| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9997828447339848, | |
| "eval_steps": 500, | |
| "global_step": 1151, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0008686210640608035, | |
| "grad_norm": 17.493449382349226, | |
| "learning_rate": 3.3333333333333335e-07, | |
| "loss": 8.7443, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.001737242128121607, | |
| "grad_norm": 14.990450099593053, | |
| "learning_rate": 6.666666666666667e-07, | |
| "loss": 8.6957, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0026058631921824105, | |
| "grad_norm": 16.423535165287575, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 8.692, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.003474484256243214, | |
| "grad_norm": 13.143676841573155, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 8.6702, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.004343105320304018, | |
| "grad_norm": 8.964338427609599, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 8.6436, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.005211726384364821, | |
| "grad_norm": 5.711486770337647, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 8.6472, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.006080347448425625, | |
| "grad_norm": 8.527323895494213, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 8.7054, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.006948968512486428, | |
| "grad_norm": 9.12482767652379, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 8.7084, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.007817589576547232, | |
| "grad_norm": 5.337796907422001, | |
| "learning_rate": 3e-06, | |
| "loss": 8.5973, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.008686210640608035, | |
| "grad_norm": 3.563435103977848, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 8.594, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.009554831704668839, | |
| "grad_norm": 2.8337150046011836, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 8.562, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.010423452768729642, | |
| "grad_norm": 5.305490346258808, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 8.5552, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.011292073832790446, | |
| "grad_norm": 3.8824803788805204, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 8.5319, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.01216069489685125, | |
| "grad_norm": 2.779267764600503, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 8.5131, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.013029315960912053, | |
| "grad_norm": 2.8850795560965943, | |
| "learning_rate": 5e-06, | |
| "loss": 8.4689, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.013897937024972856, | |
| "grad_norm": 2.7572282850785186, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 8.574, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.01476655808903366, | |
| "grad_norm": 2.573141583562948, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 8.4873, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.015635179153094463, | |
| "grad_norm": 3.180322337220607, | |
| "learning_rate": 6e-06, | |
| "loss": 8.4521, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.016503800217155265, | |
| "grad_norm": 3.491770308520874, | |
| "learning_rate": 6.333333333333333e-06, | |
| "loss": 8.3846, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.01737242128121607, | |
| "grad_norm": 2.540124589419276, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 8.4451, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.018241042345276872, | |
| "grad_norm": 3.121998583339167, | |
| "learning_rate": 7e-06, | |
| "loss": 8.2468, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.019109663409337677, | |
| "grad_norm": 4.290808656004097, | |
| "learning_rate": 7.333333333333333e-06, | |
| "loss": 8.3109, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.01997828447339848, | |
| "grad_norm": 4.450357829892193, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 8.2419, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.020846905537459284, | |
| "grad_norm": 3.6215464038427188, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 8.2952, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.021715526601520086, | |
| "grad_norm": 4.442035772364645, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 8.377, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02258414766558089, | |
| "grad_norm": 4.843618321464985, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 8.2759, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.023452768729641693, | |
| "grad_norm": 2.3677599699211487, | |
| "learning_rate": 9e-06, | |
| "loss": 8.2335, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0243213897937025, | |
| "grad_norm": 2.7402708582612223, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 8.1205, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0251900108577633, | |
| "grad_norm": 3.367587996037935, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 8.1712, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.026058631921824105, | |
| "grad_norm": 3.7838438130817647, | |
| "learning_rate": 1e-05, | |
| "loss": 8.155, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.026927252985884907, | |
| "grad_norm": 4.375332206336344, | |
| "learning_rate": 9.999982328608275e-06, | |
| "loss": 8.143, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.027795874049945712, | |
| "grad_norm": 6.575802280327978, | |
| "learning_rate": 9.999929314571892e-06, | |
| "loss": 8.1258, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.028664495114006514, | |
| "grad_norm": 9.035297148846244, | |
| "learning_rate": 9.99984095830722e-06, | |
| "loss": 8.0348, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.02953311617806732, | |
| "grad_norm": 6.634311793226191, | |
| "learning_rate": 9.999717260508202e-06, | |
| "loss": 8.0355, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.03040173724212812, | |
| "grad_norm": 8.124508838108945, | |
| "learning_rate": 9.999558222146359e-06, | |
| "loss": 7.9736, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.031270358306188926, | |
| "grad_norm": 5.8285745211892515, | |
| "learning_rate": 9.999363844470767e-06, | |
| "loss": 7.9337, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.03213897937024973, | |
| "grad_norm": 9.094040714600302, | |
| "learning_rate": 9.999134129008061e-06, | |
| "loss": 7.8325, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.03300760043431053, | |
| "grad_norm": 8.468740095393665, | |
| "learning_rate": 9.998869077562416e-06, | |
| "loss": 7.8086, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.033876221498371335, | |
| "grad_norm": 6.732505858210668, | |
| "learning_rate": 9.998568692215532e-06, | |
| "loss": 7.6726, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.03474484256243214, | |
| "grad_norm": 21.57246390769929, | |
| "learning_rate": 9.998232975326619e-06, | |
| "loss": 7.8017, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.035613463626492946, | |
| "grad_norm": 12.672871887843293, | |
| "learning_rate": 9.997861929532384e-06, | |
| "loss": 7.6013, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.036482084690553744, | |
| "grad_norm": 9.395510661131551, | |
| "learning_rate": 9.997455557747002e-06, | |
| "loss": 7.4226, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.03735070575461455, | |
| "grad_norm": 12.735314205401677, | |
| "learning_rate": 9.9970138631621e-06, | |
| "loss": 7.3855, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.038219326818675355, | |
| "grad_norm": 10.285907675522205, | |
| "learning_rate": 9.99653684924672e-06, | |
| "loss": 7.3198, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.03908794788273615, | |
| "grad_norm": 9.965187203305367, | |
| "learning_rate": 9.996024519747312e-06, | |
| "loss": 7.1805, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.03995656894679696, | |
| "grad_norm": 17.926836165323984, | |
| "learning_rate": 9.995476878687687e-06, | |
| "loss": 6.9857, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.04082519001085776, | |
| "grad_norm": 23.24576845075726, | |
| "learning_rate": 9.994893930368987e-06, | |
| "loss": 7.0586, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.04169381107491857, | |
| "grad_norm": 10.789226700023386, | |
| "learning_rate": 9.994275679369664e-06, | |
| "loss": 6.8635, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.04256243213897937, | |
| "grad_norm": 20.819248588210748, | |
| "learning_rate": 9.99362213054543e-06, | |
| "loss": 6.8014, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.04343105320304017, | |
| "grad_norm": 13.054140096829588, | |
| "learning_rate": 9.992933289029225e-06, | |
| "loss": 6.6718, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04429967426710098, | |
| "grad_norm": 31.11350513451145, | |
| "learning_rate": 9.992209160231182e-06, | |
| "loss": 6.7444, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.04516829533116178, | |
| "grad_norm": 18.66011619667928, | |
| "learning_rate": 9.991449749838567e-06, | |
| "loss": 6.4907, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.04603691639522258, | |
| "grad_norm": 30.236445483678942, | |
| "learning_rate": 9.990655063815758e-06, | |
| "loss": 6.4475, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.046905537459283386, | |
| "grad_norm": 21.993099100665887, | |
| "learning_rate": 9.989825108404178e-06, | |
| "loss": 6.4676, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.04777415852334419, | |
| "grad_norm": 19.83551768600084, | |
| "learning_rate": 9.988959890122257e-06, | |
| "loss": 6.2759, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.048642779587405, | |
| "grad_norm": 13.170118097146572, | |
| "learning_rate": 9.988059415765378e-06, | |
| "loss": 6.1095, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.049511400651465795, | |
| "grad_norm": 22.67994571341824, | |
| "learning_rate": 9.987123692405825e-06, | |
| "loss": 6.0883, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0503800217155266, | |
| "grad_norm": 13.038280028112723, | |
| "learning_rate": 9.986152727392721e-06, | |
| "loss": 6.0249, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.051248642779587406, | |
| "grad_norm": 14.901505032251393, | |
| "learning_rate": 9.985146528351983e-06, | |
| "loss": 6.1762, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.05211726384364821, | |
| "grad_norm": 26.832853386529305, | |
| "learning_rate": 9.984105103186245e-06, | |
| "loss": 6.08, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.05298588490770901, | |
| "grad_norm": 15.17632623872157, | |
| "learning_rate": 9.983028460074811e-06, | |
| "loss": 5.9952, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.053854505971769814, | |
| "grad_norm": 24.418647213861092, | |
| "learning_rate": 9.981916607473589e-06, | |
| "loss": 5.9961, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.05472312703583062, | |
| "grad_norm": 18.746932536918226, | |
| "learning_rate": 9.98076955411501e-06, | |
| "loss": 5.8426, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.055591748099891425, | |
| "grad_norm": 21.558720998266843, | |
| "learning_rate": 9.97958730900798e-06, | |
| "loss": 5.7789, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.05646036916395222, | |
| "grad_norm": 28.510478407780514, | |
| "learning_rate": 9.97836988143779e-06, | |
| "loss": 5.7467, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.05732899022801303, | |
| "grad_norm": 15.832636672012686, | |
| "learning_rate": 9.977117280966065e-06, | |
| "loss": 5.8336, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.058197611292073834, | |
| "grad_norm": 29.18156046853244, | |
| "learning_rate": 9.975829517430662e-06, | |
| "loss": 5.8591, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.05906623235613464, | |
| "grad_norm": 18.247831823792946, | |
| "learning_rate": 9.974506600945618e-06, | |
| "loss": 5.7356, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.05993485342019544, | |
| "grad_norm": 23.96714109884484, | |
| "learning_rate": 9.973148541901053e-06, | |
| "loss": 5.7721, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.06080347448425624, | |
| "grad_norm": 18.93369316823687, | |
| "learning_rate": 9.9717553509631e-06, | |
| "loss": 5.7324, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.06167209554831705, | |
| "grad_norm": 22.87674608218244, | |
| "learning_rate": 9.97032703907381e-06, | |
| "loss": 5.7795, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.06254071661237785, | |
| "grad_norm": 22.61900902318279, | |
| "learning_rate": 9.968863617451078e-06, | |
| "loss": 5.7367, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.06340933767643865, | |
| "grad_norm": 17.381975236661205, | |
| "learning_rate": 9.967365097588548e-06, | |
| "loss": 5.4779, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.06427795874049946, | |
| "grad_norm": 16.21661033530993, | |
| "learning_rate": 9.965831491255521e-06, | |
| "loss": 5.7334, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.06514657980456026, | |
| "grad_norm": 13.791459531600026, | |
| "learning_rate": 9.964262810496867e-06, | |
| "loss": 5.5777, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.06601520086862106, | |
| "grad_norm": 20.390991279830846, | |
| "learning_rate": 9.962659067632933e-06, | |
| "loss": 5.5614, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.06688382193268187, | |
| "grad_norm": 16.095238532163904, | |
| "learning_rate": 9.961020275259433e-06, | |
| "loss": 5.6062, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.06775244299674267, | |
| "grad_norm": 9.877965182839311, | |
| "learning_rate": 9.959346446247367e-06, | |
| "loss": 5.6125, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.06862106406080347, | |
| "grad_norm": 17.028948309558636, | |
| "learning_rate": 9.957637593742905e-06, | |
| "loss": 5.5812, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.06948968512486428, | |
| "grad_norm": 8.979583407927759, | |
| "learning_rate": 9.955893731167295e-06, | |
| "loss": 5.5746, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.07035830618892508, | |
| "grad_norm": 12.03923638076693, | |
| "learning_rate": 9.95411487221675e-06, | |
| "loss": 5.6653, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.07122692725298589, | |
| "grad_norm": 12.390003162693889, | |
| "learning_rate": 9.952301030862337e-06, | |
| "loss": 5.4735, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.07209554831704669, | |
| "grad_norm": 9.315040920552768, | |
| "learning_rate": 9.950452221349887e-06, | |
| "loss": 5.5041, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.07296416938110749, | |
| "grad_norm": 10.904257522944858, | |
| "learning_rate": 9.948568458199856e-06, | |
| "loss": 5.4586, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.0738327904451683, | |
| "grad_norm": 12.975062616231535, | |
| "learning_rate": 9.94664975620723e-06, | |
| "loss": 5.3053, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.0747014115092291, | |
| "grad_norm": 20.113827062713174, | |
| "learning_rate": 9.944696130441399e-06, | |
| "loss": 5.5913, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.0755700325732899, | |
| "grad_norm": 12.517008621704704, | |
| "learning_rate": 9.942707596246051e-06, | |
| "loss": 5.337, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.07643865363735071, | |
| "grad_norm": 9.758313745988033, | |
| "learning_rate": 9.940684169239035e-06, | |
| "loss": 5.3173, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.07730727470141151, | |
| "grad_norm": 10.515920909580231, | |
| "learning_rate": 9.938625865312252e-06, | |
| "loss": 5.4537, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0781758957654723, | |
| "grad_norm": 14.20391683761756, | |
| "learning_rate": 9.936532700631518e-06, | |
| "loss": 5.5486, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.07904451682953312, | |
| "grad_norm": 6.514270269870405, | |
| "learning_rate": 9.93440469163645e-06, | |
| "loss": 5.538, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.07991313789359392, | |
| "grad_norm": 24.656805213333442, | |
| "learning_rate": 9.932241855040328e-06, | |
| "loss": 5.3992, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.08078175895765473, | |
| "grad_norm": 15.24785447513823, | |
| "learning_rate": 9.930044207829966e-06, | |
| "loss": 5.4491, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.08165038002171553, | |
| "grad_norm": 17.034838291407535, | |
| "learning_rate": 9.927811767265581e-06, | |
| "loss": 5.5654, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.08251900108577633, | |
| "grad_norm": 16.66779749233834, | |
| "learning_rate": 9.925544550880653e-06, | |
| "loss": 5.3037, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.08338762214983714, | |
| "grad_norm": 13.53801819272121, | |
| "learning_rate": 9.92324257648179e-06, | |
| "loss": 5.1831, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.08425624321389794, | |
| "grad_norm": 13.925819342029348, | |
| "learning_rate": 9.920905862148586e-06, | |
| "loss": 5.3506, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.08512486427795873, | |
| "grad_norm": 11.595896262884168, | |
| "learning_rate": 9.918534426233486e-06, | |
| "loss": 5.3379, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.08599348534201955, | |
| "grad_norm": 10.793238449505045, | |
| "learning_rate": 9.916128287361634e-06, | |
| "loss": 5.2807, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.08686210640608034, | |
| "grad_norm": 9.100694836617444, | |
| "learning_rate": 9.913687464430727e-06, | |
| "loss": 5.3927, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08773072747014116, | |
| "grad_norm": 10.570344665746836, | |
| "learning_rate": 9.91121197661087e-06, | |
| "loss": 5.3416, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.08859934853420195, | |
| "grad_norm": 8.054172083905852, | |
| "learning_rate": 9.908701843344427e-06, | |
| "loss": 5.4551, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.08946796959826275, | |
| "grad_norm": 9.349388343495786, | |
| "learning_rate": 9.906157084345865e-06, | |
| "loss": 5.3192, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.09033659066232357, | |
| "grad_norm": 7.475662385329425, | |
| "learning_rate": 9.903577719601597e-06, | |
| "loss": 5.2733, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.09120521172638436, | |
| "grad_norm": 7.943968157184266, | |
| "learning_rate": 9.900963769369827e-06, | |
| "loss": 5.3985, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.09207383279044516, | |
| "grad_norm": 7.531045700331419, | |
| "learning_rate": 9.8983152541804e-06, | |
| "loss": 5.3622, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.09294245385450597, | |
| "grad_norm": 15.575311318718667, | |
| "learning_rate": 9.895632194834625e-06, | |
| "loss": 5.1931, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.09381107491856677, | |
| "grad_norm": 11.029594288829802, | |
| "learning_rate": 9.892914612405117e-06, | |
| "loss": 5.2938, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.09467969598262758, | |
| "grad_norm": 11.455193260474546, | |
| "learning_rate": 9.890162528235641e-06, | |
| "loss": 5.1901, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.09554831704668838, | |
| "grad_norm": 11.086046323655495, | |
| "learning_rate": 9.887375963940936e-06, | |
| "loss": 5.4614, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.09641693811074918, | |
| "grad_norm": 7.999590835458638, | |
| "learning_rate": 9.884554941406539e-06, | |
| "loss": 5.2932, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.09728555917481, | |
| "grad_norm": 11.326641474544164, | |
| "learning_rate": 9.881699482788627e-06, | |
| "loss": 5.1322, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.09815418023887079, | |
| "grad_norm": 7.633538464939616, | |
| "learning_rate": 9.878809610513836e-06, | |
| "loss": 5.1649, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.09902280130293159, | |
| "grad_norm": 9.626046071760667, | |
| "learning_rate": 9.87588534727908e-06, | |
| "loss": 5.3233, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.0998914223669924, | |
| "grad_norm": 9.133312938258387, | |
| "learning_rate": 9.872926716051387e-06, | |
| "loss": 5.1132, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.1007600434310532, | |
| "grad_norm": 9.505514994492872, | |
| "learning_rate": 9.869933740067703e-06, | |
| "loss": 5.1579, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.10162866449511401, | |
| "grad_norm": 8.791259591754573, | |
| "learning_rate": 9.866906442834713e-06, | |
| "loss": 5.2137, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.10249728555917481, | |
| "grad_norm": 10.053006259051758, | |
| "learning_rate": 9.863844848128668e-06, | |
| "loss": 5.2787, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.10336590662323561, | |
| "grad_norm": 13.131490348069805, | |
| "learning_rate": 9.860748979995183e-06, | |
| "loss": 4.9206, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.10423452768729642, | |
| "grad_norm": 6.631095589009018, | |
| "learning_rate": 9.857618862749062e-06, | |
| "loss": 5.2466, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.10510314875135722, | |
| "grad_norm": 13.217875916209636, | |
| "learning_rate": 9.85445452097409e-06, | |
| "loss": 5.2414, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.10597176981541802, | |
| "grad_norm": 8.029250043406146, | |
| "learning_rate": 9.851255979522856e-06, | |
| "loss": 5.2792, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.10684039087947883, | |
| "grad_norm": 17.54497296410541, | |
| "learning_rate": 9.848023263516552e-06, | |
| "loss": 5.4227, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.10770901194353963, | |
| "grad_norm": 11.789384484790114, | |
| "learning_rate": 9.844756398344773e-06, | |
| "loss": 5.2564, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.10857763300760044, | |
| "grad_norm": 12.89494048195122, | |
| "learning_rate": 9.841455409665322e-06, | |
| "loss": 5.173, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.10944625407166124, | |
| "grad_norm": 10.08813144239029, | |
| "learning_rate": 9.838120323404004e-06, | |
| "loss": 5.2173, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.11031487513572204, | |
| "grad_norm": 8.840110668733463, | |
| "learning_rate": 9.834751165754428e-06, | |
| "loss": 5.237, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.11118349619978285, | |
| "grad_norm": 10.088550995951381, | |
| "learning_rate": 9.831347963177794e-06, | |
| "loss": 5.2767, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.11205211726384365, | |
| "grad_norm": 8.195942675919266, | |
| "learning_rate": 9.827910742402693e-06, | |
| "loss": 5.2096, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.11292073832790445, | |
| "grad_norm": 11.245032884796654, | |
| "learning_rate": 9.824439530424888e-06, | |
| "loss": 5.1405, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.11378935939196526, | |
| "grad_norm": 10.023313323111374, | |
| "learning_rate": 9.820934354507113e-06, | |
| "loss": 5.2706, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.11465798045602606, | |
| "grad_norm": 5.266854203910289, | |
| "learning_rate": 9.817395242178854e-06, | |
| "loss": 5.2976, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.11552660152008687, | |
| "grad_norm": 13.356121226198322, | |
| "learning_rate": 9.813822221236125e-06, | |
| "loss": 5.1784, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.11639522258414767, | |
| "grad_norm": 7.209004502332046, | |
| "learning_rate": 9.810215319741262e-06, | |
| "loss": 5.2083, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.11726384364820847, | |
| "grad_norm": 9.476007633968416, | |
| "learning_rate": 9.806574566022693e-06, | |
| "loss": 5.1428, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.11813246471226928, | |
| "grad_norm": 8.330598001164512, | |
| "learning_rate": 9.802899988674728e-06, | |
| "loss": 5.0473, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.11900108577633008, | |
| "grad_norm": 6.96008730119411, | |
| "learning_rate": 9.799191616557317e-06, | |
| "loss": 5.2599, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.11986970684039087, | |
| "grad_norm": 7.736962765205919, | |
| "learning_rate": 9.795449478795836e-06, | |
| "loss": 5.2652, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.12073832790445169, | |
| "grad_norm": 6.749695837587577, | |
| "learning_rate": 9.791673604780856e-06, | |
| "loss": 5.238, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.12160694896851248, | |
| "grad_norm": 10.329369152037573, | |
| "learning_rate": 9.787864024167911e-06, | |
| "loss": 5.1537, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.12247557003257328, | |
| "grad_norm": 6.905477249008306, | |
| "learning_rate": 9.78402076687726e-06, | |
| "loss": 5.2558, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.1233441910966341, | |
| "grad_norm": 11.627003258310488, | |
| "learning_rate": 9.780143863093663e-06, | |
| "loss": 5.2708, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.1242128121606949, | |
| "grad_norm": 7.001100481839455, | |
| "learning_rate": 9.776233343266138e-06, | |
| "loss": 5.1064, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.1250814332247557, | |
| "grad_norm": 11.285622964490454, | |
| "learning_rate": 9.772289238107717e-06, | |
| "loss": 4.9967, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.1259500542888165, | |
| "grad_norm": 7.568164194677143, | |
| "learning_rate": 9.768311578595212e-06, | |
| "loss": 5.014, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.1268186753528773, | |
| "grad_norm": 9.381543499879124, | |
| "learning_rate": 9.764300395968969e-06, | |
| "loss": 5.0742, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.1276872964169381, | |
| "grad_norm": 8.76903592042971, | |
| "learning_rate": 9.760255721732626e-06, | |
| "loss": 4.9568, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.12855591748099893, | |
| "grad_norm": 7.998187265875589, | |
| "learning_rate": 9.756177587652857e-06, | |
| "loss": 5.1517, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.12942453854505973, | |
| "grad_norm": 10.119399253599614, | |
| "learning_rate": 9.752066025759132e-06, | |
| "loss": 5.0338, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.13029315960912052, | |
| "grad_norm": 6.547018882815891, | |
| "learning_rate": 9.74792106834346e-06, | |
| "loss": 5.0788, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.13116178067318132, | |
| "grad_norm": 8.201656095763651, | |
| "learning_rate": 9.743742747960138e-06, | |
| "loss": 5.193, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.13203040173724212, | |
| "grad_norm": 6.751595098502055, | |
| "learning_rate": 9.739531097425493e-06, | |
| "loss": 5.0545, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.13289902280130292, | |
| "grad_norm": 11.073060727681726, | |
| "learning_rate": 9.735286149817623e-06, | |
| "loss": 4.9874, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.13376764386536374, | |
| "grad_norm": 6.413004734395765, | |
| "learning_rate": 9.731007938476145e-06, | |
| "loss": 5.1605, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.13463626492942454, | |
| "grad_norm": 9.902861982329972, | |
| "learning_rate": 9.726696497001923e-06, | |
| "loss": 5.0762, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.13550488599348534, | |
| "grad_norm": 9.135620944817346, | |
| "learning_rate": 9.722351859256815e-06, | |
| "loss": 5.1326, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.13637350705754614, | |
| "grad_norm": 5.094932628448521, | |
| "learning_rate": 9.717974059363392e-06, | |
| "loss": 5.0871, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.13724212812160694, | |
| "grad_norm": 13.8159898342583, | |
| "learning_rate": 9.713563131704685e-06, | |
| "loss": 5.0887, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.13811074918566776, | |
| "grad_norm": 7.776188941220426, | |
| "learning_rate": 9.709119110923911e-06, | |
| "loss": 5.0136, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.13897937024972856, | |
| "grad_norm": 14.140298714913962, | |
| "learning_rate": 9.70464203192419e-06, | |
| "loss": 5.1441, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.13984799131378936, | |
| "grad_norm": 7.397703190515995, | |
| "learning_rate": 9.700131929868289e-06, | |
| "loss": 5.0998, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.14071661237785016, | |
| "grad_norm": 13.939460570037966, | |
| "learning_rate": 9.695588840178331e-06, | |
| "loss": 5.0197, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.14158523344191096, | |
| "grad_norm": 10.270022957942269, | |
| "learning_rate": 9.691012798535524e-06, | |
| "loss": 4.8067, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.14245385450597178, | |
| "grad_norm": 10.393393548384978, | |
| "learning_rate": 9.686403840879877e-06, | |
| "loss": 5.1063, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.14332247557003258, | |
| "grad_norm": 9.993662702844784, | |
| "learning_rate": 9.681762003409926e-06, | |
| "loss": 5.1959, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.14419109663409338, | |
| "grad_norm": 7.693407344731637, | |
| "learning_rate": 9.677087322582434e-06, | |
| "loss": 5.0605, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.14505971769815418, | |
| "grad_norm": 9.965954257327015, | |
| "learning_rate": 9.672379835112124e-06, | |
| "loss": 5.1052, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.14592833876221498, | |
| "grad_norm": 6.623907098912022, | |
| "learning_rate": 9.667639577971372e-06, | |
| "loss": 5.2321, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.14679695982627577, | |
| "grad_norm": 12.30970638465487, | |
| "learning_rate": 9.662866588389931e-06, | |
| "loss": 4.9307, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.1476655808903366, | |
| "grad_norm": 9.854445105895712, | |
| "learning_rate": 9.658060903854633e-06, | |
| "loss": 5.2573, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.1485342019543974, | |
| "grad_norm": 9.733745744797229, | |
| "learning_rate": 9.653222562109093e-06, | |
| "loss": 5.1312, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.1494028230184582, | |
| "grad_norm": 7.148106476376709, | |
| "learning_rate": 9.64835160115341e-06, | |
| "loss": 4.9459, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.150271444082519, | |
| "grad_norm": 9.910271107591148, | |
| "learning_rate": 9.643448059243881e-06, | |
| "loss": 4.9896, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.1511400651465798, | |
| "grad_norm": 6.743151416523807, | |
| "learning_rate": 9.638511974892689e-06, | |
| "loss": 5.0992, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.15200868621064062, | |
| "grad_norm": 9.278378594694823, | |
| "learning_rate": 9.633543386867599e-06, | |
| "loss": 5.0205, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.15287730727470142, | |
| "grad_norm": 8.006244545055509, | |
| "learning_rate": 9.628542334191665e-06, | |
| "loss": 5.1466, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.15374592833876222, | |
| "grad_norm": 6.372417255435819, | |
| "learning_rate": 9.623508856142914e-06, | |
| "loss": 5.078, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.15461454940282301, | |
| "grad_norm": 8.38636496529513, | |
| "learning_rate": 9.61844299225404e-06, | |
| "loss": 5.1724, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.1554831704668838, | |
| "grad_norm": 7.43768649121031, | |
| "learning_rate": 9.613344782312093e-06, | |
| "loss": 4.8699, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.1563517915309446, | |
| "grad_norm": 12.808038040940353, | |
| "learning_rate": 9.608214266358171e-06, | |
| "loss": 5.0929, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.15722041259500544, | |
| "grad_norm": 10.021409419376392, | |
| "learning_rate": 9.603051484687096e-06, | |
| "loss": 4.9412, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.15808903365906624, | |
| "grad_norm": 8.383987107054914, | |
| "learning_rate": 9.597856477847111e-06, | |
| "loss": 5.0407, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.15895765472312703, | |
| "grad_norm": 11.839795913237127, | |
| "learning_rate": 9.592629286639545e-06, | |
| "loss": 4.9284, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.15982627578718783, | |
| "grad_norm": 7.10734457748747, | |
| "learning_rate": 9.58736995211851e-06, | |
| "loss": 5.058, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.16069489685124863, | |
| "grad_norm": 9.11261944633933, | |
| "learning_rate": 9.582078515590565e-06, | |
| "loss": 4.8771, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.16156351791530946, | |
| "grad_norm": 9.824025734718676, | |
| "learning_rate": 9.5767550186144e-06, | |
| "loss": 4.8962, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.16243213897937026, | |
| "grad_norm": 8.437193122982974, | |
| "learning_rate": 9.571399503000499e-06, | |
| "loss": 4.9084, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.16330076004343105, | |
| "grad_norm": 5.687318397304989, | |
| "learning_rate": 9.566012010810826e-06, | |
| "loss": 5.08, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.16416938110749185, | |
| "grad_norm": 14.433519675550484, | |
| "learning_rate": 9.560592584358489e-06, | |
| "loss": 4.8611, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.16503800217155265, | |
| "grad_norm": 6.794400971865543, | |
| "learning_rate": 9.555141266207398e-06, | |
| "loss": 4.936, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.16590662323561348, | |
| "grad_norm": 14.402134569349881, | |
| "learning_rate": 9.549658099171945e-06, | |
| "loss": 4.8904, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.16677524429967427, | |
| "grad_norm": 9.083194622310913, | |
| "learning_rate": 9.544143126316661e-06, | |
| "loss": 5.0294, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.16764386536373507, | |
| "grad_norm": 10.639070515809655, | |
| "learning_rate": 9.538596390955876e-06, | |
| "loss": 4.9204, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.16851248642779587, | |
| "grad_norm": 7.267347749331012, | |
| "learning_rate": 9.533017936653385e-06, | |
| "loss": 5.0803, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.16938110749185667, | |
| "grad_norm": 8.011322892386996, | |
| "learning_rate": 9.527407807222096e-06, | |
| "loss": 5.0817, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.17024972855591747, | |
| "grad_norm": 8.33354841480774, | |
| "learning_rate": 9.521766046723699e-06, | |
| "loss": 5.0201, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.1711183496199783, | |
| "grad_norm": 4.646575968820735, | |
| "learning_rate": 9.516092699468305e-06, | |
| "loss": 5.1351, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.1719869706840391, | |
| "grad_norm": 9.81795639706777, | |
| "learning_rate": 9.510387810014115e-06, | |
| "loss": 4.8496, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.1728555917480999, | |
| "grad_norm": 8.118086500928438, | |
| "learning_rate": 9.504651423167056e-06, | |
| "loss": 4.7814, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.1737242128121607, | |
| "grad_norm": 8.90163622203598, | |
| "learning_rate": 9.49888358398043e-06, | |
| "loss": 4.9746, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1745928338762215, | |
| "grad_norm": 9.581229012117431, | |
| "learning_rate": 9.493084337754573e-06, | |
| "loss": 4.9412, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.1754614549402823, | |
| "grad_norm": 5.4797873048345185, | |
| "learning_rate": 9.487253730036484e-06, | |
| "loss": 5.0215, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.1763300760043431, | |
| "grad_norm": 13.799396637492386, | |
| "learning_rate": 9.481391806619475e-06, | |
| "loss": 5.1105, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.1771986970684039, | |
| "grad_norm": 10.196454355029717, | |
| "learning_rate": 9.475498613542808e-06, | |
| "loss": 5.0285, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.1780673181324647, | |
| "grad_norm": 13.961082886424126, | |
| "learning_rate": 9.469574197091345e-06, | |
| "loss": 4.986, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.1789359391965255, | |
| "grad_norm": 9.989829598419863, | |
| "learning_rate": 9.46361860379516e-06, | |
| "loss": 5.021, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.17980456026058633, | |
| "grad_norm": 12.997591407480371, | |
| "learning_rate": 9.4576318804292e-06, | |
| "loss": 4.9919, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.18067318132464713, | |
| "grad_norm": 9.372234356116795, | |
| "learning_rate": 9.451614074012905e-06, | |
| "loss": 5.1485, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.18154180238870793, | |
| "grad_norm": 10.142709831697546, | |
| "learning_rate": 9.445565231809832e-06, | |
| "loss": 5.0171, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.18241042345276873, | |
| "grad_norm": 9.226181067395016, | |
| "learning_rate": 9.439485401327296e-06, | |
| "loss": 5.0061, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.18327904451682953, | |
| "grad_norm": 8.279353289891953, | |
| "learning_rate": 9.433374630315997e-06, | |
| "loss": 4.9115, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.18414766558089032, | |
| "grad_norm": 11.929279731379703, | |
| "learning_rate": 9.427232966769634e-06, | |
| "loss": 4.6847, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.18501628664495115, | |
| "grad_norm": 5.482572771970624, | |
| "learning_rate": 9.421060458924539e-06, | |
| "loss": 4.9443, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.18588490770901195, | |
| "grad_norm": 7.427326598312704, | |
| "learning_rate": 9.414857155259289e-06, | |
| "loss": 4.9023, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.18675352877307275, | |
| "grad_norm": 8.360757486256176, | |
| "learning_rate": 9.408623104494336e-06, | |
| "loss": 4.984, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.18762214983713354, | |
| "grad_norm": 7.208982202085643, | |
| "learning_rate": 9.402358355591609e-06, | |
| "loss": 4.963, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.18849077090119434, | |
| "grad_norm": 5.301087365446961, | |
| "learning_rate": 9.39606295775415e-06, | |
| "loss": 4.998, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.18935939196525517, | |
| "grad_norm": 7.799837692221157, | |
| "learning_rate": 9.389736960425708e-06, | |
| "loss": 4.9741, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.19022801302931597, | |
| "grad_norm": 7.151033616802984, | |
| "learning_rate": 9.383380413290361e-06, | |
| "loss": 4.8871, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.19109663409337677, | |
| "grad_norm": 5.787982552484058, | |
| "learning_rate": 9.376993366272128e-06, | |
| "loss": 4.9864, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.19196525515743756, | |
| "grad_norm": 7.905445175850379, | |
| "learning_rate": 9.370575869534565e-06, | |
| "loss": 5.0129, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.19283387622149836, | |
| "grad_norm": 4.339162197971099, | |
| "learning_rate": 9.364127973480385e-06, | |
| "loss": 5.0006, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.1937024972855592, | |
| "grad_norm": 9.706062286895799, | |
| "learning_rate": 9.35764972875105e-06, | |
| "loss": 4.8678, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.19457111834962, | |
| "grad_norm": 6.364386062061041, | |
| "learning_rate": 9.351141186226387e-06, | |
| "loss": 4.8405, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.19543973941368079, | |
| "grad_norm": 8.97734562182805, | |
| "learning_rate": 9.344602397024172e-06, | |
| "loss": 4.9155, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.19630836047774158, | |
| "grad_norm": 7.362245536098451, | |
| "learning_rate": 9.338033412499743e-06, | |
| "loss": 4.7595, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.19717698154180238, | |
| "grad_norm": 7.54540953421405, | |
| "learning_rate": 9.331434284245585e-06, | |
| "loss": 4.8956, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.19804560260586318, | |
| "grad_norm": 7.423100629581046, | |
| "learning_rate": 9.324805064090939e-06, | |
| "loss": 4.9622, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.198914223669924, | |
| "grad_norm": 6.020905403406819, | |
| "learning_rate": 9.318145804101377e-06, | |
| "loss": 4.959, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.1997828447339848, | |
| "grad_norm": 4.670252852600942, | |
| "learning_rate": 9.31145655657841e-06, | |
| "loss": 5.0506, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.2006514657980456, | |
| "grad_norm": 7.795934999989379, | |
| "learning_rate": 9.30473737405906e-06, | |
| "loss": 4.9765, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.2015200868621064, | |
| "grad_norm": 6.01240795364037, | |
| "learning_rate": 9.29798830931547e-06, | |
| "loss": 4.9211, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.2023887079261672, | |
| "grad_norm": 5.698601126510405, | |
| "learning_rate": 9.291209415354466e-06, | |
| "loss": 4.7963, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.20325732899022803, | |
| "grad_norm": 12.382329963742128, | |
| "learning_rate": 9.284400745417154e-06, | |
| "loss": 4.8329, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.20412595005428882, | |
| "grad_norm": 6.0887020552234645, | |
| "learning_rate": 9.277562352978504e-06, | |
| "loss": 4.9987, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.20499457111834962, | |
| "grad_norm": 9.258532830886514, | |
| "learning_rate": 9.270694291746918e-06, | |
| "loss": 4.7606, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.20586319218241042, | |
| "grad_norm": 5.467188874710136, | |
| "learning_rate": 9.26379661566382e-06, | |
| "loss": 4.8058, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.20673181324647122, | |
| "grad_norm": 8.326138039007624, | |
| "learning_rate": 9.256869378903226e-06, | |
| "loss": 4.8709, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.20760043431053202, | |
| "grad_norm": 6.953137100468736, | |
| "learning_rate": 9.249912635871317e-06, | |
| "loss": 4.877, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.20846905537459284, | |
| "grad_norm": 7.746260501101959, | |
| "learning_rate": 9.242926441206024e-06, | |
| "loss": 4.866, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.20933767643865364, | |
| "grad_norm": 5.662860561699009, | |
| "learning_rate": 9.235910849776578e-06, | |
| "loss": 4.9476, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.21020629750271444, | |
| "grad_norm": 9.471074354024374, | |
| "learning_rate": 9.2288659166831e-06, | |
| "loss": 4.8759, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.21107491856677524, | |
| "grad_norm": 5.294803045493394, | |
| "learning_rate": 9.221791697256152e-06, | |
| "loss": 4.8172, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.21194353963083604, | |
| "grad_norm": 9.913380058373116, | |
| "learning_rate": 9.214688247056316e-06, | |
| "loss": 4.9476, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.21281216069489686, | |
| "grad_norm": 6.941540338932768, | |
| "learning_rate": 9.207555621873748e-06, | |
| "loss": 4.8194, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.21368078175895766, | |
| "grad_norm": 9.460762282346524, | |
| "learning_rate": 9.20039387772774e-06, | |
| "loss": 4.9644, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.21454940282301846, | |
| "grad_norm": 7.830249212526504, | |
| "learning_rate": 9.19320307086629e-06, | |
| "loss": 4.9704, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.21541802388707926, | |
| "grad_norm": 6.925252918501548, | |
| "learning_rate": 9.185983257765648e-06, | |
| "loss": 4.9168, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.21628664495114006, | |
| "grad_norm": 7.233949141897864, | |
| "learning_rate": 9.178734495129876e-06, | |
| "loss": 4.8646, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.21715526601520088, | |
| "grad_norm": 5.897422894008084, | |
| "learning_rate": 9.171456839890408e-06, | |
| "loss": 5.0017, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.21802388707926168, | |
| "grad_norm": 8.435251394805453, | |
| "learning_rate": 9.1641503492056e-06, | |
| "loss": 4.8114, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.21889250814332248, | |
| "grad_norm": 5.760910243649614, | |
| "learning_rate": 9.156815080460277e-06, | |
| "loss": 4.8976, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.21976112920738328, | |
| "grad_norm": 7.520437903960479, | |
| "learning_rate": 9.149451091265286e-06, | |
| "loss": 4.7124, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.22062975027144408, | |
| "grad_norm": 7.013652737318022, | |
| "learning_rate": 9.142058439457044e-06, | |
| "loss": 4.9533, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.22149837133550487, | |
| "grad_norm": 7.159149521810479, | |
| "learning_rate": 9.134637183097083e-06, | |
| "loss": 4.8566, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.2223669923995657, | |
| "grad_norm": 6.499343085246989, | |
| "learning_rate": 9.127187380471595e-06, | |
| "loss": 4.6676, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.2232356134636265, | |
| "grad_norm": 5.21337701054524, | |
| "learning_rate": 9.11970909009097e-06, | |
| "loss": 4.922, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.2241042345276873, | |
| "grad_norm": 6.349567353113073, | |
| "learning_rate": 9.112202370689337e-06, | |
| "loss": 4.8532, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.2249728555917481, | |
| "grad_norm": 6.818694420622126, | |
| "learning_rate": 9.104667281224114e-06, | |
| "loss": 4.9436, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.2258414766558089, | |
| "grad_norm": 7.214238229211862, | |
| "learning_rate": 9.09710388087553e-06, | |
| "loss": 5.0028, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.22671009771986972, | |
| "grad_norm": 5.209070263470396, | |
| "learning_rate": 9.089512229046167e-06, | |
| "loss": 4.5818, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.22757871878393052, | |
| "grad_norm": 10.360050755845188, | |
| "learning_rate": 9.08189238536049e-06, | |
| "loss": 4.8978, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.22844733984799132, | |
| "grad_norm": 6.643043793388568, | |
| "learning_rate": 9.07424440966439e-06, | |
| "loss": 4.7172, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.2293159609120521, | |
| "grad_norm": 7.816417260568394, | |
| "learning_rate": 9.066568362024697e-06, | |
| "loss": 4.7677, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.2301845819761129, | |
| "grad_norm": 7.24101938038275, | |
| "learning_rate": 9.058864302728722e-06, | |
| "loss": 4.9232, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.23105320304017374, | |
| "grad_norm": 5.882755828325563, | |
| "learning_rate": 9.051132292283772e-06, | |
| "loss": 4.8006, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.23192182410423454, | |
| "grad_norm": 7.033058546114696, | |
| "learning_rate": 9.043372391416687e-06, | |
| "loss": 4.8018, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.23279044516829533, | |
| "grad_norm": 6.638958562180367, | |
| "learning_rate": 9.035584661073357e-06, | |
| "loss": 4.8614, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.23365906623235613, | |
| "grad_norm": 5.164804077792831, | |
| "learning_rate": 9.02776916241824e-06, | |
| "loss": 4.6965, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.23452768729641693, | |
| "grad_norm": 7.8053882165369055, | |
| "learning_rate": 9.019925956833884e-06, | |
| "loss": 4.8867, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.23539630836047773, | |
| "grad_norm": 5.302096025761552, | |
| "learning_rate": 9.012055105920452e-06, | |
| "loss": 4.895, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.23626492942453856, | |
| "grad_norm": 5.851233842251707, | |
| "learning_rate": 9.004156671495224e-06, | |
| "loss": 4.9754, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.23713355048859935, | |
| "grad_norm": 6.290732292956187, | |
| "learning_rate": 8.996230715592129e-06, | |
| "loss": 4.8171, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.23800217155266015, | |
| "grad_norm": 6.760963585753358, | |
| "learning_rate": 8.988277300461238e-06, | |
| "loss": 4.6128, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.23887079261672095, | |
| "grad_norm": 5.437936675855679, | |
| "learning_rate": 8.980296488568296e-06, | |
| "loss": 4.8612, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.23973941368078175, | |
| "grad_norm": 8.546817288458861, | |
| "learning_rate": 8.972288342594211e-06, | |
| "loss": 4.7358, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.24060803474484257, | |
| "grad_norm": 5.168018360375808, | |
| "learning_rate": 8.96425292543458e-06, | |
| "loss": 4.8366, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.24147665580890337, | |
| "grad_norm": 8.339946348434717, | |
| "learning_rate": 8.95619030019918e-06, | |
| "loss": 4.9408, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.24234527687296417, | |
| "grad_norm": 4.812823437728942, | |
| "learning_rate": 8.94810053021148e-06, | |
| "loss": 4.8325, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.24321389793702497, | |
| "grad_norm": 9.00150446795358, | |
| "learning_rate": 8.939983679008147e-06, | |
| "loss": 4.766, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.24408251900108577, | |
| "grad_norm": 6.85159808721181, | |
| "learning_rate": 8.931839810338541e-06, | |
| "loss": 4.9044, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.24495114006514657, | |
| "grad_norm": 5.419252004137471, | |
| "learning_rate": 8.923668988164213e-06, | |
| "loss": 4.7154, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.2458197611292074, | |
| "grad_norm": 8.042101516011865, | |
| "learning_rate": 8.915471276658405e-06, | |
| "loss": 4.8391, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.2466883821932682, | |
| "grad_norm": 6.113748424857356, | |
| "learning_rate": 8.907246740205553e-06, | |
| "loss": 4.7499, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.247557003257329, | |
| "grad_norm": 7.989470062325969, | |
| "learning_rate": 8.898995443400767e-06, | |
| "loss": 4.8469, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.2484256243213898, | |
| "grad_norm": 4.646955530626333, | |
| "learning_rate": 8.890717451049335e-06, | |
| "loss": 4.7358, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.24929424538545059, | |
| "grad_norm": 7.744655477388046, | |
| "learning_rate": 8.882412828166213e-06, | |
| "loss": 4.7807, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.2501628664495114, | |
| "grad_norm": 5.999315404456719, | |
| "learning_rate": 8.874081639975508e-06, | |
| "loss": 4.7731, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.2510314875135722, | |
| "grad_norm": 7.366537198808325, | |
| "learning_rate": 8.865723951909972e-06, | |
| "loss": 4.9502, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.251900108577633, | |
| "grad_norm": 5.915226962483226, | |
| "learning_rate": 8.857339829610483e-06, | |
| "loss": 4.7959, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.25276872964169383, | |
| "grad_norm": 9.035720639467963, | |
| "learning_rate": 8.848929338925536e-06, | |
| "loss": 4.7853, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.2536373507057546, | |
| "grad_norm": 5.777665004211188, | |
| "learning_rate": 8.84049254591072e-06, | |
| "loss": 4.8299, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.25450597176981543, | |
| "grad_norm": 9.211239169808724, | |
| "learning_rate": 8.8320295168282e-06, | |
| "loss": 4.6564, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.2553745928338762, | |
| "grad_norm": 6.121012460613405, | |
| "learning_rate": 8.8235403181462e-06, | |
| "loss": 4.6625, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.256243213897937, | |
| "grad_norm": 7.943211261154958, | |
| "learning_rate": 8.815025016538477e-06, | |
| "loss": 4.9183, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.25711183496199785, | |
| "grad_norm": 7.520085515355943, | |
| "learning_rate": 8.806483678883803e-06, | |
| "loss": 4.7658, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.2579804560260586, | |
| "grad_norm": 6.255511074855573, | |
| "learning_rate": 8.79791637226543e-06, | |
| "loss": 4.8817, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.25884907709011945, | |
| "grad_norm": 8.574508707280637, | |
| "learning_rate": 8.789323163970573e-06, | |
| "loss": 4.8496, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.2597176981541802, | |
| "grad_norm": 5.222979665146013, | |
| "learning_rate": 8.780704121489876e-06, | |
| "loss": 4.7925, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.26058631921824105, | |
| "grad_norm": 9.382159121398075, | |
| "learning_rate": 8.772059312516883e-06, | |
| "loss": 4.88, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.2614549402823019, | |
| "grad_norm": 6.8357894514157245, | |
| "learning_rate": 8.76338880494751e-06, | |
| "loss": 4.6685, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.26232356134636264, | |
| "grad_norm": 6.517300496516373, | |
| "learning_rate": 8.754692666879504e-06, | |
| "loss": 4.8814, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.26319218241042347, | |
| "grad_norm": 9.702116008549124, | |
| "learning_rate": 8.745970966611917e-06, | |
| "loss": 4.6584, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.26406080347448424, | |
| "grad_norm": 6.260530698612518, | |
| "learning_rate": 8.737223772644562e-06, | |
| "loss": 4.6473, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.26492942453854507, | |
| "grad_norm": 9.829705946999965, | |
| "learning_rate": 8.72845115367748e-06, | |
| "loss": 4.771, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.26579804560260584, | |
| "grad_norm": 6.69025005399437, | |
| "learning_rate": 8.719653178610396e-06, | |
| "loss": 4.778, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.26666666666666666, | |
| "grad_norm": 6.70318175356247, | |
| "learning_rate": 8.710829916542184e-06, | |
| "loss": 4.8116, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.2675352877307275, | |
| "grad_norm": 7.751822349780793, | |
| "learning_rate": 8.701981436770322e-06, | |
| "loss": 4.6695, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.26840390879478826, | |
| "grad_norm": 7.108916006280296, | |
| "learning_rate": 8.69310780879034e-06, | |
| "loss": 4.7317, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.2692725298588491, | |
| "grad_norm": 4.328672737830481, | |
| "learning_rate": 8.684209102295292e-06, | |
| "loss": 4.6794, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.27014115092290986, | |
| "grad_norm": 8.322594402892245, | |
| "learning_rate": 8.675285387175183e-06, | |
| "loss": 4.8747, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.2710097719869707, | |
| "grad_norm": 4.508165025699175, | |
| "learning_rate": 8.666336733516447e-06, | |
| "loss": 4.9213, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.2718783930510315, | |
| "grad_norm": 6.5835338898059055, | |
| "learning_rate": 8.657363211601375e-06, | |
| "loss": 4.7728, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.2727470141150923, | |
| "grad_norm": 8.537697102858328, | |
| "learning_rate": 8.64836489190758e-06, | |
| "loss": 4.7095, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.2736156351791531, | |
| "grad_norm": 8.365612338039389, | |
| "learning_rate": 8.639341845107432e-06, | |
| "loss": 4.8467, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.2744842562432139, | |
| "grad_norm": 6.665043705485368, | |
| "learning_rate": 8.630294142067505e-06, | |
| "loss": 4.8167, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.2753528773072747, | |
| "grad_norm": 7.82967483334624, | |
| "learning_rate": 8.621221853848022e-06, | |
| "loss": 4.7616, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.2762214983713355, | |
| "grad_norm": 4.138871108394252, | |
| "learning_rate": 8.6121250517023e-06, | |
| "loss": 4.7277, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.2770901194353963, | |
| "grad_norm": 5.604431351464519, | |
| "learning_rate": 8.603003807076184e-06, | |
| "loss": 4.6664, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.2779587404994571, | |
| "grad_norm": 8.601258868804189, | |
| "learning_rate": 8.593858191607492e-06, | |
| "loss": 4.8013, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.2788273615635179, | |
| "grad_norm": 4.906802688431626, | |
| "learning_rate": 8.584688277125446e-06, | |
| "loss": 4.6898, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.2796959826275787, | |
| "grad_norm": 10.116257183106686, | |
| "learning_rate": 8.575494135650115e-06, | |
| "loss": 4.8905, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.28056460369163955, | |
| "grad_norm": 5.885384217150105, | |
| "learning_rate": 8.566275839391842e-06, | |
| "loss": 4.8431, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.2814332247557003, | |
| "grad_norm": 7.268094086044695, | |
| "learning_rate": 8.557033460750685e-06, | |
| "loss": 4.8863, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.28230184581976114, | |
| "grad_norm": 7.115813734582664, | |
| "learning_rate": 8.547767072315835e-06, | |
| "loss": 4.8921, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.2831704668838219, | |
| "grad_norm": 5.851149266761853, | |
| "learning_rate": 8.538476746865066e-06, | |
| "loss": 4.6311, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.28403908794788274, | |
| "grad_norm": 11.032345343462614, | |
| "learning_rate": 8.529162557364148e-06, | |
| "loss": 4.751, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.28490770901194357, | |
| "grad_norm": 5.0859521313383045, | |
| "learning_rate": 8.519824576966274e-06, | |
| "loss": 4.7293, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.28577633007600434, | |
| "grad_norm": 13.803163681395665, | |
| "learning_rate": 8.510462879011492e-06, | |
| "loss": 4.8309, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.28664495114006516, | |
| "grad_norm": 10.999396408035773, | |
| "learning_rate": 8.50107753702613e-06, | |
| "loss": 4.7266, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.28751357220412593, | |
| "grad_norm": 9.25286217238097, | |
| "learning_rate": 8.49166862472221e-06, | |
| "loss": 4.7683, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.28838219326818676, | |
| "grad_norm": 6.4323672461710135, | |
| "learning_rate": 8.482236215996881e-06, | |
| "loss": 4.7786, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.28925081433224753, | |
| "grad_norm": 11.742176348664781, | |
| "learning_rate": 8.47278038493182e-06, | |
| "loss": 4.7423, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.29011943539630836, | |
| "grad_norm": 7.418561647305186, | |
| "learning_rate": 8.463301205792675e-06, | |
| "loss": 4.7125, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.2909880564603692, | |
| "grad_norm": 13.421309741389427, | |
| "learning_rate": 8.45379875302846e-06, | |
| "loss": 4.7378, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.29185667752442995, | |
| "grad_norm": 12.62311847789775, | |
| "learning_rate": 8.444273101270982e-06, | |
| "loss": 4.6425, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.2927252985884908, | |
| "grad_norm": 7.036866908269808, | |
| "learning_rate": 8.434724325334252e-06, | |
| "loss": 4.7556, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.29359391965255155, | |
| "grad_norm": 8.910330385460947, | |
| "learning_rate": 8.425152500213898e-06, | |
| "loss": 4.6621, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.2944625407166124, | |
| "grad_norm": 6.680041055622366, | |
| "learning_rate": 8.415557701086572e-06, | |
| "loss": 4.58, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.2953311617806732, | |
| "grad_norm": 5.917224901051049, | |
| "learning_rate": 8.405940003309366e-06, | |
| "loss": 4.764, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.29619978284473397, | |
| "grad_norm": 6.025660992004857, | |
| "learning_rate": 8.396299482419213e-06, | |
| "loss": 4.9539, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.2970684039087948, | |
| "grad_norm": 6.8290925321533855, | |
| "learning_rate": 8.386636214132303e-06, | |
| "loss": 4.7968, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.29793702497285557, | |
| "grad_norm": 6.766388648711616, | |
| "learning_rate": 8.376950274343476e-06, | |
| "loss": 4.7316, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.2988056460369164, | |
| "grad_norm": 5.199438881706867, | |
| "learning_rate": 8.367241739125645e-06, | |
| "loss": 4.6717, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.2996742671009772, | |
| "grad_norm": 5.537976824790018, | |
| "learning_rate": 8.35751068472917e-06, | |
| "loss": 4.5794, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.300542888165038, | |
| "grad_norm": 7.434218743498804, | |
| "learning_rate": 8.347757187581288e-06, | |
| "loss": 4.5797, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.3014115092290988, | |
| "grad_norm": 3.7410734562690102, | |
| "learning_rate": 8.337981324285495e-06, | |
| "loss": 4.6053, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.3022801302931596, | |
| "grad_norm": 7.494539433625744, | |
| "learning_rate": 8.328183171620953e-06, | |
| "loss": 4.6358, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.3031487513572204, | |
| "grad_norm": 8.577268240575261, | |
| "learning_rate": 8.318362806541878e-06, | |
| "loss": 4.6568, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.30401737242128124, | |
| "grad_norm": 5.870073757957237, | |
| "learning_rate": 8.308520306176948e-06, | |
| "loss": 4.8527, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.304885993485342, | |
| "grad_norm": 10.403543599580171, | |
| "learning_rate": 8.298655747828685e-06, | |
| "loss": 4.6002, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.30575461454940284, | |
| "grad_norm": 5.381680003218319, | |
| "learning_rate": 8.288769208972858e-06, | |
| "loss": 4.7809, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.3066232356134636, | |
| "grad_norm": 9.877174040443599, | |
| "learning_rate": 8.278860767257865e-06, | |
| "loss": 4.7145, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.30749185667752443, | |
| "grad_norm": 6.336637930519997, | |
| "learning_rate": 8.26893050050413e-06, | |
| "loss": 4.8645, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.30836047774158526, | |
| "grad_norm": 9.02647409091144, | |
| "learning_rate": 8.258978486703493e-06, | |
| "loss": 4.6869, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.30922909880564603, | |
| "grad_norm": 7.279729496607072, | |
| "learning_rate": 8.24900480401859e-06, | |
| "loss": 4.883, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.31009771986970686, | |
| "grad_norm": 6.615700438254423, | |
| "learning_rate": 8.239009530782244e-06, | |
| "loss": 4.774, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.3109663409337676, | |
| "grad_norm": 8.777698668573509, | |
| "learning_rate": 8.228992745496851e-06, | |
| "loss": 4.9262, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.31183496199782845, | |
| "grad_norm": 5.9973840840887505, | |
| "learning_rate": 8.21895452683376e-06, | |
| "loss": 4.7127, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.3127035830618892, | |
| "grad_norm": 10.806900681675172, | |
| "learning_rate": 8.20889495363266e-06, | |
| "loss": 4.7254, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.31357220412595005, | |
| "grad_norm": 8.826153430021765, | |
| "learning_rate": 8.198814104900951e-06, | |
| "loss": 4.5981, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.3144408251900109, | |
| "grad_norm": 9.306758043660382, | |
| "learning_rate": 8.188712059813135e-06, | |
| "loss": 4.6332, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.31530944625407165, | |
| "grad_norm": 5.858792200617635, | |
| "learning_rate": 8.178588897710189e-06, | |
| "loss": 4.58, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.31617806731813247, | |
| "grad_norm": 6.68100229149933, | |
| "learning_rate": 8.16844469809894e-06, | |
| "loss": 4.7159, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.31704668838219324, | |
| "grad_norm": 5.735283629069225, | |
| "learning_rate": 8.158279540651446e-06, | |
| "loss": 4.8563, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.31791530944625407, | |
| "grad_norm": 6.584386859485198, | |
| "learning_rate": 8.14809350520436e-06, | |
| "loss": 4.751, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.3187839305103149, | |
| "grad_norm": 6.582333473666524, | |
| "learning_rate": 8.137886671758317e-06, | |
| "loss": 4.7886, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.31965255157437567, | |
| "grad_norm": 6.4883303364815275, | |
| "learning_rate": 8.127659120477296e-06, | |
| "loss": 4.6657, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.3205211726384365, | |
| "grad_norm": 5.918831827629274, | |
| "learning_rate": 8.117410931687992e-06, | |
| "loss": 4.8691, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.32138979370249726, | |
| "grad_norm": 5.283964365819055, | |
| "learning_rate": 8.107142185879185e-06, | |
| "loss": 4.8074, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.3222584147665581, | |
| "grad_norm": 3.8890066265932184, | |
| "learning_rate": 8.096852963701113e-06, | |
| "loss": 4.5405, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.3231270358306189, | |
| "grad_norm": 6.990269124593145, | |
| "learning_rate": 8.086543345964833e-06, | |
| "loss": 4.6027, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.3239956568946797, | |
| "grad_norm": 8.43721076775689, | |
| "learning_rate": 8.07621341364158e-06, | |
| "loss": 4.647, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.3248642779587405, | |
| "grad_norm": 5.58990373211778, | |
| "learning_rate": 8.065863247862153e-06, | |
| "loss": 4.6015, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.3257328990228013, | |
| "grad_norm": 6.97196709970691, | |
| "learning_rate": 8.05549292991625e-06, | |
| "loss": 4.6307, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.3266015200868621, | |
| "grad_norm": 6.1054642976234215, | |
| "learning_rate": 8.045102541251855e-06, | |
| "loss": 4.5934, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.32747014115092293, | |
| "grad_norm": 5.762512151008954, | |
| "learning_rate": 8.034692163474576e-06, | |
| "loss": 4.5073, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.3283387622149837, | |
| "grad_norm": 7.3544384303060575, | |
| "learning_rate": 8.02426187834702e-06, | |
| "loss": 4.7555, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.32920738327904453, | |
| "grad_norm": 4.541115018669931, | |
| "learning_rate": 8.013811767788144e-06, | |
| "loss": 4.5251, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.3300760043431053, | |
| "grad_norm": 7.061819858205978, | |
| "learning_rate": 8.003341913872616e-06, | |
| "loss": 4.6566, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.3309446254071661, | |
| "grad_norm": 5.949506061458709, | |
| "learning_rate": 7.992852398830164e-06, | |
| "loss": 4.6047, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.33181324647122695, | |
| "grad_norm": 8.186210627662106, | |
| "learning_rate": 7.982343305044932e-06, | |
| "loss": 4.5416, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.3326818675352877, | |
| "grad_norm": 5.5358466371451645, | |
| "learning_rate": 7.971814715054837e-06, | |
| "loss": 4.6173, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.33355048859934855, | |
| "grad_norm": 7.682761597446245, | |
| "learning_rate": 7.961266711550922e-06, | |
| "loss": 4.5791, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.3344191096634093, | |
| "grad_norm": 5.430465140999319, | |
| "learning_rate": 7.950699377376696e-06, | |
| "loss": 4.4514, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.33528773072747015, | |
| "grad_norm": 7.022253684816494, | |
| "learning_rate": 7.940112795527493e-06, | |
| "loss": 4.7673, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.33615635179153097, | |
| "grad_norm": 6.175962941485549, | |
| "learning_rate": 7.929507049149817e-06, | |
| "loss": 4.5196, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.33702497285559174, | |
| "grad_norm": 7.343612570596667, | |
| "learning_rate": 7.918882221540692e-06, | |
| "loss": 4.6791, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.33789359391965257, | |
| "grad_norm": 5.998391268508065, | |
| "learning_rate": 7.908238396147002e-06, | |
| "loss": 4.8548, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.33876221498371334, | |
| "grad_norm": 7.019440378415661, | |
| "learning_rate": 7.897575656564836e-06, | |
| "loss": 4.6303, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.33963083604777417, | |
| "grad_norm": 6.839076558244222, | |
| "learning_rate": 7.886894086538841e-06, | |
| "loss": 4.6871, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.34049945711183494, | |
| "grad_norm": 5.40567329714043, | |
| "learning_rate": 7.876193769961555e-06, | |
| "loss": 4.6996, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.34136807817589576, | |
| "grad_norm": 7.504268933018013, | |
| "learning_rate": 7.865474790872749e-06, | |
| "loss": 4.6356, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.3422366992399566, | |
| "grad_norm": 4.823115754529949, | |
| "learning_rate": 7.854737233458764e-06, | |
| "loss": 4.5891, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.34310532030401736, | |
| "grad_norm": 7.282830148401436, | |
| "learning_rate": 7.843981182051866e-06, | |
| "loss": 4.5814, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.3439739413680782, | |
| "grad_norm": 7.366691458186102, | |
| "learning_rate": 7.83320672112956e-06, | |
| "loss": 4.6604, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.34484256243213895, | |
| "grad_norm": 4.92486295778556, | |
| "learning_rate": 7.822413935313947e-06, | |
| "loss": 4.6913, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.3457111834961998, | |
| "grad_norm": 8.210710382886061, | |
| "learning_rate": 7.811602909371044e-06, | |
| "loss": 4.5104, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.3465798045602606, | |
| "grad_norm": 5.177887473136741, | |
| "learning_rate": 7.800773728210133e-06, | |
| "loss": 4.7158, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.3474484256243214, | |
| "grad_norm": 6.985161948689458, | |
| "learning_rate": 7.789926476883079e-06, | |
| "loss": 4.5611, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.3483170466883822, | |
| "grad_norm": 6.148668513601841, | |
| "learning_rate": 7.779061240583669e-06, | |
| "loss": 4.7723, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.349185667752443, | |
| "grad_norm": 5.763986287210963, | |
| "learning_rate": 7.768178104646953e-06, | |
| "loss": 4.6559, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.3500542888165038, | |
| "grad_norm": 5.7385002364185915, | |
| "learning_rate": 7.757277154548552e-06, | |
| "loss": 4.6866, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.3509229098805646, | |
| "grad_norm": 5.490491868927748, | |
| "learning_rate": 7.746358475904006e-06, | |
| "loss": 4.4739, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.3517915309446254, | |
| "grad_norm": 7.746007698882787, | |
| "learning_rate": 7.735422154468087e-06, | |
| "loss": 4.6623, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.3526601520086862, | |
| "grad_norm": 5.140117986756119, | |
| "learning_rate": 7.724468276134143e-06, | |
| "loss": 4.5921, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.353528773072747, | |
| "grad_norm": 5.822784492855372, | |
| "learning_rate": 7.713496926933405e-06, | |
| "loss": 4.5497, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.3543973941368078, | |
| "grad_norm": 4.504836312173092, | |
| "learning_rate": 7.70250819303432e-06, | |
| "loss": 4.6658, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.35526601520086865, | |
| "grad_norm": 10.535092606649714, | |
| "learning_rate": 7.691502160741879e-06, | |
| "loss": 4.5037, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.3561346362649294, | |
| "grad_norm": 6.1030400833165555, | |
| "learning_rate": 7.680478916496927e-06, | |
| "loss": 4.7045, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.35700325732899024, | |
| "grad_norm": 7.459890595821586, | |
| "learning_rate": 7.669438546875495e-06, | |
| "loss": 4.7035, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.357871878393051, | |
| "grad_norm": 8.136016337416066, | |
| "learning_rate": 7.658381138588111e-06, | |
| "loss": 4.5822, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.35874049945711184, | |
| "grad_norm": 7.256847852406638, | |
| "learning_rate": 7.647306778479135e-06, | |
| "loss": 4.5594, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.35960912052117266, | |
| "grad_norm": 7.032905775099178, | |
| "learning_rate": 7.636215553526054e-06, | |
| "loss": 4.6578, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.36047774158523344, | |
| "grad_norm": 6.984223271346296, | |
| "learning_rate": 7.625107550838813e-06, | |
| "loss": 4.5949, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.36134636264929426, | |
| "grad_norm": 8.223177636534274, | |
| "learning_rate": 7.613982857659134e-06, | |
| "loss": 4.5468, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.36221498371335503, | |
| "grad_norm": 5.980657620361127, | |
| "learning_rate": 7.602841561359822e-06, | |
| "loss": 4.495, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.36308360477741586, | |
| "grad_norm": 6.056077053853974, | |
| "learning_rate": 7.591683749444077e-06, | |
| "loss": 4.7391, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.36395222584147663, | |
| "grad_norm": 7.884243252755605, | |
| "learning_rate": 7.5805095095448245e-06, | |
| "loss": 4.5782, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.36482084690553745, | |
| "grad_norm": 8.363070322855709, | |
| "learning_rate": 7.569318929424002e-06, | |
| "loss": 4.5914, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.3656894679695983, | |
| "grad_norm": 6.163289948356193, | |
| "learning_rate": 7.558112096971889e-06, | |
| "loss": 4.6895, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.36655808903365905, | |
| "grad_norm": 6.6531465448851135, | |
| "learning_rate": 7.5468891002064045e-06, | |
| "loss": 4.5072, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.3674267100977199, | |
| "grad_norm": 6.718432675278386, | |
| "learning_rate": 7.535650027272432e-06, | |
| "loss": 4.5402, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.36829533116178065, | |
| "grad_norm": 6.071906565126357, | |
| "learning_rate": 7.5243949664411035e-06, | |
| "loss": 4.6823, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.3691639522258415, | |
| "grad_norm": 7.151230840698435, | |
| "learning_rate": 7.5131240061091285e-06, | |
| "loss": 4.5005, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.3700325732899023, | |
| "grad_norm": 5.338683591643838, | |
| "learning_rate": 7.501837234798084e-06, | |
| "loss": 4.5487, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.37090119435396307, | |
| "grad_norm": 6.10815615028733, | |
| "learning_rate": 7.490534741153733e-06, | |
| "loss": 4.6367, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.3717698154180239, | |
| "grad_norm": 6.115503279804057, | |
| "learning_rate": 7.47921661394531e-06, | |
| "loss": 4.5094, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.37263843648208467, | |
| "grad_norm": 5.269106785049928, | |
| "learning_rate": 7.46788294206485e-06, | |
| "loss": 4.5126, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.3735070575461455, | |
| "grad_norm": 7.458849192557889, | |
| "learning_rate": 7.4565338145264595e-06, | |
| "loss": 4.7316, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.3743756786102063, | |
| "grad_norm": 4.724897099377871, | |
| "learning_rate": 7.445169320465645e-06, | |
| "loss": 4.7243, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.3752442996742671, | |
| "grad_norm": 7.229325419079616, | |
| "learning_rate": 7.433789549138592e-06, | |
| "loss": 4.4991, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.3761129207383279, | |
| "grad_norm": 6.540185537119028, | |
| "learning_rate": 7.42239458992148e-06, | |
| "loss": 4.4751, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.3769815418023887, | |
| "grad_norm": 7.538652093884085, | |
| "learning_rate": 7.410984532309768e-06, | |
| "loss": 4.5819, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.3778501628664495, | |
| "grad_norm": 10.09025026929928, | |
| "learning_rate": 7.399559465917499e-06, | |
| "loss": 4.5811, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.37871878393051034, | |
| "grad_norm": 5.99150737185514, | |
| "learning_rate": 7.3881194804765975e-06, | |
| "loss": 4.6508, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.3795874049945711, | |
| "grad_norm": 10.294753301023693, | |
| "learning_rate": 7.376664665836156e-06, | |
| "loss": 4.6694, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.38045602605863194, | |
| "grad_norm": 10.09163312709563, | |
| "learning_rate": 7.3651951119617415e-06, | |
| "loss": 4.515, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.3813246471226927, | |
| "grad_norm": 8.660200561326562, | |
| "learning_rate": 7.353710908934672e-06, | |
| "loss": 4.7109, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.38219326818675353, | |
| "grad_norm": 5.980585172331498, | |
| "learning_rate": 7.342212146951329e-06, | |
| "loss": 4.62, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.38306188925081436, | |
| "grad_norm": 7.2092542635747705, | |
| "learning_rate": 7.3306989163224365e-06, | |
| "loss": 4.6266, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.38393051031487513, | |
| "grad_norm": 8.245957827229953, | |
| "learning_rate": 7.319171307472355e-06, | |
| "loss": 4.5792, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.38479913137893595, | |
| "grad_norm": 4.686151630126687, | |
| "learning_rate": 7.307629410938364e-06, | |
| "loss": 4.6665, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.3856677524429967, | |
| "grad_norm": 7.408155624716384, | |
| "learning_rate": 7.296073317369967e-06, | |
| "loss": 4.3742, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.38653637350705755, | |
| "grad_norm": 7.497505710273664, | |
| "learning_rate": 7.284503117528167e-06, | |
| "loss": 4.5901, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.3874049945711184, | |
| "grad_norm": 4.224052938744046, | |
| "learning_rate": 7.272918902284758e-06, | |
| "loss": 4.625, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.38827361563517915, | |
| "grad_norm": 9.130148186812361, | |
| "learning_rate": 7.261320762621605e-06, | |
| "loss": 4.5953, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.38914223669924, | |
| "grad_norm": 5.04037207826654, | |
| "learning_rate": 7.249708789629944e-06, | |
| "loss": 4.3398, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.39001085776330074, | |
| "grad_norm": 10.934078227299317, | |
| "learning_rate": 7.2380830745096474e-06, | |
| "loss": 4.8164, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.39087947882736157, | |
| "grad_norm": 5.288203351054606, | |
| "learning_rate": 7.226443708568525e-06, | |
| "loss": 4.6176, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.39174809989142234, | |
| "grad_norm": 13.494419416996744, | |
| "learning_rate": 7.214790783221596e-06, | |
| "loss": 4.4502, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.39261672095548317, | |
| "grad_norm": 5.947583732503299, | |
| "learning_rate": 7.2031243899903755e-06, | |
| "loss": 4.6519, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.393485342019544, | |
| "grad_norm": 13.629266769082431, | |
| "learning_rate": 7.191444620502154e-06, | |
| "loss": 4.6178, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.39435396308360476, | |
| "grad_norm": 8.435112519451463, | |
| "learning_rate": 7.17975156648928e-06, | |
| "loss": 4.3477, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.3952225841476656, | |
| "grad_norm": 13.097395961934241, | |
| "learning_rate": 7.168045319788436e-06, | |
| "loss": 4.5158, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.39609120521172636, | |
| "grad_norm": 9.326564191975844, | |
| "learning_rate": 7.1563259723399204e-06, | |
| "loss": 4.6681, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.3969598262757872, | |
| "grad_norm": 12.31102530036286, | |
| "learning_rate": 7.144593616186925e-06, | |
| "loss": 4.4425, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.397828447339848, | |
| "grad_norm": 9.61855296186821, | |
| "learning_rate": 7.13284834347481e-06, | |
| "loss": 4.755, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.3986970684039088, | |
| "grad_norm": 11.132702885829065, | |
| "learning_rate": 7.121090246450381e-06, | |
| "loss": 4.6984, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.3995656894679696, | |
| "grad_norm": 9.438660704133007, | |
| "learning_rate": 7.1093194174611665e-06, | |
| "loss": 4.5854, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.4004343105320304, | |
| "grad_norm": 10.460987068330683, | |
| "learning_rate": 7.0975359489546914e-06, | |
| "loss": 4.6308, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.4013029315960912, | |
| "grad_norm": 8.135876347998378, | |
| "learning_rate": 7.0857399334777525e-06, | |
| "loss": 4.5529, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.40217155266015203, | |
| "grad_norm": 12.65605978727452, | |
| "learning_rate": 7.073931463675685e-06, | |
| "loss": 4.5816, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.4030401737242128, | |
| "grad_norm": 10.229434009840299, | |
| "learning_rate": 7.062110632291641e-06, | |
| "loss": 4.5845, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.40390879478827363, | |
| "grad_norm": 9.117061503179459, | |
| "learning_rate": 7.0502775321658655e-06, | |
| "loss": 4.5246, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.4047774158523344, | |
| "grad_norm": 8.569940050056474, | |
| "learning_rate": 7.038432256234956e-06, | |
| "loss": 4.5735, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.4056460369163952, | |
| "grad_norm": 7.752296760070429, | |
| "learning_rate": 7.026574897531137e-06, | |
| "loss": 4.3389, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.40651465798045605, | |
| "grad_norm": 6.650253151223029, | |
| "learning_rate": 7.014705549181537e-06, | |
| "loss": 4.559, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.4073832790445168, | |
| "grad_norm": 6.044551087774109, | |
| "learning_rate": 7.0028243044074425e-06, | |
| "loss": 4.565, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.40825190010857765, | |
| "grad_norm": 7.013354857314451, | |
| "learning_rate": 6.990931256523583e-06, | |
| "loss": 4.4699, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.4091205211726384, | |
| "grad_norm": 7.694869076287624, | |
| "learning_rate": 6.97902649893738e-06, | |
| "loss": 4.5624, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.40998914223669924, | |
| "grad_norm": 4.8526449692385905, | |
| "learning_rate": 6.96711012514823e-06, | |
| "loss": 4.5316, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.41085776330076007, | |
| "grad_norm": 6.982180222415533, | |
| "learning_rate": 6.955182228746757e-06, | |
| "loss": 4.4836, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.41172638436482084, | |
| "grad_norm": 5.126018865451303, | |
| "learning_rate": 6.943242903414087e-06, | |
| "loss": 4.6034, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.41259500542888167, | |
| "grad_norm": 5.649991787266421, | |
| "learning_rate": 6.9312922429211065e-06, | |
| "loss": 4.576, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.41346362649294244, | |
| "grad_norm": 6.484179499960972, | |
| "learning_rate": 6.9193303411277265e-06, | |
| "loss": 4.4123, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.41433224755700326, | |
| "grad_norm": 6.697110498760649, | |
| "learning_rate": 6.907357291982148e-06, | |
| "loss": 4.5219, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.41520086862106403, | |
| "grad_norm": 5.161776605561941, | |
| "learning_rate": 6.895373189520124e-06, | |
| "loss": 4.6172, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.41606948968512486, | |
| "grad_norm": 6.744775916139352, | |
| "learning_rate": 6.883378127864218e-06, | |
| "loss": 4.4805, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.4169381107491857, | |
| "grad_norm": 5.164686600631932, | |
| "learning_rate": 6.871372201223068e-06, | |
| "loss": 4.5713, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.41780673181324646, | |
| "grad_norm": 6.684936475739929, | |
| "learning_rate": 6.859355503890643e-06, | |
| "loss": 4.5636, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.4186753528773073, | |
| "grad_norm": 4.963603500277838, | |
| "learning_rate": 6.847328130245506e-06, | |
| "loss": 4.4979, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.41954397394136805, | |
| "grad_norm": 6.455524904363139, | |
| "learning_rate": 6.83529017475007e-06, | |
| "loss": 4.5704, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.4204125950054289, | |
| "grad_norm": 5.327238544989892, | |
| "learning_rate": 6.8232417319498585e-06, | |
| "loss": 4.5729, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.4212812160694897, | |
| "grad_norm": 7.5408057979505525, | |
| "learning_rate": 6.811182896472764e-06, | |
| "loss": 4.5542, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.4221498371335505, | |
| "grad_norm": 5.148215274061414, | |
| "learning_rate": 6.799113763028296e-06, | |
| "loss": 4.5727, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.4230184581976113, | |
| "grad_norm": 5.126262165864149, | |
| "learning_rate": 6.78703442640685e-06, | |
| "loss": 4.392, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.4238870792616721, | |
| "grad_norm": 6.5704973294653986, | |
| "learning_rate": 6.774944981478953e-06, | |
| "loss": 4.7496, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.4247557003257329, | |
| "grad_norm": 7.6270736279053715, | |
| "learning_rate": 6.762845523194527e-06, | |
| "loss": 4.4043, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.4256243213897937, | |
| "grad_norm": 4.7832965680424975, | |
| "learning_rate": 6.750736146582129e-06, | |
| "loss": 4.4942, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.4264929424538545, | |
| "grad_norm": 8.637265182760746, | |
| "learning_rate": 6.738616946748229e-06, | |
| "loss": 4.6536, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.4273615635179153, | |
| "grad_norm": 5.5509816327572, | |
| "learning_rate": 6.726488018876431e-06, | |
| "loss": 4.5646, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.4282301845819761, | |
| "grad_norm": 7.976660262443447, | |
| "learning_rate": 6.7143494582267565e-06, | |
| "loss": 4.3377, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.4290988056460369, | |
| "grad_norm": 5.8635253190574925, | |
| "learning_rate": 6.702201360134874e-06, | |
| "loss": 4.6456, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.42996742671009774, | |
| "grad_norm": 6.191768041893744, | |
| "learning_rate": 6.690043820011362e-06, | |
| "loss": 4.4138, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.4308360477741585, | |
| "grad_norm": 6.21848167204258, | |
| "learning_rate": 6.677876933340952e-06, | |
| "loss": 4.5893, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.43170466883821934, | |
| "grad_norm": 6.974213096926323, | |
| "learning_rate": 6.665700795681795e-06, | |
| "loss": 4.488, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.4325732899022801, | |
| "grad_norm": 4.90870178244335, | |
| "learning_rate": 6.65351550266468e-06, | |
| "loss": 4.4874, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.43344191096634094, | |
| "grad_norm": 8.263746621881054, | |
| "learning_rate": 6.64132114999232e-06, | |
| "loss": 4.3805, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.43431053203040176, | |
| "grad_norm": 7.323957132091585, | |
| "learning_rate": 6.6291178334385695e-06, | |
| "loss": 4.4165, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.43517915309446253, | |
| "grad_norm": 6.480701522877974, | |
| "learning_rate": 6.616905648847693e-06, | |
| "loss": 4.3516, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.43604777415852336, | |
| "grad_norm": 7.079757216369442, | |
| "learning_rate": 6.604684692133597e-06, | |
| "loss": 4.6046, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.43691639522258413, | |
| "grad_norm": 7.375345908223872, | |
| "learning_rate": 6.5924550592790894e-06, | |
| "loss": 4.5464, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.43778501628664496, | |
| "grad_norm": 7.222616053445717, | |
| "learning_rate": 6.580216846335118e-06, | |
| "loss": 4.5919, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.4386536373507057, | |
| "grad_norm": 4.689789965685166, | |
| "learning_rate": 6.567970149420018e-06, | |
| "loss": 4.3752, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.43952225841476655, | |
| "grad_norm": 5.714230313496368, | |
| "learning_rate": 6.555715064718756e-06, | |
| "loss": 4.5501, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.4403908794788274, | |
| "grad_norm": 7.038923018037996, | |
| "learning_rate": 6.543451688482182e-06, | |
| "loss": 4.4103, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.44125950054288815, | |
| "grad_norm": 7.943178319434065, | |
| "learning_rate": 6.531180117026258e-06, | |
| "loss": 4.5398, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.442128121606949, | |
| "grad_norm": 5.455050769628667, | |
| "learning_rate": 6.518900446731319e-06, | |
| "loss": 4.5569, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.44299674267100975, | |
| "grad_norm": 9.260245349881815, | |
| "learning_rate": 6.506612774041302e-06, | |
| "loss": 4.6123, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.4438653637350706, | |
| "grad_norm": 8.258092780806443, | |
| "learning_rate": 6.494317195462999e-06, | |
| "loss": 4.4937, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.4447339847991314, | |
| "grad_norm": 7.575744472833341, | |
| "learning_rate": 6.482013807565292e-06, | |
| "loss": 4.479, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.44560260586319217, | |
| "grad_norm": 6.848928494852384, | |
| "learning_rate": 6.469702706978397e-06, | |
| "loss": 4.6304, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.446471226927253, | |
| "grad_norm": 6.429893016072357, | |
| "learning_rate": 6.457383990393105e-06, | |
| "loss": 4.4752, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.44733984799131377, | |
| "grad_norm": 6.645971348847954, | |
| "learning_rate": 6.445057754560025e-06, | |
| "loss": 4.4823, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.4482084690553746, | |
| "grad_norm": 7.649197834317788, | |
| "learning_rate": 6.432724096288818e-06, | |
| "loss": 4.4119, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.4490770901194354, | |
| "grad_norm": 4.915321016306434, | |
| "learning_rate": 6.420383112447446e-06, | |
| "loss": 4.5927, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.4499457111834962, | |
| "grad_norm": 11.680853631510265, | |
| "learning_rate": 6.408034899961398e-06, | |
| "loss": 4.6962, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.450814332247557, | |
| "grad_norm": 8.30571918662033, | |
| "learning_rate": 6.395679555812942e-06, | |
| "loss": 4.4227, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.4516829533116178, | |
| "grad_norm": 9.811824442503674, | |
| "learning_rate": 6.383317177040357e-06, | |
| "loss": 4.5434, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.4525515743756786, | |
| "grad_norm": 8.160240990996138, | |
| "learning_rate": 6.370947860737173e-06, | |
| "loss": 4.5247, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.45342019543973944, | |
| "grad_norm": 9.499135895845495, | |
| "learning_rate": 6.358571704051401e-06, | |
| "loss": 4.5605, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.4542888165038002, | |
| "grad_norm": 8.5708728035736, | |
| "learning_rate": 6.346188804184782e-06, | |
| "loss": 4.4378, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.45515743756786103, | |
| "grad_norm": 8.020265826405886, | |
| "learning_rate": 6.333799258392015e-06, | |
| "loss": 4.3991, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.4560260586319218, | |
| "grad_norm": 8.224933821515378, | |
| "learning_rate": 6.3214031639799975e-06, | |
| "loss": 4.5, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.45689467969598263, | |
| "grad_norm": 7.476113586399829, | |
| "learning_rate": 6.309000618307058e-06, | |
| "loss": 4.5767, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.45776330076004346, | |
| "grad_norm": 7.324699664493708, | |
| "learning_rate": 6.296591718782193e-06, | |
| "loss": 4.6139, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.4586319218241042, | |
| "grad_norm": 8.277908852422534, | |
| "learning_rate": 6.284176562864303e-06, | |
| "loss": 4.4586, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.45950054288816505, | |
| "grad_norm": 6.4211734472674005, | |
| "learning_rate": 6.271755248061425e-06, | |
| "loss": 4.4396, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.4603691639522258, | |
| "grad_norm": 8.333565942506247, | |
| "learning_rate": 6.259327871929968e-06, | |
| "loss": 4.4319, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.46123778501628665, | |
| "grad_norm": 5.765983896353204, | |
| "learning_rate": 6.246894532073945e-06, | |
| "loss": 4.5144, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.4621064060803475, | |
| "grad_norm": 8.432302157463148, | |
| "learning_rate": 6.234455326144208e-06, | |
| "loss": 4.4091, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.46297502714440825, | |
| "grad_norm": 6.536784604924449, | |
| "learning_rate": 6.222010351837684e-06, | |
| "loss": 4.438, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.4638436482084691, | |
| "grad_norm": 7.775568222041834, | |
| "learning_rate": 6.209559706896603e-06, | |
| "loss": 4.4694, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.46471226927252984, | |
| "grad_norm": 7.053360240600179, | |
| "learning_rate": 6.197103489107726e-06, | |
| "loss": 4.5039, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.46558089033659067, | |
| "grad_norm": 6.449780887258137, | |
| "learning_rate": 6.184641796301596e-06, | |
| "loss": 4.4826, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.46644951140065144, | |
| "grad_norm": 6.784307145245734, | |
| "learning_rate": 6.172174726351743e-06, | |
| "loss": 4.6193, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.46731813246471227, | |
| "grad_norm": 6.2467877899636575, | |
| "learning_rate": 6.159702377173935e-06, | |
| "loss": 4.4363, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.4681867535287731, | |
| "grad_norm": 7.5393857303754155, | |
| "learning_rate": 6.147224846725402e-06, | |
| "loss": 4.4544, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.46905537459283386, | |
| "grad_norm": 6.6525514703420585, | |
| "learning_rate": 6.134742233004073e-06, | |
| "loss": 4.5408, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.4699239956568947, | |
| "grad_norm": 7.973280585624998, | |
| "learning_rate": 6.122254634047787e-06, | |
| "loss": 4.5086, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.47079261672095546, | |
| "grad_norm": 6.6800439107137635, | |
| "learning_rate": 6.109762147933553e-06, | |
| "loss": 4.3572, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.4716612377850163, | |
| "grad_norm": 6.076538888634451, | |
| "learning_rate": 6.097264872776749e-06, | |
| "loss": 4.5751, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.4725298588490771, | |
| "grad_norm": 8.835225489595787, | |
| "learning_rate": 6.084762906730379e-06, | |
| "loss": 4.5004, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.4733984799131379, | |
| "grad_norm": 4.829662247046767, | |
| "learning_rate": 6.0722563479842764e-06, | |
| "loss": 4.3311, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.4742671009771987, | |
| "grad_norm": 6.986298812339827, | |
| "learning_rate": 6.059745294764359e-06, | |
| "loss": 4.365, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.4751357220412595, | |
| "grad_norm": 7.574390765696185, | |
| "learning_rate": 6.04722984533183e-06, | |
| "loss": 4.4365, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.4760043431053203, | |
| "grad_norm": 5.7402675424404395, | |
| "learning_rate": 6.034710097982432e-06, | |
| "loss": 4.2018, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.47687296416938113, | |
| "grad_norm": 7.703977261086262, | |
| "learning_rate": 6.022186151045652e-06, | |
| "loss": 4.4048, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.4777415852334419, | |
| "grad_norm": 6.289071645181934, | |
| "learning_rate": 6.009658102883974e-06, | |
| "loss": 4.5359, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.4786102062975027, | |
| "grad_norm": 6.972078426295843, | |
| "learning_rate": 5.997126051892082e-06, | |
| "loss": 4.4349, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.4794788273615635, | |
| "grad_norm": 7.66830398869564, | |
| "learning_rate": 5.984590096496099e-06, | |
| "loss": 4.5563, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.4803474484256243, | |
| "grad_norm": 6.0227616340295365, | |
| "learning_rate": 5.972050335152819e-06, | |
| "loss": 4.2982, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.48121606948968515, | |
| "grad_norm": 8.678277229194649, | |
| "learning_rate": 5.959506866348924e-06, | |
| "loss": 4.4519, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.4820846905537459, | |
| "grad_norm": 4.787842386763273, | |
| "learning_rate": 5.94695978860021e-06, | |
| "loss": 4.438, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.48295331161780675, | |
| "grad_norm": 10.597087001950817, | |
| "learning_rate": 5.934409200450828e-06, | |
| "loss": 4.3361, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.4838219326818675, | |
| "grad_norm": 6.769261685937347, | |
| "learning_rate": 5.9218552004724895e-06, | |
| "loss": 4.2451, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.48469055374592834, | |
| "grad_norm": 8.774187421231344, | |
| "learning_rate": 5.909297887263708e-06, | |
| "loss": 4.4513, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.48555917480998917, | |
| "grad_norm": 6.831684880334275, | |
| "learning_rate": 5.896737359449015e-06, | |
| "loss": 4.5108, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.48642779587404994, | |
| "grad_norm": 7.414601371689213, | |
| "learning_rate": 5.884173715678193e-06, | |
| "loss": 4.4822, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.48729641693811077, | |
| "grad_norm": 6.332769305151126, | |
| "learning_rate": 5.871607054625497e-06, | |
| "loss": 4.3, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.48816503800217154, | |
| "grad_norm": 7.518053075962911, | |
| "learning_rate": 5.859037474988875e-06, | |
| "loss": 4.2958, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.48903365906623236, | |
| "grad_norm": 7.791263799758281, | |
| "learning_rate": 5.846465075489202e-06, | |
| "loss": 4.6028, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.48990228013029313, | |
| "grad_norm": 7.99797189120995, | |
| "learning_rate": 5.8338899548695004e-06, | |
| "loss": 4.362, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.49077090119435396, | |
| "grad_norm": 7.553380214852644, | |
| "learning_rate": 5.821312211894159e-06, | |
| "loss": 4.5026, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.4916395222584148, | |
| "grad_norm": 5.392709057685767, | |
| "learning_rate": 5.808731945348168e-06, | |
| "loss": 4.3386, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.49250814332247556, | |
| "grad_norm": 7.863510875655094, | |
| "learning_rate": 5.7961492540363365e-06, | |
| "loss": 4.564, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.4933767643865364, | |
| "grad_norm": 7.618164313464631, | |
| "learning_rate": 5.783564236782514e-06, | |
| "loss": 4.4442, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.49424538545059715, | |
| "grad_norm": 8.728366265826342, | |
| "learning_rate": 5.770976992428821e-06, | |
| "loss": 4.4746, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.495114006514658, | |
| "grad_norm": 7.8924314705792025, | |
| "learning_rate": 5.758387619834872e-06, | |
| "loss": 4.4972, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.4959826275787188, | |
| "grad_norm": 6.344676137880645, | |
| "learning_rate": 5.74579621787699e-06, | |
| "loss": 4.4232, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.4968512486427796, | |
| "grad_norm": 5.552402668916067, | |
| "learning_rate": 5.73320288544744e-06, | |
| "loss": 4.46, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.4977198697068404, | |
| "grad_norm": 7.611624637980265, | |
| "learning_rate": 5.720607721453651e-06, | |
| "loss": 4.5376, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.49858849077090117, | |
| "grad_norm": 7.077190569309273, | |
| "learning_rate": 5.708010824817432e-06, | |
| "loss": 4.3802, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.499457111834962, | |
| "grad_norm": 6.943034504994782, | |
| "learning_rate": 5.695412294474208e-06, | |
| "loss": 4.4052, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.5003257328990228, | |
| "grad_norm": 7.847100098403116, | |
| "learning_rate": 5.682812229372225e-06, | |
| "loss": 4.3222, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.5011943539630836, | |
| "grad_norm": 4.926170240297532, | |
| "learning_rate": 5.67021072847179e-06, | |
| "loss": 4.3407, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.5020629750271444, | |
| "grad_norm": 4.81827513225177, | |
| "learning_rate": 5.657607890744485e-06, | |
| "loss": 4.5482, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.5029315960912052, | |
| "grad_norm": 9.69284798196995, | |
| "learning_rate": 5.64500381517239e-06, | |
| "loss": 4.5489, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.503800217155266, | |
| "grad_norm": 6.4397567914279525, | |
| "learning_rate": 5.632398600747307e-06, | |
| "loss": 4.5095, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.5046688382193268, | |
| "grad_norm": 7.564842572267876, | |
| "learning_rate": 5.619792346469988e-06, | |
| "loss": 4.3857, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.5055374592833877, | |
| "grad_norm": 5.717273086517985, | |
| "learning_rate": 5.607185151349342e-06, | |
| "loss": 4.601, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.5064060803474484, | |
| "grad_norm": 5.765808842291691, | |
| "learning_rate": 5.594577114401677e-06, | |
| "loss": 4.4892, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.5072747014115092, | |
| "grad_norm": 8.975257143861999, | |
| "learning_rate": 5.581968334649906e-06, | |
| "loss": 4.3199, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.50814332247557, | |
| "grad_norm": 5.004707622689579, | |
| "learning_rate": 5.56935891112278e-06, | |
| "loss": 4.5351, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.5090119435396309, | |
| "grad_norm": 12.075805744578952, | |
| "learning_rate": 5.5567489428541035e-06, | |
| "loss": 4.5282, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.5098805646036917, | |
| "grad_norm": 7.78389463909419, | |
| "learning_rate": 5.54413852888196e-06, | |
| "loss": 4.3581, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.5107491856677524, | |
| "grad_norm": 7.658427280297901, | |
| "learning_rate": 5.531527768247935e-06, | |
| "loss": 4.2974, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.5116178067318132, | |
| "grad_norm": 7.231975397294993, | |
| "learning_rate": 5.518916759996337e-06, | |
| "loss": 4.3415, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.512486427795874, | |
| "grad_norm": 6.751836330853958, | |
| "learning_rate": 5.506305603173414e-06, | |
| "loss": 4.3004, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.5133550488599349, | |
| "grad_norm": 8.355115868210659, | |
| "learning_rate": 5.493694396826589e-06, | |
| "loss": 4.4837, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.5142236699239957, | |
| "grad_norm": 6.193051769061237, | |
| "learning_rate": 5.481083240003665e-06, | |
| "loss": 4.4574, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.5150922909880564, | |
| "grad_norm": 8.721016177095263, | |
| "learning_rate": 5.468472231752065e-06, | |
| "loss": 4.5605, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.5159609120521172, | |
| "grad_norm": 8.690619765588671, | |
| "learning_rate": 5.455861471118041e-06, | |
| "loss": 4.4022, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.5168295331161781, | |
| "grad_norm": 6.149157342377077, | |
| "learning_rate": 5.443251057145899e-06, | |
| "loss": 4.3339, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.5176981541802389, | |
| "grad_norm": 9.505811782297348, | |
| "learning_rate": 5.430641088877221e-06, | |
| "loss": 4.2946, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.5185667752442997, | |
| "grad_norm": 7.387109532510519, | |
| "learning_rate": 5.418031665350096e-06, | |
| "loss": 4.4647, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.5194353963083604, | |
| "grad_norm": 8.365180622825434, | |
| "learning_rate": 5.405422885598324e-06, | |
| "loss": 4.376, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.5203040173724213, | |
| "grad_norm": 7.489996252960465, | |
| "learning_rate": 5.3928148486506584e-06, | |
| "loss": 4.452, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.5211726384364821, | |
| "grad_norm": 7.038197518326514, | |
| "learning_rate": 5.380207653530014e-06, | |
| "loss": 4.433, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.5220412595005429, | |
| "grad_norm": 6.5440132590329725, | |
| "learning_rate": 5.367601399252694e-06, | |
| "loss": 4.339, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.5229098805646037, | |
| "grad_norm": 7.590698363002852, | |
| "learning_rate": 5.354996184827612e-06, | |
| "loss": 4.5249, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.5237785016286645, | |
| "grad_norm": 5.330185549680669, | |
| "learning_rate": 5.3423921092555184e-06, | |
| "loss": 4.2862, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.5246471226927253, | |
| "grad_norm": 11.75232584431798, | |
| "learning_rate": 5.329789271528212e-06, | |
| "loss": 4.5025, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.5255157437567861, | |
| "grad_norm": 7.036744842599782, | |
| "learning_rate": 5.3171877706277785e-06, | |
| "loss": 4.5778, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.5263843648208469, | |
| "grad_norm": 14.767373768502868, | |
| "learning_rate": 5.304587705525795e-06, | |
| "loss": 4.6429, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.5272529858849077, | |
| "grad_norm": 13.155922558039405, | |
| "learning_rate": 5.291989175182569e-06, | |
| "loss": 4.308, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.5281216069489685, | |
| "grad_norm": 10.112650168109823, | |
| "learning_rate": 5.2793922785463515e-06, | |
| "loss": 4.484, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.5289902280130293, | |
| "grad_norm": 11.115807165127777, | |
| "learning_rate": 5.266797114552562e-06, | |
| "loss": 4.2531, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.5298588490770901, | |
| "grad_norm": 8.789007652727864, | |
| "learning_rate": 5.254203782123013e-06, | |
| "loss": 4.4873, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.530727470141151, | |
| "grad_norm": 8.584208445591985, | |
| "learning_rate": 5.241612380165131e-06, | |
| "loss": 4.474, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.5315960912052117, | |
| "grad_norm": 8.322437053520234, | |
| "learning_rate": 5.229023007571179e-06, | |
| "loss": 4.3145, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.5324647122692725, | |
| "grad_norm": 7.166099901687578, | |
| "learning_rate": 5.216435763217487e-06, | |
| "loss": 4.3139, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.5333333333333333, | |
| "grad_norm": 10.11706910820565, | |
| "learning_rate": 5.203850745963666e-06, | |
| "loss": 4.4996, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.5342019543973942, | |
| "grad_norm": 7.536732454472386, | |
| "learning_rate": 5.191268054651833e-06, | |
| "loss": 4.4032, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.535070575461455, | |
| "grad_norm": 9.963574957880162, | |
| "learning_rate": 5.178687788105842e-06, | |
| "loss": 4.5332, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.5359391965255157, | |
| "grad_norm": 7.28030489177068, | |
| "learning_rate": 5.166110045130503e-06, | |
| "loss": 4.4073, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.5368078175895765, | |
| "grad_norm": 9.51169465858539, | |
| "learning_rate": 5.153534924510799e-06, | |
| "loss": 4.3795, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.5376764386536373, | |
| "grad_norm": 6.905433659455053, | |
| "learning_rate": 5.1409625250111265e-06, | |
| "loss": 4.4542, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.5385450597176982, | |
| "grad_norm": 9.924412663417344, | |
| "learning_rate": 5.1283929453745055e-06, | |
| "loss": 4.2425, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.539413680781759, | |
| "grad_norm": 7.187624451476708, | |
| "learning_rate": 5.1158262843218076e-06, | |
| "loss": 4.509, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.5402823018458197, | |
| "grad_norm": 7.551854215214278, | |
| "learning_rate": 5.103262640550986e-06, | |
| "loss": 4.4906, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.5411509229098805, | |
| "grad_norm": 7.066809081049507, | |
| "learning_rate": 5.090702112736295e-06, | |
| "loss": 4.5479, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.5420195439739414, | |
| "grad_norm": 6.869001279033655, | |
| "learning_rate": 5.078144799527513e-06, | |
| "loss": 4.335, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.5428881650380022, | |
| "grad_norm": 6.184389891480793, | |
| "learning_rate": 5.0655907995491726e-06, | |
| "loss": 4.4417, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.543756786102063, | |
| "grad_norm": 5.725304760538053, | |
| "learning_rate": 5.053040211399792e-06, | |
| "loss": 4.4285, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.5446254071661237, | |
| "grad_norm": 6.858465748404722, | |
| "learning_rate": 5.0404931336510785e-06, | |
| "loss": 4.5386, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.5454940282301846, | |
| "grad_norm": 5.26607902680547, | |
| "learning_rate": 5.027949664847182e-06, | |
| "loss": 4.3599, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.5463626492942454, | |
| "grad_norm": 7.76854318501132, | |
| "learning_rate": 5.015409903503903e-06, | |
| "loss": 4.4496, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.5472312703583062, | |
| "grad_norm": 6.476156741121861, | |
| "learning_rate": 5.00287394810792e-06, | |
| "loss": 4.5308, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.548099891422367, | |
| "grad_norm": 5.759923281782965, | |
| "learning_rate": 4.9903418971160276e-06, | |
| "loss": 4.3798, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.5489685124864278, | |
| "grad_norm": 7.291747973328974, | |
| "learning_rate": 4.977813848954349e-06, | |
| "loss": 4.4205, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.5498371335504886, | |
| "grad_norm": 6.271882053028793, | |
| "learning_rate": 4.9652899020175706e-06, | |
| "loss": 4.4581, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.5507057546145494, | |
| "grad_norm": 7.284670931214164, | |
| "learning_rate": 4.952770154668173e-06, | |
| "loss": 4.5707, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.5515743756786102, | |
| "grad_norm": 4.328191558276064, | |
| "learning_rate": 4.940254705235643e-06, | |
| "loss": 4.5398, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.552442996742671, | |
| "grad_norm": 5.190845151993087, | |
| "learning_rate": 4.927743652015723e-06, | |
| "loss": 4.4066, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.5533116178067318, | |
| "grad_norm": 5.037059057279867, | |
| "learning_rate": 4.915237093269624e-06, | |
| "loss": 4.4995, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.5541802388707926, | |
| "grad_norm": 5.593175005468861, | |
| "learning_rate": 4.902735127223251e-06, | |
| "loss": 4.4209, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.5550488599348534, | |
| "grad_norm": 7.109606325525756, | |
| "learning_rate": 4.890237852066449e-06, | |
| "loss": 4.4592, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.5559174809989142, | |
| "grad_norm": 7.103955309366861, | |
| "learning_rate": 4.877745365952214e-06, | |
| "loss": 4.3799, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.5567861020629751, | |
| "grad_norm": 6.9509815611789785, | |
| "learning_rate": 4.865257766995929e-06, | |
| "loss": 4.4195, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.5576547231270358, | |
| "grad_norm": 5.206700465162741, | |
| "learning_rate": 4.852775153274597e-06, | |
| "loss": 4.4101, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.5585233441910966, | |
| "grad_norm": 9.493231366877055, | |
| "learning_rate": 4.8402976228260665e-06, | |
| "loss": 4.4044, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.5593919652551574, | |
| "grad_norm": 7.354474990771557, | |
| "learning_rate": 4.827825273648259e-06, | |
| "loss": 4.4041, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.5602605863192183, | |
| "grad_norm": 7.477143176038921, | |
| "learning_rate": 4.8153582036984055e-06, | |
| "loss": 4.2882, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.5611292073832791, | |
| "grad_norm": 6.891158599557778, | |
| "learning_rate": 4.802896510892274e-06, | |
| "loss": 4.4556, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.5619978284473398, | |
| "grad_norm": 6.181299599357454, | |
| "learning_rate": 4.790440293103399e-06, | |
| "loss": 4.5159, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.5628664495114006, | |
| "grad_norm": 6.219602356707123, | |
| "learning_rate": 4.7779896481623165e-06, | |
| "loss": 4.305, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.5637350705754615, | |
| "grad_norm": 7.689806502498994, | |
| "learning_rate": 4.765544673855793e-06, | |
| "loss": 4.588, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.5646036916395223, | |
| "grad_norm": 5.222351629918664, | |
| "learning_rate": 4.753105467926058e-06, | |
| "loss": 4.5114, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.5654723127035831, | |
| "grad_norm": 6.0422096833801255, | |
| "learning_rate": 4.740672128070033e-06, | |
| "loss": 4.4576, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.5663409337676438, | |
| "grad_norm": 6.35521507705001, | |
| "learning_rate": 4.728244751938576e-06, | |
| "loss": 4.2865, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.5672095548317047, | |
| "grad_norm": 5.486777370518167, | |
| "learning_rate": 4.715823437135698e-06, | |
| "loss": 4.3724, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.5680781758957655, | |
| "grad_norm": 6.005852070933836, | |
| "learning_rate": 4.703408281217808e-06, | |
| "loss": 4.3402, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.5689467969598263, | |
| "grad_norm": 5.5827526578094595, | |
| "learning_rate": 4.690999381692943e-06, | |
| "loss": 4.38, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.5698154180238871, | |
| "grad_norm": 4.834434200631698, | |
| "learning_rate": 4.678596836020003e-06, | |
| "loss": 4.3628, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.5706840390879478, | |
| "grad_norm": 5.54557315267325, | |
| "learning_rate": 4.666200741607986e-06, | |
| "loss": 4.423, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.5715526601520087, | |
| "grad_norm": 5.839997452850654, | |
| "learning_rate": 4.6538111958152195e-06, | |
| "loss": 4.4357, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.5724212812160695, | |
| "grad_norm": 5.117404141092577, | |
| "learning_rate": 4.6414282959486015e-06, | |
| "loss": 4.2396, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.5732899022801303, | |
| "grad_norm": 4.960346982680198, | |
| "learning_rate": 4.62905213926283e-06, | |
| "loss": 4.299, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.5741585233441912, | |
| "grad_norm": 5.666130352112717, | |
| "learning_rate": 4.616682822959644e-06, | |
| "loss": 4.3401, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.5750271444082519, | |
| "grad_norm": 8.838421793842883, | |
| "learning_rate": 4.604320444187058e-06, | |
| "loss": 4.5139, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.5758957654723127, | |
| "grad_norm": 6.818030373290299, | |
| "learning_rate": 4.591965100038604e-06, | |
| "loss": 4.3741, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.5767643865363735, | |
| "grad_norm": 5.238826628516235, | |
| "learning_rate": 4.579616887552556e-06, | |
| "loss": 4.3466, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.5776330076004343, | |
| "grad_norm": 6.774064034828669, | |
| "learning_rate": 4.567275903711182e-06, | |
| "loss": 4.5341, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.5785016286644951, | |
| "grad_norm": 10.5839617130325, | |
| "learning_rate": 4.554942245439977e-06, | |
| "loss": 4.55, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.5793702497285559, | |
| "grad_norm": 4.4556947282323875, | |
| "learning_rate": 4.542616009606896e-06, | |
| "loss": 4.4169, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.5802388707926167, | |
| "grad_norm": 7.295340591592647, | |
| "learning_rate": 4.5302972930216035e-06, | |
| "loss": 4.1999, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.5811074918566775, | |
| "grad_norm": 6.167555266777589, | |
| "learning_rate": 4.5179861924347105e-06, | |
| "loss": 4.4135, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.5819761129207384, | |
| "grad_norm": 4.262024109429587, | |
| "learning_rate": 4.505682804537002e-06, | |
| "loss": 4.3303, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.5828447339847991, | |
| "grad_norm": 5.436615646031177, | |
| "learning_rate": 4.493387225958698e-06, | |
| "loss": 4.2423, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.5837133550488599, | |
| "grad_norm": 7.121365514253581, | |
| "learning_rate": 4.481099553268683e-06, | |
| "loss": 4.3157, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.5845819761129207, | |
| "grad_norm": 4.638764375663413, | |
| "learning_rate": 4.468819882973743e-06, | |
| "loss": 4.591, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.5854505971769816, | |
| "grad_norm": 5.718150177859018, | |
| "learning_rate": 4.456548311517818e-06, | |
| "loss": 4.4703, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.5863192182410424, | |
| "grad_norm": 6.5924354876426445, | |
| "learning_rate": 4.444284935281245e-06, | |
| "loss": 4.4582, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.5871878393051031, | |
| "grad_norm": 5.619242120861036, | |
| "learning_rate": 4.432029850579983e-06, | |
| "loss": 4.2845, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.5880564603691639, | |
| "grad_norm": 8.851755293802896, | |
| "learning_rate": 4.419783153664885e-06, | |
| "loss": 4.4555, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.5889250814332248, | |
| "grad_norm": 4.981757116839891, | |
| "learning_rate": 4.407544940720912e-06, | |
| "loss": 4.4267, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.5897937024972856, | |
| "grad_norm": 8.597721909600045, | |
| "learning_rate": 4.395315307866404e-06, | |
| "loss": 4.3179, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.5906623235613464, | |
| "grad_norm": 6.495128819820371, | |
| "learning_rate": 4.383094351152309e-06, | |
| "loss": 4.4839, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.5915309446254071, | |
| "grad_norm": 7.608989760198819, | |
| "learning_rate": 4.370882166561432e-06, | |
| "loss": 4.4416, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.5923995656894679, | |
| "grad_norm": 5.854433397474583, | |
| "learning_rate": 4.358678850007681e-06, | |
| "loss": 4.3324, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.5932681867535288, | |
| "grad_norm": 7.407730937666916, | |
| "learning_rate": 4.3464844973353215e-06, | |
| "loss": 4.3833, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.5941368078175896, | |
| "grad_norm": 4.968500771960475, | |
| "learning_rate": 4.334299204318208e-06, | |
| "loss": 4.2341, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.5950054288816504, | |
| "grad_norm": 7.558518400906316, | |
| "learning_rate": 4.322123066659048e-06, | |
| "loss": 4.4444, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.5958740499457111, | |
| "grad_norm": 4.962150548530693, | |
| "learning_rate": 4.309956179988641e-06, | |
| "loss": 4.3576, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.596742671009772, | |
| "grad_norm": 6.938240211755133, | |
| "learning_rate": 4.2977986398651285e-06, | |
| "loss": 4.3146, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.5976112920738328, | |
| "grad_norm": 7.587763936061048, | |
| "learning_rate": 4.285650541773243e-06, | |
| "loss": 4.4289, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.5984799131378936, | |
| "grad_norm": 5.969538038161447, | |
| "learning_rate": 4.273511981123569e-06, | |
| "loss": 4.3759, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.5993485342019544, | |
| "grad_norm": 7.899549221564559, | |
| "learning_rate": 4.261383053251773e-06, | |
| "loss": 4.3071, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.6002171552660152, | |
| "grad_norm": 5.2445353058223505, | |
| "learning_rate": 4.2492638534178695e-06, | |
| "loss": 4.281, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.601085776330076, | |
| "grad_norm": 7.6245739463456985, | |
| "learning_rate": 4.237154476805475e-06, | |
| "loss": 4.2956, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.6019543973941368, | |
| "grad_norm": 5.470116425098481, | |
| "learning_rate": 4.225055018521048e-06, | |
| "loss": 4.2829, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.6028230184581976, | |
| "grad_norm": 8.639572149060786, | |
| "learning_rate": 4.2129655735931514e-06, | |
| "loss": 4.4738, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.6036916395222585, | |
| "grad_norm": 6.152417088405707, | |
| "learning_rate": 4.200886236971707e-06, | |
| "loss": 4.4568, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.6045602605863192, | |
| "grad_norm": 9.3108808124201, | |
| "learning_rate": 4.188817103527238e-06, | |
| "loss": 4.6181, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.60542888165038, | |
| "grad_norm": 6.812741933127771, | |
| "learning_rate": 4.176758268050141e-06, | |
| "loss": 4.3743, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.6062975027144408, | |
| "grad_norm": 7.946624143488305, | |
| "learning_rate": 4.164709825249931e-06, | |
| "loss": 4.3947, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.6071661237785017, | |
| "grad_norm": 7.52895947435545, | |
| "learning_rate": 4.152671869754496e-06, | |
| "loss": 4.4524, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.6080347448425625, | |
| "grad_norm": 6.78338934618417, | |
| "learning_rate": 4.140644496109358e-06, | |
| "loss": 4.4503, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.6089033659066232, | |
| "grad_norm": 7.448080226128123, | |
| "learning_rate": 4.128627798776933e-06, | |
| "loss": 4.353, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.609771986970684, | |
| "grad_norm": 5.9613691312207475, | |
| "learning_rate": 4.116621872135782e-06, | |
| "loss": 4.1222, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.6106406080347448, | |
| "grad_norm": 7.5790067996481625, | |
| "learning_rate": 4.104626810479878e-06, | |
| "loss": 4.3692, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.6115092290988057, | |
| "grad_norm": 5.453670887522505, | |
| "learning_rate": 4.092642708017853e-06, | |
| "loss": 4.491, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.6123778501628665, | |
| "grad_norm": 9.305782789439998, | |
| "learning_rate": 4.080669658872275e-06, | |
| "loss": 4.5958, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.6132464712269272, | |
| "grad_norm": 5.699148615360036, | |
| "learning_rate": 4.068707757078895e-06, | |
| "loss": 4.2548, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.614115092290988, | |
| "grad_norm": 7.911694729153124, | |
| "learning_rate": 4.056757096585914e-06, | |
| "loss": 4.4262, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.6149837133550489, | |
| "grad_norm": 6.872921512476096, | |
| "learning_rate": 4.044817771253243e-06, | |
| "loss": 4.3627, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.6158523344191097, | |
| "grad_norm": 7.476546715366341, | |
| "learning_rate": 4.0328898748517715e-06, | |
| "loss": 4.3003, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.6167209554831705, | |
| "grad_norm": 7.118852675240264, | |
| "learning_rate": 4.020973501062621e-06, | |
| "loss": 4.2213, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.6175895765472312, | |
| "grad_norm": 6.777324416098178, | |
| "learning_rate": 4.009068743476418e-06, | |
| "loss": 4.4026, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.6184581976112921, | |
| "grad_norm": 6.267261071143136, | |
| "learning_rate": 3.997175695592558e-06, | |
| "loss": 4.3549, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.6193268186753529, | |
| "grad_norm": 5.469188635248625, | |
| "learning_rate": 3.985294450818465e-06, | |
| "loss": 4.5069, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.6201954397394137, | |
| "grad_norm": 5.5464442937088645, | |
| "learning_rate": 3.973425102468864e-06, | |
| "loss": 4.3273, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.6210640608034745, | |
| "grad_norm": 6.3365698216315485, | |
| "learning_rate": 3.961567743765047e-06, | |
| "loss": 4.2548, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.6219326818675353, | |
| "grad_norm": 6.762739461810207, | |
| "learning_rate": 3.949722467834136e-06, | |
| "loss": 4.2804, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.6228013029315961, | |
| "grad_norm": 6.867786850485367, | |
| "learning_rate": 3.9378893677083585e-06, | |
| "loss": 4.4824, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.6236699239956569, | |
| "grad_norm": 4.426890287359104, | |
| "learning_rate": 3.926068536324318e-06, | |
| "loss": 4.2629, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.6245385450597177, | |
| "grad_norm": 5.764947147703928, | |
| "learning_rate": 3.914260066522249e-06, | |
| "loss": 4.4748, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.6254071661237784, | |
| "grad_norm": 6.449173453157844, | |
| "learning_rate": 3.902464051045308e-06, | |
| "loss": 4.2288, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.6262757871878393, | |
| "grad_norm": 7.156025732803798, | |
| "learning_rate": 3.890680582538835e-06, | |
| "loss": 4.471, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.6271444082519001, | |
| "grad_norm": 6.044472987310762, | |
| "learning_rate": 3.878909753549621e-06, | |
| "loss": 4.3782, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.6280130293159609, | |
| "grad_norm": 6.041979384384623, | |
| "learning_rate": 3.867151656525191e-06, | |
| "loss": 4.3686, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.6288816503800218, | |
| "grad_norm": 4.965615098879875, | |
| "learning_rate": 3.8554063838130774e-06, | |
| "loss": 4.3014, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.6297502714440825, | |
| "grad_norm": 6.077620537282477, | |
| "learning_rate": 3.84367402766008e-06, | |
| "loss": 4.5319, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.6306188925081433, | |
| "grad_norm": 7.348008590308795, | |
| "learning_rate": 3.831954680211567e-06, | |
| "loss": 4.4329, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.6314875135722041, | |
| "grad_norm": 4.148156488850514, | |
| "learning_rate": 3.820248433510721e-06, | |
| "loss": 4.3825, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.6323561346362649, | |
| "grad_norm": 5.868839788368692, | |
| "learning_rate": 3.8085553794978464e-06, | |
| "loss": 4.4061, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.6332247557003258, | |
| "grad_norm": 5.4956452972488465, | |
| "learning_rate": 3.7968756100096264e-06, | |
| "loss": 4.4363, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.6340933767643865, | |
| "grad_norm": 6.885854053993367, | |
| "learning_rate": 3.7852092167784057e-06, | |
| "loss": 4.4167, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.6349619978284473, | |
| "grad_norm": 3.9881851549507723, | |
| "learning_rate": 3.7735562914314753e-06, | |
| "loss": 4.2648, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.6358306188925081, | |
| "grad_norm": 3.4678484500146927, | |
| "learning_rate": 3.761916925490355e-06, | |
| "loss": 4.3543, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.636699239956569, | |
| "grad_norm": 5.980995299387367, | |
| "learning_rate": 3.7502912103700573e-06, | |
| "loss": 4.346, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.6375678610206298, | |
| "grad_norm": 6.151689157843326, | |
| "learning_rate": 3.738679237378395e-06, | |
| "loss": 4.4269, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.6384364820846905, | |
| "grad_norm": 5.540758455448802, | |
| "learning_rate": 3.7270810977152437e-06, | |
| "loss": 4.5648, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.6393051031487513, | |
| "grad_norm": 4.261398640280459, | |
| "learning_rate": 3.7154968824718335e-06, | |
| "loss": 4.1601, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.6401737242128122, | |
| "grad_norm": 5.048852018013335, | |
| "learning_rate": 3.703926682630034e-06, | |
| "loss": 4.1836, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.641042345276873, | |
| "grad_norm": 6.223488653795751, | |
| "learning_rate": 3.6923705890616385e-06, | |
| "loss": 4.2932, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.6419109663409338, | |
| "grad_norm": 4.00296364061148, | |
| "learning_rate": 3.6808286925276476e-06, | |
| "loss": 4.5271, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.6427795874049945, | |
| "grad_norm": 7.6337502276759, | |
| "learning_rate": 3.669301083677563e-06, | |
| "loss": 4.3391, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.6436482084690553, | |
| "grad_norm": 3.648965509349653, | |
| "learning_rate": 3.657787853048671e-06, | |
| "loss": 4.3708, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.6445168295331162, | |
| "grad_norm": 6.430179601615156, | |
| "learning_rate": 3.6462890910653287e-06, | |
| "loss": 4.2363, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.645385450597177, | |
| "grad_norm": 3.8810246258525414, | |
| "learning_rate": 3.6348048880382603e-06, | |
| "loss": 4.4739, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.6462540716612378, | |
| "grad_norm": 6.8458334527081615, | |
| "learning_rate": 3.6233353341638434e-06, | |
| "loss": 4.348, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.6471226927252985, | |
| "grad_norm": 3.560922995570656, | |
| "learning_rate": 3.611880519523403e-06, | |
| "loss": 4.3628, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.6479913137893594, | |
| "grad_norm": 5.703893391333462, | |
| "learning_rate": 3.600440534082501e-06, | |
| "loss": 4.2112, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.6488599348534202, | |
| "grad_norm": 5.21895941854659, | |
| "learning_rate": 3.5890154676902346e-06, | |
| "loss": 4.3305, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.649728555917481, | |
| "grad_norm": 4.447049515180316, | |
| "learning_rate": 3.5776054100785223e-06, | |
| "loss": 4.3878, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.6505971769815418, | |
| "grad_norm": 4.949990794413519, | |
| "learning_rate": 3.56621045086141e-06, | |
| "loss": 4.2244, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.6514657980456026, | |
| "grad_norm": 5.258986448566508, | |
| "learning_rate": 3.554830679534357e-06, | |
| "loss": 4.4186, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.6523344191096634, | |
| "grad_norm": 5.064550853101523, | |
| "learning_rate": 3.5434661854735406e-06, | |
| "loss": 4.3671, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.6532030401737242, | |
| "grad_norm": 5.6966971937871635, | |
| "learning_rate": 3.532117057935151e-06, | |
| "loss": 4.1835, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.654071661237785, | |
| "grad_norm": 5.913092142311183, | |
| "learning_rate": 3.520783386054689e-06, | |
| "loss": 4.414, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.6549402823018459, | |
| "grad_norm": 4.6289425679321985, | |
| "learning_rate": 3.5094652588462685e-06, | |
| "loss": 4.3397, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.6558089033659066, | |
| "grad_norm": 4.196557296390773, | |
| "learning_rate": 3.498162765201918e-06, | |
| "loss": 4.4643, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.6566775244299674, | |
| "grad_norm": 4.276537515669203, | |
| "learning_rate": 3.486875993890874e-06, | |
| "loss": 4.316, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.6575461454940282, | |
| "grad_norm": 3.638394175512923, | |
| "learning_rate": 3.475605033558896e-06, | |
| "loss": 4.2535, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.6584147665580891, | |
| "grad_norm": 3.7547136720071044, | |
| "learning_rate": 3.4643499727275704e-06, | |
| "loss": 4.4355, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.6592833876221499, | |
| "grad_norm": 4.529779878719585, | |
| "learning_rate": 3.4531108997935956e-06, | |
| "loss": 4.3798, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.6601520086862106, | |
| "grad_norm": 4.5346637615839205, | |
| "learning_rate": 3.4418879030281133e-06, | |
| "loss": 4.4147, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.6610206297502714, | |
| "grad_norm": 5.652353695284788, | |
| "learning_rate": 3.430681070575999e-06, | |
| "loss": 4.3372, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.6618892508143323, | |
| "grad_norm": 7.170934682558012, | |
| "learning_rate": 3.419490490455176e-06, | |
| "loss": 4.3881, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.6627578718783931, | |
| "grad_norm": 5.047705198747088, | |
| "learning_rate": 3.408316250555922e-06, | |
| "loss": 4.3122, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.6636264929424539, | |
| "grad_norm": 7.07536095984844, | |
| "learning_rate": 3.3971584386401816e-06, | |
| "loss": 4.3548, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.6644951140065146, | |
| "grad_norm": 3.3336263913275572, | |
| "learning_rate": 3.386017142340867e-06, | |
| "loss": 4.3743, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.6653637350705754, | |
| "grad_norm": 7.065332392195253, | |
| "learning_rate": 3.374892449161187e-06, | |
| "loss": 4.4953, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.6662323561346363, | |
| "grad_norm": 3.4501029280252937, | |
| "learning_rate": 3.3637844464739492e-06, | |
| "loss": 4.3351, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.6671009771986971, | |
| "grad_norm": 6.687004143766434, | |
| "learning_rate": 3.352693221520867e-06, | |
| "loss": 4.4546, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.6679695982627579, | |
| "grad_norm": 4.50307043857389, | |
| "learning_rate": 3.341618861411887e-06, | |
| "loss": 4.3398, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.6688382193268186, | |
| "grad_norm": 7.8703135527519, | |
| "learning_rate": 3.3305614531245077e-06, | |
| "loss": 4.2638, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.6697068403908795, | |
| "grad_norm": 3.922590217758762, | |
| "learning_rate": 3.319521083503075e-06, | |
| "loss": 4.2913, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.6705754614549403, | |
| "grad_norm": 8.346093676191689, | |
| "learning_rate": 3.3084978392581223e-06, | |
| "loss": 4.4285, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.6714440825190011, | |
| "grad_norm": 6.650194245577298, | |
| "learning_rate": 3.2974918069656797e-06, | |
| "loss": 4.2699, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.6723127035830619, | |
| "grad_norm": 6.560011908235111, | |
| "learning_rate": 3.286503073066596e-06, | |
| "loss": 4.4655, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.6731813246471227, | |
| "grad_norm": 7.055749636651488, | |
| "learning_rate": 3.2755317238658585e-06, | |
| "loss": 4.4159, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.6740499457111835, | |
| "grad_norm": 5.354746730373342, | |
| "learning_rate": 3.2645778455319143e-06, | |
| "loss": 4.3549, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.6749185667752443, | |
| "grad_norm": 5.146550880864029, | |
| "learning_rate": 3.2536415240959954e-06, | |
| "loss": 4.4694, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.6757871878393051, | |
| "grad_norm": 5.831672070421321, | |
| "learning_rate": 3.2427228454514496e-06, | |
| "loss": 4.1214, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.6766558089033659, | |
| "grad_norm": 4.521596862758205, | |
| "learning_rate": 3.2318218953530485e-06, | |
| "loss": 4.2677, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.6775244299674267, | |
| "grad_norm": 6.314265698622917, | |
| "learning_rate": 3.2209387594163316e-06, | |
| "loss": 4.412, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.6783930510314875, | |
| "grad_norm": 4.866756540705389, | |
| "learning_rate": 3.2100735231169238e-06, | |
| "loss": 4.4095, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.6792616720955483, | |
| "grad_norm": 6.493844920886384, | |
| "learning_rate": 3.1992262717898687e-06, | |
| "loss": 4.4263, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.6801302931596092, | |
| "grad_norm": 7.051427175831958, | |
| "learning_rate": 3.1883970906289568e-06, | |
| "loss": 4.3525, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.6809989142236699, | |
| "grad_norm": 6.86360370814285, | |
| "learning_rate": 3.1775860646860566e-06, | |
| "loss": 4.2127, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.6818675352877307, | |
| "grad_norm": 5.647923819665503, | |
| "learning_rate": 3.1667932788704414e-06, | |
| "loss": 4.252, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.6827361563517915, | |
| "grad_norm": 4.838224572668243, | |
| "learning_rate": 3.1560188179481356e-06, | |
| "loss": 4.3531, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.6836047774158523, | |
| "grad_norm": 6.70434806859771, | |
| "learning_rate": 3.1452627665412384e-06, | |
| "loss": 4.2778, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.6844733984799132, | |
| "grad_norm": 3.8436097880252174, | |
| "learning_rate": 3.134525209127255e-06, | |
| "loss": 4.2968, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.6853420195439739, | |
| "grad_norm": 6.343428669639347, | |
| "learning_rate": 3.1238062300384464e-06, | |
| "loss": 4.2696, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.6862106406080347, | |
| "grad_norm": 5.042121553913117, | |
| "learning_rate": 3.1131059134611595e-06, | |
| "loss": 4.3773, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.6870792616720955, | |
| "grad_norm": 5.935535940636119, | |
| "learning_rate": 3.1024243434351653e-06, | |
| "loss": 4.3472, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.6879478827361564, | |
| "grad_norm": 4.211630727724998, | |
| "learning_rate": 3.0917616038530006e-06, | |
| "loss": 4.3975, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.6888165038002172, | |
| "grad_norm": 5.39953776616512, | |
| "learning_rate": 3.0811177784593086e-06, | |
| "loss": 4.2212, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.6896851248642779, | |
| "grad_norm": 6.2670789214250355, | |
| "learning_rate": 3.0704929508501836e-06, | |
| "loss": 4.2709, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.6905537459283387, | |
| "grad_norm": 5.3978340661584285, | |
| "learning_rate": 3.059887204472508e-06, | |
| "loss": 4.3752, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.6914223669923996, | |
| "grad_norm": 4.222514400176097, | |
| "learning_rate": 3.0493006226233067e-06, | |
| "loss": 4.283, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.6922909880564604, | |
| "grad_norm": 6.363789170503971, | |
| "learning_rate": 3.03873328844908e-06, | |
| "loss": 4.2572, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.6931596091205212, | |
| "grad_norm": 4.996564390552643, | |
| "learning_rate": 3.028185284945164e-06, | |
| "loss": 4.2321, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.6940282301845819, | |
| "grad_norm": 3.7088045379057424, | |
| "learning_rate": 3.01765669495507e-06, | |
| "loss": 4.5402, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.6948968512486428, | |
| "grad_norm": 4.7675482950474475, | |
| "learning_rate": 3.0071476011698387e-06, | |
| "loss": 4.247, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6957654723127036, | |
| "grad_norm": 5.639346110912645, | |
| "learning_rate": 2.9966580861273847e-06, | |
| "loss": 4.4099, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.6966340933767644, | |
| "grad_norm": 4.770187045539365, | |
| "learning_rate": 2.9861882322118565e-06, | |
| "loss": 4.1644, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.6975027144408252, | |
| "grad_norm": 3.6627706854689106, | |
| "learning_rate": 2.9757381216529814e-06, | |
| "loss": 4.3601, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.698371335504886, | |
| "grad_norm": 3.9453583597610256, | |
| "learning_rate": 2.9653078365254267e-06, | |
| "loss": 4.3775, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.6992399565689468, | |
| "grad_norm": 4.121043954930214, | |
| "learning_rate": 2.954897458748147e-06, | |
| "loss": 4.2391, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.7001085776330076, | |
| "grad_norm": 3.5024329761434974, | |
| "learning_rate": 2.9445070700837486e-06, | |
| "loss": 4.4102, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.7009771986970684, | |
| "grad_norm": 5.1118475735166085, | |
| "learning_rate": 2.934136752137849e-06, | |
| "loss": 4.3506, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.7018458197611293, | |
| "grad_norm": 4.68701671319949, | |
| "learning_rate": 2.9237865863584204e-06, | |
| "loss": 4.273, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.70271444082519, | |
| "grad_norm": 4.207677749842511, | |
| "learning_rate": 2.9134566540351695e-06, | |
| "loss": 4.4052, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.7035830618892508, | |
| "grad_norm": 5.588319045300857, | |
| "learning_rate": 2.903147036298888e-06, | |
| "loss": 4.3446, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.7044516829533116, | |
| "grad_norm": 4.694144532657754, | |
| "learning_rate": 2.892857814120815e-06, | |
| "loss": 4.3963, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.7053203040173724, | |
| "grad_norm": 5.261041999651582, | |
| "learning_rate": 2.8825890683120087e-06, | |
| "loss": 4.1755, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.7061889250814333, | |
| "grad_norm": 3.537979646262889, | |
| "learning_rate": 2.8723408795227063e-06, | |
| "loss": 4.3071, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.707057546145494, | |
| "grad_norm": 3.8989008087932753, | |
| "learning_rate": 2.8621133282416836e-06, | |
| "loss": 4.262, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.7079261672095548, | |
| "grad_norm": 3.9902100422123175, | |
| "learning_rate": 2.8519064947956403e-06, | |
| "loss": 4.4904, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.7087947882736156, | |
| "grad_norm": 5.206036718500678, | |
| "learning_rate": 2.8417204593485566e-06, | |
| "loss": 4.3229, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.7096634093376765, | |
| "grad_norm": 5.794935718355797, | |
| "learning_rate": 2.831555301901061e-06, | |
| "loss": 4.3783, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.7105320304017373, | |
| "grad_norm": 4.0421897856037114, | |
| "learning_rate": 2.82141110228981e-06, | |
| "loss": 4.3945, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.711400651465798, | |
| "grad_norm": 4.6514330236834, | |
| "learning_rate": 2.811287940186866e-06, | |
| "loss": 4.3617, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.7122692725298588, | |
| "grad_norm": 3.6423689253885194, | |
| "learning_rate": 2.80118589509905e-06, | |
| "loss": 4.2173, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.7131378935939197, | |
| "grad_norm": 3.6396624178466896, | |
| "learning_rate": 2.791105046367341e-06, | |
| "loss": 4.3947, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 0.7140065146579805, | |
| "grad_norm": 3.8347475901940054, | |
| "learning_rate": 2.781045473166239e-06, | |
| "loss": 4.1539, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.7148751357220413, | |
| "grad_norm": 3.9705888799663027, | |
| "learning_rate": 2.771007254503149e-06, | |
| "loss": 4.4397, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 0.715743756786102, | |
| "grad_norm": 3.6196867847154164, | |
| "learning_rate": 2.7609904692177573e-06, | |
| "loss": 4.0627, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.7166123778501629, | |
| "grad_norm": 3.1130920798740362, | |
| "learning_rate": 2.750995195981412e-06, | |
| "loss": 4.2884, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.7174809989142237, | |
| "grad_norm": 3.288307548785416, | |
| "learning_rate": 2.7410215132965074e-06, | |
| "loss": 4.3433, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.7183496199782845, | |
| "grad_norm": 3.5010329845706596, | |
| "learning_rate": 2.7310694994958713e-06, | |
| "loss": 4.2592, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 0.7192182410423453, | |
| "grad_norm": 3.0547168595737504, | |
| "learning_rate": 2.721139232742137e-06, | |
| "loss": 4.3653, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.720086862106406, | |
| "grad_norm": 3.1039092159267074, | |
| "learning_rate": 2.711230791027144e-06, | |
| "loss": 4.319, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 0.7209554831704669, | |
| "grad_norm": 3.000221643527686, | |
| "learning_rate": 2.7013442521713157e-06, | |
| "loss": 4.3168, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.7218241042345277, | |
| "grad_norm": 3.56612108454193, | |
| "learning_rate": 2.691479693823053e-06, | |
| "loss": 4.2912, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 0.7226927252985885, | |
| "grad_norm": 4.045486521817849, | |
| "learning_rate": 2.6816371934581224e-06, | |
| "loss": 4.4497, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.7235613463626493, | |
| "grad_norm": 5.06756019279877, | |
| "learning_rate": 2.6718168283790502e-06, | |
| "loss": 4.2895, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 0.7244299674267101, | |
| "grad_norm": 5.209971490598599, | |
| "learning_rate": 2.6620186757145055e-06, | |
| "loss": 4.3888, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.7252985884907709, | |
| "grad_norm": 5.759379090394831, | |
| "learning_rate": 2.652242812418712e-06, | |
| "loss": 4.3862, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.7261672095548317, | |
| "grad_norm": 4.223016994293722, | |
| "learning_rate": 2.642489315270832e-06, | |
| "loss": 4.3214, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.7270358306188925, | |
| "grad_norm": 4.5544158674627955, | |
| "learning_rate": 2.632758260874358e-06, | |
| "loss": 4.2358, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.7279044516829533, | |
| "grad_norm": 4.589085818499639, | |
| "learning_rate": 2.6230497256565234e-06, | |
| "loss": 4.1757, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.7287730727470141, | |
| "grad_norm": 4.716082178521095, | |
| "learning_rate": 2.613363785867699e-06, | |
| "loss": 4.2912, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 0.7296416938110749, | |
| "grad_norm": 4.0122133897679975, | |
| "learning_rate": 2.6037005175807883e-06, | |
| "loss": 4.2088, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.7305103148751357, | |
| "grad_norm": 4.519123777136473, | |
| "learning_rate": 2.594059996690636e-06, | |
| "loss": 4.3983, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 0.7313789359391966, | |
| "grad_norm": 4.366992105875965, | |
| "learning_rate": 2.5844422989134294e-06, | |
| "loss": 4.3502, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.7322475570032573, | |
| "grad_norm": 4.902796136843425, | |
| "learning_rate": 2.574847499786103e-06, | |
| "loss": 4.3827, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 0.7331161780673181, | |
| "grad_norm": 3.7309923778017335, | |
| "learning_rate": 2.5652756746657474e-06, | |
| "loss": 4.1343, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.7339847991313789, | |
| "grad_norm": 4.633964698796071, | |
| "learning_rate": 2.5557268987290196e-06, | |
| "loss": 4.2807, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.7348534201954398, | |
| "grad_norm": 3.158778304410855, | |
| "learning_rate": 2.546201246971542e-06, | |
| "loss": 4.1983, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.7357220412595006, | |
| "grad_norm": 4.280110117256575, | |
| "learning_rate": 2.536698794207327e-06, | |
| "loss": 4.1502, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 0.7365906623235613, | |
| "grad_norm": 4.019320202479935, | |
| "learning_rate": 2.527219615068181e-06, | |
| "loss": 4.269, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.7374592833876221, | |
| "grad_norm": 4.684956718222349, | |
| "learning_rate": 2.517763784003121e-06, | |
| "loss": 4.2774, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 0.738327904451683, | |
| "grad_norm": 4.72042770271221, | |
| "learning_rate": 2.5083313752777893e-06, | |
| "loss": 4.3012, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.7391965255157438, | |
| "grad_norm": 3.9457131710176028, | |
| "learning_rate": 2.4989224629738705e-06, | |
| "loss": 4.3457, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.7400651465798046, | |
| "grad_norm": 3.86871572131023, | |
| "learning_rate": 2.4895371209885082e-06, | |
| "loss": 4.2702, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.7409337676438653, | |
| "grad_norm": 4.162164409779829, | |
| "learning_rate": 2.4801754230337287e-06, | |
| "loss": 4.2957, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 0.7418023887079261, | |
| "grad_norm": 5.002185565803835, | |
| "learning_rate": 2.4708374426358543e-06, | |
| "loss": 4.2632, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.742671009771987, | |
| "grad_norm": 5.4293250327800004, | |
| "learning_rate": 2.4615232531349332e-06, | |
| "loss": 4.2441, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.7435396308360478, | |
| "grad_norm": 4.995810113032375, | |
| "learning_rate": 2.452232927684166e-06, | |
| "loss": 4.2673, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.7444082519001086, | |
| "grad_norm": 3.2917198811992803, | |
| "learning_rate": 2.442966539249318e-06, | |
| "loss": 4.1662, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 0.7452768729641693, | |
| "grad_norm": 4.814530155931897, | |
| "learning_rate": 2.4337241606081587e-06, | |
| "loss": 4.2989, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.7461454940282302, | |
| "grad_norm": 5.299603931940639, | |
| "learning_rate": 2.424505864349886e-06, | |
| "loss": 4.3474, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 0.747014115092291, | |
| "grad_norm": 5.282979213255703, | |
| "learning_rate": 2.4153117228745543e-06, | |
| "loss": 4.2909, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.7478827361563518, | |
| "grad_norm": 4.536156542709376, | |
| "learning_rate": 2.4061418083925085e-06, | |
| "loss": 4.3692, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 0.7487513572204126, | |
| "grad_norm": 5.026058475783841, | |
| "learning_rate": 2.396996192923818e-06, | |
| "loss": 4.3141, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.7496199782844734, | |
| "grad_norm": 5.465713560303808, | |
| "learning_rate": 2.387874948297701e-06, | |
| "loss": 4.3735, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 0.7504885993485342, | |
| "grad_norm": 3.9399735021333, | |
| "learning_rate": 2.3787781461519786e-06, | |
| "loss": 4.2175, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.751357220412595, | |
| "grad_norm": 5.576777900877066, | |
| "learning_rate": 2.3697058579324976e-06, | |
| "loss": 4.1978, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.7522258414766558, | |
| "grad_norm": 5.197291281490897, | |
| "learning_rate": 2.3606581548925696e-06, | |
| "loss": 4.3048, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.7530944625407167, | |
| "grad_norm": 3.767610507301271, | |
| "learning_rate": 2.3516351080924206e-06, | |
| "loss": 4.257, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 0.7539630836047774, | |
| "grad_norm": 5.305437866922308, | |
| "learning_rate": 2.3426367883986254e-06, | |
| "loss": 4.3791, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.7548317046688382, | |
| "grad_norm": 4.698829389303069, | |
| "learning_rate": 2.333663266483555e-06, | |
| "loss": 4.4556, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 0.755700325732899, | |
| "grad_norm": 4.351560277566826, | |
| "learning_rate": 2.3247146128248183e-06, | |
| "loss": 4.3409, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.7565689467969599, | |
| "grad_norm": 3.103240864167305, | |
| "learning_rate": 2.3157908977047096e-06, | |
| "loss": 4.1706, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 0.7574375678610207, | |
| "grad_norm": 5.956940596200794, | |
| "learning_rate": 2.3068921912096585e-06, | |
| "loss": 4.3036, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.7583061889250814, | |
| "grad_norm": 4.030266382079339, | |
| "learning_rate": 2.2980185632296797e-06, | |
| "loss": 4.3916, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 0.7591748099891422, | |
| "grad_norm": 4.907585920293589, | |
| "learning_rate": 2.2891700834578175e-06, | |
| "loss": 4.1033, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.760043431053203, | |
| "grad_norm": 3.297402834871266, | |
| "learning_rate": 2.2803468213896063e-06, | |
| "loss": 4.3624, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.7609120521172639, | |
| "grad_norm": 4.662439533859811, | |
| "learning_rate": 2.2715488463225228e-06, | |
| "loss": 4.3034, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.7617806731813247, | |
| "grad_norm": 5.165185490895554, | |
| "learning_rate": 2.262776227355439e-06, | |
| "loss": 4.2157, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 0.7626492942453854, | |
| "grad_norm": 7.506559487228573, | |
| "learning_rate": 2.254029033388084e-06, | |
| "loss": 4.4301, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.7635179153094462, | |
| "grad_norm": 3.9252824887426554, | |
| "learning_rate": 2.2453073331204957e-06, | |
| "loss": 4.2351, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 0.7643865363735071, | |
| "grad_norm": 5.290959184706457, | |
| "learning_rate": 2.2366111950524906e-06, | |
| "loss": 4.2538, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.7652551574375679, | |
| "grad_norm": 4.7171299694367175, | |
| "learning_rate": 2.2279406874831164e-06, | |
| "loss": 4.4357, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 0.7661237785016287, | |
| "grad_norm": 29.979748449798375, | |
| "learning_rate": 2.2192958785101258e-06, | |
| "loss": 4.3621, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.7669923995656894, | |
| "grad_norm": 7.860730063348475, | |
| "learning_rate": 2.210676836029429e-06, | |
| "loss": 4.3751, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 0.7678610206297503, | |
| "grad_norm": 4.334356696505269, | |
| "learning_rate": 2.20208362773457e-06, | |
| "loss": 4.2966, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.7687296416938111, | |
| "grad_norm": 7.332346221259919, | |
| "learning_rate": 2.193516321116198e-06, | |
| "loss": 4.2862, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.7695982627578719, | |
| "grad_norm": 5.9624071677426125, | |
| "learning_rate": 2.1849749834615235e-06, | |
| "loss": 4.3632, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.7704668838219327, | |
| "grad_norm": 4.954161674367398, | |
| "learning_rate": 2.176459681853801e-06, | |
| "loss": 4.2059, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 0.7713355048859935, | |
| "grad_norm": 6.378179042127337, | |
| "learning_rate": 2.167970483171801e-06, | |
| "loss": 4.166, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.7722041259500543, | |
| "grad_norm": 4.9618200288070575, | |
| "learning_rate": 2.1595074540892815e-06, | |
| "loss": 4.3356, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 0.7730727470141151, | |
| "grad_norm": 6.856421677907798, | |
| "learning_rate": 2.1510706610744654e-06, | |
| "loss": 4.3511, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.7739413680781759, | |
| "grad_norm": 4.410108580533019, | |
| "learning_rate": 2.1426601703895195e-06, | |
| "loss": 4.2937, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 0.7748099891422368, | |
| "grad_norm": 7.583217716536857, | |
| "learning_rate": 2.13427604809003e-06, | |
| "loss": 4.3183, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.7756786102062975, | |
| "grad_norm": 5.9361144172296685, | |
| "learning_rate": 2.125918360024493e-06, | |
| "loss": 4.374, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 0.7765472312703583, | |
| "grad_norm": 6.875470826753328, | |
| "learning_rate": 2.117587171833789e-06, | |
| "loss": 4.4312, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.7774158523344191, | |
| "grad_norm": 5.096832067096056, | |
| "learning_rate": 2.109282548950667e-06, | |
| "loss": 4.278, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.77828447339848, | |
| "grad_norm": 5.729522521702186, | |
| "learning_rate": 2.1010045565992363e-06, | |
| "loss": 4.2937, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.7791530944625407, | |
| "grad_norm": 6.742811437853955, | |
| "learning_rate": 2.0927532597944496e-06, | |
| "loss": 4.2172, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 0.7800217155266015, | |
| "grad_norm": 5.6301478715572975, | |
| "learning_rate": 2.0845287233415963e-06, | |
| "loss": 4.176, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.7808903365906623, | |
| "grad_norm": 5.235841496697818, | |
| "learning_rate": 2.0763310118357893e-06, | |
| "loss": 4.3747, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 0.7817589576547231, | |
| "grad_norm": 5.049217344169785, | |
| "learning_rate": 2.06816018966146e-06, | |
| "loss": 4.2626, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.782627578718784, | |
| "grad_norm": 4.5457661427037195, | |
| "learning_rate": 2.060016320991853e-06, | |
| "loss": 4.2773, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 0.7834961997828447, | |
| "grad_norm": 4.117834715315518, | |
| "learning_rate": 2.051899469788522e-06, | |
| "loss": 4.3088, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.7843648208469055, | |
| "grad_norm": 5.135053442236565, | |
| "learning_rate": 2.043809699800824e-06, | |
| "loss": 4.2918, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 0.7852334419109663, | |
| "grad_norm": 7.82388906554306, | |
| "learning_rate": 2.0357470745654213e-06, | |
| "loss": 4.3529, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.7861020629750272, | |
| "grad_norm": 4.837908722893104, | |
| "learning_rate": 2.0277116574057905e-06, | |
| "loss": 4.1562, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.786970684039088, | |
| "grad_norm": 3.68904796267135, | |
| "learning_rate": 2.0197035114317056e-06, | |
| "loss": 4.2128, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.7878393051031487, | |
| "grad_norm": 4.534016411952663, | |
| "learning_rate": 2.0117226995387625e-06, | |
| "loss": 4.3172, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 0.7887079261672095, | |
| "grad_norm": 3.526628215688476, | |
| "learning_rate": 2.0037692844078728e-06, | |
| "loss": 4.2893, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.7895765472312704, | |
| "grad_norm": 3.8435396264813457, | |
| "learning_rate": 1.9958433285047766e-06, | |
| "loss": 4.2629, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 0.7904451682953312, | |
| "grad_norm": 12.416597855589442, | |
| "learning_rate": 1.9879448940795496e-06, | |
| "loss": 4.1928, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.791313789359392, | |
| "grad_norm": 3.4260543752536354, | |
| "learning_rate": 1.980074043166118e-06, | |
| "loss": 4.3879, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 0.7921824104234527, | |
| "grad_norm": 4.024334100744361, | |
| "learning_rate": 1.9722308375817616e-06, | |
| "loss": 4.3358, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.7930510314875135, | |
| "grad_norm": 3.8937964884137704, | |
| "learning_rate": 1.9644153389266428e-06, | |
| "loss": 4.2585, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 0.7939196525515744, | |
| "grad_norm": 4.112423538802947, | |
| "learning_rate": 1.9566276085833137e-06, | |
| "loss": 4.3666, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.7947882736156352, | |
| "grad_norm": 3.361914528770081, | |
| "learning_rate": 1.9488677077162294e-06, | |
| "loss": 4.3395, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.795656894679696, | |
| "grad_norm": 3.201261785374445, | |
| "learning_rate": 1.9411356972712802e-06, | |
| "loss": 4.1024, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.7965255157437567, | |
| "grad_norm": 3.1715636963506193, | |
| "learning_rate": 1.9334316379753037e-06, | |
| "loss": 4.3268, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 0.7973941368078176, | |
| "grad_norm": 4.702760089867412, | |
| "learning_rate": 1.92575559033561e-06, | |
| "loss": 4.2872, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.7982627578718784, | |
| "grad_norm": 4.301084643411801, | |
| "learning_rate": 1.91810761463951e-06, | |
| "loss": 4.2491, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 0.7991313789359392, | |
| "grad_norm": 4.368620526838832, | |
| "learning_rate": 1.9104877709538346e-06, | |
| "loss": 4.1825, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 4.541552734260731, | |
| "learning_rate": 1.902896119124471e-06, | |
| "loss": 4.2936, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 0.8008686210640608, | |
| "grad_norm": 3.753504642364837, | |
| "learning_rate": 1.8953327187758872e-06, | |
| "loss": 4.2807, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.8017372421281216, | |
| "grad_norm": 4.1076335957004035, | |
| "learning_rate": 1.8877976293106645e-06, | |
| "loss": 4.3054, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 0.8026058631921824, | |
| "grad_norm": 5.022940647754388, | |
| "learning_rate": 1.8802909099090328e-06, | |
| "loss": 4.2605, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.8034744842562432, | |
| "grad_norm": 4.601415915720413, | |
| "learning_rate": 1.8728126195284063e-06, | |
| "loss": 4.2974, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.8043431053203041, | |
| "grad_norm": 3.5937684601582505, | |
| "learning_rate": 1.8653628169029172e-06, | |
| "loss": 4.4372, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.8052117263843648, | |
| "grad_norm": 4.087141421454951, | |
| "learning_rate": 1.8579415605429566e-06, | |
| "loss": 4.3578, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 0.8060803474484256, | |
| "grad_norm": 3.768303031259501, | |
| "learning_rate": 1.850548908734715e-06, | |
| "loss": 4.3954, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.8069489685124864, | |
| "grad_norm": 4.631014557901584, | |
| "learning_rate": 1.843184919539724e-06, | |
| "loss": 4.3692, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 0.8078175895765473, | |
| "grad_norm": 5.588991832935504, | |
| "learning_rate": 1.8358496507944004e-06, | |
| "loss": 4.3679, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.8086862106406081, | |
| "grad_norm": 4.760700300967851, | |
| "learning_rate": 1.8285431601095932e-06, | |
| "loss": 4.2824, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 0.8095548317046688, | |
| "grad_norm": 7.110512990765382, | |
| "learning_rate": 1.8212655048701263e-06, | |
| "loss": 4.276, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.8104234527687296, | |
| "grad_norm": 4.84933914886446, | |
| "learning_rate": 1.8140167422343536e-06, | |
| "loss": 4.3264, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 0.8112920738327905, | |
| "grad_norm": 6.250495292898315, | |
| "learning_rate": 1.8067969291337111e-06, | |
| "loss": 4.2783, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.8121606948968513, | |
| "grad_norm": 5.081084041830173, | |
| "learning_rate": 1.7996061222722602e-06, | |
| "loss": 4.1333, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.8130293159609121, | |
| "grad_norm": 4.764801399944141, | |
| "learning_rate": 1.7924443781262537e-06, | |
| "loss": 4.258, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.8138979370249728, | |
| "grad_norm": 12.564644718015563, | |
| "learning_rate": 1.7853117529436853e-06, | |
| "loss": 4.4282, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 0.8147665580890336, | |
| "grad_norm": 9.219100670998142, | |
| "learning_rate": 1.7782083027438493e-06, | |
| "loss": 4.4036, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.8156351791530945, | |
| "grad_norm": 3.513961026121951, | |
| "learning_rate": 1.7711340833169027e-06, | |
| "loss": 4.2192, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 0.8165038002171553, | |
| "grad_norm": 7.5648391009618985, | |
| "learning_rate": 1.7640891502234242e-06, | |
| "loss": 4.3509, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.8173724212812161, | |
| "grad_norm": 7.457786533569347, | |
| "learning_rate": 1.7570735587939774e-06, | |
| "loss": 4.4761, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 0.8182410423452768, | |
| "grad_norm": 6.24994158926114, | |
| "learning_rate": 1.7500873641286822e-06, | |
| "loss": 4.3524, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.8191096634093377, | |
| "grad_norm": 4.706302961561861, | |
| "learning_rate": 1.7431306210967757e-06, | |
| "loss": 4.4049, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 0.8199782844733985, | |
| "grad_norm": 5.2319379724074375, | |
| "learning_rate": 1.7362033843361808e-06, | |
| "loss": 4.3184, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.8208469055374593, | |
| "grad_norm": 3.9931677964283336, | |
| "learning_rate": 1.7293057082530823e-06, | |
| "loss": 4.3872, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.8217155266015201, | |
| "grad_norm": 5.122675213766239, | |
| "learning_rate": 1.7224376470214965e-06, | |
| "loss": 4.2414, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.8225841476655809, | |
| "grad_norm": 4.8271711783672595, | |
| "learning_rate": 1.7155992545828459e-06, | |
| "loss": 4.3155, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 0.8234527687296417, | |
| "grad_norm": 5.08549920131417, | |
| "learning_rate": 1.708790584645536e-06, | |
| "loss": 4.3494, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.8243213897937025, | |
| "grad_norm": 3.2828478853516265, | |
| "learning_rate": 1.7020116906845314e-06, | |
| "loss": 4.1676, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 0.8251900108577633, | |
| "grad_norm": 4.217032686568317, | |
| "learning_rate": 1.6952626259409403e-06, | |
| "loss": 4.2612, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.826058631921824, | |
| "grad_norm": 3.3093948746453967, | |
| "learning_rate": 1.6885434434215928e-06, | |
| "loss": 4.2619, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 0.8269272529858849, | |
| "grad_norm": 3.4071908680921816, | |
| "learning_rate": 1.681854195898624e-06, | |
| "loss": 4.3616, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.8277958740499457, | |
| "grad_norm": 3.950423518377903, | |
| "learning_rate": 1.6751949359090608e-06, | |
| "loss": 4.4618, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 0.8286644951140065, | |
| "grad_norm": 3.113460449412214, | |
| "learning_rate": 1.6685657157544152e-06, | |
| "loss": 4.2883, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.8295331161780674, | |
| "grad_norm": 3.978159856987336, | |
| "learning_rate": 1.6619665875002589e-06, | |
| "loss": 4.2919, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.8304017372421281, | |
| "grad_norm": 3.627190156440042, | |
| "learning_rate": 1.655397602975829e-06, | |
| "loss": 4.4444, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.8312703583061889, | |
| "grad_norm": 3.501429507504989, | |
| "learning_rate": 1.6488588137736142e-06, | |
| "loss": 4.3162, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 0.8321389793702497, | |
| "grad_norm": 3.912898082889782, | |
| "learning_rate": 1.6423502712489498e-06, | |
| "loss": 4.1927, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.8330076004343105, | |
| "grad_norm": 2.9544066919092002, | |
| "learning_rate": 1.6358720265196162e-06, | |
| "loss": 4.2873, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 0.8338762214983714, | |
| "grad_norm": 3.7766990336064423, | |
| "learning_rate": 1.629424130465436e-06, | |
| "loss": 4.2311, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.8347448425624321, | |
| "grad_norm": 3.187557348988122, | |
| "learning_rate": 1.6230066337278721e-06, | |
| "loss": 4.3319, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 0.8356134636264929, | |
| "grad_norm": 2.9035478405536184, | |
| "learning_rate": 1.6166195867096379e-06, | |
| "loss": 4.2798, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.8364820846905537, | |
| "grad_norm": 3.581073588055323, | |
| "learning_rate": 1.6102630395742936e-06, | |
| "loss": 4.2378, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 0.8373507057546146, | |
| "grad_norm": 2.9717432186991313, | |
| "learning_rate": 1.603937042245851e-06, | |
| "loss": 4.1643, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.8382193268186754, | |
| "grad_norm": 3.4682548739965195, | |
| "learning_rate": 1.5976416444083919e-06, | |
| "loss": 4.2895, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.8390879478827361, | |
| "grad_norm": 2.8257749424306917, | |
| "learning_rate": 1.5913768955056669e-06, | |
| "loss": 4.2387, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.8399565689467969, | |
| "grad_norm": 3.4045296792179585, | |
| "learning_rate": 1.585142844740712e-06, | |
| "loss": 4.2493, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 0.8408251900108578, | |
| "grad_norm": 3.632548936442941, | |
| "learning_rate": 1.5789395410754624e-06, | |
| "loss": 4.0125, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.8416938110749186, | |
| "grad_norm": 3.458367314452796, | |
| "learning_rate": 1.5727670332303662e-06, | |
| "loss": 4.2027, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 0.8425624321389794, | |
| "grad_norm": 3.5230657596459563, | |
| "learning_rate": 1.5666253696840039e-06, | |
| "loss": 4.2408, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.8434310532030401, | |
| "grad_norm": 3.0474045364092475, | |
| "learning_rate": 1.5605145986727055e-06, | |
| "loss": 4.1706, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 0.844299674267101, | |
| "grad_norm": 2.9068561874294243, | |
| "learning_rate": 1.5544347681901708e-06, | |
| "loss": 4.2883, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.8451682953311618, | |
| "grad_norm": 3.2628088807621385, | |
| "learning_rate": 1.548385925987097e-06, | |
| "loss": 4.2064, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 0.8460369163952226, | |
| "grad_norm": 2.877609574460853, | |
| "learning_rate": 1.5423681195707995e-06, | |
| "loss": 4.2439, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.8469055374592834, | |
| "grad_norm": 3.6963360738924007, | |
| "learning_rate": 1.5363813962048404e-06, | |
| "loss": 4.4093, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.8477741585233441, | |
| "grad_norm": 3.6396595948485686, | |
| "learning_rate": 1.530425802908657e-06, | |
| "loss": 4.2486, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.848642779587405, | |
| "grad_norm": 3.0025316303653593, | |
| "learning_rate": 1.5245013864571915e-06, | |
| "loss": 4.2436, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 0.8495114006514658, | |
| "grad_norm": 3.075143454138421, | |
| "learning_rate": 1.518608193380527e-06, | |
| "loss": 4.1952, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.8503800217155266, | |
| "grad_norm": 3.837382455029135, | |
| "learning_rate": 1.5127462699635175e-06, | |
| "loss": 4.3256, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 0.8512486427795874, | |
| "grad_norm": 3.912314406379558, | |
| "learning_rate": 1.5069156622454286e-06, | |
| "loss": 4.1805, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.8521172638436482, | |
| "grad_norm": 3.556579266906855, | |
| "learning_rate": 1.5011164160195713e-06, | |
| "loss": 4.187, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 0.852985884907709, | |
| "grad_norm": 4.767048077361486, | |
| "learning_rate": 1.495348576832945e-06, | |
| "loss": 4.3401, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.8538545059717698, | |
| "grad_norm": 3.34570276142316, | |
| "learning_rate": 1.4896121899858855e-06, | |
| "loss": 4.3152, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 0.8547231270358306, | |
| "grad_norm": 3.928790482520009, | |
| "learning_rate": 1.4839073005316954e-06, | |
| "loss": 4.4026, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.8555917480998915, | |
| "grad_norm": 3.896817042687782, | |
| "learning_rate": 1.4782339532763035e-06, | |
| "loss": 4.3404, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.8564603691639522, | |
| "grad_norm": 3.5320973494543972, | |
| "learning_rate": 1.4725921927779053e-06, | |
| "loss": 4.2046, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.857328990228013, | |
| "grad_norm": 4.983206804496437, | |
| "learning_rate": 1.466982063346617e-06, | |
| "loss": 4.2724, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 0.8581976112920738, | |
| "grad_norm": 2.6576025845538602, | |
| "learning_rate": 1.4614036090441242e-06, | |
| "loss": 4.2774, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.8590662323561347, | |
| "grad_norm": 3.890820577234005, | |
| "learning_rate": 1.4558568736833403e-06, | |
| "loss": 4.3634, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 0.8599348534201955, | |
| "grad_norm": 4.047293622617586, | |
| "learning_rate": 1.450341900828055e-06, | |
| "loss": 4.2452, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.8608034744842562, | |
| "grad_norm": 3.082152289805242, | |
| "learning_rate": 1.4448587337926029e-06, | |
| "loss": 4.2638, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 0.861672095548317, | |
| "grad_norm": 3.279358197940446, | |
| "learning_rate": 1.4394074156415127e-06, | |
| "loss": 4.4121, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.8625407166123779, | |
| "grad_norm": 3.527792553396294, | |
| "learning_rate": 1.4339879891891745e-06, | |
| "loss": 4.1801, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 0.8634093376764387, | |
| "grad_norm": 3.505923077019055, | |
| "learning_rate": 1.4286004969995026e-06, | |
| "loss": 4.2853, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.8642779587404995, | |
| "grad_norm": 3.286205070282355, | |
| "learning_rate": 1.4232449813856024e-06, | |
| "loss": 4.375, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.8651465798045602, | |
| "grad_norm": 3.931680228410995, | |
| "learning_rate": 1.4179214844094354e-06, | |
| "loss": 4.2502, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.866015200868621, | |
| "grad_norm": 2.4039398272576897, | |
| "learning_rate": 1.4126300478814912e-06, | |
| "loss": 4.208, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 0.8668838219326819, | |
| "grad_norm": 3.7568390710068518, | |
| "learning_rate": 1.4073707133604553e-06, | |
| "loss": 4.1628, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.8677524429967427, | |
| "grad_norm": 3.3414877541755748, | |
| "learning_rate": 1.4021435221528907e-06, | |
| "loss": 4.1127, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 0.8686210640608035, | |
| "grad_norm": 3.458776188975169, | |
| "learning_rate": 1.3969485153129052e-06, | |
| "loss": 4.1204, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.8694896851248642, | |
| "grad_norm": 4.919567300721427, | |
| "learning_rate": 1.3917857336418311e-06, | |
| "loss": 4.3408, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 0.8703583061889251, | |
| "grad_norm": 3.065606325345494, | |
| "learning_rate": 1.3866552176879073e-06, | |
| "loss": 3.9601, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 0.8712269272529859, | |
| "grad_norm": 3.761007586350327, | |
| "learning_rate": 1.3815570077459616e-06, | |
| "loss": 4.2438, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 0.8720955483170467, | |
| "grad_norm": 2.885986718619506, | |
| "learning_rate": 1.3764911438570873e-06, | |
| "loss": 4.26, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 0.8729641693811075, | |
| "grad_norm": 3.826731301340684, | |
| "learning_rate": 1.3714576658083356e-06, | |
| "loss": 4.3967, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.8738327904451683, | |
| "grad_norm": 2.7458285420431805, | |
| "learning_rate": 1.366456613132402e-06, | |
| "loss": 4.3103, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 0.8747014115092291, | |
| "grad_norm": 3.0626194989917996, | |
| "learning_rate": 1.3614880251073126e-06, | |
| "loss": 4.3979, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 0.8755700325732899, | |
| "grad_norm": 2.904930109177129, | |
| "learning_rate": 1.356551940756119e-06, | |
| "loss": 4.1057, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 0.8764386536373507, | |
| "grad_norm": 2.8060273606690864, | |
| "learning_rate": 1.3516483988465911e-06, | |
| "loss": 4.2066, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 0.8773072747014115, | |
| "grad_norm": 2.9968484430903852, | |
| "learning_rate": 1.3467774378909088e-06, | |
| "loss": 4.2804, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.8781758957654723, | |
| "grad_norm": 2.4009421672150264, | |
| "learning_rate": 1.3419390961453673e-06, | |
| "loss": 4.2054, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 0.8790445168295331, | |
| "grad_norm": 2.9054455101472914, | |
| "learning_rate": 1.3371334116100692e-06, | |
| "loss": 4.2161, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 0.8799131378935939, | |
| "grad_norm": 2.7097147432315474, | |
| "learning_rate": 1.332360422028629e-06, | |
| "loss": 4.2635, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 0.8807817589576548, | |
| "grad_norm": 3.4351233135843895, | |
| "learning_rate": 1.3276201648878778e-06, | |
| "loss": 4.235, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 0.8816503800217155, | |
| "grad_norm": 3.021256304492239, | |
| "learning_rate": 1.3229126774175663e-06, | |
| "loss": 4.1398, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.8825190010857763, | |
| "grad_norm": 4.561864873305928, | |
| "learning_rate": 1.3182379965900755e-06, | |
| "loss": 4.2196, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 0.8833876221498371, | |
| "grad_norm": 2.25054567072425, | |
| "learning_rate": 1.3135961591201234e-06, | |
| "loss": 4.3461, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 0.884256243213898, | |
| "grad_norm": 4.349538372833775, | |
| "learning_rate": 1.3089872014644772e-06, | |
| "loss": 4.1902, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 0.8851248642779588, | |
| "grad_norm": 5.211240766793744, | |
| "learning_rate": 1.3044111598216697e-06, | |
| "loss": 4.127, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 0.8859934853420195, | |
| "grad_norm": 3.915181006258746, | |
| "learning_rate": 1.2998680701317116e-06, | |
| "loss": 4.3165, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.8868621064060803, | |
| "grad_norm": 2.634220197635655, | |
| "learning_rate": 1.2953579680758102e-06, | |
| "loss": 4.0896, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 0.8877307274701411, | |
| "grad_norm": 4.444464060721951, | |
| "learning_rate": 1.2908808890760898e-06, | |
| "loss": 4.3389, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 0.888599348534202, | |
| "grad_norm": 3.6060226328219223, | |
| "learning_rate": 1.2864368682953144e-06, | |
| "loss": 4.3551, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 0.8894679695982628, | |
| "grad_norm": 5.166782973498534, | |
| "learning_rate": 1.2820259406366086e-06, | |
| "loss": 4.343, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.8903365906623235, | |
| "grad_norm": 2.8773053428878397, | |
| "learning_rate": 1.2776481407431858e-06, | |
| "loss": 4.2031, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.8912052117263843, | |
| "grad_norm": 4.339800538926299, | |
| "learning_rate": 1.2733035029980764e-06, | |
| "loss": 4.348, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 0.8920738327904452, | |
| "grad_norm": 3.237109745257262, | |
| "learning_rate": 1.2689920615238564e-06, | |
| "loss": 4.4058, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 0.892942453854506, | |
| "grad_norm": 3.5663716818909514, | |
| "learning_rate": 1.2647138501823787e-06, | |
| "loss": 4.3298, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 0.8938110749185668, | |
| "grad_norm": 3.7872661083743613, | |
| "learning_rate": 1.2604689025745097e-06, | |
| "loss": 4.3178, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 0.8946796959826275, | |
| "grad_norm": 3.555878208510988, | |
| "learning_rate": 1.2562572520398636e-06, | |
| "loss": 4.429, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.8955483170466884, | |
| "grad_norm": 3.548292136180123, | |
| "learning_rate": 1.2520789316565407e-06, | |
| "loss": 4.2436, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 0.8964169381107492, | |
| "grad_norm": 2.8754548971872453, | |
| "learning_rate": 1.247933974240869e-06, | |
| "loss": 4.2454, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 0.89728555917481, | |
| "grad_norm": 3.1391372015122534, | |
| "learning_rate": 1.2438224123471442e-06, | |
| "loss": 4.2209, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 0.8981541802388708, | |
| "grad_norm": 4.08265544167836, | |
| "learning_rate": 1.2397442782673751e-06, | |
| "loss": 4.2561, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 0.8990228013029316, | |
| "grad_norm": 2.8484362391754097, | |
| "learning_rate": 1.2356996040310312e-06, | |
| "loss": 4.2269, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.8998914223669924, | |
| "grad_norm": 3.802284151154688, | |
| "learning_rate": 1.231688421404789e-06, | |
| "loss": 4.149, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 0.9007600434310532, | |
| "grad_norm": 2.8793289723328175, | |
| "learning_rate": 1.2277107618922843e-06, | |
| "loss": 4.2458, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 0.901628664495114, | |
| "grad_norm": 3.515343781948761, | |
| "learning_rate": 1.2237666567338632e-06, | |
| "loss": 4.0914, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 0.9024972855591749, | |
| "grad_norm": 3.5760188126248926, | |
| "learning_rate": 1.2198561369063366e-06, | |
| "loss": 4.1833, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 0.9033659066232356, | |
| "grad_norm": 2.8127761697413787, | |
| "learning_rate": 1.2159792331227404e-06, | |
| "loss": 4.1005, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.9042345276872964, | |
| "grad_norm": 3.4802083558560195, | |
| "learning_rate": 1.212135975832091e-06, | |
| "loss": 4.1616, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 0.9051031487513572, | |
| "grad_norm": 3.3333225446939205, | |
| "learning_rate": 1.2083263952191446e-06, | |
| "loss": 4.3845, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 0.905971769815418, | |
| "grad_norm": 2.88436141121272, | |
| "learning_rate": 1.2045505212041644e-06, | |
| "loss": 4.3108, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 0.9068403908794789, | |
| "grad_norm": 3.6801662883359194, | |
| "learning_rate": 1.200808383442684e-06, | |
| "loss": 4.2905, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 0.9077090119435396, | |
| "grad_norm": 3.019629347981631, | |
| "learning_rate": 1.1971000113252726e-06, | |
| "loss": 4.1987, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.9085776330076004, | |
| "grad_norm": 4.409221785598703, | |
| "learning_rate": 1.1934254339773074e-06, | |
| "loss": 4.3084, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 0.9094462540716612, | |
| "grad_norm": 2.709116535172474, | |
| "learning_rate": 1.1897846802587395e-06, | |
| "loss": 4.1717, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 0.9103148751357221, | |
| "grad_norm": 3.8179262996054235, | |
| "learning_rate": 1.1861777787638762e-06, | |
| "loss": 4.3074, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 0.9111834961997829, | |
| "grad_norm": 3.281229721724311, | |
| "learning_rate": 1.1826047578211473e-06, | |
| "loss": 4.5075, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 0.9120521172638436, | |
| "grad_norm": 3.0456938909447113, | |
| "learning_rate": 1.1790656454928866e-06, | |
| "loss": 4.4835, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.9129207383279044, | |
| "grad_norm": 3.2615534456726123, | |
| "learning_rate": 1.1755604695751134e-06, | |
| "loss": 4.3037, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 0.9137893593919653, | |
| "grad_norm": 2.9749791444621163, | |
| "learning_rate": 1.1720892575973095e-06, | |
| "loss": 4.3342, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 0.9146579804560261, | |
| "grad_norm": 3.3527992702942577, | |
| "learning_rate": 1.1686520368222066e-06, | |
| "loss": 4.2419, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 0.9155266015200869, | |
| "grad_norm": 2.4945121180562797, | |
| "learning_rate": 1.1652488342455726e-06, | |
| "loss": 4.1137, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 0.9163952225841476, | |
| "grad_norm": 3.2304134555988897, | |
| "learning_rate": 1.161879676595996e-06, | |
| "loss": 4.3529, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.9172638436482085, | |
| "grad_norm": 6.349732062570278, | |
| "learning_rate": 1.1585445903346784e-06, | |
| "loss": 4.1944, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 0.9181324647122693, | |
| "grad_norm": 2.9360851966959536, | |
| "learning_rate": 1.1552436016552273e-06, | |
| "loss": 4.1959, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 0.9190010857763301, | |
| "grad_norm": 3.205644542692188, | |
| "learning_rate": 1.1519767364834494e-06, | |
| "loss": 4.3528, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 0.9198697068403909, | |
| "grad_norm": 3.2880341446977126, | |
| "learning_rate": 1.1487440204771454e-06, | |
| "loss": 4.3303, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 0.9207383279044516, | |
| "grad_norm": 3.6622193093061615, | |
| "learning_rate": 1.1455454790259118e-06, | |
| "loss": 4.2606, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.9216069489685125, | |
| "grad_norm": 3.0436045507003873, | |
| "learning_rate": 1.14238113725094e-06, | |
| "loss": 4.3319, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 0.9224755700325733, | |
| "grad_norm": 2.946350336011715, | |
| "learning_rate": 1.1392510200048167e-06, | |
| "loss": 4.3378, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 0.9233441910966341, | |
| "grad_norm": 2.864144280137868, | |
| "learning_rate": 1.1361551518713331e-06, | |
| "loss": 4.3061, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 0.924212812160695, | |
| "grad_norm": 2.742188676823669, | |
| "learning_rate": 1.133093557165288e-06, | |
| "loss": 4.4552, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 0.9250814332247557, | |
| "grad_norm": 2.961292413358245, | |
| "learning_rate": 1.1300662599322992e-06, | |
| "loss": 4.2678, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.9259500542888165, | |
| "grad_norm": 3.237248386245042, | |
| "learning_rate": 1.1270732839486137e-06, | |
| "loss": 4.3326, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 0.9268186753528773, | |
| "grad_norm": 3.4990748997732877, | |
| "learning_rate": 1.1241146527209192e-06, | |
| "loss": 4.3578, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 0.9276872964169381, | |
| "grad_norm": 2.895322090554506, | |
| "learning_rate": 1.1211903894861655e-06, | |
| "loss": 4.2311, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 0.9285559174809989, | |
| "grad_norm": 3.687438443063813, | |
| "learning_rate": 1.1183005172113743e-06, | |
| "loss": 4.2228, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 0.9294245385450597, | |
| "grad_norm": 2.6961355880795748, | |
| "learning_rate": 1.1154450585934625e-06, | |
| "loss": 4.0498, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.9302931596091205, | |
| "grad_norm": 3.688749799792538, | |
| "learning_rate": 1.1126240360590658e-06, | |
| "loss": 4.1798, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 0.9311617806731813, | |
| "grad_norm": 2.8110003232896625, | |
| "learning_rate": 1.1098374717643587e-06, | |
| "loss": 4.3087, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 0.9320304017372422, | |
| "grad_norm": 2.5887554177535943, | |
| "learning_rate": 1.1070853875948837e-06, | |
| "loss": 4.2262, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 0.9328990228013029, | |
| "grad_norm": 2.847071399167459, | |
| "learning_rate": 1.1043678051653768e-06, | |
| "loss": 4.2708, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 0.9337676438653637, | |
| "grad_norm": 4.142401135834692, | |
| "learning_rate": 1.1016847458195999e-06, | |
| "loss": 4.1147, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.9346362649294245, | |
| "grad_norm": 2.9501471418720557, | |
| "learning_rate": 1.0990362306301725e-06, | |
| "loss": 4.2384, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 0.9355048859934854, | |
| "grad_norm": 2.501434571608401, | |
| "learning_rate": 1.0964222803984048e-06, | |
| "loss": 4.3386, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 0.9363735070575462, | |
| "grad_norm": 3.0424764452901956, | |
| "learning_rate": 1.0938429156541364e-06, | |
| "loss": 4.3864, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 0.9372421281216069, | |
| "grad_norm": 3.904570325899996, | |
| "learning_rate": 1.0912981566555736e-06, | |
| "loss": 4.1481, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 0.9381107491856677, | |
| "grad_norm": 3.3812480967472127, | |
| "learning_rate": 1.0887880233891307e-06, | |
| "loss": 4.4023, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.9389793702497286, | |
| "grad_norm": 3.5320678055096018, | |
| "learning_rate": 1.0863125355692749e-06, | |
| "loss": 4.1932, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 0.9398479913137894, | |
| "grad_norm": 3.2937159160580904, | |
| "learning_rate": 1.0838717126383676e-06, | |
| "loss": 4.1515, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 0.9407166123778502, | |
| "grad_norm": 3.9965990047935973, | |
| "learning_rate": 1.081465573766515e-06, | |
| "loss": 4.2552, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 0.9415852334419109, | |
| "grad_norm": 2.6702475745108014, | |
| "learning_rate": 1.079094137851415e-06, | |
| "loss": 4.2498, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 0.9424538545059717, | |
| "grad_norm": 4.745534492210504, | |
| "learning_rate": 1.0767574235182125e-06, | |
| "loss": 4.2509, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.9433224755700326, | |
| "grad_norm": 2.632042835693663, | |
| "learning_rate": 1.0744554491193483e-06, | |
| "loss": 4.2512, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 0.9441910966340934, | |
| "grad_norm": 4.902218185640278, | |
| "learning_rate": 1.0721882327344199e-06, | |
| "loss": 4.2946, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 0.9450597176981542, | |
| "grad_norm": 3.611290466221663, | |
| "learning_rate": 1.0699557921700337e-06, | |
| "loss": 4.3242, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 0.9459283387622149, | |
| "grad_norm": 4.159732582837829, | |
| "learning_rate": 1.0677581449596724e-06, | |
| "loss": 4.3335, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 0.9467969598262758, | |
| "grad_norm": 3.3609545882421, | |
| "learning_rate": 1.0655953083635507e-06, | |
| "loss": 4.3966, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.9476655808903366, | |
| "grad_norm": 3.43860026090556, | |
| "learning_rate": 1.0634672993684828e-06, | |
| "loss": 4.2277, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 0.9485342019543974, | |
| "grad_norm": 3.0705569428179884, | |
| "learning_rate": 1.0613741346877496e-06, | |
| "loss": 4.2567, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 0.9494028230184582, | |
| "grad_norm": 3.3793501615647195, | |
| "learning_rate": 1.0593158307609649e-06, | |
| "loss": 4.237, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 0.950271444082519, | |
| "grad_norm": 2.929105704044937, | |
| "learning_rate": 1.0572924037539494e-06, | |
| "loss": 4.2117, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 0.9511400651465798, | |
| "grad_norm": 2.71792555170425, | |
| "learning_rate": 1.0553038695586018e-06, | |
| "loss": 4.1614, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.9520086862106406, | |
| "grad_norm": 3.78382633594629, | |
| "learning_rate": 1.0533502437927722e-06, | |
| "loss": 4.3529, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 0.9528773072747014, | |
| "grad_norm": 2.607501362679155, | |
| "learning_rate": 1.0514315418001456e-06, | |
| "loss": 4.3052, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 0.9537459283387623, | |
| "grad_norm": 3.663360468883577, | |
| "learning_rate": 1.0495477786501138e-06, | |
| "loss": 4.1874, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 0.954614549402823, | |
| "grad_norm": 2.630120856617326, | |
| "learning_rate": 1.0476989691376622e-06, | |
| "loss": 4.242, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 0.9554831704668838, | |
| "grad_norm": 3.823941956325378, | |
| "learning_rate": 1.045885127783252e-06, | |
| "loss": 4.2043, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.9563517915309446, | |
| "grad_norm": 2.312883220474864, | |
| "learning_rate": 1.0441062688327051e-06, | |
| "loss": 4.1473, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 0.9572204125950055, | |
| "grad_norm": 4.094343889130672, | |
| "learning_rate": 1.0423624062570952e-06, | |
| "loss": 4.222, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 0.9580890336590663, | |
| "grad_norm": 2.209599527243729, | |
| "learning_rate": 1.0406535537526343e-06, | |
| "loss": 4.1423, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 0.958957654723127, | |
| "grad_norm": 3.382906333029422, | |
| "learning_rate": 1.0389797247405677e-06, | |
| "loss": 4.2021, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 0.9598262757871878, | |
| "grad_norm": 2.1618780172948004, | |
| "learning_rate": 1.0373409323670688e-06, | |
| "loss": 4.2185, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.9606948968512486, | |
| "grad_norm": 2.734044937681216, | |
| "learning_rate": 1.0357371895031331e-06, | |
| "loss": 4.1892, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 0.9615635179153095, | |
| "grad_norm": 3.0779816158117237, | |
| "learning_rate": 1.0341685087444804e-06, | |
| "loss": 4.1711, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 0.9624321389793703, | |
| "grad_norm": 3.4747018762690973, | |
| "learning_rate": 1.0326349024114533e-06, | |
| "loss": 4.1973, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 0.963300760043431, | |
| "grad_norm": 36.83010937108777, | |
| "learning_rate": 1.0311363825489225e-06, | |
| "loss": 4.3603, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 0.9641693811074918, | |
| "grad_norm": 4.281579688730939, | |
| "learning_rate": 1.0296729609261914e-06, | |
| "loss": 4.2369, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.9650380021715527, | |
| "grad_norm": 3.6658084904084647, | |
| "learning_rate": 1.0282446490369017e-06, | |
| "loss": 4.328, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 0.9659066232356135, | |
| "grad_norm": 3.1538713927243083, | |
| "learning_rate": 1.0268514580989476e-06, | |
| "loss": 4.224, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 0.9667752442996743, | |
| "grad_norm": 2.9277894965124895, | |
| "learning_rate": 1.025493399054383e-06, | |
| "loss": 4.1622, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 0.967643865363735, | |
| "grad_norm": 2.792456827835999, | |
| "learning_rate": 1.0241704825693384e-06, | |
| "loss": 4.1866, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 0.9685124864277959, | |
| "grad_norm": 3.0405461234372178, | |
| "learning_rate": 1.0228827190339365e-06, | |
| "loss": 4.2563, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.9693811074918567, | |
| "grad_norm": 2.456925962421864, | |
| "learning_rate": 1.0216301185622093e-06, | |
| "loss": 4.2406, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 0.9702497285559175, | |
| "grad_norm": 2.556980498653978, | |
| "learning_rate": 1.0204126909920216e-06, | |
| "loss": 4.2689, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 0.9711183496199783, | |
| "grad_norm": 2.494568998681425, | |
| "learning_rate": 1.0192304458849905e-06, | |
| "loss": 4.1788, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 0.971986970684039, | |
| "grad_norm": 2.414367486391019, | |
| "learning_rate": 1.0180833925264123e-06, | |
| "loss": 4.3123, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 0.9728555917480999, | |
| "grad_norm": 2.5637239329618797, | |
| "learning_rate": 1.0169715399251886e-06, | |
| "loss": 4.1474, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.9737242128121607, | |
| "grad_norm": 2.1036887441385, | |
| "learning_rate": 1.0158948968137562e-06, | |
| "loss": 4.1269, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 0.9745928338762215, | |
| "grad_norm": 2.922835583562659, | |
| "learning_rate": 1.0148534716480189e-06, | |
| "loss": 4.3625, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 0.9754614549402822, | |
| "grad_norm": 9.43946034741677, | |
| "learning_rate": 1.013847272607279e-06, | |
| "loss": 4.1683, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 0.9763300760043431, | |
| "grad_norm": 4.165841700664373, | |
| "learning_rate": 1.0128763075941765e-06, | |
| "loss": 4.2542, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 0.9771986970684039, | |
| "grad_norm": 2.839928915059915, | |
| "learning_rate": 1.0119405842346225e-06, | |
| "loss": 4.3801, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.9780673181324647, | |
| "grad_norm": 2.8834509246127547, | |
| "learning_rate": 1.0110401098777443e-06, | |
| "loss": 4.1791, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 0.9789359391965256, | |
| "grad_norm": 3.323021602743116, | |
| "learning_rate": 1.010174891595824e-06, | |
| "loss": 4.2308, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 0.9798045602605863, | |
| "grad_norm": 2.8595797575000574, | |
| "learning_rate": 1.0093449361842436e-06, | |
| "loss": 4.163, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 0.9806731813246471, | |
| "grad_norm": 2.6315144637856416, | |
| "learning_rate": 1.0085502501614336e-06, | |
| "loss": 4.2709, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 0.9815418023887079, | |
| "grad_norm": 3.743912990675417, | |
| "learning_rate": 1.00779083976882e-06, | |
| "loss": 4.378, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.9824104234527687, | |
| "grad_norm": 2.5760124356799627, | |
| "learning_rate": 1.007066710970775e-06, | |
| "loss": 4.3711, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 0.9832790445168296, | |
| "grad_norm": 3.24899714712198, | |
| "learning_rate": 1.0063778694545713e-06, | |
| "loss": 4.413, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 0.9841476655808903, | |
| "grad_norm": 2.651957802148301, | |
| "learning_rate": 1.0057243206303377e-06, | |
| "loss": 4.1993, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 0.9850162866449511, | |
| "grad_norm": 2.2387986170054557, | |
| "learning_rate": 1.0051060696310137e-06, | |
| "loss": 4.0838, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 0.9858849077090119, | |
| "grad_norm": 3.0411821712384324, | |
| "learning_rate": 1.0045231213123148e-06, | |
| "loss": 4.3682, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.9867535287730728, | |
| "grad_norm": 4.325765653367129, | |
| "learning_rate": 1.0039754802526882e-06, | |
| "loss": 4.2435, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 0.9876221498371336, | |
| "grad_norm": 2.974916977057162, | |
| "learning_rate": 1.00346315075328e-06, | |
| "loss": 4.1753, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 0.9884907709011943, | |
| "grad_norm": 2.953803498662762, | |
| "learning_rate": 1.002986136837902e-06, | |
| "loss": 4.1284, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 0.9893593919652551, | |
| "grad_norm": 3.8781438570753104, | |
| "learning_rate": 1.0025444422529981e-06, | |
| "loss": 4.2912, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 0.990228013029316, | |
| "grad_norm": 2.751967293645226, | |
| "learning_rate": 1.0021380704676165e-06, | |
| "loss": 4.2889, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.9910966340933768, | |
| "grad_norm": 3.6019606680877354, | |
| "learning_rate": 1.001767024673382e-06, | |
| "loss": 4.3346, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 0.9919652551574376, | |
| "grad_norm": 4.860956653468777, | |
| "learning_rate": 1.00143130778447e-06, | |
| "loss": 4.4012, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 0.9928338762214983, | |
| "grad_norm": 2.978905973214082, | |
| "learning_rate": 1.0011309224375852e-06, | |
| "loss": 4.2442, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 0.9937024972855591, | |
| "grad_norm": 3.8701647090673466, | |
| "learning_rate": 1.0008658709919392e-06, | |
| "loss": 4.1964, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 0.99457111834962, | |
| "grad_norm": 3.288978549917795, | |
| "learning_rate": 1.0006361555292336e-06, | |
| "loss": 4.3262, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.9954397394136808, | |
| "grad_norm": 2.960844019977527, | |
| "learning_rate": 1.0004417778536422e-06, | |
| "loss": 4.3462, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 0.9963083604777416, | |
| "grad_norm": 3.7853750361182774, | |
| "learning_rate": 1.0002827394917987e-06, | |
| "loss": 4.1935, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 0.9971769815418023, | |
| "grad_norm": 3.018359866548487, | |
| "learning_rate": 1.000159041692782e-06, | |
| "loss": 4.2258, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 0.9980456026058632, | |
| "grad_norm": 3.5491363826838827, | |
| "learning_rate": 1.0000706854281087e-06, | |
| "loss": 4.3101, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 0.998914223669924, | |
| "grad_norm": 2.819477984506924, | |
| "learning_rate": 1.000017671391725e-06, | |
| "loss": 4.2138, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.9997828447339848, | |
| "grad_norm": 4.035987734141358, | |
| "learning_rate": 1e-06, | |
| "loss": 4.2367, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 0.9997828447339848, | |
| "step": 1151, | |
| "total_flos": 995560529133568.0, | |
| "train_loss": 4.727850101182608, | |
| "train_runtime": 34373.7603, | |
| "train_samples_per_second": 2.143, | |
| "train_steps_per_second": 0.033 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 1151, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 995560529133568.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |