 6.176234770092187e-07,
      "loss": 0.2342,
      "step": 40250
    },
    {
      "epoch": 2.1459982409850484,
      "grad_norm": 1.9128290810427413,
      "learning_rate": 6.169108760081794e-07,
      "loss": 0.2363,
      "step": 40260
    },
    {
      "epoch": 2.146531275818875,
      "grad_norm": 1.7790798262436074,
      "learning_rate": 6.161985799011426e-07,
      "loss": 0.2394,
      "step": 40270
    },
    {
      "epoch": 2.147064310652701,
      "grad_norm": 1.8445577079913038,
      "learning_rate": 6.154865889340334e-07,
      "loss": 0.23,
      "step": 40280
    },
    {
      "epoch": 2.1475973454865276,
      "grad_norm": 1.9163465124993924,
      "learning_rate": 6.147749033526736e-07,
      "loss": 0.2349,
      "step": 40290
    },
    {
      "epoch": 2.148130380320354,
      "grad_norm": 2.0614654944564856,
      "learning_rate": 6.140635234027785e-07,
      "loss": 0.2337,
      "step": 40300
    },
    {
      "epoch": 2.1486634151541804,
      "grad_norm": 1.8389780451693138,
      "learning_rate": 6.133524493299579e-07,
      "loss": 0.2339,
      "step": 40310
    },
    {
      "epoch": 2.1491964499880067,
      "grad_norm": 2.008779633227989,
      "learning_rate": 6.126416813797164e-07,
      "loss": 0.2438,
      "step": 40320
    },
    {
      "epoch": 2.149729484821833,
      "grad_norm": 1.706666831851398,
      "learning_rate": 6.119312197974525e-07,
      "loss": 0.2298,
      "step": 40330
    },
    {
      "epoch": 2.1502625196556595,
      "grad_norm": 2.0249086731345045,
      "learning_rate": 6.112210648284591e-07,
      "loss": 0.2405,
      "step": 40340
    },
    {
      "epoch": 2.150795554489486,
      "grad_norm": 1.9574147972801035,
      "learning_rate": 6.105112167179234e-07,
      "loss": 0.2356,
      "step": 40350
    },
    {
      "epoch": 2.1513285893233123,
      "grad_norm": 1.8816226527137219,
      "learning_rate": 6.098016757109266e-07,
      "loss": 0.2346,
      "step": 40360
    },
    {
      "epoch": 2.1518616241571387,
      "grad_norm": 1.837837352088897,
      "learning_rate": 6.090924420524431e-07,
      "loss": 0.2416,
      "step": 40370
    },
    {
      "epoch": 2.152394658990965,
      "grad_norm": 1.7478395417047299,
      "learning_rate": 6.083835159873421e-07,
      "loss": 0.2419,
      "step": 40380
    },
    {
      "epoch": 2.1529276938247914,
      "grad_norm": 1.8591808472201339,
      "learning_rate": 6.076748977603866e-07,
      "loss": 0.2391,
      "step": 40390
    },
    {
      "epoch": 2.153460728658618,
      "grad_norm": 1.9222457471360213,
      "learning_rate": 6.069665876162327e-07,
      "loss": 0.2329,
      "step": 40400
    },
    {
      "epoch": 2.153993763492444,
      "grad_norm": 1.9202576321301945,
      "learning_rate": 6.062585857994308e-07,
      "loss": 0.2184,
      "step": 40410
    },
    {
      "epoch": 2.1545267983262706,
      "grad_norm": 1.830743858035785,
      "learning_rate": 6.055508925544243e-07,
      "loss": 0.241,
      "step": 40420
    },
    {
      "epoch": 2.155059833160097,
      "grad_norm": 1.9105412717135142,
      "learning_rate": 6.048435081255504e-07,
      "loss": 0.2347,
      "step": 40430
    },
    {
      "epoch": 2.1555928679939234,
      "grad_norm": 1.6642688016971123,
      "learning_rate": 6.0413643275704e-07,
      "loss": 0.2271,
      "step": 40440
    },
    {
      "epoch": 2.1561259028277497,
      "grad_norm": 1.623355846812178,
      "learning_rate": 6.034296666930158e-07,
      "loss": 0.2431,
      "step": 40450
    },
    {
      "epoch": 2.156658937661576,
      "grad_norm": 1.701916761226239,
      "learning_rate": 6.027232101774955e-07,
      "loss": 0.2395,
      "step": 40460
    },
    {
      "epoch": 2.1571919724954025,
      "grad_norm": 1.6361426929807774,
      "learning_rate": 6.020170634543891e-07,
      "loss": 0.2319,
      "step": 40470
    },
    {
      "epoch": 2.157725007329229,
      "grad_norm": 1.8521042090478255,
      "learning_rate": 6.013112267674995e-07,
      "loss": 0.232,
      "step": 40480
    },
    {
      "epoch": 2.1582580421630553,
      "grad_norm": 1.9075258483782722,
      "learning_rate": 6.006057003605232e-07,
      "loss": 0.2477,
      "step": 40490
    },
    {
      "epoch": 2.1587910769968817,
      "grad_norm": 1.7874511827360193,
      "learning_rate": 5.999004844770489e-07,
      "loss": 0.2352,
      "step": 40500
    },
    {
      "epoch": 2.159324111830708,
      "grad_norm": 1.7542222965378522,
      "learning_rate": 5.991955793605585e-07,
      "loss": 0.2334,
      "step": 40510
    },
    {
      "epoch": 2.1598571466645344,
      "grad_norm": 1.8406409662779866,
      "learning_rate": 5.984909852544269e-07,
      "loss": 0.2305,
      "step": 40520
    },
    {
      "epoch": 2.160390181498361,
      "grad_norm": 2.2688464626274043,
      "learning_rate": 5.9778670240192e-07,
      "loss": 0.2372,
      "step": 40530
    },
    {
      "epoch": 2.160923216332187,
      "grad_norm": 1.9001663884456879,
      "learning_rate": 5.970827310461982e-07,
      "loss": 0.2301,
      "step": 40540
    },
    {
      "epoch": 2.1614562511660136,
      "grad_norm": 1.832299857326013,
      "learning_rate": 5.963790714303135e-07,
      "loss": 0.2362,
      "step": 40550
    },
    {
      "epoch": 2.16198928599984,
      "grad_norm": 1.764636082158625,
      "learning_rate": 5.956757237972104e-07,
      "loss": 0.2283,
      "step": 40560
    },
    {
      "epoch": 2.1625223208336664,
      "grad_norm": 1.8422444657459665,
      "learning_rate": 5.949726883897254e-07,
      "loss": 0.2398,
      "step": 40570
    },
    {
      "epoch": 2.1630553556674927,
      "grad_norm": 2.0173068093101176,
      "learning_rate": 5.942699654505876e-07,
      "loss": 0.2484,
      "step": 40580
    },
    {
      "epoch": 2.163588390501319,
      "grad_norm": 1.8589623409673304,
      "learning_rate": 5.93567555222418e-07,
      "loss": 0.2323,
      "step": 40590
    },
    {
      "epoch": 2.1641214253351455,
      "grad_norm": 1.800587857582897,
      "learning_rate": 5.928654579477298e-07,
      "loss": 0.228,
      "step": 40600
    },
    {
      "epoch": 2.164654460168972,
      "grad_norm": 1.7929840766407423,
      "learning_rate": 5.921636738689284e-07,
      "loss": 0.2293,
      "step": 40610
    },
    {
      "epoch": 2.1651874950027983,
      "grad_norm": 1.8451060628561284,
      "learning_rate": 5.914622032283097e-07,
      "loss": 0.2358,
      "step": 40620
    },
    {
      "epoch": 2.1657205298366247,
      "grad_norm": 1.5921258544611028,
      "learning_rate": 5.90761046268063e-07,
      "loss": 0.228,
      "step": 40630
    },
    {
      "epoch": 2.166253564670451,
      "grad_norm": 1.7890295621623638,
      "learning_rate": 5.900602032302685e-07,
      "loss": 0.2353,
      "step": 40640
    },
    {
      "epoch": 2.1667865995042774,
      "grad_norm": 1.9393112504629715,
      "learning_rate": 5.893596743568985e-07,
      "loss": 0.2322,
      "step": 40650
    },
    {
      "epoch": 2.167319634338104,
      "grad_norm": 1.9181713578254165,
      "learning_rate": 5.886594598898161e-07,
      "loss": 0.2372,
      "step": 40660
    },
    {
      "epoch": 2.16785266917193,
      "grad_norm": 1.7836954206546396,
      "learning_rate": 5.879595600707768e-07,
      "loss": 0.2396,
      "step": 40670
    },
    {
      "epoch": 2.1683857040057566,
      "grad_norm": 1.9014215755651191,
      "learning_rate": 5.872599751414268e-07,
      "loss": 0.2392,
      "step": 40680
    },
    {
      "epoch": 2.168918738839583,
      "grad_norm": 1.794862747026264,
      "learning_rate": 5.865607053433032e-07,
      "loss": 0.2401,
      "step": 40690
    },
    {
      "epoch": 2.1694517736734094,
      "grad_norm": 1.7694071044096837,
      "learning_rate": 5.858617509178352e-07,
      "loss": 0.2308,
      "step": 40700
    },
    {
      "epoch": 2.1699848085072357,
      "grad_norm": 1.8194583706889857,
      "learning_rate": 5.851631121063433e-07,
      "loss": 0.236,
      "step": 40710
    },
    {
      "epoch": 2.170517843341062,
      "grad_norm": 1.8572292630540166,
      "learning_rate": 5.844647891500374e-07,
      "loss": 0.2385,
      "step": 40720
    },
    {
      "epoch": 2.1710508781748885,
      "grad_norm": 1.989948841682253,
      "learning_rate": 5.837667822900199e-07,
      "loss": 0.2378,
      "step": 40730
    },
    {
      "epoch": 2.171583913008715,
      "grad_norm": 1.967804161061831,
      "learning_rate": 5.830690917672835e-07,
      "loss": 0.2394,
      "step": 40740
    },
    {
      "epoch": 2.1721169478425413,
      "grad_norm": 1.963925464836842,
      "learning_rate": 5.82371717822712e-07,
      "loss": 0.2371,
      "step": 40750
    },
    {
      "epoch": 2.172649982676368,
      "grad_norm": 1.9148295542836957,
      "learning_rate": 5.816746606970795e-07,
      "loss": 0.235,
      "step": 40760
    },
    {
      "epoch": 2.1731830175101945,
      "grad_norm": 1.789270293579564,
      "learning_rate": 5.809779206310509e-07,
      "loss": 0.2279,
      "step": 40770
    },
    {
      "epoch": 2.173716052344021,
      "grad_norm": 1.9674705963494186,
      "learning_rate": 5.802814978651817e-07,
      "loss": 0.2296,
      "step": 40780
    },
    {
      "epoch": 2.1742490871778473,
      "grad_norm": 1.9664260384733157,
      "learning_rate": 5.795853926399175e-07,
      "loss": 0.2294,
      "step": 40790
    },
    {
      "epoch": 2.1747821220116736,
      "grad_norm": 1.7356793118636098,
      "learning_rate": 5.788896051955953e-07,
      "loss": 0.235,
      "step": 40800
    },
    {
      "epoch": 2.1753151568455,
      "grad_norm": 1.7205912341988363,
      "learning_rate": 5.781941357724404e-07,
      "loss": 0.2409,
      "step": 40810
    },
    {
      "epoch": 2.1758481916793264,
      "grad_norm": 1.787702738502243,
      "learning_rate": 5.774989846105704e-07,
      "loss": 0.238,
      "step": 40820
    },
    {
      "epoch": 2.176381226513153,
      "grad_norm": 1.8703477157717947,
      "learning_rate": 5.768041519499915e-07,
      "loss": 0.2286,
      "step": 40830
    },
    {
      "epoch": 2.176914261346979,
      "grad_norm": 1.9274239309754442,
      "learning_rate": 5.761096380306011e-07,
      "loss": 0.2289,
      "step": 40840
    },
    {
      "epoch": 2.1774472961808056,
      "grad_norm": 2.017726288014313,
      "learning_rate": 5.754154430921858e-07,
      "loss": 0.2323,
      "step": 40850
    },
    {
      "epoch": 2.177980331014632,
      "grad_norm": 1.8697963268100288,
      "learning_rate": 5.747215673744224e-07,
      "loss": 0.2186,
      "step": 40860
    },
    {
      "epoch": 2.1785133658484583,
      "grad_norm": 1.6623568098838089,
      "learning_rate": 5.740280111168772e-07,
      "loss": 0.2376,
      "step": 40870
    },
    {
      "epoch": 2.1790464006822847,
      "grad_norm": 2.0152712054603596,
      "learning_rate": 5.73334774559007e-07,
      "loss": 0.2377,
      "step": 40880
    },
    {
      "epoch": 2.179579435516111,
      "grad_norm": 1.8907127585324865,
      "learning_rate": 5.726418579401565e-07,
      "loss": 0.2304,
      "step": 40890
    },
    {
      "epoch": 2.1801124703499375,
      "grad_norm": 1.8945059120727403,
      "learning_rate": 5.719492614995617e-07,
      "loss": 0.2399,
      "step": 40900
    },
    {
      "epoch": 2.180645505183764,
      "grad_norm": 1.7565273368230947,
      "learning_rate": 5.712569854763473e-07,
      "loss": 0.2327,
      "step": 40910
    },
    {
      "epoch": 2.1811785400175903,
      "grad_norm": 2.1596659349803193,
      "learning_rate": 5.705650301095273e-07,
      "loss": 0.2441,
      "step": 40920
    },
    {
      "epoch": 2.1817115748514166,
      "grad_norm": 1.9228147554650465,
      "learning_rate": 5.698733956380053e-07,
      "loss": 0.2355,
      "step": 40930
    },
    {
      "epoch": 2.182244609685243,
      "grad_norm": 1.658515462375666,
      "learning_rate": 5.691820823005741e-07,
      "loss": 0.2331,
      "step": 40940
    },
    {
      "epoch": 2.1827776445190694,
      "grad_norm": 1.8195605936564088,
      "learning_rate": 5.684910903359152e-07,
      "loss": 0.2386,
      "step": 40950
    },
    {
      "epoch": 2.183310679352896,
      "grad_norm": 1.7378493426447967,
      "learning_rate": 5.678004199825996e-07,
      "loss": 0.2359,
      "step": 40960
    },
    {
      "epoch": 2.183843714186722,
      "grad_norm": 1.6851770971682727,
      "learning_rate": 5.671100714790876e-07,
      "loss": 0.2352,
      "step": 40970
    },
    {
      "epoch": 2.1843767490205486,
      "grad_norm": 1.729961895177647,
      "learning_rate": 5.664200450637268e-07,
      "loss": 0.2389,
      "step": 40980
    },
    {
      "epoch": 2.184909783854375,
      "grad_norm": 1.7377335399456209,
      "learning_rate": 5.657303409747553e-07,
      "loss": 0.2327,
      "step": 40990
    },
    {
      "epoch": 2.1854428186882013,
      "grad_norm": 1.8816422231335426,
      "learning_rate": 5.65040959450299e-07,
      "loss": 0.231,
      "step": 41000
    },
    {
      "epoch": 2.1859758535220277,
      "grad_norm": 1.7219088073980446,
      "learning_rate": 5.64351900728373e-07,
      "loss": 0.2362,
      "step": 41010
    },
    {
      "epoch": 2.186508888355854,
      "grad_norm": 1.774979949990636,
      "learning_rate": 5.636631650468805e-07,
      "loss": 0.2333,
      "step": 41020
    },
    {
      "epoch": 2.1870419231896805,
      "grad_norm": 1.8455125136844523,
      "learning_rate": 5.629747526436134e-07,
      "loss": 0.2432,
      "step": 41030
    },
    {
      "epoch": 2.187574958023507,
      "grad_norm": 1.8011976764329316,
      "learning_rate": 5.622866637562521e-07,
      "loss": 0.2459,
      "step": 41040
    },
    {
      "epoch": 2.1881079928573333,
      "grad_norm": 2.2607482696987216,
      "learning_rate": 5.615988986223652e-07,
      "loss": 0.2369,
      "step": 41050
    },
    {
      "epoch": 2.1886410276911596,
      "grad_norm": 1.8643421860625518,
      "learning_rate": 5.609114574794086e-07,
      "loss": 0.2355,
      "step": 41060
    },
    {
      "epoch": 2.189174062524986,
      "grad_norm": 1.8478859369761511,
      "learning_rate": 5.602243405647279e-07,
      "loss": 0.2301,
      "step": 41070
    },
    {
      "epoch": 2.1897070973588124,
      "grad_norm": 1.7790646393456686,
      "learning_rate": 5.595375481155558e-07,
      "loss": 0.2276,
      "step": 41080
    },
    {
      "epoch": 2.190240132192639,
      "grad_norm": 1.7567693055835896,
      "learning_rate": 5.588510803690131e-07,
      "loss": 0.2247,
      "step": 41090
    },
    {
      "epoch": 2.190773167026465,
      "grad_norm": 1.8364057387463932,
      "learning_rate": 5.58164937562109e-07,
      "loss": 0.2233,
      "step": 41100
    },
    {
      "epoch": 2.1913062018602916,
      "grad_norm": 1.8249598384629306,
      "learning_rate": 5.5747911993174e-07,
      "loss": 0.2394,
      "step": 41110
    },
    {
      "epoch": 2.191839236694118,
      "grad_norm": 1.8999814847080447,
      "learning_rate": 5.567936277146903e-07,
      "loss": 0.2326,
      "step": 41120
    },
    {
      "epoch": 2.1923722715279443,
      "grad_norm": 1.9969572751593117,
      "learning_rate": 5.561084611476318e-07,
      "loss": 0.2364,
      "step": 41130
    },
    {
      "epoch": 2.1929053063617707,
      "grad_norm": 1.6343258671086636,
      "learning_rate": 5.554236204671247e-07,
      "loss": 0.2379,
      "step": 41140
    },
    {
      "epoch": 2.193438341195597,
      "grad_norm": 2.048404009499756,
      "learning_rate": 5.547391059096153e-07,
      "loss": 0.231,
      "step": 41150
    },
    {
      "epoch": 2.1939713760294235,
      "grad_norm": 1.8255090738376567,
      "learning_rate": 5.54054917711438e-07,
      "loss": 0.2336,
      "step": 41160
    },
    {
      "epoch": 2.19450441086325,
      "grad_norm": 1.9027710365258481,
      "learning_rate": 5.53371056108815e-07,
      "loss": 0.2428,
      "step": 41170
    },
    {
      "epoch": 2.1950374456970763,
      "grad_norm": 1.9699946266229345,
      "learning_rate": 5.526875213378554e-07,
      "loss": 0.2377,
      "step": 41180
    },
    {
      "epoch": 2.1955704805309026,
      "grad_norm": 1.8604479866525307,
      "learning_rate": 5.520043136345549e-07,
      "loss": 0.2366,
      "step": 41190
    },
    {
      "epoch": 2.196103515364729,
      "grad_norm": 1.9590906199732714,
      "learning_rate": 5.513214332347971e-07,
      "loss": 0.2424,
      "step": 41200
    },
    {
      "epoch": 2.1966365501985554,
      "grad_norm": 1.815167864508418,
      "learning_rate": 5.50638880374352e-07,
      "loss": 0.234,
      "step": 41210
    },
    {
      "epoch": 2.197169585032382,
      "grad_norm": 2.045426959205902,
      "learning_rate": 5.499566552888778e-07,
      "loss": 0.2299,
      "step": 41220
    },
    {
      "epoch": 2.197702619866208,
      "grad_norm": 1.8629147751819057,
      "learning_rate": 5.492747582139168e-07,
      "loss": 0.2385,
      "step": 41230
    },
    {
      "epoch": 2.1982356547000346,
      "grad_norm": 1.953497543100977,
      "learning_rate": 5.485931893849007e-07,
      "loss": 0.241,
      "step": 41240
    },
    {
      "epoch": 2.198768689533861,
      "grad_norm": 1.855971223680553,
      "learning_rate": 5.479119490371465e-07,
      "loss": 0.2356,
      "step": 41250
    },
    {
      "epoch": 2.1993017243676873,
      "grad_norm": 1.821970187769495,
      "learning_rate": 5.472310374058588e-07,
      "loss": 0.2341,
      "step": 41260
    },
    {
      "epoch": 2.1998347592015137,
      "grad_norm": 1.8890834742967249,
      "learning_rate": 5.465504547261275e-07,
      "loss": 0.237,
      "step": 41270
    },
    {
      "epoch": 2.20036779403534,
      "grad_norm": 1.885150850743481,
      "learning_rate": 5.4587020123293e-07,
      "loss": 0.2365,
      "step": 41280
    },
    {
      "epoch": 2.2009008288691665,
      "grad_norm": 1.8907399258728894,
      "learning_rate": 5.451902771611293e-07,
      "loss": 0.2305,
      "step": 41290
    },
    {
      "epoch": 2.201433863702993,
      "grad_norm": 1.7072965913464644,
      "learning_rate": 5.44510682745475e-07,
      "loss": 0.2266,
      "step": 41300
    },
    {
      "epoch": 2.2019668985368193,
      "grad_norm": 1.7210767756656278,
      "learning_rate": 5.438314182206034e-07,
      "loss": 0.2355,
      "step": 41310
    },
    {
      "epoch": 2.2024999333706456,
      "grad_norm": 1.792967201798984,
      "learning_rate": 5.431524838210353e-07,
      "loss": 0.2332,
      "step": 41320
    },
    {
      "epoch": 2.203032968204472,
      "grad_norm": 1.9881998980737898,
      "learning_rate": 5.424738797811792e-07,
      "loss": 0.2313,
      "step": 41330
    },
    {
      "epoch": 2.2035660030382984,
      "grad_norm": 1.7870206056336637,
      "learning_rate": 5.417956063353286e-07,
      "loss": 0.2365,
      "step": 41340
    },
    {
      "epoch": 2.204099037872125,
      "grad_norm": 1.994831152860379,
      "learning_rate": 5.411176637176633e-07,
      "loss": 0.238,
      "step": 41350
    },
    {
      "epoch": 2.204632072705951,
      "grad_norm": 1.7070181368201565,
      "learning_rate": 5.404400521622487e-07,
      "loss": 0.2347,
      "step": 41360
    },
    {
      "epoch": 2.2051651075397776,
      "grad_norm": 1.9823732922440238,
      "learning_rate": 5.397627719030361e-07,
      "loss": 0.2294,
      "step": 41370
    },
    {
      "epoch": 2.205698142373604,
      "grad_norm": 1.9182536798723082,
      "learning_rate": 5.390858231738619e-07,
      "loss": 0.2452,
      "step": 41380
    },
    {
      "epoch": 2.2062311772074303,
      "grad_norm": 1.929649682275285,
      "learning_rate": 5.384092062084492e-07,
      "loss": 0.2524,
      "step": 41390
    },
    {
      "epoch": 2.2067642120412567,
      "grad_norm": 2.0523633893198703,
      "learning_rate": 5.377329212404047e-07,
      "loss": 0.2317,
      "step": 41400
    },
    {
      "epoch": 2.207297246875083,
      "grad_norm": 1.8742211349317188,
      "learning_rate": 5.370569685032219e-07,
      "loss": 0.2362,
      "step": 41410
    },
    {
      "epoch": 2.2078302817089095,
      "grad_norm": 1.848327079031136,
      "learning_rate": 5.363813482302793e-07,
      "loss": 0.2441,
      "step": 41420
    },
    {
      "epoch": 2.208363316542736,
      "grad_norm": 1.8637500405563077,
      "learning_rate": 5.357060606548405e-07,
      "loss": 0.2399,
      "step": 41430
    },
    {
      "epoch": 2.2088963513765623,
      "grad_norm": 1.8348135464263529,
      "learning_rate": 5.350311060100541e-07,
      "loss": 0.2369,
      "step": 41440
    },
    {
      "epoch": 2.2094293862103886,
      "grad_norm": 1.8620144851029274,
      "learning_rate": 5.343564845289542e-07,
      "loss": 0.2359,
      "step": 41450
    },
    {
      "epoch": 2.2099624210442155,
      "grad_norm": 2.205022243332574,
      "learning_rate": 5.336821964444593e-07,
      "loss": 0.2271,
      "step": 41460
    },
    {
      "epoch": 2.210495455878042,
      "grad_norm": 1.8624232532760654,
      "learning_rate": 5.330082419893739e-07,
      "loss": 0.2324,
      "step": 41470
    },
    {
      "epoch": 2.2110284907118682,
      "grad_norm": 2.189729306612113,
      "learning_rate": 5.323346213963851e-07,
      "loss": 0.2353,
      "step": 41480
    },
    {
      "epoch": 2.2115615255456946,
      "grad_norm": 1.876837513457624,
      "learning_rate": 5.316613348980668e-07,
      "loss": 0.2397,
      "step": 41490
    },
    {
      "epoch": 2.212094560379521,
      "grad_norm": 1.7472747201796073,
      "learning_rate": 5.30988382726877e-07,
      "loss": 0.2276,
      "step": 41500
    },
    {
      "epoch": 2.2126275952133474,
      "grad_norm": 1.942813019035239,
      "learning_rate": 5.303157651151581e-07,
      "loss": 0.2322,
      "step": 41510
    },
    {
      "epoch": 2.213160630047174,
      "grad_norm": 1.8815600048930214,
      "learning_rate": 5.296434822951368e-07,
      "loss": 0.2333,
      "step": 41520
    },
    {
      "epoch": 2.213693664881,
      "grad_norm": 2.084876747021146,
      "learning_rate": 5.289715344989247e-07,
      "loss": 0.2397,
      "step": 41530
    },
    {
      "epoch": 2.2142266997148266,
      "grad_norm": 1.8256909397652323,
      "learning_rate": 5.282999219585172e-07,
      "loss": 0.2327,
      "step": 41540
    },
    {
      "epoch": 2.214759734548653,
      "grad_norm": 1.773299026851792,
      "learning_rate": 5.276286449057943e-07,
      "loss": 0.23,
      "step": 41550
    },
    {
      "epoch": 2.2152927693824793,
      "grad_norm": 1.7732854459541896,
      "learning_rate": 5.269577035725208e-07,
      "loss": 0.2448,
      "step": 41560
    },
    {
      "epoch": 2.2158258042163057,
      "grad_norm": 1.7331647752262196,
      "learning_rate": 5.262870981903439e-07,
      "loss": 0.2311,
      "step": 41570
    },
    {
      "epoch": 2.216358839050132,
      "grad_norm": 2.0154771642934586,
      "learning_rate": 5.256168289907959e-07,
      "loss": 0.2275,
      "step": 41580
    },
    {
      "epoch": 2.2168918738839585,
      "grad_norm": 1.861152621679387,
      "learning_rate": 5.249468962052933e-07,
      "loss": 0.2276,
      "step": 41590
    },
    {
      "epoch": 2.217424908717785,
      "grad_norm": 1.8022987639062718,
      "learning_rate": 5.242773000651357e-07,
      "loss": 0.233,
      "step": 41600
    },
    {
      "epoch": 2.2179579435516112,
      "grad_norm": 1.7446226513462637,
      "learning_rate": 5.236080408015073e-07,
      "loss": 0.2246,
      "step": 41610
    },
    {
      "epoch": 2.2184909783854376,
      "grad_norm": 1.714695028136221,
      "learning_rate": 5.229391186454751e-07,
      "loss": 0.2243,
      "step": 41620
    },
    {
      "epoch": 2.219024013219264,
      "grad_norm": 1.6550759358329208,
      "learning_rate": 5.222705338279906e-07,
      "loss": 0.2326,
      "step": 41630
    },
    {
      "epoch": 2.2195570480530904,
      "grad_norm": 1.8502921022397325,
      "learning_rate": 5.216022865798884e-07,
      "loss": 0.2259,
      "step": 41640
    },
    {
      "epoch": 2.220090082886917,
      "grad_norm": 1.8028172807314409,
      "learning_rate": 5.209343771318857e-07,
      "loss": 0.2293,
      "step": 41650
    },
    {
      "epoch": 2.220623117720743,
      "grad_norm": 1.6723396627343226,
      "learning_rate": 5.202668057145846e-07,
      "loss": 0.2328,
      "step": 41660
    },
    {
      "epoch": 2.2211561525545696,
      "grad_norm": 1.797152027330531,
      "learning_rate": 5.195995725584696e-07,
      "loss": 0.2346,
      "step": 41670
    },
    {
      "epoch": 2.221689187388396,
      "grad_norm": 1.8658059936380529,
      "learning_rate": 5.189326778939085e-07,
      "loss": 0.2265,
      "step": 41680
    },
    {
      "epoch": 2.2222222222222223,
      "grad_norm": 2.097308845740392,
      "learning_rate": 5.182661219511525e-07,
      "loss": 0.2354,
      "step": 41690
    },
    {
      "epoch": 2.2227552570560487,
      "grad_norm": 1.688084102481659,
      "learning_rate": 5.175999049603357e-07,
      "loss": 0.2345,
      "step": 41700
    },
    {
      "epoch": 2.223288291889875,
      "grad_norm": 1.806918329911081,
      "learning_rate": 5.169340271514751e-07,
      "loss": 0.2334,
      "step": 41710
    },
    {
      "epoch": 2.2238213267237015,
      "grad_norm": 1.8065267488088768,
      "learning_rate": 5.162684887544705e-07,
      "loss": 0.2341,
      "step": 41720
    },
    {
      "epoch": 2.224354361557528,
      "grad_norm": 1.6287845961942662,
      "learning_rate": 5.156032899991055e-07,
      "loss": 0.2321,
      "step": 41730
    },
    {
      "epoch": 2.2248873963913542,
      "grad_norm": 1.7870541928190262,
      "learning_rate": 5.149384311150442e-07,
      "loss": 0.2361,
      "step": 41740
    },
    {
      "epoch": 2.2254204312251806,
      "grad_norm": 1.8632902957098911,
      "learning_rate": 5.142739123318358e-07,
      "loss": 0.2406,
      "step": 41750
    },
    {
      "epoch": 2.225953466059007,
      "grad_norm": 1.7904847228036171,
      "learning_rate": 5.136097338789105e-07,
      "loss": 0.2413,
      "step": 41760
    },
    {
      "epoch": 2.2264865008928334,
      "grad_norm": 1.8578067378892353,
      "learning_rate": 5.129458959855818e-07,
      "loss": 0.2315,
      "step": 41770
    },
    {
      "epoch": 2.22701953572666,
      "grad_norm": 1.7122818821507455,
      "learning_rate": 5.122823988810454e-07,
      "loss": 0.2333,
      "step": 41780
    },
    {
      "epoch": 2.227552570560486,
      "grad_norm": 1.8435971896283132,
      "learning_rate": 5.116192427943792e-07,
      "loss": 0.2323,
      "step": 41790
    },
    {
      "epoch": 2.2280856053943126,
      "grad_norm": 1.7828977196808466,
      "learning_rate": 5.109564279545435e-07,
      "loss": 0.2263,
      "step": 41800
    },
    {
      "epoch": 2.228618640228139,
      "grad_norm": 1.7134832623698832,
      "learning_rate": 5.102939545903812e-07,
      "loss": 0.229,
      "step": 41810
    },
    {
      "epoch": 2.2291516750619653,
      "grad_norm": 2.0030313148881174,
      "learning_rate": 5.09631822930616e-07,
      "loss": 0.2308,
      "step": 41820
    },
    {
      "epoch": 2.2296847098957917,
      "grad_norm": 1.838416068085716,
      "learning_rate": 5.089700332038545e-07,
      "loss": 0.2249,
      "step": 41830
    },
    {
      "epoch": 2.230217744729618,
      "grad_norm": 1.8378146046419626,
      "learning_rate": 5.083085856385858e-07,
      "loss": 0.232,
      "step": 41840
    },
    {
      "epoch": 2.2307507795634445,
      "grad_norm": 1.8746900208680886,
      "learning_rate": 5.0764748046318e-07,
      "loss": 0.2337,
      "step": 41850
    },
    {
      "epoch": 2.231283814397271,
      "grad_norm": 1.7921558846555603,
      "learning_rate": 5.069867179058892e-07,
      "loss": 0.2367,
      "step": 41860
    },
    {
      "epoch": 2.2318168492310972,
      "grad_norm": 1.8708803751035725,
      "learning_rate": 5.063262981948475e-07,
      "loss": 0.2407,
      "step": 41870
    },
    {
      "epoch": 2.2323498840649236,
      "grad_norm": 1.8768091311257802,
      "learning_rate": 5.0566622155807e-07,
      "loss": 0.2362,
      "step": 41880
    },
    {
      "epoch": 2.23288291889875,
      "grad_norm": 1.8554853006993999,
      "learning_rate": 5.050064882234542e-07,
      "loss": 0.2254,
      "step": 41890
    },
    {
      "epoch": 2.2334159537325764,
      "grad_norm": 1.8157190078437806,
      "learning_rate": 5.043470984187787e-07,
      "loss": 0.2281,
      "step": 41900
    },
    {
      "epoch": 2.233948988566403,
      "grad_norm": 1.8774213062895782,
      "learning_rate": 5.036880523717029e-07,
      "loss": 0.2308,
      "step": 41910
    },
    {
      "epoch": 2.234482023400229,
      "grad_norm": 1.870961308820473,
      "learning_rate": 5.030293503097682e-07,
      "loss": 0.2286,
      "step": 41920
    },
    {
      "epoch": 2.2350150582340556,
      "grad_norm": 1.7131023886177141,
      "learning_rate": 5.023709924603971e-07,
      "loss": 0.2245,
      "step": 41930
    },
    {
      "epoch": 2.235548093067882,
      "grad_norm": 1.9098054247283778,
      "learning_rate": 5.017129790508932e-07,
      "loss": 0.2339,
      "step": 41940
    },
    {
      "epoch": 2.2360811279017083,
      "grad_norm": 1.834768247406661,
      "learning_rate": 5.010553103084414e-07,
      "loss": 0.2392,
      "step": 41950
    },
    {
      "epoch": 2.2366141627355347,
      "grad_norm": 1.8753774032794372,
      "learning_rate": 5.003979864601072e-07,
      "loss": 0.2356,
      "step": 41960
    },
    {
      "epoch": 2.237147197569361,
      "grad_norm": 2.113872789190751,
      "learning_rate": 4.997410077328374e-07,
      "loss": 0.2397,
      "step": 41970
    },
    {
      "epoch": 2.2376802324031875,
      "grad_norm": 2.063779534936551,
      "learning_rate": 4.990843743534598e-07,
      "loss": 0.2193,
      "step": 41980
    },
    {
      "epoch": 2.238213267237014,
      "grad_norm": 1.8658433844855866,
      "learning_rate": 4.984280865486817e-07,
      "loss": 0.2281,
      "step": 41990
    },
    {
      "epoch": 2.2387463020708402,
      "grad_norm": 1.9347700022938346,
      "learning_rate": 4.977721445450926e-07,
      "loss": 0.2317,
      "step": 42000
    },
    {
      "epoch": 2.2392793369046666,
      "grad_norm": 1.6694069475944362,
      "learning_rate": 4.971165485691618e-07,
      "loss": 0.2211,
      "step": 42010
    },
    {
      "epoch": 2.239812371738493,
      "grad_norm": 1.7828251662860373,
      "learning_rate": 4.964612988472398e-07,
      "loss": 0.2316,
      "step": 42020
    },
    {
      "epoch": 2.2403454065723194,
      "grad_norm": 2.1127487130410025,
      "learning_rate": 4.958063956055567e-07,
      "loss": 0.2291,
      "step": 42030
    },
    {
      "epoch": 2.240878441406146,
      "grad_norm": 1.8674174795271359,
      "learning_rate": 4.95151839070224e-07,
      "loss": 0.2314,
      "step": 42040
    },
    {
      "epoch": 2.241411476239972,
      "grad_norm": 1.9101220646138868,
      "learning_rate": 4.944976294672321e-07,
      "loss": 0.229,
      "step": 42050
    },
    {
      "epoch": 2.2419445110737986,
      "grad_norm": 1.85493508855506,
      "learning_rate": 4.938437670224526e-07,
      "loss": 0.2311,
      "step": 42060
    },
    {
      "epoch": 2.242477545907625,
      "grad_norm": 1.8403463607809167,
      "learning_rate": 4.931902519616373e-07,
      "loss": 0.2346,
      "step": 42070
    },
    {
      "epoch": 2.2430105807414513,
      "grad_norm": 1.849465526135925,
      "learning_rate": 4.925370845104177e-07,
      "loss": 0.2363,
      "step": 42080
    },
    {
      "epoch": 2.2435436155752777,
      "grad_norm": 1.760393165622021,
      "learning_rate": 4.918842648943058e-07,
      "loss": 0.2406,
      "step": 42090
    },
    {
      "epoch": 2.244076650409104,
      "grad_norm": 1.8126003007161957,
      "learning_rate": 4.912317933386922e-07,
      "loss": 0.2375,
      "step": 42100
    },
    {
      "epoch": 2.2446096852429305,
      "grad_norm": 1.737873700508675,
      "learning_rate": 4.905796700688489e-07,
      "loss": 0.2351,
      "step": 42110
    },
    {
      "epoch": 2.245142720076757,
      "grad_norm": 2.091441772225788,
      "learning_rate": 4.899278953099266e-07,
      "loss": 0.2322,
      "step": 42120
    },
    {
      "epoch": 2.2456757549105832,
      "grad_norm": 1.8487757068908606,
      "learning_rate": 4.892764692869566e-07,
      "loss": 0.2388,
      "step": 42130
    },
    {
      "epoch": 2.2462087897444096,
      "grad_norm": 1.9634130849931193,
      "learning_rate": 4.886253922248488e-07,
      "loss": 0.2317,
      "step": 42140
    },
    {
      "epoch": 2.246741824578236,
      "grad_norm": 1.8722736397728355,
      "learning_rate": 4.87974664348393e-07,
      "loss": 0.2304,
      "step": 42150
    },
    {
      "epoch": 2.2472748594120624,
      "grad_norm": 1.9453801643001631,
      "learning_rate": 4.873242858822589e-07,
      "loss": 0.2334,
      "step": 42160
    },
    {
      "epoch": 2.247807894245889,
      "grad_norm": 1.938295100944413,
      "learning_rate": 4.866742570509956e-07,
      "loss": 0.2265,
      "step": 42170
    },
    {
      "epoch": 2.248340929079715,
      "grad_norm": 1.9087832182764721,
      "learning_rate": 4.860245780790298e-07,
      "loss": 0.2431,
      "step": 42180
    },
    {
      "epoch": 2.2488739639135416,
      "grad_norm": 1.947170735985685,
      "learning_rate": 4.853752491906693e-07,
      "loss": 0.2294,
      "step": 42190
    },
    {
      "epoch": 2.249406998747368,
      "grad_norm": 1.743819283349934,
      "learning_rate": 4.847262706101004e-07,
      "loss": 0.234,
      "step": 42200
    },
    {
      "epoch": 2.2499400335811943,
      "grad_norm": 1.672139376639322,
      "learning_rate": 4.840776425613887e-07,
      "loss": 0.2315,
      "step": 42210
    },
    {
      "epoch": 2.2504730684150207,
      "grad_norm": 1.8625421344128692,
      "learning_rate": 4.834293652684781e-07,
      "loss": 0.2371,
      "step": 42220
    },
    {
      "epoch": 2.251006103248847,
      "grad_norm": 1.9270640605908427,
      "learning_rate": 4.827814389551921e-07,
      "loss": 0.2346,
      "step": 42230
    },
    {
      "epoch": 2.2515391380826735,
      "grad_norm": 1.939093175886227,
      "learning_rate": 4.821338638452327e-07,
      "loss": 0.227,
      "step": 42240
    },
    {
      "epoch": 2.2520721729165,
      "grad_norm": 1.8811465193785428,
      "learning_rate": 4.814866401621808e-07,
      "loss": 0.2311,
      "step": 42250
    },
    {
      "epoch": 2.2526052077503262,
      "grad_norm": 1.6655322973270514,
      "learning_rate": 4.80839768129496e-07,
      "loss": 0.2331,
      "step": 42260
    },
    {
      "epoch": 2.2531382425841526,
      "grad_norm": 1.8888896409919291,
      "learning_rate": 4.801932479705157e-07,
      "loss": 0.2389,
      "step": 42270
    },
    {
      "epoch": 2.2536712774179795,
      "grad_norm": 1.7826633957754323,
      "learning_rate": 4.795470799084569e-07,
      "loss": 0.2458,
      "step": 42280
    },
    {
      "epoch": 2.254204312251806,
      "grad_norm": 2.0284305155667903,
      "learning_rate": 4.789012641664145e-07,
      "loss": 0.234,
      "step": 42290
    },
    {
      "epoch": 2.2547373470856322,
      "grad_norm": 2.120019229343177,
      "learning_rate": 4.782558009673619e-07,
      "loss": 0.2346,
      "step": 42300
    },
    {
      "epoch": 2.2552703819194586,
      "grad_norm": 1.7441441734890615,
      "learning_rate": 4.776106905341507e-07,
      "loss": 0.2224,
      "step": 42310
    },
    {
      "epoch": 2.255803416753285,
      "grad_norm": 1.8788600256722958,
      "learning_rate": 4.769659330895109e-07,
      "loss": 0.2312,
      "step": 42320
    },
    {
      "epoch": 2.2563364515871114,
      "grad_norm": 1.8752322772145442,
      "learning_rate": 4.7632152885605025e-07,
      "loss": 0.2385,
      "step": 42330
    },
    {
      "epoch": 2.2568694864209378,
      "grad_norm": 1.7736666883910046,
      "learning_rate": 4.756774780562554e-07,
      "loss": 0.2354,
      "step": 42340
    },
    {
      "epoch": 2.257402521254764,
      "grad_norm": 1.7587975426282534,
      "learning_rate": 4.7503378091248947e-07,
      "loss": 0.2338,
      "step": 42350
    },
    {
      "epoch": 2.2579355560885905,
      "grad_norm": 1.978167736350367,
      "learning_rate": 4.7439043764699447e-07,
      "loss": 0.2285,
      "step": 42360
    },
    {
      "epoch": 2.258468590922417,
      "grad_norm": 1.7672834261207926,
      "learning_rate": 4.737474484818905e-07,
      "loss": 0.2344,
      "step": 42370
    },
    {
      "epoch": 2.2590016257562433,
      "grad_norm": 1.8658786813493369,
      "learning_rate": 4.7310481363917494e-07,
      "loss": 0.2371,
      "step": 42380
    },
    {
      "epoch": 2.2595346605900697,
      "grad_norm": 1.8627629990173118,
      "learning_rate": 4.724625333407227e-07,
      "loss": 0.2286,
      "step": 42390
    },
    {
      "epoch": 2.260067695423896,
      "grad_norm": 1.6903194752508697,
      "learning_rate": 4.7182060780828667e-07,
      "loss": 0.2376,
      "step": 42400
    },
    {
      "epoch": 2.2606007302577225,
      "grad_norm": 2.107018056391333,
      "learning_rate": 4.711790372634971e-07,
      "loss": 0.2349,
      "step": 42410
    },
    {
      "epoch": 2.261133765091549,
      "grad_norm": 1.9341597934084553,
      "learning_rate": 4.7053782192786193e-07,
      "loss": 0.2288,
      "step": 42420
    },
    {
      "epoch": 2.2616667999253752,
      "grad_norm": 1.9583370673619234,
      "learning_rate": 4.698969620227656e-07,
      "loss": 0.228,
      "step": 42430
    },
    {
      "epoch": 2.2621998347592016,
      "grad_norm": 1.6745201940271226,
      "learning_rate": 4.692564577694705e-07,
      "loss": 0.2406,
      "step": 42440
    },
    {
      "epoch": 2.262732869593028,
      "grad_norm": 1.851860195602235,
      "learning_rate": 4.6861630938911643e-07,
      "loss": 0.2271,
      "step": 42450
    },
    {
      "epoch": 2.2632659044268544,
      "grad_norm": 1.710943831084753,
      "learning_rate": 4.679765171027199e-07,
      "loss": 0.2208,
      "step": 42460
    },
    {
      "epoch": 2.2637989392606808,
      "grad_norm": 1.7970447050825435,
      "learning_rate": 4.6733708113117465e-07,
      "loss": 0.2338,
      "step": 42470
    },
    {
      "epoch": 2.264331974094507,
      "grad_norm": 1.8789122968307477,
      "learning_rate": 4.666980016952514e-07,
      "loss": 0.2323,
      "step": 42480
    },
    {
      "epoch": 2.2648650089283335,
      "grad_norm": 1.7167903847047923,
      "learning_rate": 4.660592790155977e-07,
      "loss": 0.2308,
      "step": 42490
    },
    {
      "epoch": 2.26539804376216,
      "grad_norm": 1.8706389285247746,
      "learning_rate": 4.6542091331273814e-07,
      "loss": 0.2468,
      "step": 42500
    },
    {
      "epoch": 2.2659310785959863,
      "grad_norm": 1.6938167223822291,
      "learning_rate": 4.64782904807074e-07,
      "loss": 0.2306,
      "step": 42510
    },
    {
      "epoch": 2.2664641134298127,
      "grad_norm": 1.8478433154322607,
      "learning_rate": 4.641452537188824e-07,
      "loss": 0.2308,
      "step": 42520
    },
    {
      "epoch": 2.266997148263639,
      "grad_norm": 1.9754760861091705,
      "learning_rate": 4.635079602683183e-07,
      "loss": 0.243,
      "step": 42530
    },
    {
      "epoch": 2.2675301830974655,
      "grad_norm": 1.8397041044859581,
      "learning_rate": 4.6287102467541266e-07,
      "loss": 0.2408,
      "step": 42540
    },
    {
      "epoch": 2.268063217931292,
      "grad_norm": 1.8496143151326303,
      "learning_rate": 4.622344471600728e-07,
      "loss": 0.2282,
      "step": 42550
    },
    {
      "epoch": 2.2685962527651182,
      "grad_norm": 1.9766410879762266,
      "learning_rate": 4.615982279420824e-07,
      "loss": 0.2419,
      "step": 42560
    },
    {
      "epoch": 2.2691292875989446,
      "grad_norm": 1.8251922862618457,
      "learning_rate": 4.6096236724110176e-07,
      "loss": 0.2313,
      "step": 42570
    },
    {
      "epoch": 2.269662322432771,
      "grad_norm": 1.6797407330918674,
      "learning_rate": 4.6032686527666706e-07,
      "loss": 0.2332,
      "step": 42580
    },
    {
      "epoch": 2.2701953572665974,
      "grad_norm": 2.10280804168582,
      "learning_rate": 4.596917222681912e-07,
      "loss": 0.2361,
      "step": 42590
    },
    {
      "epoch": 2.2707283921004238,
      "grad_norm": 1.8013404493877903,
      "learning_rate": 4.5905693843496184e-07,
      "loss": 0.2321,
      "step": 42600
    },
    {
      "epoch": 2.27126142693425,
      "grad_norm": 1.7743338907602508,
      "learning_rate": 4.5842251399614374e-07,
      "loss": 0.2368,
      "step": 42610
    },
    {
      "epoch": 2.2717944617680765,
      "grad_norm": 1.8691707983203496,
      "learning_rate": 4.5778844917077763e-07,
      "loss": 0.2297,
      "step": 42620
    },
    {
      "epoch": 2.272327496601903,
      "grad_norm": 1.857685041241757,
      "learning_rate": 4.571547441777794e-07,
      "loss": 0.2322,
      "step": 42630
    },
    {
      "epoch": 2.2728605314357293,
      "grad_norm": 1.8790157892044763,
      "learning_rate": 4.5652139923594146e-07,
      "loss": 0.2352,
      "step": 42640
    },
    {
      "epoch": 2.2733935662695557,
      "grad_norm": 1.8117625637832968,
      "learning_rate": 4.558884145639311e-07,
      "loss": 0.2375,
      "step": 42650
    },
    {
      "epoch": 2.273926601103382,
      "grad_norm": 1.7990210539400835,
      "learning_rate": 4.552557903802919e-07,
      "loss": 0.2294,
      "step": 42660
    },
    {
      "epoch": 2.2744596359372085,
      "grad_norm": 1.7693751564385993,
      "learning_rate": 4.546235269034428e-07,
      "loss": 0.2317,
      "step": 42670
    },
    {
      "epoch": 2.274992670771035,
      "grad_norm": 1.8876326125492982,
      "learning_rate": 4.539916243516784e-07,
      "loss": 0.2371,
      "step": 42680
    },
    {
      "epoch": 2.2755257056048612,
      "grad_norm": 1.947364438316141,
      "learning_rate": 4.5336008294316744e-07,
      "loss": 0.232,
      "step": 42690
    },
    {
      "epoch": 2.2760587404386876,
      "grad_norm": 1.7971182646760888,
      "learning_rate": 4.5272890289595555e-07,
      "loss": 0.239,
      "step": 42700
    },
    {
      "epoch": 2.276591775272514,
      "grad_norm": 1.999241739908646,
      "learning_rate": 4.5209808442796283e-07,
      "loss": 0.2316,
      "step": 42710
    },
    {
      "epoch": 2.2771248101063404,
      "grad_norm": 1.7267725470441606,
      "learning_rate": 4.51467627756985e-07,
      "loss": 0.2315,
      "step": 42720
    },
    {
      "epoch": 2.2776578449401668,
      "grad_norm": 1.8409026628441352,
      "learning_rate": 4.5083753310069245e-07,
      "loss": 0.2344,
      "step": 42730
    },
    {
      "epoch": 2.278190879773993,
      "grad_norm": 1.7260263407615797,
      "learning_rate": 4.5020780067663037e-07,
      "loss": 0.2248,
      "step": 42740
    },
    {
      "epoch": 2.2787239146078195,
      "grad_norm": 1.795052678220571,
      "learning_rate": 4.495784307022196e-07,
      "loss": 0.2375,
      "step": 42750
    },
    {
      "epoch": 2.279256949441646,
      "grad_norm": 1.8149960275508512,
      "learning_rate": 4.489494233947558e-07,
      "loss": 0.2272,
      "step": 42760
    },
    {
      "epoch": 2.2797899842754723,
      "grad_norm": 1.9428445748313783,
      "learning_rate": 4.483207789714081e-07,
      "loss": 0.239,
      "step": 42770
    },
    {
      "epoch": 2.2803230191092987,
      "grad_norm": 1.8301121821728776,
      "learning_rate": 4.4769249764922186e-07,
      "loss": 0.2284,
      "step": 42780
    },
    {
      "epoch": 2.280856053943125,
      "grad_norm": 1.8378815741989343,
      "learning_rate": 4.470645796451164e-07,
      "loss": 0.2292,
      "step": 42790
    },
    {
      "epoch": 2.2813890887769515,
      "grad_norm": 2.1277228878300356,
      "learning_rate": 4.4643702517588584e-07,
      "loss": 0.2317,
      "step": 42800
    },
    {
      "epoch": 2.281922123610778,
      "grad_norm": 1.771124578339705,
      "learning_rate": 4.458098344581987e-07,
      "loss": 0.2264,
      "step": 42810
    },
    {
      "epoch": 2.2824551584446042,
      "grad_norm": 1.8387020335790323,
      "learning_rate": 4.451830077085978e-07,
      "loss": 0.2335,
      "step": 42820
    },
    {
      "epoch": 2.2829881932784306,
      "grad_norm": 1.8135657261329514,
      "learning_rate": 4.4455654514350053e-07,
      "loss": 0.2254,
      "step": 42830
    },
    {
      "epoch": 2.283521228112257,
      "grad_norm": 1.902499338620403,
      "learning_rate": 4.4393044697919826e-07,
      "loss": 0.2391,
      "step": 42840
    },
    {
      "epoch": 2.2840542629460834,
      "grad_norm": 1.9089766654981986,
      "learning_rate": 4.4330471343185716e-07,
      "loss": 0.2332,
      "step": 42850
    },
    {
      "epoch": 2.2845872977799098,
      "grad_norm": 1.8034748005744983,
      "learning_rate": 4.42679344717516e-07,
      "loss": 0.2302,
      "step": 42860
    },
    {
      "epoch": 2.285120332613736,
      "grad_norm": 1.8426650090008008,
      "learning_rate": 4.420543410520894e-07,
      "loss": 0.2359,
      "step": 42870
    },
    {
      "epoch": 2.2856533674475625,
      "grad_norm": 2.033473062759244,
      "learning_rate": 4.414297026513648e-07,
      "loss": 0.2368,
      "step": 42880
    },
    {
      "epoch": 2.286186402281389,
      "grad_norm": 1.5954590444033023,
      "learning_rate": 4.408054297310044e-07,
      "loss": 0.2303,
      "step": 42890
    },
    {
      "epoch": 2.2867194371152153,
      "grad_norm": 1.7093432020464265,
      "learning_rate": 4.401815225065432e-07,
      "loss": 0.2361,
      "step": 42900
    },
    {
      "epoch": 2.2872524719490417,
      "grad_norm": 1.8502023256520532,
      "learning_rate": 4.3955798119339073e-07,
      "loss": 0.2257,
      "step": 42910
    },
    {
      "epoch": 2.287785506782868,
      "grad_norm": 1.735471106677551,
      "learning_rate": 4.3893480600682983e-07,
      "loss": 0.2385,
      "step": 42920
    },
    {
      "epoch": 2.2883185416166945,
      "grad_norm": 1.662535673374314,
      "learning_rate": 4.3831199716201755e-07,
      "loss": 0.2286,
      "step": 42930
    },
    {
      "epoch": 2.2888515764505213,
      "grad_norm": 1.7316375105576984,
      "learning_rate": 4.376895548739828e-07,
      "loss": 0.2323,
      "step": 42940
    },
    {
      "epoch": 2.2893846112843477,
      "grad_norm": 1.7395390591431699,
      "learning_rate": 4.370674793576298e-07,
      "loss": 0.2293,
      "step": 42950
    },
    {
      "epoch": 2.289917646118174,
      "grad_norm": 1.9915582547965862,
      "learning_rate": 4.364457708277352e-07,
      "loss": 0.2246,
      "step": 42960
    },
    {
      "epoch": 2.2904506809520004,
      "grad_norm": 1.9231096806475911,
      "learning_rate": 4.358244294989492e-07,
      "loss": 0.2345,
      "step": 42970
    },
    {
      "epoch": 2.290983715785827,
      "grad_norm": 1.9004509022148988,
      "learning_rate": 4.3520345558579503e-07,
      "loss": 0.2312,
      "step": 42980
    },
    {
      "epoch": 2.291516750619653,
      "grad_norm": 2.0325938137160837,
      "learning_rate": 4.3458284930266947e-07,
      "loss": 0.2337,
      "step": 42990
    },
    {
      "epoch": 2.2920497854534796,
      "grad_norm": 1.7956676692823648,
      "learning_rate": 4.339626108638419e-07,
      "loss": 0.2424,
      "step": 43000
    },
    {
      "epoch": 2.292582820287306,
      "grad_norm": 1.9228644954546887,
      "learning_rate": 4.3334274048345483e-07,
      "loss": 0.2375,
      "step": 43010
    },
    {
      "epoch": 2.2931158551211324,
      "grad_norm": 2.0439539585184088,
      "learning_rate": 4.327232383755245e-07,
      "loss": 0.2268,
      "step": 43020
    },
    {
      "epoch": 2.2936488899549587,
      "grad_norm": 1.9244388483606711,
      "learning_rate": 4.3210410475393796e-07,
      "loss": 0.2305,
      "step": 43030
    },
    {
      "epoch": 2.294181924788785,
      "grad_norm": 1.7583810860383617,
      "learning_rate": 4.314853398324572e-07,
      "loss": 0.2348,
      "step": 43040
    },
    {
      "epoch": 2.2947149596226115,
      "grad_norm": 1.8156478766029809,
      "learning_rate": 4.3086694382471566e-07,
      "loss": 0.2268,
      "step": 43050
    },
    {
      "epoch": 2.295247994456438,
      "grad_norm": 1.885166228652532,
      "learning_rate": 4.302489169442202e-07,
      "loss": 0.2371,
      "step": 43060
    },
    {
      "epoch": 2.2957810292902643,
      "grad_norm": 1.8584555041910376,
      "learning_rate": 4.2963125940434947e-07,
      "loss": 0.2327,
      "step": 43070
    },
    {
      "epoch": 2.2963140641240907,
      "grad_norm": 1.8611022220316689,
      "learning_rate": 4.2901397141835514e-07,
      "loss": 0.2335,
      "step": 43080
    },
    {
      "epoch": 2.296847098957917,
      "grad_norm": 2.0853339141325935,
      "learning_rate": 4.283970531993613e-07,
      "loss": 0.2317,
      "step": 43090
    },
    {
      "epoch": 2.2973801337917434,
      "grad_norm": 1.6461176827686417,
      "learning_rate": 4.2778050496036427e-07,
      "loss": 0.2395,
      "step": 43100
    },
    {
      "epoch": 2.29791316862557,
      "grad_norm": 1.9070429360533467,
      "learning_rate": 4.271643269142319e-07,
      "loss": 0.2275,
      "step": 43110
    },
    {
      "epoch": 2.298446203459396,
      "grad_norm": 1.8666758070460823,
      "learning_rate": 4.2654851927370524e-07,
      "loss": 0.2391,
      "step": 43120
    },
    {
      "epoch": 2.2989792382932226,
      "grad_norm": 1.7049743747771595,
      "learning_rate": 4.2593308225139706e-07,
      "loss": 0.2309,
      "step": 43130
    },
    {
      "epoch": 2.299512273127049,
      "grad_norm": 1.715478967492305,
      "learning_rate": 4.253180160597923e-07,
      "loss": 0.2394,
      "step": 43140
    },
    {
      "epoch": 2.3000453079608754,
      "grad_norm": 1.7614232042048674,
      "learning_rate": 4.2470332091124785e-07,
      "loss": 0.2356,
      "step": 43150
    },
    {
      "epoch": 2.3005783427947017,
      "grad_norm": 2.1090699973170675,
      "learning_rate": 4.240889970179921e-07,
      "loss": 0.2372,
      "step": 43160
    },
    {
      "epoch": 2.301111377628528,
      "grad_norm": 1.9380812697639274,
      "learning_rate": 4.23475044592126e-07,
      "loss": 0.2288,
      "step": 43170
    },
    {
      "epoch": 2.3016444124623545,
      "grad_norm": 1.890436193855489,
      "learning_rate": 4.228614638456215e-07,
      "loss": 0.2282,
      "step": 43180
    },
    {
      "epoch": 2.302177447296181,
      "grad_norm": 1.9718528479527335,
      "learning_rate": 4.222482549903232e-07,
      "loss": 0.2348,
      "step": 43190
    },
    {
      "epoch": 2.3027104821300073,
      "grad_norm": 1.7844388520509364,
      "learning_rate": 4.2163541823794554e-07,
      "loss": 0.2405,
      "step": 43200
    },
    {
      "epoch": 2.3032435169638337,
      "grad_norm": 1.8660275891759122,
      "learning_rate": 4.2102295380007626e-07,
      "loss": 0.22,
      "step": 43210
    },
    {
      "epoch": 2.30377655179766,
      "grad_norm": 1.7614593371642695,
      "learning_rate": 4.2041086188817375e-07,
      "loss": 0.2287,
      "step": 43220
    },
    {
      "epoch": 2.3043095866314864,
      "grad_norm": 1.9126634574050185,
      "learning_rate": 4.197991427135681e-07,
      "loss": 0.2355,
      "step": 43230
    },
    {
      "epoch": 2.304842621465313,
      "grad_norm": 1.9077801620819665,
      "learning_rate": 4.191877964874603e-07,
      "loss": 0.2286,
      "step": 43240
    },
    {
      "epoch": 2.305375656299139,
      "grad_norm": 1.9181295595242365,
      "learning_rate": 4.185768234209231e-07,
      "loss": 0.2328,
      "step": 43250
    },
    {
      "epoch": 2.3059086911329656,
      "grad_norm": 2.079126291961258,
      "learning_rate": 4.179662237248998e-07,
      "loss": 0.2349,
      "step": 43260
    },
    {
      "epoch": 2.306441725966792,
      "grad_norm": 2.077282808355432,
      "learning_rate": 4.1735599761020583e-07,
      "loss": 0.235,
      "step": 43270
    },
    {
      "epoch": 2.3069747608006184,
      "grad_norm": 1.8201828178811226,
      "learning_rate": 4.1674614528752604e-07,
      "loss": 0.2375,
      "step": 43280
    },
    {
      "epoch": 2.3075077956344447,
      "grad_norm": 1.7940395279305905,
      "learning_rate": 4.161366669674173e-07,
      "loss": 0.2326,
      "step": 43290
    },
    {
      "epoch": 2.308040830468271,
      "grad_norm": 1.9452268109562,
      "learning_rate": 4.1552756286030734e-07,
      "loss": 0.2303,
      "step": 43300
    },
    {
      "epoch": 2.3085738653020975,
      "grad_norm": 1.8500163335966857,
      "learning_rate": 4.149188331764946e-07,
      "loss": 0.234,
      "step": 43310
    },
    {
      "epoch": 2.309106900135924,
      "grad_norm": 1.9389897917401564,
      "learning_rate": 4.143104781261479e-07,
      "loss": 0.2196,
      "step": 43320
    },
    {
      "epoch": 2.3096399349697503,
      "grad_norm": 1.836229719492679,
      "learning_rate": 4.137024979193073e-07,
      "loss": 0.2321,
      "step": 43330
    },
    {
      "epoch": 2.3101729698035767,
      "grad_norm": 2.003268492653125,
      "learning_rate": 4.1309489276588286e-07,
      "loss": 0.2272,
      "step": 43340
    },
    {
      "epoch": 2.310706004637403,
      "grad_norm": 1.9834491676416712,
      "learning_rate": 4.1248766287565557e-07,
      "loss": 0.242,
      "step": 43350
    },
    {
      "epoch": 2.3112390394712294,
      "grad_norm": 1.6075310032902455,
      "learning_rate": 4.11880808458277e-07,
      "loss": 0.2277,
      "step": 43360
    },
    {
      "epoch": 2.311772074305056,
      "grad_norm": 1.843460793192278,
      "learning_rate": 4.112743297232681e-07,
      "loss": 0.2271,
      "step": 43370
    },
    {
      "epoch": 2.312305109138882,
      "grad_norm": 1.764649918319197,
      "learning_rate": 4.106682268800209e-07,
      "loss": 0.2342,
      "step": 43380
    },
    {
      "epoch": 2.3128381439727086,
      "grad_norm": 1.882159740560088,
      "learning_rate": 4.1006250013779815e-07,
      "loss": 0.2379,
      "step": 43390
    },
    {
      "epoch": 2.313371178806535,
      "grad_norm": 1.7846466642214778,
      "learning_rate": 4.094571497057315e-07,
      "loss": 0.2436,
      "step": 43400
    },
    {
      "epoch": 2.3139042136403614,
      "grad_norm": 1.7503959927259862,
      "learning_rate": 4.088521757928234e-07,
      "loss": 0.2282,
      "step": 43410
    },
    {
      "epoch": 2.3144372484741877,
      "grad_norm": 1.9495438425812188,
      "learning_rate": 4.082475786079463e-07,
      "loss": 0.2181,
      "step": 43420
    },
    {
      "epoch": 2.314970283308014,
      "grad_norm": 2.067940658477297,
      "learning_rate": 4.0764335835984254e-07,
      "loss": 0.2316,
      "step": 43430
    },
    {
      "epoch": 2.3155033181418405,
      "grad_norm": 1.9238285432771345,
      "learning_rate": 4.070395152571244e-07,
      "loss": 0.2303,
      "step": 43440
    },
    {
      "epoch": 2.316036352975667,
      "grad_norm": 1.7596627263325204,
      "learning_rate": 4.064360495082737e-07,
      "loss": 0.2391,
      "step": 43450
    },
    {
      "epoch": 2.3165693878094933,
      "grad_norm": 1.8634308869698077,
      "learning_rate": 4.0583296132164273e-07,
      "loss": 0.2313,
      "step": 43460
    },
    {
      "epoch": 2.3171024226433197,
      "grad_norm": 1.9099722976756353,
      "learning_rate": 4.0523025090545163e-07,
      "loss": 0.2272,
      "step": 43470
    },
    {
      "epoch": 2.317635457477146,
      "grad_norm": 1.9128345573092822,
      "learning_rate": 4.0462791846779185e-07,
      "loss": 0.2336,
      "step": 43480
    },
    {
      "epoch": 2.3181684923109724,
      "grad_norm": 2.0177637064665435,
      "learning_rate": 4.0402596421662367e-07,
      "loss": 0.2292,
      "step": 43490
    },
    {
      "epoch": 2.318701527144799,
      "grad_norm": 1.980972372806886,
      "learning_rate": 4.0342438835977706e-07,
      "loss": 0.2358,
      "step": 43500
    },
    {
      "epoch": 2.319234561978625,
      "grad_norm": 1.9184681073209102,
      "learning_rate": 4.02823191104951e-07,
      "loss": 0.2349,
      "step": 43510
    },
    {
      "epoch": 2.3197675968124516,
      "grad_norm": 1.7722290851541211,
      "learning_rate": 4.022223726597142e-07,
      "loss": 0.2316,
      "step": 43520
    },
    {
      "epoch": 2.320300631646278,
      "grad_norm": 1.8864562719843854,
      "learning_rate": 4.016219332315041e-07,
      "loss": 0.2326,
      "step": 43530
    },
    {
      "epoch": 2.3208336664801044,
      "grad_norm": 1.8069693763879173,
      "learning_rate": 4.010218730276281e-07,
      "loss": 0.2275,
      "step": 43540
    },
    {
      "epoch": 2.3213667013139307,
      "grad_norm": 2.0232109121801005,
      "learning_rate": 4.0042219225526084e-07,
      "loss": 0.2376,
      "step": 43550
    },
    {
      "epoch": 2.321899736147757,
      "grad_norm": 2.003864463839099,
      "learning_rate": 3.9982289112144816e-07,
      "loss": 0.2452,
      "step": 43560
    },
    {
      "epoch": 2.3224327709815835,
      "grad_norm": 1.9419991368143972,
      "learning_rate": 3.992239698331034e-07,
      "loss": 0.2249,
      "step": 43570
    },
    {
      "epoch": 2.32296580581541,
      "grad_norm": 1.9471153096981155,
      "learning_rate": 3.9862542859700947e-07,
      "loss": 0.2285,
      "step": 43580
    },
    {
      "epoch": 2.3234988406492363,
      "grad_norm": 2.000337669290855,
      "learning_rate": 3.9802726761981753e-07,
      "loss": 0.2359,
      "step": 43590
    },
    {
      "epoch": 2.3240318754830627,
      "grad_norm": 1.8254717959218814,
      "learning_rate": 3.974294871080482e-07,
      "loss": 0.2361,
      "step": 43600
    },
    {
      "epoch": 2.324564910316889,
      "grad_norm": 1.8650594242645446,
      "learning_rate": 3.968320872680897e-07,
      "loss": 0.2375,
      "step": 43610
    },
    {
      "epoch": 2.3250979451507154,
      "grad_norm": 1.725944562856312,
      "learning_rate": 3.9623506830619977e-07,
      "loss": 0.2246,
      "step": 43620
    },
    {
      "epoch": 2.325630979984542,
      "grad_norm": 1.91347340765011,
      "learning_rate": 3.9563843042850474e-07,
      "loss": 0.2488,
      "step": 43630
    },
    {
      "epoch": 2.326164014818368,
      "grad_norm": 1.6147853364370393,
      "learning_rate": 3.950421738409978e-07,
      "loss": 0.2291,
      "step": 43640
    },
    {
      "epoch": 2.3266970496521946,
      "grad_norm": 1.9493235160935352,
      "learning_rate": 3.944462987495419e-07,
      "loss": 0.2422,
      "step": 43650
    },
    {
      "epoch": 2.327230084486021,
      "grad_norm": 1.8332262940892994,
      "learning_rate": 3.938508053598684e-07,
      "loss": 0.2316,
      "step": 43660
    },
    {
      "epoch": 2.3277631193198474,
      "grad_norm": 1.7974237308584196,
      "learning_rate": 3.932556938775761e-07,
      "loss": 0.2303,
      "step": 43670
    },
    {
      "epoch": 2.3282961541536737,
      "grad_norm": 1.7566820981179891,
      "learning_rate": 3.9266096450813234e-07,
      "loss": 0.227,
      "step": 43680
    },
    {
      "epoch": 2.3288291889875,
      "grad_norm": 1.8340465834411588,
      "learning_rate": 3.920666174568725e-07,
      "loss": 0.2266,
      "step": 43690
    },
    {
      "epoch": 2.3293622238213265,
      "grad_norm": 1.839618194147695,
      "learning_rate": 3.9147265292900015e-07,
      "loss": 0.239,
      "step": 43700
    },
    {
      "epoch": 2.329895258655153,
      "grad_norm": 1.703268206510014,
      "learning_rate": 3.9087907112958655e-07,
      "loss": 0.233,
      "step": 43710
    },
    {
      "epoch": 2.3304282934889793,
      "grad_norm": 1.741300518072399,
      "learning_rate": 3.9028587226357044e-07,
      "loss": 0.2335,
      "step": 43720
    },
    {
      "epoch": 2.3309613283228057,
      "grad_norm": 2.047601950855497,
      "learning_rate": 3.8969305653575883e-07,
      "loss": 0.2269,
      "step": 43730
    },
    {
      "epoch": 2.331494363156632,
      "grad_norm": 1.8594699461092736,
      "learning_rate": 3.8910062415082665e-07,
      "loss": 0.228,
      "step": 43740
    },
    {
      "epoch": 2.3320273979904584,
      "grad_norm": 1.9306105713331234,
      "learning_rate": 3.8850857531331595e-07,
      "loss": 0.2404,
      "step": 43750
    },
    {
      "epoch": 2.332560432824285,
      "grad_norm": 1.8770415640069713,
      "learning_rate": 3.8791691022763694e-07,
      "loss": 0.2403,
      "step": 43760
    },
    {
      "epoch": 2.333093467658111,
      "grad_norm": 1.9193973920535186,
      "learning_rate": 3.873256290980666e-07,
      "loss": 0.2374,
      "step": 43770
    },
    {
      "epoch": 2.333626502491938,
      "grad_norm": 1.853722157096408,
      "learning_rate": 3.867347321287501e-07,
      "loss": 0.234,
      "step": 43780
    },
    {
      "epoch": 2.3341595373257644,
      "grad_norm": 1.7944476421511701,
      "learning_rate": 3.8614421952369934e-07,
      "loss": 0.2338,
      "step": 43790
    },
    {
      "epoch": 2.334692572159591,
      "grad_norm": 1.7157673748499487,
      "learning_rate": 3.8555409148679436e-07,
      "loss": 0.2322,
      "step": 43800
    },
    {
      "epoch": 2.335225606993417,
      "grad_norm": 1.971296922817286,
      "learning_rate": 3.849643482217811e-07,
      "loss": 0.2252,
      "step": 43810
    },
    {
      "epoch": 2.3357586418272436,
      "grad_norm": 1.8794398375986827,
      "learning_rate": 3.8437498993227383e-07,
      "loss": 0.2235,
      "step": 43820
    },
    {
      "epoch": 2.33629167666107,
      "grad_norm": 1.7166018963175234,
      "learning_rate": 3.837860168217532e-07,
      "loss": 0.2302,
      "step": 43830
    },
    {
      "epoch": 2.3368247114948963,
      "grad_norm": 2.1268001172631745,
      "learning_rate": 3.8319742909356753e-07,
      "loss": 0.2319,
      "step": 43840
    },
    {
      "epoch": 2.3373577463287227,
      "grad_norm": 1.858222160507147,
      "learning_rate": 3.826092269509315e-07,
      "loss": 0.2369,
      "step": 43850
    },
    {
      "epoch": 2.337890781162549,
      "grad_norm": 1.786878946659758,
      "learning_rate": 3.8202141059692715e-07,
      "loss": 0.2359,
      "step": 43860
    },
    {
      "epoch": 2.3384238159963755,
      "grad_norm": 1.7700973961809217,
      "learning_rate": 3.8143398023450264e-07,
      "loss": 0.2392,
      "step": 43870
    },
    {
      "epoch": 2.338956850830202,
      "grad_norm": 1.8965529849909888,
      "learning_rate": 3.8084693606647396e-07,
      "loss": 0.2387,
      "step": 43880
    },
    {
      "epoch": 2.3394898856640283,
      "grad_norm": 2.0219125392641604,
      "learning_rate": 3.8026027829552227e-07,
      "loss": 0.229,
      "step": 43890
    },
    {
      "epoch": 2.3400229204978547,
      "grad_norm": 1.575383109924614,
      "learning_rate": 3.7967400712419656e-07,
      "loss": 0.2367,
      "step": 43900
    },
    {
      "epoch": 2.340555955331681,
      "grad_norm": 1.7119252406645349,
      "learning_rate": 3.7908812275491177e-07,
      "loss": 0.2298,
      "step": 43910
    },
    {
      "epoch": 2.3410889901655074,
      "grad_norm": 1.9026835919537153,
      "learning_rate": 3.7850262538994956e-07,
      "loss": 0.2394,
      "step": 43920
    },
    {
      "epoch": 2.341622024999334,
      "grad_norm": 1.60726293277227,
      "learning_rate": 3.7791751523145776e-07,
      "loss": 0.227,
      "step": 43930
    },
    {
      "epoch": 2.34215505983316,
      "grad_norm": 1.8636574122114802,
      "learning_rate": 3.773327924814507e-07,
      "loss": 0.2438,
      "step": 43940
    },
    {
      "epoch": 2.3426880946669866,
      "grad_norm": 2.0283459079073896,
      "learning_rate": 3.767484573418088e-07,
      "loss": 0.2335,
      "step": 43950
    },
    {
      "epoch": 2.343221129500813,
      "grad_norm": 1.7168210068373488,
      "learning_rate": 3.761645100142788e-07,
      "loss": 0.2338,
      "step": 43960
    },
    {
      "epoch": 2.3437541643346393,
      "grad_norm": 1.708005975434955,
      "learning_rate": 3.75580950700474e-07,
      "loss": 0.2248,
      "step": 43970
    },
    {
      "epoch": 2.3442871991684657,
      "grad_norm": 1.9192579123528652,
      "learning_rate": 3.7499777960187197e-07,
      "loss": 0.2323,
      "step": 43980
    },
    {
      "epoch": 2.344820234002292,
      "grad_norm": 2.0162790446700325,
      "learning_rate": 3.7441499691981806e-07,
      "loss": 0.234,
      "step": 43990
    },
    {
      "epoch": 2.3453532688361185,
      "grad_norm": 1.9992310121658998,
      "learning_rate": 3.738326028555231e-07,
      "loss": 0.2351,
      "step": 44000
    },
    {
      "epoch": 2.345886303669945,
      "grad_norm": 2.081128203600095,
      "learning_rate": 3.732505976100635e-07,
      "loss": 0.2346,
      "step": 44010
    },
    {
      "epoch": 2.3464193385037713,
      "grad_norm": 2.0993624655991776,
      "learning_rate": 3.726689813843814e-07,
      "loss": 0.2296,
      "step": 44020
    },
    {
      "epoch": 2.3469523733375977,
      "grad_norm": 1.9986459316718512,
      "learning_rate": 3.720877543792849e-07,
      "loss": 0.2269,
      "step": 44030
    },
    {
      "epoch": 2.347485408171424,
      "grad_norm": 1.6067567539726733,
      "learning_rate": 3.7150691679544727e-07,
      "loss": 0.2315,
      "step": 44040
    },
    {
      "epoch": 2.3480184430052504,
      "grad_norm": 1.9850228955740072,
      "learning_rate": 3.7092646883340837e-07,
      "loss": 0.2399,
      "step": 44050
    },
    {
      "epoch": 2.348551477839077,
      "grad_norm": 1.895731229034648,
      "learning_rate": 3.703464106935716e-07,
      "loss": 0.2373,
      "step": 44060
    },
    {
      "epoch": 2.349084512672903,
      "grad_norm": 2.138490873402173,
      "learning_rate": 3.697667425762075e-07,
      "loss": 0.2333,
      "step": 44070
    },
    {
      "epoch": 2.3496175475067296,
      "grad_norm": 1.8225858237580619,
      "learning_rate": 3.6918746468145166e-07,
      "loss": 0.236,
      "step": 44080
    },
    {
      "epoch": 2.350150582340556,
      "grad_norm": 1.8938366352265967,
      "learning_rate": 3.686085772093042e-07,
      "loss": 0.2483,
      "step": 44090
    },
    {
      "epoch": 2.3506836171743823,
      "grad_norm": 1.9139577165775592,
      "learning_rate": 3.6803008035963126e-07,
      "loss": 0.2321,
      "step": 44100
    },
    {
      "epoch": 2.3512166520082087,
      "grad_norm": 2.053253716019063,
      "learning_rate": 3.6745197433216374e-07,
      "loss": 0.2354,
      "step": 44110
    },
    {
      "epoch": 2.351749686842035,
      "grad_norm": 1.8316551235267415,
      "learning_rate": 3.668742593264975e-07,
      "loss": 0.2391,
      "step": 44120
    },
    {
      "epoch": 2.3522827216758615,
      "grad_norm": 1.904036068525322,
      "learning_rate": 3.662969355420939e-07,
      "loss": 0.2308,
      "step": 44130
    },
    {
      "epoch": 2.352815756509688,
      "grad_norm": 1.8338111282682406,
      "learning_rate": 3.657200031782788e-07,
      "loss": 0.2338,
      "step": 44140
    },
    {
      "epoch": 2.3533487913435143,
      "grad_norm": 1.8796984246894208,
      "learning_rate": 3.6514346243424236e-07,
      "loss": 0.2402,
      "step": 44150
    },
    {
      "epoch": 2.3538818261773407,
      "grad_norm": 1.729343016486441,
      "learning_rate": 3.645673135090406e-07,
      "loss": 0.2279,
      "step": 44160
    },
    {
      "epoch": 2.354414861011167,
      "grad_norm": 1.942982037980061,
      "learning_rate": 3.639915566015939e-07,
      "loss": 0.2378,
      "step": 44170
    },
    {
      "epoch": 2.3549478958449934,
      "grad_norm": 1.8061841086698494,
      "learning_rate": 3.6341619191068706e-07,
      "loss": 0.2354,
      "step": 44180
    },
    {
      "epoch": 2.35548093067882,
      "grad_norm": 1.7595150339179584,
      "learning_rate": 3.6284121963496963e-07,
      "loss": 0.2398,
      "step": 44190
    },
    {
      "epoch": 2.356013965512646,
      "grad_norm": 1.80208043235444,
      "learning_rate": 3.622666399729558e-07,
      "loss": 0.2408,
      "step": 44200
    },
    {
      "epoch": 2.3565470003464726,
      "grad_norm": 1.9068596203402068,
      "learning_rate": 3.616924531230238e-07,
      "loss": 0.236,
      "step": 44210
    },
    {
      "epoch": 2.357080035180299,
      "grad_norm": 1.827092880708015,
      "learning_rate": 3.611186592834173e-07,
      "loss": 0.2336,
      "step": 44220
    },
    {
      "epoch": 2.3576130700141253,
      "grad_norm": 1.586436840595486,
      "learning_rate": 3.6054525865224236e-07,
      "loss": 0.2293,
      "step": 44230
    },
    {
      "epoch": 2.3581461048479517,
      "grad_norm": 1.866734222890412,
      "learning_rate": 3.5997225142747077e-07,
      "loss": 0.231,
      "step": 44240
    },
    {
      "epoch": 2.358679139681778,
      "grad_norm": 1.9085351986533177,
      "learning_rate": 3.5939963780693833e-07,
      "loss": 0.2411,
      "step": 44250
    },
    {
      "epoch": 2.3592121745156045,
      "grad_norm": 1.941876936676392,
      "learning_rate": 3.588274179883448e-07,
      "loss": 0.2398,
      "step": 44260
    },
    {
      "epoch": 2.359745209349431,
      "grad_norm": 1.7552977939621326,
      "learning_rate": 3.582555921692537e-07,
      "loss": 0.2338,
      "step": 44270
    },
    {
      "epoch": 2.3602782441832573,
      "grad_norm": 1.87750006585796,
      "learning_rate": 3.576841605470928e-07,
      "loss": 0.2365,
      "step": 44280
    },
    {
      "epoch": 2.3608112790170837,
      "grad_norm": 1.8335900004663377,
      "learning_rate": 3.5711312331915386e-07,
      "loss": 0.2349,
      "step": 44290
    },
    {
      "epoch": 2.36134431385091,
      "grad_norm": 1.7586720884536853,
      "learning_rate": 3.5654248068259197e-07,
      "loss": 0.2331,
      "step": 44300
    },
    {
      "epoch": 2.3618773486847364,
      "grad_norm": 1.8497322933294977,
      "learning_rate": 3.5597223283442727e-07,
      "loss": 0.2325,
      "step": 44310
    },
    {
      "epoch": 2.362410383518563,
      "grad_norm": 1.948999700885997,
      "learning_rate": 3.5540237997154136e-07,
      "loss": 0.2347,
      "step": 44320
    },
    {
      "epoch": 2.362943418352389,
      "grad_norm": 2.089930263418612,
      "learning_rate": 3.5483292229068135e-07,
      "loss": 0.233,
      "step": 44330
    },
    {
      "epoch": 2.3634764531862156,
      "grad_norm": 1.8252607559134169,
      "learning_rate": 3.5426385998845756e-07,
      "loss": 0.2257,
      "step": 44340
    },
    {
      "epoch": 2.364009488020042,
      "grad_norm": 1.910239711368906,
      "learning_rate": 3.5369519326134326e-07,
      "loss": 0.2414,
      "step": 44350
    },
    {
      "epoch": 2.3645425228538683,
      "grad_norm": 1.9372822163075252,
      "learning_rate": 3.531269223056754e-07,
      "loss": 0.2369,
      "step": 44360
    },
    {
      "epoch": 2.3650755576876947,
      "grad_norm": 1.7658933635224532,
      "learning_rate": 3.525590473176547e-07,
      "loss": 0.2282,
      "step": 44370
    },
    {
      "epoch": 2.365608592521521,
      "grad_norm": 1.8203418278159127,
      "learning_rate": 3.5199156849334453e-07,
      "loss": 0.2285,
      "step": 44380
    },
    {
      "epoch": 2.3661416273553475,
      "grad_norm": 1.903342823333316,
      "learning_rate": 3.5142448602867225e-07,
      "loss": 0.2301,
      "step": 44390
    },
    {
      "epoch": 2.366674662189174,
      "grad_norm": 1.914778583067269,
      "learning_rate": 3.50857800119427e-07,
      "loss": 0.2337,
      "step": 44400
    },
    {
      "epoch": 2.3672076970230003,
      "grad_norm": 1.7224811445103003,
      "learning_rate": 3.5029151096126245e-07,
      "loss": 0.2384,
      "step": 44410
    },
    {
      "epoch": 2.3677407318568267,
      "grad_norm": 1.82220783002562,
      "learning_rate": 3.497256187496946e-07,
      "loss": 0.2307,
      "step": 44420
    },
    {
      "epoch": 2.3682737666906535,
      "grad_norm": 1.7308138081951363,
      "learning_rate": 3.4916012368010256e-07,
      "loss": 0.2266,
      "step": 44430
    },
    {
      "epoch": 2.36880680152448,
      "grad_norm": 1.8043586888479446,
      "learning_rate": 3.4859502594772853e-07,
      "loss": 0.23,
      "step": 44440
    },
    {
      "epoch": 2.3693398363583063,
      "grad_norm": 1.8561944883823565,
      "learning_rate": 3.4803032574767687e-07,
      "loss": 0.231,
      "step": 44450
    },
    {
      "epoch": 2.3698728711921326,
      "grad_norm": 1.8076154914225688,
      "learning_rate": 3.474660232749153e-07,
      "loss": 0.2213,
      "step": 44460
    },
    {
      "epoch": 2.370405906025959,
      "grad_norm": 1.927947948267964,
      "learning_rate": 3.469021187242741e-07,
      "loss": 0.227,
      "step": 44470
    },
    {
      "epoch": 2.3709389408597854,
      "grad_norm": 1.9109891051459649,
      "learning_rate": 3.4633861229044634e-07,
      "loss": 0.2304,
      "step": 44480
    },
    {
      "epoch": 2.371471975693612,
      "grad_norm": 1.933937067531113,
      "learning_rate": 3.4577550416798676e-07,
      "loss": 0.2315,
      "step": 44490
    },
    {
      "epoch": 2.372005010527438,
      "grad_norm": 1.8457800873991088,
      "learning_rate": 3.4521279455131355e-07,
      "loss": 0.235,
      "step": 44500
    },
    {
      "epoch": 2.3725380453612646,
      "grad_norm": 1.9410414547652006,
      "learning_rate": 3.4465048363470694e-07,
      "loss": 0.2375,
      "step": 44510
    },
    {
      "epoch": 2.373071080195091,
      "grad_norm": 1.727851345591523,
      "learning_rate": 3.4408857161230954e-07,
      "loss": 0.2261,
      "step": 44520
    },
    {
      "epoch": 2.3736041150289173,
      "grad_norm": 1.896459098773258,
      "learning_rate": 3.435270586781263e-07,
      "loss": 0.2408,
      "step": 44530
    },
    {
      "epoch": 2.3741371498627437,
      "grad_norm": 1.9590896210390245,
      "learning_rate": 3.4296594502602425e-07,
      "loss": 0.2342,
      "step": 44540
    },
    {
      "epoch": 2.37467018469657,
      "grad_norm": 1.8361239227182435,
      "learning_rate": 3.424052308497326e-07,
      "loss": 0.2373,
      "step": 44550
    },
    {
      "epoch": 2.3752032195303965,
      "grad_norm": 1.8046521580889512,
      "learning_rate": 3.4184491634284324e-07,
      "loss": 0.2288,
      "step": 44560
    },
    {
      "epoch": 2.375736254364223,
      "grad_norm": 1.868681825665443,
      "learning_rate": 3.4128500169880857e-07,
      "loss": 0.2333,
      "step": 44570
    },
    {
      "epoch": 2.3762692891980493,
      "grad_norm": 1.8731354530554936,
      "learning_rate": 3.407254871109442e-07,
      "loss": 0.2329,
      "step": 44580
    },
    {
      "epoch": 2.3768023240318756,
      "grad_norm": 1.8840990667162818,
      "learning_rate": 3.4016637277242737e-07,
      "loss": 0.2326,
      "step": 44590
    },
    {
      "epoch": 2.377335358865702,
      "grad_norm": 1.7767063171137862,
      "learning_rate": 3.396076588762969e-07,
      "loss": 0.2325,
      "step": 44600
    },
    {
      "epoch": 2.3778683936995284,
      "grad_norm": 1.9163964727088443,
      "learning_rate": 3.3904934561545373e-07,
      "loss": 0.2372,
      "step": 44610
    },
    {
      "epoch": 2.378401428533355,
      "grad_norm": 1.836263090050185,
      "learning_rate": 3.3849143318266034e-07,
      "loss": 0.2261,
      "step": 44620
    },
    {
      "epoch": 2.378934463367181,
      "grad_norm": 1.8265048798532915,
      "learning_rate": 3.3793392177054025e-07,
      "loss": 0.2406,
      "step": 44630
    },
    {
      "epoch": 2.3794674982010076,
      "grad_norm": 2.0797589363804834,
      "learning_rate": 3.373768115715799e-07,
      "loss": 0.2323,
      "step": 44640
    },
    {
      "epoch": 2.380000533034834,
      "grad_norm": 1.8147188310523423,
      "learning_rate": 3.3682010277812504e-07,
      "loss": 0.2422,
      "step": 44650
    },
    {
      "epoch": 2.3805335678686603,
      "grad_norm": 2.01982338281718,
      "learning_rate": 3.362637955823849e-07,
      "loss": 0.2361,
      "step": 44660
    },
    {
      "epoch": 2.3810666027024867,
      "grad_norm": 1.8751944560296212,
      "learning_rate": 3.357078901764291e-07,
      "loss": 0.2364,
      "step": 44670
    },
    {
      "epoch": 2.381599637536313,
      "grad_norm": 1.75777002273749,
      "learning_rate": 3.351523867521886e-07,
      "loss": 0.2325,
      "step": 44680
    },
    {
      "epoch": 2.3821326723701395,
      "grad_norm": 1.7277700396005058,
      "learning_rate": 3.3459728550145583e-07,
      "loss": 0.2367,
      "step": 44690
    },
    {
      "epoch": 2.382665707203966,
      "grad_norm": 2.297881216656287,
      "learning_rate": 3.340425866158841e-07,
      "loss": 0.2417,
      "step": 44700
    },
    {
      "epoch": 2.3831987420377923,
      "grad_norm": 1.761500407236202,
      "learning_rate": 3.3348829028698783e-07,
      "loss": 0.2333,
      "step": 44710
    },
    {
      "epoch": 2.3837317768716186,
      "grad_norm": 1.8977749758015268,
      "learning_rate": 3.329343967061427e-07,
      "loss": 0.2286,
      "step": 44720
    },
    {
      "epoch": 2.384264811705445,
      "grad_norm": 2.0101267240735914,
      "learning_rate": 3.3238090606458544e-07,
      "loss": 0.234,
      "step": 44730
    },
    {
      "epoch": 2.3847978465392714,
      "grad_norm": 1.8596815489904357,
      "learning_rate": 3.318278185534126e-07,
      "loss": 0.227,
      "step": 44740
    },
    {
      "epoch": 2.385330881373098,
      "grad_norm": 1.6824458830117548,
      "learning_rate": 3.3127513436358324e-07,
      "loss": 0.2389,
      "step": 44750
    },
    {
      "epoch": 2.385863916206924,
      "grad_norm": 1.6644788570158902,
      "learning_rate": 3.307228536859153e-07,
      "loss": 0.2419,
      "step": 44760
    },
    {
      "epoch": 2.3863969510407506,
      "grad_norm": 1.8699469264626025,
      "learning_rate": 3.3017097671108876e-07,
      "loss": 0.2446,
      "step": 44770
    },
    {
      "epoch": 2.386929985874577,
      "grad_norm": 2.0404994297497745,
      "learning_rate": 3.29619503629644e-07,
      "loss": 0.2275,
      "step": 44780
    },
    {
      "epoch": 2.3874630207084033,
      "grad_norm": 1.7679521272542127,
      "learning_rate": 3.290684346319815e-07,
      "loss": 0.235,
      "step": 44790
    },
    {
      "epoch": 2.3879960555422297,
      "grad_norm": 1.8270346535223767,
      "learning_rate": 3.285177699083629e-07,
      "loss": 0.236,
      "step": 44800
    },
    {
      "epoch": 2.388529090376056,
      "grad_norm": 1.7096500971707846,
      "learning_rate": 3.2796750964890946e-07,
      "loss": 0.2326,
      "step": 44810
    },
    {
      "epoch": 2.3890621252098825,
      "grad_norm": 2.0496169854011987,
      "learning_rate": 3.274176540436034e-07,
      "loss": 0.2379,
      "step": 44820
    },
    {
      "epoch": 2.389595160043709,
      "grad_norm": 1.6942251010521172,
      "learning_rate": 3.2686820328228734e-07,
      "loss": 0.2338,
      "step": 44830
    },
    {
      "epoch": 2.3901281948775353,
      "grad_norm": 2.0017609267722407,
      "learning_rate": 3.263191575546632e-07,
      "loss": 0.2375,
      "step": 44840
    },
    {
      "epoch": 2.3906612297113616,
      "grad_norm": 2.077435475975477,
      "learning_rate": 3.2577051705029365e-07,
      "loss": 0.229,
      "step": 44850
    },
    {
      "epoch": 2.391194264545188,
      "grad_norm": 1.7167448395510585,
      "learning_rate": 3.25222281958602e-07,
      "loss": 0.2325,
      "step": 44860
    },
    {
      "epoch": 2.3917272993790144,
      "grad_norm": 1.9193049587787723,
      "learning_rate": 3.246744524688708e-07,
      "loss": 0.227,
      "step": 44870
    },
    {
      "epoch": 2.392260334212841,
      "grad_norm": 1.9681925185217501,
      "learning_rate": 3.2412702877024274e-07,
      "loss": 0.2368,
      "step": 44880
    },
    {
      "epoch": 2.392793369046667,
      "grad_norm": 1.74948522690138,
      "learning_rate": 3.235800110517206e-07,
      "loss": 0.2387,
      "step": 44890
    },
    {
      "epoch": 2.3933264038804936,
      "grad_norm": 1.921881904525258,
      "learning_rate": 3.2303339950216687e-07,
      "loss": 0.2352,
      "step": 44900
    },
    {
      "epoch": 2.39385943871432,
      "grad_norm": 2.0156324457037624,
      "learning_rate": 3.224871943103038e-07,
      "loss": 0.2255,
      "step": 44910
    },
    {
      "epoch": 2.3943924735481463,
      "grad_norm": 2.0184942248269238,
      "learning_rate": 3.219413956647138e-07,
      "loss": 0.2395,
      "step": 44920
    },
    {
      "epoch": 2.3949255083819727,
      "grad_norm": 1.977277192837676,
      "learning_rate": 3.213960037538376e-07,
      "loss": 0.2382,
      "step": 44930
    },
    {
      "epoch": 2.395458543215799,
      "grad_norm": 1.9722131989752463,
      "learning_rate": 3.208510187659769e-07,
      "loss": 0.2255,
      "step": 44940
    },
    {
      "epoch": 2.3959915780496255,
      "grad_norm": 1.7815611620797984,
      "learning_rate": 3.2030644088929226e-07,
      "loss": 0.2379,
      "step": 44950
    },
    {
      "epoch": 2.396524612883452,
      "grad_norm": 1.7455627849084154,
      "learning_rate": 3.1976227031180387e-07,
      "loss": 0.2239,
      "step": 44960
    },
    {
      "epoch": 2.3970576477172783,
      "grad_norm": 1.6978419794211683,
      "learning_rate": 3.1921850722139123e-07,
      "loss": 0.2334,
      "step": 44970
    },
    {
      "epoch": 2.3975906825511046,
      "grad_norm": 2.4973798112075034,
      "learning_rate": 3.1867515180579317e-07,
      "loss": 0.2302,
      "step": 44980
    },
    {
      "epoch": 2.398123717384931,
      "grad_norm": 1.9424031813213634,
      "learning_rate": 3.1813220425260757e-07,
      "loss": 0.2343,
      "step": 44990
    },
    {
      "epoch": 2.3986567522187574,
      "grad_norm": 1.870039107482244,
      "learning_rate": 3.175896647492924e-07,
      "loss": 0.2325,
      "step": 45000
    },
    {
      "epoch": 2.399189787052584,
      "grad_norm": 1.7563257979622962,
      "learning_rate": 3.170475334831631e-07,
      "loss": 0.2307,
      "step": 45010
    },
    {
      "epoch": 2.39972282188641,
      "grad_norm": 1.9611417379053857,
      "learning_rate": 3.1650581064139533e-07,
      "loss": 0.231,
      "step": 45020
    }
  ],
  "logging_steps": 10,
  "max_steps": 56280,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 5628,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3453884985475072.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}