diff --git "a/checkpoint-502/trainer_state.json" "b/checkpoint-502/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-502/trainer_state.json"
@@ -0,0 +1,3970 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.21253175275190517,
+  "global_step": 502,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1666666666666665e-05,
+      "loss": 3.0643,
+      "theoretical_loss": 3.321567680436603,
+      "tokens_seen": 2990538752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333333e-05,
+      "loss": 3.0798,
+      "theoretical_loss": 3.3215564803546,
+      "tokens_seen": 2990669824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000125,
+      "loss": 2.8688,
+      "theoretical_loss": 3.321545280900887,
+      "tokens_seen": 2990800896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016666666666666666,
+      "loss": 2.7194,
+      "theoretical_loss": 3.3215340820754022,
+      "tokens_seen": 2990931968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00020833333333333335,
+      "loss": 2.6193,
+      "theoretical_loss": 3.3215228838780817,
+      "tokens_seen": 2991063040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00025,
+      "loss": 2.8571,
+      "theoretical_loss": 3.3215116863088636,
+      "tokens_seen": 2991194112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002916666666666667,
+      "loss": 2.7571,
+      "theoretical_loss": 3.3215004893676854,
+      "tokens_seen": 2991325184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0003333333333333333,
+      "loss": 2.8877,
+      "theoretical_loss": 3.321489293054483,
+      "tokens_seen": 2991456256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000375,
+      "loss": 2.714,
+      "theoretical_loss": 3.321478097369195,
+      "tokens_seen": 2991587328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004166666666666667,
+      "loss": 2.6564,
+      "theoretical_loss": 3.321466902311758,
+      "tokens_seen": 2991718400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004583333333333333,
+      "loss": 2.5638,
+      "theoretical_loss": 3.3214557078821096,
+      "tokens_seen": 2991849472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005,
+      "loss": 2.4354,
+      "theoretical_loss": 3.321444514080187,
+      "tokens_seen": 2991980544
+    },
+    {
+      "epoch": 0.01,
+      "objective/train/docs_used": 1640856,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.5749809741973877,
+      "objective/train/theoretical_loss": 3.321438917414603,
+      "objective/train/tokens_used": 22097376,
+      "theoretical_loss": 3.321438917414603,
+      "tokens_seen": 2992046080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005416666666666666,
+      "loss": 2.5713,
+      "theoretical_loss": 3.321433320905927,
+      "tokens_seen": 2992111616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005833333333333334,
+      "loss": 2.4812,
+      "theoretical_loss": 3.3214221283592678,
+      "tokens_seen": 2992242688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000625,
+      "loss": 2.7622,
+      "theoretical_loss": 3.321410936440146,
+      "tokens_seen": 2992373760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006666666666666666,
+      "loss": 2.609,
+      "theoretical_loss": 3.3213997451485,
+      "tokens_seen": 2992504832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007083333333333334,
+      "loss": 2.4537,
+      "theoretical_loss": 3.3213885544842654,
+      "tokens_seen": 2992635904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00075,
+      "loss": 2.4831,
+      "theoretical_loss": 3.321377364447381,
+      "tokens_seen": 2992766976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007916666666666666,
+      "loss": 2.5607,
+      "theoretical_loss": 3.3213661750377836,
+      "tokens_seen": 2992898048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008333333333333334,
+      "loss": 2.5759,
+      "theoretical_loss": 3.3213549862554106,
+      "tokens_seen": 2993029120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000875,
+      "loss": 2.3106,
+      "theoretical_loss": 3.3213437981001994,
+      "tokens_seen": 2993160192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009166666666666666,
+      "loss": 2.5471,
+      "theoretical_loss": 3.3213326105720875,
+      "tokens_seen": 2993291264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009583333333333334,
+      "loss": 2.602,
+      "theoretical_loss": 3.3213214236710122,
+      "tokens_seen": 2993422336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.001,
+      "loss": 2.5877,
+      "theoretical_loss": 3.321310237396911,
+      "tokens_seen": 2993553408
+    },
+    {
+      "epoch": 0.01,
+      "objective/train/docs_used": 1641461,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.557373523712158,
+      "objective/train/theoretical_loss": 3.3212990517497207,
+      "objective/train/tokens_used": 23735776,
+      "theoretical_loss": 3.3212990517497207,
+      "tokens_seen": 2993684480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009995722840034217,
+      "loss": 2.7042,
+      "theoretical_loss": 3.3212990517497207,
+      "tokens_seen": 2993684480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009991445680068436,
+      "loss": 2.5234,
+      "theoretical_loss": 3.3212878667293797,
+      "tokens_seen": 2993815552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009987168520102653,
+      "loss": 2.5502,
+      "theoretical_loss": 3.321276682335825,
+      "tokens_seen": 2993946624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000998289136013687,
+      "loss": 2.6849,
+      "theoretical_loss": 3.3212654985689936,
+      "tokens_seen": 2994077696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009978614200171086,
+      "loss": 2.6348,
+      "theoretical_loss": 3.3212543154288237,
+      "tokens_seen": 2994208768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009974337040205303,
+      "loss": 2.6793,
+      "theoretical_loss": 3.3212431329152525,
+      "tokens_seen": 2994339840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009970059880239522,
+      "loss": 2.6212,
+      "theoretical_loss": 3.321231951028217,
+      "tokens_seen": 2994470912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009965782720273739,
+      "loss": 2.629,
+      "theoretical_loss": 3.3212207697676552,
+      "tokens_seen": 2994601984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009961505560307955,
+      "loss": 2.5865,
+      "theoretical_loss": 3.3212095891335043,
+      "tokens_seen": 2994733056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009957228400342174,
+      "loss": 2.667,
+      "theoretical_loss": 3.321198409125702,
+      "tokens_seen": 2994864128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000995295124037639,
+      "loss": 2.55,
+      "theoretical_loss": 3.321187229744186,
+      "tokens_seen": 2994995200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009948674080410608,
+      "loss": 2.6324,
+      "theoretical_loss": 3.321176050988893,
+      "tokens_seen": 2995126272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009944396920444824,
+      "loss": 2.8406,
+      "theoretical_loss": 3.3211648728597614,
+      "tokens_seen": 2995257344
+    },
+    {
+      "epoch": 0.02,
+      "objective/train/docs_used": 1642666,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.8021185398101807,
+      "objective/train/theoretical_loss": 3.3211592840299864,
+      "objective/train/tokens_used": 25374176,
+      "theoretical_loss": 3.3211592840299864,
+      "tokens_seen": 2995322880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009940119760479041,
+      "loss": 2.5438,
+      "theoretical_loss": 3.3211536953567284,
+      "tokens_seen": 2995388416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000993584260051326,
+      "loss": 2.7618,
+      "theoretical_loss": 3.321142518479731,
+      "tokens_seen": 2995519488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009931565440547477,
+      "loss": 2.6638,
+      "theoretical_loss": 3.321131342228708,
+      "tokens_seen": 2995650560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009927288280581694,
+      "loss": 2.7287,
+      "theoretical_loss": 3.321120166603596,
+      "tokens_seen": 2995781632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000992301112061591,
+      "loss": 2.5832,
+      "theoretical_loss": 3.3211089916043326,
+      "tokens_seen": 2995912704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009918733960650127,
+      "loss": 2.6355,
+      "theoretical_loss": 3.3210978172308554,
+      "tokens_seen": 2996043776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009914456800684346,
+      "loss": 2.5139,
+      "theoretical_loss": 3.3210866434831026,
+      "tokens_seen": 2996174848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009910179640718563,
+      "loss": 2.5879,
+      "theoretical_loss": 3.3210754703610106,
+      "tokens_seen": 2996305920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009905902480752782,
+      "loss": 2.5935,
+      "theoretical_loss": 3.321064297864518,
+      "tokens_seen": 2996436992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009901625320786998,
+      "loss": 2.6176,
+      "theoretical_loss": 3.3210531259935627,
+      "tokens_seen": 2996568064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009897348160821215,
+      "loss": 2.6405,
+      "theoretical_loss": 3.321041954748081,
+      "tokens_seen": 2996699136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009893071000855432,
+      "loss": 2.7274,
+      "theoretical_loss": 3.321030784128012,
+      "tokens_seen": 2996830208
+    },
+    {
+      "epoch": 0.02,
+      "objective/train/docs_used": 1643300,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 3.023165464401245,
+      "objective/train/theoretical_loss": 3.321019614133292,
+      "objective/train/tokens_used": 27012576,
+      "theoretical_loss": 3.321019614133292,
+      "tokens_seen": 2996961280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009888793840889649,
+      "loss": 2.7003,
+      "theoretical_loss": 3.321019614133292,
+      "tokens_seen": 2996961280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009884516680923865,
+      "loss": 2.7436,
+      "theoretical_loss": 3.3210084447638595,
+      "tokens_seen": 2997092352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009880239520958084,
+      "loss": 2.5873,
+      "theoretical_loss": 3.320997276019652,
+      "tokens_seen": 2997223424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00098759623609923,
+      "loss": 2.5974,
+      "theoretical_loss": 3.3209861079006067,
+      "tokens_seen": 2997354496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000987168520102652,
+      "loss": 2.5806,
+      "theoretical_loss": 3.320974940406662,
+      "tokens_seen": 2997485568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009867408041060737,
+      "loss": 2.6771,
+      "theoretical_loss": 3.320963773537755,
+      "tokens_seen": 2997616640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009863130881094953,
+      "loss": 2.7313,
+      "theoretical_loss": 3.320952607293824,
+      "tokens_seen": 2997747712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000985885372112917,
+      "loss": 2.7302,
+      "theoretical_loss": 3.320941441674806,
+      "tokens_seen": 2997878784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009854576561163387,
+      "loss": 2.893,
+      "theoretical_loss": 3.320930276680639,
+      "tokens_seen": 2998009856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009850299401197606,
+      "loss": 2.6886,
+      "theoretical_loss": 3.3209191123112607,
+      "tokens_seen": 2998140928
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009846022241231823,
+      "loss": 2.7801,
+      "theoretical_loss": 3.320907948566609,
+      "tokens_seen": 2998272000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000984174508126604,
+      "loss": 2.6538,
+      "theoretical_loss": 3.3208967854466214,
+      "tokens_seen": 2998403072
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009837467921300258,
+      "loss": 2.5705,
+      "theoretical_loss": 3.3208856229512356,
+      "tokens_seen": 2998534144
+    },
+    {
+      "epoch": 0.03,
+      "objective/train/docs_used": 1644380,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.542499542236328,
+      "objective/train/theoretical_loss": 3.320880041937749,
+      "objective/train/tokens_used": 28650976,
+      "theoretical_loss": 3.320880041937749,
+      "tokens_seen": 2998599680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009833190761334475,
+      "loss": 2.5074,
+      "theoretical_loss": 3.3208744610803898,
+      "tokens_seen": 2998665216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009828913601368692,
+      "loss": 2.6151,
+      "theoretical_loss": 3.320863299834021,
+      "tokens_seen": 2998796288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009824636441402908,
+      "loss": 2.6994,
+      "theoretical_loss": 3.320852139212068,
+      "tokens_seen": 2998927360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009820359281437125,
+      "loss": 2.568,
+      "theoretical_loss": 3.3208409792144677,
+      "tokens_seen": 2999058432
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009816082121471344,
+      "loss": 2.5552,
+      "theoretical_loss": 3.320829819841158,
+      "tokens_seen": 2999189504
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000981180496150556,
+      "loss": 2.6719,
+      "theoretical_loss": 3.320818661092077,
+      "tokens_seen": 2999320576
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009807527801539778,
+      "loss": 2.5567,
+      "theoretical_loss": 3.3208075029671624,
+      "tokens_seen": 2999451648
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009803250641573994,
+      "loss": 2.5511,
+      "theoretical_loss": 3.320796345466352,
+      "tokens_seen": 2999582720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009798973481608211,
+      "loss": 2.6608,
+      "theoretical_loss": 3.320785188589584,
+      "tokens_seen": 2999713792
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000979469632164243,
+      "loss": 2.5947,
+      "theoretical_loss": 3.3207740323367956,
+      "tokens_seen": 2999844864
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009790419161676647,
+      "loss": 2.6511,
+      "theoretical_loss": 3.3207628767079242,
+      "tokens_seen": 2999975936
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009786142001710863,
+      "loss": 2.5573,
+      "theoretical_loss": 3.3207517217029094,
+      "tokens_seen": 3000107008
+    },
+    {
+      "epoch": 0.03,
+      "objective/train/docs_used": 1645056,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.4763801097869873,
+      "objective/train/theoretical_loss": 3.3207405673216877,
+      "objective/train/tokens_used": 30289376,
+      "theoretical_loss": 3.3207405673216877,
+      "tokens_seen": 3000238080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009781864841745082,
+      "loss": 2.6745,
+      "theoretical_loss": 3.3207405673216877,
+      "tokens_seen": 3000238080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00097775876817793,
+      "loss": 2.697,
+      "theoretical_loss": 3.320729413564197,
+      "tokens_seen": 3000369152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009773310521813516,
+      "loss": 2.6853,
+      "theoretical_loss": 3.3207182604303753,
+      "tokens_seen": 3000500224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009769033361847733,
+      "loss": 2.3445,
+      "theoretical_loss": 3.320707107920161,
+      "tokens_seen": 3000631296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000976475620188195,
+      "loss": 2.6763,
+      "theoretical_loss": 3.3206959560334917,
+      "tokens_seen": 3000762368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009760479041916168,
+      "loss": 2.5198,
+      "theoretical_loss": 3.320684804770305,
+      "tokens_seen": 3000893440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009756201881950385,
+      "loss": 2.7,
+      "theoretical_loss": 3.3206736541305393,
+      "tokens_seen": 3001024512
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009751924721984602,
+      "loss": 2.6958,
+      "theoretical_loss": 3.3206625041141318,
+      "tokens_seen": 3001155584
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000974764756201882,
+      "loss": 2.6457,
+      "theoretical_loss": 3.3206513547210212,
+      "tokens_seen": 3001286656
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009743370402053036,
+      "loss": 2.7946,
+      "theoretical_loss": 3.320640205951145,
+      "tokens_seen": 3001417728
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009739093242087254,
+      "loss": 2.6682,
+      "theoretical_loss": 3.3206290578044415,
+      "tokens_seen": 3001548800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009734816082121472,
+      "loss": 2.5484,
+      "theoretical_loss": 3.3206179102808484,
+      "tokens_seen": 3001679872
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009730538922155689,
+      "loss": 2.6724,
+      "theoretical_loss": 3.3206067633803036,
+      "tokens_seen": 3001810944
+    },
+    {
+      "epoch": 0.04,
+      "objective/train/docs_used": 1646327,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.547891855239868,
+      "objective/train/theoretical_loss": 3.320601190163655,
+      "objective/train/tokens_used": 31927776,
+      "theoretical_loss": 3.320601190163655,
+      "tokens_seen": 3001876480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009726261762189907,
+      "loss": 2.5267,
+      "theoretical_loss": 3.320595617102745,
+      "tokens_seen": 3001942016
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009721984602224123,
+      "loss": 2.6548,
+      "theoretical_loss": 3.320584471448111,
+      "tokens_seen": 3002073088
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009717707442258341,
+      "loss": 2.5147,
+      "theoretical_loss": 3.3205733264163393,
+      "tokens_seen": 3002204160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009713430282292558,
+      "loss": 2.4505,
+      "theoretical_loss": 3.320562182007368,
+      "tokens_seen": 3002335232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009709153122326775,
+      "loss": 2.6305,
+      "theoretical_loss": 3.320551038221135,
+      "tokens_seen": 3002466304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009704875962360993,
+      "loss": 2.5482,
+      "theoretical_loss": 3.3205398950575784,
+      "tokens_seen": 3002597376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009700598802395209,
+      "loss": 2.7266,
+      "theoretical_loss": 3.320528752516636,
+      "tokens_seen": 3002728448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009696321642429427,
+      "loss": 2.5155,
+      "theoretical_loss": 3.3205176105982463,
+      "tokens_seen": 3002859520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009692044482463645,
+      "loss": 2.7628,
+      "theoretical_loss": 3.320506469302347,
+      "tokens_seen": 3002990592
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009687767322497862,
+      "loss": 2.6802,
+      "theoretical_loss": 3.3204953286288763,
+      "tokens_seen": 3003121664
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000968349016253208,
+      "loss": 2.7921,
+      "theoretical_loss": 3.3204841885777725,
+      "tokens_seen": 3003252736
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009679213002566296,
+      "loss": 2.6088,
+      "theoretical_loss": 3.3204730491489727,
+      "tokens_seen": 3003383808
+    },
+    {
+      "epoch": 0.04,
+      "objective/train/docs_used": 1647543,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.5402848720550537,
+      "objective/train/theoretical_loss": 3.3204619103424164,
+      "objective/train/tokens_used": 33566176,
+      "theoretical_loss": 3.3204619103424164,
+      "tokens_seen": 3003514880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009674935842600513,
+      "loss": 2.6446,
+      "theoretical_loss": 3.3204619103424164,
+      "tokens_seen": 3003514880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009670658682634731,
+      "loss": 2.5202,
+      "theoretical_loss": 3.3204507721580403,
+      "tokens_seen": 3003645952
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009666381522668948,
+      "loss": 2.4832,
+      "theoretical_loss": 3.3204396345957834,
+      "tokens_seen": 3003777024
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009662104362703165,
+      "loss": 2.6544,
+      "theoretical_loss": 3.320428497655584,
+      "tokens_seen": 3003908096
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009657827202737382,
+      "loss": 2.7507,
+      "theoretical_loss": 3.320417361337379,
+      "tokens_seen": 3004039168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00096535500427716,
+      "loss": 2.5786,
+      "theoretical_loss": 3.3204062256411078,
+      "tokens_seen": 3004170240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009649272882805818,
+      "loss": 2.6649,
+      "theoretical_loss": 3.320395090566708,
+      "tokens_seen": 3004301312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009644995722840035,
+      "loss": 2.6338,
+      "theoretical_loss": 3.3203839561141173,
+      "tokens_seen": 3004432384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009640718562874252,
+      "loss": 2.664,
+      "theoretical_loss": 3.320372822283275,
+      "tokens_seen": 3004563456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009636441402908469,
+      "loss": 2.6909,
+      "theoretical_loss": 3.3203616890741183,
+      "tokens_seen": 3004694528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009632164242942686,
+      "loss": 2.4478,
+      "theoretical_loss": 3.3203505564865856,
+      "tokens_seen": 3004825600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009627887082976904,
+      "loss": 2.6265,
+      "theoretical_loss": 3.3203394245206153,
+      "tokens_seen": 3004956672
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000962360992301112,
+      "loss": 2.5755,
+      "theoretical_loss": 3.320328293176145,
+      "tokens_seen": 3005087744
+    },
+    {
+      "epoch": 0.05,
+      "objective/train/docs_used": 1648109,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.5592408180236816,
+      "objective/train/theoretical_loss": 3.3203227277369534,
+      "objective/train/tokens_used": 35204576,
+      "theoretical_loss": 3.3203227277369534,
+      "tokens_seen": 3005153280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009619332763045337,
+      "loss": 2.567,
+      "theoretical_loss": 3.320317162453114,
+      "tokens_seen": 3005218816
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009615055603079555,
+      "loss": 2.6035,
+      "theoretical_loss": 3.3203060323514593,
+      "tokens_seen": 3005349888
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009610778443113773,
+      "loss": 2.6209,
+      "theoretical_loss": 3.3202949028711197,
+      "tokens_seen": 3005480960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009606501283147991,
+      "loss": 2.522,
+      "theoretical_loss": 3.3202837740120335,
+      "tokens_seen": 3005612032
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009602224123182207,
+      "loss": 2.5764,
+      "theoretical_loss": 3.3202726457741387,
+      "tokens_seen": 3005743104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009597946963216424,
+      "loss": 2.6805,
+      "theoretical_loss": 3.320261518157374,
+      "tokens_seen": 3005874176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009593669803250642,
+      "loss": 2.4835,
+      "theoretical_loss": 3.3202503911616765,
+      "tokens_seen": 3006005248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009589392643284859,
+      "loss": 2.5907,
+      "theoretical_loss": 3.320239264786986,
+      "tokens_seen": 3006136320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009585115483319077,
+      "loss": 2.617,
+      "theoretical_loss": 3.3202281390332393,
+      "tokens_seen": 3006267392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009580838323353293,
+      "loss": 2.5027,
+      "theoretical_loss": 3.320217013900376,
+      "tokens_seen": 3006398464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000957656116338751,
+      "loss": 2.6857,
+      "theoretical_loss": 3.3202058893883333,
+      "tokens_seen": 3006529536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009572284003421729,
+      "loss": 2.6411,
+      "theoretical_loss": 3.3201947654970505,
+      "tokens_seen": 3006660608
+    },
+    {
+      "epoch": 0.05,
+      "objective/train/docs_used": 1649212,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.7101857662200928,
+      "objective/train/theoretical_loss": 3.320183642226465,
+      "objective/train/tokens_used": 36842976,
+      "theoretical_loss": 3.320183642226465,
+      "tokens_seen": 3006791680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009568006843455946,
+      "loss": 2.5369,
+      "theoretical_loss": 3.320183642226465,
+      "tokens_seen": 3006791680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009563729683490164,
+      "loss": 2.5836,
+      "theoretical_loss": 3.3201725195765155,
+      "tokens_seen": 3006922752
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000955945252352438,
+      "loss": 2.5358,
+      "theoretical_loss": 3.3201613975471402,
+      "tokens_seen": 3007053824
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009555175363558597,
+      "loss": 2.6035,
+      "theoretical_loss": 3.3201502761382775,
+      "tokens_seen": 3007184896
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009550898203592815,
+      "loss": 2.4594,
+      "theoretical_loss": 3.320139155349866,
+      "tokens_seen": 3007315968
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009546621043627032,
+      "loss": 2.6376,
+      "theoretical_loss": 3.3201280351818436,
+      "tokens_seen": 3007447040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009542343883661248,
+      "loss": 2.5904,
+      "theoretical_loss": 3.320116915634149,
+      "tokens_seen": 3007578112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009538066723695466,
+      "loss": 2.7616,
+      "theoretical_loss": 3.3201057967067205,
+      "tokens_seen": 3007709184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009533789563729683,
+      "loss": 2.6076,
+      "theoretical_loss": 3.3200946783994962,
+      "tokens_seen": 3007840256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009529512403763902,
+      "loss": 2.6121,
+      "theoretical_loss": 3.3200835607124146,
+      "tokens_seen": 3007971328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009525235243798119,
+      "loss": 2.6083,
+      "theoretical_loss": 3.3200724436454143,
+      "tokens_seen": 3008102400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009520958083832335,
+      "loss": 2.6831,
+      "theoretical_loss": 3.3200613271984336,
+      "tokens_seen": 3008233472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009516680923866553,
+      "loss": 2.6156,
+      "theoretical_loss": 3.3200502113714108,
+      "tokens_seen": 3008364544
+    },
+    {
+      "epoch": 0.06,
+      "objective/train/docs_used": 1649940,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.545426368713379,
+      "objective/train/theoretical_loss": 3.3200446536903643,
+      "objective/train/tokens_used": 38481376,
+      "theoretical_loss": 3.3200446536903643,
+      "tokens_seen": 3008430080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000951240376390077,
+      "loss": 2.6849,
+      "theoretical_loss": 3.3200390961642845,
+      "tokens_seen": 3008495616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009508126603934988,
+      "loss": 2.654,
+      "theoretical_loss": 3.3200279815769926,
+      "tokens_seen": 3008626688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009503849443969204,
+      "loss": 2.6215,
+      "theoretical_loss": 3.3200168676094743,
+      "tokens_seen": 3008757760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009499572284003421,
+      "loss": 2.4387,
+      "theoretical_loss": 3.320005754261668,
+      "tokens_seen": 3008888832
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009495295124037639,
+      "loss": 2.5621,
+      "theoretical_loss": 3.319994641533511,
+      "tokens_seen": 3009019904
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009491017964071857,
+      "loss": 2.731,
+      "theoretical_loss": 3.319983529424943,
+      "tokens_seen": 3009150976
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009486740804106075,
+      "loss": 2.5847,
+      "theoretical_loss": 3.3199724179359027,
+      "tokens_seen": 3009282048
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009482463644140291,
+      "loss": 2.5836,
+      "theoretical_loss": 3.319961307066327,
+      "tokens_seen": 3009413120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009478186484174508,
+      "loss": 2.6488,
+      "theoretical_loss": 3.3199501968161558,
+      "tokens_seen": 3009544192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009473909324208726,
+      "loss": 2.6244,
+      "theoretical_loss": 3.319939087185327,
+      "tokens_seen": 3009675264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009469632164242943,
+      "loss": 2.6873,
+      "theoretical_loss": 3.3199279781737796,
+      "tokens_seen": 3009806336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000946535500427716,
+      "loss": 2.5921,
+      "theoretical_loss": 3.3199168697814514,
+      "tokens_seen": 3009937408
+    },
+    {
+      "epoch": 0.06,
+      "objective/train/docs_used": 1651249,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.582075834274292,
+      "objective/train/theoretical_loss": 3.3199057620082812,
+      "objective/train/tokens_used": 40119776,
+      "theoretical_loss": 3.3199057620082812,
+      "tokens_seen": 3010068480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009461077844311377,
+      "loss": 2.5527,
+      "theoretical_loss": 3.3199057620082812,
+      "tokens_seen": 3010068480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009456800684345594,
+      "loss": 2.7818,
+      "theoretical_loss": 3.319894654854208,
+      "tokens_seen": 3010199552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009452523524379812,
+      "loss": 2.628,
+      "theoretical_loss": 3.3198835483191695,
+      "tokens_seen": 3010330624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000944824636441403,
+      "loss": 2.4649,
+      "theoretical_loss": 3.319872442403105,
+      "tokens_seen": 3010461696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009443969204448247,
+      "loss": 2.7571,
+      "theoretical_loss": 3.3198613371059524,
+      "tokens_seen": 3010592768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009439692044482464,
+      "loss": 2.5878,
+      "theoretical_loss": 3.319850232427651,
+      "tokens_seen": 3010723840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009435414884516681,
+      "loss": 2.6015,
+      "theoretical_loss": 3.3198391283681383,
+      "tokens_seen": 3010854912
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009431137724550899,
+      "loss": 2.5823,
+      "theoretical_loss": 3.3198280249273546,
+      "tokens_seen": 3010985984
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009426860564585116,
+      "loss": 2.7222,
+      "theoretical_loss": 3.319816922105237,
+      "tokens_seen": 3011117056
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009422583404619332,
+      "loss": 2.6364,
+      "theoretical_loss": 3.319805819901724,
+      "tokens_seen": 3011248128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000941830624465355,
+      "loss": 2.6811,
+      "theoretical_loss": 3.3197947183167553,
+      "tokens_seen": 3011379200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009414029084687767,
+      "loss": 2.8016,
+      "theoretical_loss": 3.319783617350269,
+      "tokens_seen": 3011510272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009409751924721985,
+      "loss": 2.5657,
+      "theoretical_loss": 3.319772517002204,
+      "tokens_seen": 3011641344
+    },
+    {
+      "epoch": 0.07,
+      "objective/train/docs_used": 1651905,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 3.0052103996276855,
+      "objective/train/theoretical_loss": 3.31976696706006,
+      "objective/train/tokens_used": 41758176,
+      "theoretical_loss": 3.31976696706006,
+      "tokens_seen": 3011706880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009405474764756203,
+      "loss": 2.6357,
+      "theoretical_loss": 3.319761417272498,
+      "tokens_seen": 3011772416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009401197604790419,
+      "loss": 2.5734,
+      "theoretical_loss": 3.319750318161091,
+      "tokens_seen": 3011903488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009396920444824637,
+      "loss": 2.424,
+      "theoretical_loss": 3.3197392196679205,
+      "tokens_seen": 3012034560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009392643284858854,
+      "loss": 2.5344,
+      "theoretical_loss": 3.3197281217929255,
+      "tokens_seen": 3012165632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009388366124893071,
+      "loss": 2.5689,
+      "theoretical_loss": 3.319717024536045,
+      "tokens_seen": 3012296704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009384088964927289,
+      "loss": 2.4989,
+      "theoretical_loss": 3.3197059278972176,
+      "tokens_seen": 3012427776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009379811804961505,
+      "loss": 2.6272,
+      "theoretical_loss": 3.3196948318763817,
+      "tokens_seen": 3012558848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009375534644995723,
+      "loss": 2.5959,
+      "theoretical_loss": 3.319683736473476,
+      "tokens_seen": 3012689920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000937125748502994,
+      "loss": 2.5228,
+      "theoretical_loss": 3.3196726416884395,
+      "tokens_seen": 3012820992
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009366980325064158,
+      "loss": 2.7357,
+      "theoretical_loss": 3.3196615475212106,
+      "tokens_seen": 3012952064
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009362703165098376,
+      "loss": 2.6946,
+      "theoretical_loss": 3.3196504539717284,
+      "tokens_seen": 3013083136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009358426005132592,
+      "loss": 2.7344,
+      "theoretical_loss": 3.3196393610399317,
+      "tokens_seen": 3013214208
+    },
+    {
+      "epoch": 0.07,
+      "objective/train/docs_used": 1652881,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.5316507816314697,
+      "objective/train/theoretical_loss": 3.3196282687257583,
+      "objective/train/tokens_used": 43396576,
+      "theoretical_loss": 3.3196282687257583,
+      "tokens_seen": 3013345280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000935414884516681,
+      "loss": 2.7259,
+      "theoretical_loss": 3.3196282687257583,
+      "tokens_seen": 3013345280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009349871685201027,
+      "loss": 2.4882,
+      "theoretical_loss": 3.3196171770291483,
+      "tokens_seen": 3013476352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009345594525235244,
+      "loss": 2.6169,
+      "theoretical_loss": 3.3196060859500394,
+      "tokens_seen": 3013607424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009341317365269461,
+      "loss": 2.5293,
+      "theoretical_loss": 3.319594995488371,
+      "tokens_seen": 3013738496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009337040205303678,
+      "loss": 2.7782,
+      "theoretical_loss": 3.3195839056440812,
+      "tokens_seen": 3013869568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009332763045337895,
+      "loss": 2.6719,
+      "theoretical_loss": 3.3195728164171094,
+      "tokens_seen": 3014000640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009328485885372114,
+      "loss": 2.7268,
+      "theoretical_loss": 3.319561727807394,
+      "tokens_seen": 3014131712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009324208725406331,
+      "loss": 2.5997,
+      "theoretical_loss": 3.3195506398148744,
+      "tokens_seen": 3014262784
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009319931565440548,
+      "loss": 2.7602,
+      "theoretical_loss": 3.319539552439489,
+      "tokens_seen": 3014393856
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009315654405474765,
+      "loss": 2.5845,
+      "theoretical_loss": 3.3195284656811763,
+      "tokens_seen": 3014524928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009311377245508982,
+      "loss": 2.7053,
+      "theoretical_loss": 3.319517379539876,
+      "tokens_seen": 3014656000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00093071000855432,
+      "loss": 2.6045,
+      "theoretical_loss": 3.3195062940155258,
+      "tokens_seen": 3014787072
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009302822925577416,
+      "loss": 2.6324,
+      "theoretical_loss": 3.3194952091080654,
+      "tokens_seen": 3014918144
+    },
+    {
+      "epoch": 0.08,
+      "objective/train/docs_used": 1653310,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.4821465015411377,
+      "objective/train/theoretical_loss": 3.3194896668856497,
+      "objective/train/tokens_used": 45034976,
+      "theoretical_loss": 3.3194896668856497,
+      "tokens_seen": 3014983680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009298545765611634,
+      "loss": 2.532,
+      "theoretical_loss": 3.3194841248174334,
+      "tokens_seen": 3015049216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009294268605645851,
+      "loss": 2.544,
+      "theoretical_loss": 3.3194730411435684,
+      "tokens_seen": 3015180288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009289991445680068,
+      "loss": 2.826,
+      "theoretical_loss": 3.3194619580864098,
+      "tokens_seen": 3015311360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009285714285714287,
+      "loss": 2.7561,
+      "theoretical_loss": 3.3194508756458965,
+      "tokens_seen": 3015442432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009281437125748503,
+      "loss": 2.5692,
+      "theoretical_loss": 3.319439793821967,
+      "tokens_seen": 3015573504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000927715996578272,
+      "loss": 2.6322,
+      "theoretical_loss": 3.3194287126145596,
+      "tokens_seen": 3015704576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009272882805816938,
+      "loss": 2.6346,
+      "theoretical_loss": 3.3194176320236144,
+      "tokens_seen": 3015835648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009268605645851155,
+      "loss": 2.7908,
+      "theoretical_loss": 3.31940655204907,
+      "tokens_seen": 3015966720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009264328485885373,
+      "loss": 2.6439,
+      "theoretical_loss": 3.319395472690865,
+      "tokens_seen": 3016097792
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009260051325919589,
+      "loss": 2.5589,
+      "theoretical_loss": 3.3193843939489382,
+      "tokens_seen": 3016228864
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009255774165953806,
+      "loss": 2.6926,
+      "theoretical_loss": 3.319373315823229,
+      "tokens_seen": 3016359936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009251497005988024,
+      "loss": 2.7646,
+      "theoretical_loss": 3.3193622383136763,
+      "tokens_seen": 3016491008
+    },
+    {
+      "epoch": 0.08,
+      "objective/train/docs_used": 1654644,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.3390886783599854,
+      "objective/train/theoretical_loss": 3.3193511614202187,
+      "objective/train/tokens_used": 46673376,
+      "theoretical_loss": 3.3193511614202187,
+      "tokens_seen": 3016622080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009247219846022242,
+      "loss": 2.5287,
+      "theoretical_loss": 3.3193511614202187,
+      "tokens_seen": 3016622080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000924294268605646,
+      "loss": 2.7607,
+      "theoretical_loss": 3.319340085142796,
+      "tokens_seen": 3016753152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009238665526090676,
+      "loss": 2.6289,
+      "theoretical_loss": 3.319329009481346,
+      "tokens_seen": 3016884224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009234388366124893,
+      "loss": 2.7648,
+      "theoretical_loss": 3.3193179344358086,
+      "tokens_seen": 3017015296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009230111206159111,
+      "loss": 2.497,
+      "theoretical_loss": 3.319306860006122,
+      "tokens_seen": 3017146368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009225834046193328,
+      "loss": 2.4963,
+      "theoretical_loss": 3.319295786192226,
+      "tokens_seen": 3017277440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009221556886227545,
+      "loss": 2.5823,
+      "theoretical_loss": 3.319284712994059,
+      "tokens_seen": 3017408512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009217279726261762,
+      "loss": 2.6555,
+      "theoretical_loss": 3.3192736404115606,
+      "tokens_seen": 3017539584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009213002566295979,
+      "loss": 2.5264,
+      "theoretical_loss": 3.3192625684446693,
+      "tokens_seen": 3017670656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009208725406330197,
+      "loss": 2.397,
+      "theoretical_loss": 3.3192514970933242,
+      "tokens_seen": 3017801728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009204448246364415,
+      "loss": 2.6273,
+      "theoretical_loss": 3.319240426357465,
+      "tokens_seen": 3017932800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009200171086398631,
+      "loss": 2.6414,
+      "theoretical_loss": 3.31922935623703,
+      "tokens_seen": 3018063872
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009195893926432849,
+      "loss": 2.5706,
+      "theoretical_loss": 3.3192182867319584,
+      "tokens_seen": 3018194944
+    },
+    {
+      "epoch": 0.09,
+      "objective/train/docs_used": 1655335,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.3051180839538574,
+      "objective/train/theoretical_loss": 3.319212752210165,
+      "objective/train/tokens_used": 48311776,
+      "theoretical_loss": 3.319212752210165,
+      "tokens_seen": 3018260480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009191616766467066,
+      "loss": 2.5393,
+      "theoretical_loss": 3.3192072178421896,
+      "tokens_seen": 3018326016
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009187339606501284,
+      "loss": 2.6545,
+      "theoretical_loss": 3.319196149567662,
+      "tokens_seen": 3018457088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00091830624465355,
+      "loss": 2.5623,
+      "theoretical_loss": 3.3191850819083157,
+      "tokens_seen": 3018588160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009178785286569717,
+      "loss": 2.6804,
+      "theoretical_loss": 3.319174014864089,
+      "tokens_seen": 3018719232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009174508126603935,
+      "loss": 2.8051,
+      "theoretical_loss": 3.319162948434921,
+      "tokens_seen": 3018850304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009170230966638152,
+      "loss": 2.7044,
+      "theoretical_loss": 3.319151882620752,
+      "tokens_seen": 3018981376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009165953806672371,
+      "loss": 2.5421,
+      "theoretical_loss": 3.3191408174215193,
+      "tokens_seen": 3019112448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009161676646706587,
+      "loss": 2.7474,
+      "theoretical_loss": 3.3191297528371635,
+      "tokens_seen": 3019243520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009157399486740804,
+      "loss": 2.5804,
+      "theoretical_loss": 3.319118688867623,
+      "tokens_seen": 3019374592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009153122326775022,
+      "loss": 2.5145,
+      "theoretical_loss": 3.319107625512837,
+      "tokens_seen": 3019505664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009148845166809239,
+      "loss": 2.746,
+      "theoretical_loss": 3.3190965627727445,
+      "tokens_seen": 3019636736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009144568006843457,
+      "loss": 2.5949,
+      "theoretical_loss": 3.3190855006472857,
+      "tokens_seen": 3019767808
+    },
+    {
+      "epoch": 0.09,
+      "objective/train/docs_used": 1656670,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.1859257221221924,
+      "objective/train/theoretical_loss": 3.3190744391363984,
+      "objective/train/tokens_used": 49950176,
+      "theoretical_loss": 3.3190744391363984,
+      "tokens_seen": 3019898880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009140290846877673,
+      "loss": 2.6334,
+      "theoretical_loss": 3.3190744391363984,
+      "tokens_seen": 3019898880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000913601368691189,
+      "loss": 2.6976,
+      "theoretical_loss": 3.3190633782400223,
+      "tokens_seen": 3020029952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009131736526946108,
+      "loss": 2.5658,
+      "theoretical_loss": 3.3190523179580973,
+      "tokens_seen": 3020161024
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009127459366980325,
+      "loss": 2.7495,
+      "theoretical_loss": 3.3190412582905617,
+      "tokens_seen": 3020292096
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009123182207014543,
+      "loss": 2.7093,
+      "theoretical_loss": 3.319030199237355,
+      "tokens_seen": 3020423168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000911890504704876,
+      "loss": 2.5295,
+      "theoretical_loss": 3.3190191407984164,
+      "tokens_seen": 3020554240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009114627887082977,
+      "loss": 2.6081,
+      "theoretical_loss": 3.3190080829736854,
+      "tokens_seen": 3020685312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009110350727117195,
+      "loss": 2.6166,
+      "theoretical_loss": 3.318997025763101,
+      "tokens_seen": 3020816384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009106073567151412,
+      "loss": 2.7691,
+      "theoretical_loss": 3.318985969166602,
+      "tokens_seen": 3020947456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009101796407185628,
+      "loss": 2.6465,
+      "theoretical_loss": 3.3189749131841286,
+      "tokens_seen": 3021078528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009097519247219846,
+      "loss": 2.5314,
+      "theoretical_loss": 3.3189638578156195,
+      "tokens_seen": 3021209600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009093242087254063,
+      "loss": 2.8726,
+      "theoretical_loss": 3.3189528030610136,
+      "tokens_seen": 3021340672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009088964927288281,
+      "loss": 2.6762,
+      "theoretical_loss": 3.318941748920251,
+      "tokens_seen": 3021471744
+    },
+    {
+      "epoch": 0.1,
+      "objective/train/docs_used": 1657192,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.9018635749816895,
+      "objective/train/theoretical_loss": 3.318936222080042,
+      "objective/train/tokens_used": 51588576,
+      "theoretical_loss": 3.318936222080042,
+      "tokens_seen": 3021537280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009084687767322499,
+      "loss": 2.643,
+      "theoretical_loss": 3.318930695393271,
+      "tokens_seen": 3021602816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009080410607356715,
+      "loss": 2.6665,
+      "theoretical_loss": 3.3189196424800116,
+      "tokens_seen": 3021733888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009076133447390933,
+      "loss": 2.6265,
+      "theoretical_loss": 3.3189085901804134,
+      "tokens_seen": 3021864960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000907185628742515,
+      "loss": 2.5788,
+      "theoretical_loss": 3.3188975384944155,
+      "tokens_seen": 3021996032
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009067579127459367,
+      "loss": 2.6531,
+      "theoretical_loss": 3.318886487421957,
+      "tokens_seen": 3022127104
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009063301967493585,
+      "loss": 2.7177,
+      "theoretical_loss": 3.318875436962977,
+      "tokens_seen": 3022258176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009059024807527801,
+      "loss": 2.4989,
+      "theoretical_loss": 3.3188643871174155,
+      "tokens_seen": 3022389248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009054747647562019,
+      "loss": 2.5249,
+      "theoretical_loss": 3.318853337885211,
+      "tokens_seen": 3022520320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009050470487596236,
+      "loss": 2.6031,
+      "theoretical_loss": 3.318842289266304,
+      "tokens_seen": 3022651392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009046193327630453,
+      "loss": 2.5702,
+      "theoretical_loss": 3.3188312412606327,
+      "tokens_seen": 3022782464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009041916167664672,
+      "loss": 2.5348,
+      "theoretical_loss": 3.3188201938681368,
+      "tokens_seen": 3022913536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009037639007698888,
+      "loss": 2.5876,
+      "theoretical_loss": 3.318809147088756,
+      "tokens_seen": 3023044608
+    },
+    {
+      "epoch": 0.11,
+      "objective/train/docs_used": 1658380,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.8608322143554688,
+      "objective/train/theoretical_loss": 3.3187981009224297,
+      "objective/train/tokens_used": 53226976,
+      "theoretical_loss": 3.3187981009224297,
+      "tokens_seen": 3023175680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009033361847733106,
+      "loss": 2.6644,
+      "theoretical_loss": 3.3187981009224297,
+      "tokens_seen": 3023175680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009029084687767323,
+      "loss": 2.5824,
+      "theoretical_loss": 3.3187870553690972,
+      "tokens_seen": 3023306752
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000902480752780154,
+      "loss": 2.7377,
+      "theoretical_loss": 3.3187760104286976,
+      "tokens_seen": 3023437824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009020530367835757,
+      "loss": 2.6572,
+      "theoretical_loss": 3.3187649661011704,
+      "tokens_seen": 3023568896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009016253207869974,
+      "loss": 2.5447,
+      "theoretical_loss": 3.3187539223864557,
+      "tokens_seen": 3023699968
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009011976047904192,
+      "loss": 2.6011,
+      "theoretical_loss": 3.318742879284492,
+      "tokens_seen": 3023831040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009007698887938409,
+      "loss": 2.4863,
+      "theoretical_loss": 3.3187318367952194,
+      "tokens_seen": 3023962112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009003421727972627,
+      "loss": 2.4994,
+      "theoretical_loss": 3.318720794918577,
+      "tokens_seen": 3024093184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0008999144568006844,
+      "loss": 2.5443,
+      "theoretical_loss": 3.3187097536545047,
+      "tokens_seen": 3024224256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0008994867408041061,
+      "loss": 2.6808,
+      "theoretical_loss": 3.3186987130029415,
+      "tokens_seen": 3024355328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0008990590248075278,
+      "loss": 2.7511,
+      "theoretical_loss": 3.3186876729638266,
+      "tokens_seen": 3024486400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0008986313088109496,
+      "loss": 2.612,
+      "theoretical_loss": 3.3186766335371005,
+      "tokens_seen": 3024617472
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0008982035928143712,
+      "loss": 2.686,
+      "theoretical_loss": 3.318665594722702,
+      "tokens_seen": 3024748544
+    },
+    {
+      "epoch": 0.11,
+      "objective/train/docs_used": 1658975,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.8614556789398193,
+      "objective/train/theoretical_loss": 3.3186600755451066,
+      "objective/train/tokens_used": 54865376,
+      "theoretical_loss": 3.3186600755451066,
+      "tokens_seen": 3024814080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000897775876817793,
+      "loss": 2.654,
+      "theoretical_loss": 3.3186545565205705,
+      "tokens_seen": 3024879616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0008973481608212147,
+      "loss": 2.6409,
+      "theoretical_loss": 3.318643518930646,
+      "tokens_seen": 3025010688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0008969204448246364,
+      "loss": 2.5854,
+      "theoretical_loss": 3.3186324819528674,
+      "tokens_seen": 3025141760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0008964927288280582,
+      "loss": 2.6063,
+      "theoretical_loss": 3.318621445587175,
+      "tokens_seen": 3025272832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0008960650128314799,
+      "loss": 2.537,
+      "theoretical_loss": 3.3186104098335076,
+      "tokens_seen": 3025403904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0008956372968349017,
+      "loss": 2.5978,
+      "theoretical_loss": 3.318599374691805,
+      "tokens_seen": 3025534976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0008952095808383234,
+      "loss": 2.6201,
+      "theoretical_loss": 3.318588340162007,
+      "tokens_seen": 3025666048
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0008947818648417451,
+      "loss": 2.7451,
+      "theoretical_loss": 3.3185773062440527,
+      "tokens_seen": 3025797120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0008943541488451669,
+      "loss": 2.5093,
+      "theoretical_loss": 3.3185662729378826,
+      "tokens_seen": 3025928192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008939264328485885,
+      "loss": 2.517,
+      "theoretical_loss": 3.318555240243435,
+      "tokens_seen": 3026059264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008934987168520102,
+      "loss": 2.515,
+      "theoretical_loss": 3.3185442081606507,
+      "tokens_seen": 3026190336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000893071000855432,
+      "loss": 2.6143,
+      "theoretical_loss": 3.3185331766894683,
+      "tokens_seen": 3026321408
+    },
+    {
+      "epoch": 0.12,
+      "objective/train/docs_used": 1660130,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.4469985961914062,
+      "objective/train/theoretical_loss": 3.318522145829828,
+      "objective/train/tokens_used": 56503776,
+      "theoretical_loss": 3.318522145829828,
+      "tokens_seen": 3026452480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008926432848588537,
+      "loss": 2.4787,
+      "theoretical_loss": 3.318522145829828,
+      "tokens_seen": 3026452480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008922155688622756,
+      "loss": 2.6368,
+      "theoretical_loss": 3.318511115581669,
+      "tokens_seen": 3026583552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008917878528656972,
+      "loss": 2.6092,
+      "theoretical_loss": 3.3185000859449314,
+      "tokens_seen": 3026714624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008913601368691189,
+      "loss": 2.5731,
+      "theoretical_loss": 3.3184890569195544,
+      "tokens_seen": 3026845696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008909324208725407,
+      "loss": 2.5214,
+      "theoretical_loss": 3.3184780285054782,
+      "tokens_seen": 3026976768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008905047048759624,
+      "loss": 2.5803,
+      "theoretical_loss": 3.318467000702642,
+      "tokens_seen": 3027107840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008900769888793841,
+      "loss": 2.6223,
+      "theoretical_loss": 3.3184559735109853,
+      "tokens_seen": 3027238912
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008896492728828058,
+      "loss": 2.5938,
+      "theoretical_loss": 3.3184449469304482,
+      "tokens_seen": 3027369984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008892215568862275,
+      "loss": 2.6878,
+      "theoretical_loss": 3.3184339209609703,
+      "tokens_seen": 3027501056
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008887938408896493,
+      "loss": 2.4932,
+      "theoretical_loss": 3.318422895602491,
+      "tokens_seen": 3027632128
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000888366124893071,
+      "loss": 2.4597,
+      "theoretical_loss": 3.31841187085495,
+      "tokens_seen": 3027763200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008879384088964928,
+      "loss": 2.4485,
+      "theoretical_loss": 3.318400846718288,
+      "tokens_seen": 3027894272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008875106928999145,
+      "loss": 2.6036,
+      "theoretical_loss": 3.318389823192443,
+      "tokens_seen": 3028025344
+    },
+    {
+      "epoch": 0.12,
+      "objective/train/docs_used": 1661412,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.3641762733459473,
+      "objective/train/theoretical_loss": 3.3183843116585585,
+      "objective/train/tokens_used": 58142176,
+      "theoretical_loss": 3.3183843116585585,
+      "tokens_seen": 3028090880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008870829769033362,
+      "loss": 2.6277,
+      "theoretical_loss": 3.318378800277356,
+      "tokens_seen": 3028156416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000886655260906758,
+      "loss": 2.5149,
+      "theoretical_loss": 3.3183677779729663,
+      "tokens_seen": 3028287488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008862275449101797,
+      "loss": 2.6112,
+      "theoretical_loss": 3.3183567562792136,
+      "tokens_seen": 3028418560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008857998289136013,
+      "loss": 2.6564,
+      "theoretical_loss": 3.3183457351960377,
+      "tokens_seen": 3028549632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008853721129170231,
+      "loss": 2.6973,
+      "theoretical_loss": 3.3183347147233784,
+      "tokens_seen": 3028680704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008849443969204448,
+      "loss": 2.4547,
+      "theoretical_loss": 3.3183236948611756,
+      "tokens_seen": 3028811776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008845166809238666,
+      "loss": 2.3115,
+      "theoretical_loss": 3.3183126756093686,
+      "tokens_seen": 3028942848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008840889649272883,
+      "loss": 2.5859,
+      "theoretical_loss": 3.318301656967898,
+      "tokens_seen": 3029073920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00088366124893071,
+      "loss": 2.4681,
+      "theoretical_loss": 3.3182906389367024,
+      "tokens_seen": 3029204992
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008832335329341318,
+      "loss": 2.4459,
+      "theoretical_loss": 3.3182796215157224,
+      "tokens_seen": 3029336064
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008828058169375535,
+      "loss": 2.5877,
+      "theoretical_loss": 3.318268604704898,
+      "tokens_seen": 3029467136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008823781009409753,
+      "loss": 2.5306,
+      "theoretical_loss": 3.318257588504168,
+      "tokens_seen": 3029598208
+    },
+    {
+      "epoch": 0.13,
+      "objective/train/docs_used": 1662079,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.3965446949005127,
+      "objective/train/theoretical_loss": 3.318246572913474,
+      "objective/train/tokens_used": 59780576,
+      "theoretical_loss": 3.318246572913474,
+      "tokens_seen": 3029729280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008819503849443969,
+      "loss": 2.6399,
+      "theoretical_loss": 3.318246572913474,
+      "tokens_seen": 3029729280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008815226689478186,
+      "loss": 2.6156,
+      "theoretical_loss": 3.318235557932754,
+      "tokens_seen": 3029860352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008810949529512404,
+      "loss": 2.4887,
+      "theoretical_loss": 3.318224543561948,
+      "tokens_seen": 3029991424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008806672369546621,
+      "loss": 2.7212,
+      "theoretical_loss": 3.3182135298009974,
+      "tokens_seen": 3030122496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008802395209580839,
+      "loss": 2.6808,
+      "theoretical_loss": 3.3182025166498406,
+      "tokens_seen": 3030253568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008798118049615056,
+      "loss": 2.4704,
+      "theoretical_loss": 3.3181915041084182,
+      "tokens_seen": 3030384640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008793840889649273,
+      "loss": 2.4549,
+      "theoretical_loss": 3.3181804921766695,
+      "tokens_seen": 3030515712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008789563729683491,
+      "loss": 2.8214,
+      "theoretical_loss": 3.318169480854535,
+      "tokens_seen": 3030646784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008785286569717708,
+      "loss": 2.5934,
+      "theoretical_loss": 3.318158470141954,
+      "tokens_seen": 3030777856
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008781009409751924,
+      "loss": 2.6187,
+      "theoretical_loss": 3.3181474600388667,
+      "tokens_seen": 3030908928
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008776732249786142,
+      "loss": 2.3858,
+      "theoretical_loss": 3.318136450545213,
+      "tokens_seen": 3031040000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008772455089820359,
+      "loss": 2.5896,
+      "theoretical_loss": 3.318125441660933,
+      "tokens_seen": 3031171072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008768177929854577,
+      "loss": 2.388,
+      "theoretical_loss": 3.318114433385966,
+      "tokens_seen": 3031302144
+    },
+    {
+      "epoch": 0.13,
+      "objective/train/docs_used": 1662642,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.5412111282348633,
+      "objective/train/theoretical_loss": 3.3181089294769563,
+      "objective/train/tokens_used": 61418976,
+      "theoretical_loss": 3.3181089294769563,
+      "tokens_seen": 3031367680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008763900769888794,
+      "loss": 2.6377,
+      "theoretical_loss": 3.3181034257202526,
+      "tokens_seen": 3031433216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008759623609923011,
+      "loss": 2.55,
+      "theoretical_loss": 3.318092418663732,
+      "tokens_seen": 3031564288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008755346449957229,
+      "loss": 2.6605,
+      "theoretical_loss": 3.3180814122163453,
+      "tokens_seen": 3031695360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008751069289991446,
+      "loss": 2.6276,
+      "theoretical_loss": 3.3180704063780313,
+      "tokens_seen": 3031826432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008746792130025664,
+      "loss": 2.6145,
+      "theoretical_loss": 3.318059401148731,
+      "tokens_seen": 3031957504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008742514970059881,
+      "loss": 2.7501,
+      "theoretical_loss": 3.3180483965283836,
+      "tokens_seen": 3032088576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008738237810094097,
+      "loss": 2.4986,
+      "theoretical_loss": 3.318037392516929,
+      "tokens_seen": 3032219648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008733960650128315,
+      "loss": 2.5994,
+      "theoretical_loss": 3.318026389114308,
+      "tokens_seen": 3032350720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008729683490162532,
+      "loss": 2.4541,
+      "theoretical_loss": 3.3180153863204596,
+      "tokens_seen": 3032481792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008725406330196749,
+      "loss": 2.6821,
+      "theoretical_loss": 3.3180043841353246,
+      "tokens_seen": 3032612864
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008721129170230966,
+      "loss": 2.5988,
+      "theoretical_loss": 3.317993382558843,
+      "tokens_seen": 3032743936
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008716852010265184,
+      "loss": 2.6567,
+      "theoretical_loss": 3.317982381590954,
+      "tokens_seen": 3032875008
+    },
+    {
+      "epoch": 0.14,
+      "objective/train/docs_used": 1663221,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.975888252258301,
+      "objective/train/theoretical_loss": 3.3179713812315983,
+      "objective/train/tokens_used": 63057376,
+      "theoretical_loss": 3.3179713812315983,
+      "tokens_seen": 3033006080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008712574850299402,
+      "loss": 2.6197,
+      "theoretical_loss": 3.3179713812315983,
+      "tokens_seen": 3033006080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008708297690333619,
+      "loss": 2.7088,
+      "theoretical_loss": 3.317960381480716,
+      "tokens_seen": 3033137152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008704020530367836,
+      "loss": 2.582,
+      "theoretical_loss": 3.317949382338247,
+      "tokens_seen": 3033268224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008699743370402053,
+      "loss": 2.5966,
+      "theoretical_loss": 3.3179383838041314,
+      "tokens_seen": 3033399296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000869546621043627,
+      "loss": 2.8364,
+      "theoretical_loss": 3.317927385878309,
+      "tokens_seen": 3033530368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008691189050470488,
+      "loss": 2.6666,
+      "theoretical_loss": 3.31791638856072,
+      "tokens_seen": 3033661440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008686911890504705,
+      "loss": 2.6976,
+      "theoretical_loss": 3.317905391851305,
+      "tokens_seen": 3033792512
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008682634730538922,
+      "loss": 2.662,
+      "theoretical_loss": 3.3178943957500033,
+      "tokens_seen": 3033923584
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000867835757057314,
+      "loss": 2.7223,
+      "theoretical_loss": 3.317883400256756,
+      "tokens_seen": 3034054656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008674080410607357,
+      "loss": 2.754,
+      "theoretical_loss": 3.3178724053715016,
+      "tokens_seen": 3034185728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008669803250641575,
+      "loss": 2.6831,
+      "theoretical_loss": 3.3178614110941815,
+      "tokens_seen": 3034316800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008665526090675792,
+      "loss": 2.8167,
+      "theoretical_loss": 3.3178504174247356,
+      "tokens_seen": 3034447872
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008661248930710008,
+      "loss": 2.761,
+      "theoretical_loss": 3.317839424363104,
+      "tokens_seen": 3034578944
+    },
+    {
+      "epoch": 0.14,
+      "objective/train/docs_used": 1664363,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.568387746810913,
+      "objective/train/theoretical_loss": 3.3178339280602,
+      "objective/train/tokens_used": 64695776,
+      "theoretical_loss": 3.3178339280602,
+      "tokens_seen": 3034644480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008656971770744226,
+      "loss": 2.5995,
+      "theoretical_loss": 3.317828431909227,
+      "tokens_seen": 3034710016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008652694610778443,
+      "loss": 2.5506,
+      "theoretical_loss": 3.3178174400630445,
+      "tokens_seen": 3034841088
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000864841745081266,
+      "loss": 2.8138,
+      "theoretical_loss": 3.3178064488244967,
+      "tokens_seen": 3034972160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008644140290846878,
+      "loss": 2.5913,
+      "theoretical_loss": 3.3177954581935234,
+      "tokens_seen": 3035103232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008639863130881094,
+      "loss": 2.5688,
+      "theoretical_loss": 3.317784468170066,
+      "tokens_seen": 3035234304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008635585970915313,
+      "loss": 2.5422,
+      "theoretical_loss": 3.317773478754063,
+      "tokens_seen": 3035365376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000863130881094953,
+      "loss": 2.6193,
+      "theoretical_loss": 3.317762489945456,
+      "tokens_seen": 3035496448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008627031650983747,
+      "loss": 2.5738,
+      "theoretical_loss": 3.3177515017441843,
+      "tokens_seen": 3035627520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008622754491017965,
+      "loss": 2.715,
+      "theoretical_loss": 3.3177405141501883,
+      "tokens_seen": 3035758592
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008618477331052181,
+      "loss": 2.643,
+      "theoretical_loss": 3.317729527163409,
+      "tokens_seen": 3035889664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008614200171086399,
+      "loss": 2.987,
+      "theoretical_loss": 3.3177185407837855,
+      "tokens_seen": 3036020736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008609923011120616,
+      "loss": 2.594,
+      "theoretical_loss": 3.3177075550112587,
+      "tokens_seen": 3036151808
+    },
+    {
+      "epoch": 0.15,
+      "objective/train/docs_used": 1664891,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.584660291671753,
+      "objective/train/theoretical_loss": 3.3176965698457686,
+      "objective/train/tokens_used": 66334176,
+      "theoretical_loss": 3.3176965698457686,
+      "tokens_seen": 3036282880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008605645851154833,
+      "loss": 2.6228,
+      "theoretical_loss": 3.3176965698457686,
+      "tokens_seen": 3036282880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000860136869118905,
+      "loss": 2.6502,
+      "theoretical_loss": 3.3176855852872555,
+      "tokens_seen": 3036413952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008597091531223268,
+      "loss": 2.6091,
+      "theoretical_loss": 3.3176746013356597,
+      "tokens_seen": 3036545024
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008592814371257485,
+      "loss": 2.5742,
+      "theoretical_loss": 3.317663617990922,
+      "tokens_seen": 3036676096
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008588537211291703,
+      "loss": 2.573,
+      "theoretical_loss": 3.3176526352529816,
+      "tokens_seen": 3036807168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000858426005132592,
+      "loss": 2.6126,
+      "theoretical_loss": 3.31764165312178,
+      "tokens_seen": 3036938240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008579982891360137,
+      "loss": 2.6855,
+      "theoretical_loss": 3.3176306715972563,
+      "tokens_seen": 3037069312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008575705731394354,
+      "loss": 2.7764,
+      "theoretical_loss": 3.3176196906793516,
+      "tokens_seen": 3037200384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008571428571428571,
+      "loss": 2.625,
+      "theoretical_loss": 3.3176087103680056,
+      "tokens_seen": 3037331456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008567151411462789,
+      "loss": 2.6693,
+      "theoretical_loss": 3.3175977306631594,
+      "tokens_seen": 3037462528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008562874251497006,
+      "loss": 2.6201,
+      "theoretical_loss": 3.3175867515647526,
+      "tokens_seen": 3037593600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008558597091531223,
+      "loss": 2.5103,
+      "theoretical_loss": 3.317575773072726,
+      "tokens_seen": 3037724672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008554319931565441,
+      "loss": 2.4253,
+      "theoretical_loss": 3.3175647951870197,
+      "tokens_seen": 3037855744
+    },
+    {
+      "epoch": 0.15,
+      "objective/train/docs_used": 1665791,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.2302393913269043,
+      "objective/train/theoretical_loss": 3.317559306471518,
+      "objective/train/tokens_used": 67972576,
+      "theoretical_loss": 3.317559306471518,
+      "tokens_seen": 3037921280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008550042771599658,
+      "loss": 2.6287,
+      "theoretical_loss": 3.3175538179075743,
+      "tokens_seen": 3037986816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008545765611633876,
+      "loss": 2.5721,
+      "theoretical_loss": 3.31754284123433,
+      "tokens_seen": 3038117888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008541488451668093,
+      "loss": 2.5295,
+      "theoretical_loss": 3.3175318651672274,
+      "tokens_seen": 3038248960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000853721129170231,
+      "loss": 2.6531,
+      "theoretical_loss": 3.3175208897062065,
+      "tokens_seen": 3038380032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008532934131736527,
+      "loss": 2.7652,
+      "theoretical_loss": 3.317509914851208,
+      "tokens_seen": 3038511104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008528656971770744,
+      "loss": 2.4813,
+      "theoretical_loss": 3.3174989406021718,
+      "tokens_seen": 3038642176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008524379811804962,
+      "loss": 2.7802,
+      "theoretical_loss": 3.317487966959039,
+      "tokens_seen": 3038773248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008520102651839178,
+      "loss": 2.6504,
+      "theoretical_loss": 3.3174769939217494,
+      "tokens_seen": 3038904320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008515825491873395,
+      "loss": 2.6791,
+      "theoretical_loss": 3.317466021490244,
+      "tokens_seen": 3039035392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008511548331907614,
+      "loss": 2.8812,
+      "theoretical_loss": 3.3174550496644626,
+      "tokens_seen": 3039166464
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008507271171941831,
+      "loss": 2.6779,
+      "theoretical_loss": 3.317444078444346,
+      "tokens_seen": 3039297536
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008502994011976049,
+      "loss": 2.6431,
+      "theoretical_loss": 3.317433107829835,
+      "tokens_seen": 3039428608
+    },
+    {
+      "epoch": 0.16,
+      "objective/train/docs_used": 1666368,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.657705307006836,
+      "objective/train/theoretical_loss": 3.31742213782087,
+      "objective/train/tokens_used": 69610976,
+      "theoretical_loss": 3.31742213782087,
+      "tokens_seen": 3039559680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008498716852010265,
+      "loss": 2.72,
+      "theoretical_loss": 3.31742213782087,
+      "tokens_seen": 3039559680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008494439692044482,
+      "loss": 2.589,
+      "theoretical_loss": 3.3174111684173906,
+      "tokens_seen": 3039690752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00084901625320787,
+      "loss": 2.7118,
+      "theoretical_loss": 3.317400199619338,
+      "tokens_seen": 3039821824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008485885372112917,
+      "loss": 2.6093,
+      "theoretical_loss": 3.3173892314266524,
+      "tokens_seen": 3039952896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008481608212147135,
+      "loss": 2.724,
+      "theoretical_loss": 3.3173782638392746,
+      "tokens_seen": 3040083968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008477331052181351,
+      "loss": 2.6337,
+      "theoretical_loss": 3.3173672968571446,
+      "tokens_seen": 3040215040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008473053892215569,
+      "loss": 2.7281,
+      "theoretical_loss": 3.3173563304802034,
+      "tokens_seen": 3040346112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008468776732249787,
+      "loss": 2.6731,
+      "theoretical_loss": 3.3173453647083915,
+      "tokens_seen": 3040477184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008464499572284004,
+      "loss": 2.6926,
+      "theoretical_loss": 3.317334399541649,
+      "tokens_seen": 3040608256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008460222412318222,
+      "loss": 2.8351,
+      "theoretical_loss": 3.3173234349799166,
+      "tokens_seen": 3040739328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008455945252352438,
+      "loss": 2.6007,
+      "theoretical_loss": 3.317312471023135,
+      "tokens_seen": 3040870400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008451668092386655,
+      "loss": 2.7641,
+      "theoretical_loss": 3.3173015076712447,
+      "tokens_seen": 3041001472
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008447390932420873,
+      "loss": 2.7725,
+      "theoretical_loss": 3.3172905449241865,
+      "tokens_seen": 3041132544
+    },
+    {
+      "epoch": 0.16,
+      "objective/train/docs_used": 1667402,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.6917831897735596,
+      "objective/train/theoretical_loss": 3.317285063777451,
+      "objective/train/tokens_used": 71249376,
+      "theoretical_loss": 3.317285063777451,
+      "tokens_seen": 3041198080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000844311377245509,
+      "loss": 2.6852,
+      "theoretical_loss": 3.3172795827819,
+      "tokens_seen": 3041263616
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008438836612489306,
+      "loss": 2.6591,
+      "theoretical_loss": 3.3172686212443274,
+      "tokens_seen": 3041394688
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008434559452523525,
+      "loss": 2.74,
+      "theoretical_loss": 3.317257660311408,
+      "tokens_seen": 3041525760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008430282292557742,
+      "loss": 2.83,
+      "theoretical_loss": 3.3172466999830825,
+      "tokens_seen": 3041656832
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000842600513259196,
+      "loss": 2.6488,
+      "theoretical_loss": 3.317235740259292,
+      "tokens_seen": 3041787904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008421727972626177,
+      "loss": 2.5988,
+      "theoretical_loss": 3.3172247811399767,
+      "tokens_seen": 3041918976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008417450812660393,
+      "loss": 2.6984,
+      "theoretical_loss": 3.317213822625077,
+      "tokens_seen": 3042050048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008413173652694611,
+      "loss": 2.6926,
+      "theoretical_loss": 3.3172028647145346,
+      "tokens_seen": 3042181120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008408896492728828,
+      "loss": 2.6579,
+      "theoretical_loss": 3.317191907408289,
+      "tokens_seen": 3042312192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008404619332763046,
+      "loss": 2.6408,
+      "theoretical_loss": 3.3171809507062817,
+      "tokens_seen": 3042443264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008400342172797262,
+      "loss": 2.7066,
+      "theoretical_loss": 3.3171699946084523,
+      "tokens_seen": 3042574336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008396065012831479,
+      "loss": 2.7149,
+      "theoretical_loss": 3.3171590391147427,
+      "tokens_seen": 3042705408
+    },
+    {
+      "epoch": 0.17,
+      "objective/train/docs_used": 1668521,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.528013229370117,
+      "objective/train/theoretical_loss": 3.3171480842250927,
+      "objective/train/tokens_used": 72887776,
+      "theoretical_loss": 3.3171480842250927,
+      "tokens_seen": 3042836480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008391787852865698,
+      "loss": 2.6473,
+      "theoretical_loss": 3.3171480842250927,
+      "tokens_seen": 3042836480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008387510692899915,
+      "loss": 2.6857,
+      "theoretical_loss": 3.317137129939443,
+      "tokens_seen": 3042967552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008383233532934132,
+      "loss": 2.585,
+      "theoretical_loss": 3.3171261762577346,
+      "tokens_seen": 3043098624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000837895637296835,
+      "loss": 2.6799,
+      "theoretical_loss": 3.3171152231799086,
+      "tokens_seen": 3043229696
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008374679213002566,
+      "loss": 2.6625,
+      "theoretical_loss": 3.317104270705905,
+      "tokens_seen": 3043360768
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008370402053036784,
+      "loss": 2.6606,
+      "theoretical_loss": 3.3170933188356644,
+      "tokens_seen": 3043491840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008366124893071001,
+      "loss": 2.7666,
+      "theoretical_loss": 3.3170823675691277,
+      "tokens_seen": 3043622912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008361847733105218,
+      "loss": 2.7028,
+      "theoretical_loss": 3.317071416906236,
+      "tokens_seen": 3043753984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008357570573139435,
+      "loss": 2.664,
+      "theoretical_loss": 3.3170604668469297,
+      "tokens_seen": 3043885056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008353293413173653,
+      "loss": 2.7291,
+      "theoretical_loss": 3.31704951739115,
+      "tokens_seen": 3044016128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008349016253207871,
+      "loss": 2.6537,
+      "theoretical_loss": 3.317038568538837,
+      "tokens_seen": 3044147200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008344739093242088,
+      "loss": 2.5645,
+      "theoretical_loss": 3.317027620289932,
+      "tokens_seen": 3044278272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008340461933276305,
+      "loss": 2.7629,
+      "theoretical_loss": 3.317016672644375,
+      "tokens_seen": 3044409344
+    },
+    {
+      "epoch": 0.17,
+      "objective/train/docs_used": 1668980,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.177739143371582,
+      "objective/train/theoretical_loss": 3.3170111990478337,
+      "objective/train/tokens_used": 74526176,
+      "theoretical_loss": 3.3170111990478337,
+      "tokens_seen": 3044474880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0008336184773310522,
+      "loss": 2.5701,
+      "theoretical_loss": 3.3170057256021077,
+      "tokens_seen": 3044540416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008331907613344739,
+      "loss": 2.6897,
+      "theoretical_loss": 3.3169947791630703,
+      "tokens_seen": 3044671488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008327630453378957,
+      "loss": 2.7863,
+      "theoretical_loss": 3.3169838333272037,
+      "tokens_seen": 3044802560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008323353293413174,
+      "loss": 2.6692,
+      "theoretical_loss": 3.316972888094449,
+      "tokens_seen": 3044933632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000831907613344739,
+      "loss": 2.6618,
+      "theoretical_loss": 3.3169619434647464,
+      "tokens_seen": 3045064704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008314798973481608,
+      "loss": 2.5543,
+      "theoretical_loss": 3.3169509994380375,
+      "tokens_seen": 3045195776
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008310521813515826,
+      "loss": 2.7893,
+      "theoretical_loss": 3.3169400560142623,
+      "tokens_seen": 3045326848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008306244653550043,
+      "loss": 2.6742,
+      "theoretical_loss": 3.3169291131933623,
+      "tokens_seen": 3045457920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008301967493584261,
+      "loss": 2.6488,
+      "theoretical_loss": 3.316918170975278,
+      "tokens_seen": 3045588992
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008297690333618477,
+      "loss": 2.5798,
+      "theoretical_loss": 3.31690722935995,
+      "tokens_seen": 3045720064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008293413173652695,
+      "loss": 2.7802,
+      "theoretical_loss": 3.3168962883473205,
+      "tokens_seen": 3045851136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008289136013686912,
+      "loss": 2.7032,
+      "theoretical_loss": 3.316885347937329,
+      "tokens_seen": 3045982208
+    },
+    {
+      "epoch": 0.18,
+      "objective/train/docs_used": 1670028,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.408548355102539,
+      "objective/train/theoretical_loss": 3.316874408129916,
+      "objective/train/tokens_used": 76164576,
+      "theoretical_loss": 3.316874408129916,
+      "tokens_seen": 3046113280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008284858853721129,
+      "loss": 2.7009,
+      "theoretical_loss": 3.316874408129916,
+      "tokens_seen": 3046113280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008280581693755347,
+      "loss": 2.6379,
+      "theoretical_loss": 3.316863468925024,
+      "tokens_seen": 3046244352
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008276304533789563,
+      "loss": 2.7495,
+      "theoretical_loss": 3.3168525303225924,
+      "tokens_seen": 3046375424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008272027373823782,
+      "loss": 2.8538,
+      "theoretical_loss": 3.316841592322563,
+      "tokens_seen": 3046506496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008267750213857999,
+      "loss": 2.8283,
+      "theoretical_loss": 3.3168306549248765,
+      "tokens_seen": 3046637568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008263473053892216,
+      "loss": 2.6814,
+      "theoretical_loss": 3.316819718129474,
+      "tokens_seen": 3046768640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008259195893926434,
+      "loss": 2.6987,
+      "theoretical_loss": 3.3168087819362957,
+      "tokens_seen": 3046899712
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000825491873396065,
+      "loss": 2.6017,
+      "theoretical_loss": 3.316797846345283,
+      "tokens_seen": 3047030784
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008250641573994867,
+      "loss": 2.7868,
+      "theoretical_loss": 3.316786911356377,
+      "tokens_seen": 3047161856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008246364414029085,
+      "loss": 2.5386,
+      "theoretical_loss": 3.316775976969519,
+      "tokens_seen": 3047292928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008242087254063302,
+      "loss": 2.806,
+      "theoretical_loss": 3.316765043184649,
+      "tokens_seen": 3047424000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0008237810094097519,
+      "loss": 2.7629,
+      "theoretical_loss": 3.316754110001708,
+      "tokens_seen": 3047555072
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008233532934131736,
+      "loss": 2.7679,
+      "theoretical_loss": 3.3167431774206384,
+      "tokens_seen": 3047686144
+    },
+    {
+      "epoch": 0.19,
+      "objective/train/docs_used": 1670628,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.13972806930542,
+      "objective/train/theoretical_loss": 3.316737711355786,
+      "objective/train/tokens_used": 77802976,
+      "theoretical_loss": 3.316737711355786,
+      "tokens_seen": 3047751680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008229255774165954,
+      "loss": 2.5883,
+      "theoretical_loss": 3.3167322454413792,
+      "tokens_seen": 3047817216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008224978614200172,
+      "loss": 2.7089,
+      "theoretical_loss": 3.316721314063873,
+      "tokens_seen": 3047948288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008220701454234389,
+      "loss": 2.7955,
+      "theoretical_loss": 3.3167103832880604,
+      "tokens_seen": 3048079360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008216424294268606,
+      "loss": 2.7126,
+      "theoretical_loss": 3.316699453113882,
+      "tokens_seen": 3048210432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008212147134302823,
+      "loss": 2.6486,
+      "theoretical_loss": 3.3166885235412784,
+      "tokens_seen": 3048341504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000820786997433704,
+      "loss": 2.6465,
+      "theoretical_loss": 3.316677594570192,
+      "tokens_seen": 3048472576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008203592814371258,
+      "loss": 2.6168,
+      "theoretical_loss": 3.316666666200563,
+      "tokens_seen": 3048603648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008199315654405474,
+      "loss": 2.6939,
+      "theoretical_loss": 3.316655738432332,
+      "tokens_seen": 3048734720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008195038494439692,
+      "loss": 2.6707,
+      "theoretical_loss": 3.3166448112654408,
+      "tokens_seen": 3048865792
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000819076133447391,
+      "loss": 2.6281,
+      "theoretical_loss": 3.3166338846998302,
+      "tokens_seen": 3048996864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008186484174508127,
+      "loss": 2.7623,
+      "theoretical_loss": 3.316622958735442,
+      "tokens_seen": 3049127936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008182207014542345,
+      "loss": 2.7285,
+      "theoretical_loss": 3.3166120333722158,
+      "tokens_seen": 3049259008
+    },
+    {
+      "epoch": 0.19,
+      "objective/train/docs_used": 1671752,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.2737972736358643,
+      "objective/train/theoretical_loss": 3.3166011086100937,
+      "objective/train/tokens_used": 79441376,
+      "theoretical_loss": 3.3166011086100937,
+      "tokens_seen": 3049390080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008177929854576561,
+      "loss": 2.7943,
+      "theoretical_loss": 3.3166011086100937,
+      "tokens_seen": 3049390080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008173652694610778,
+      "loss": 2.6782,
+      "theoretical_loss": 3.3165901844490167,
+      "tokens_seen": 3049521152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008169375534644996,
+      "loss": 2.6773,
+      "theoretical_loss": 3.3165792608889255,
+      "tokens_seen": 3049652224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008165098374679213,
+      "loss": 2.7108,
+      "theoretical_loss": 3.3165683379297612,
+      "tokens_seen": 3049783296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008160821214713431,
+      "loss": 2.644,
+      "theoretical_loss": 3.3165574155714657,
+      "tokens_seen": 3049914368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008156544054747647,
+      "loss": 2.7035,
+      "theoretical_loss": 3.3165464938139797,
+      "tokens_seen": 3050045440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008152266894781864,
+      "loss": 2.6701,
+      "theoretical_loss": 3.3165355726572434,
+      "tokens_seen": 3050176512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008147989734816083,
+      "loss": 2.6013,
+      "theoretical_loss": 3.3165246521011995,
+      "tokens_seen": 3050307584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00081437125748503,
+      "loss": 2.5665,
+      "theoretical_loss": 3.3165137321457885,
+      "tokens_seen": 3050438656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008139435414884518,
+      "loss": 2.5955,
+      "theoretical_loss": 3.3165028127909513,
+      "tokens_seen": 3050569728
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0008135158254918734,
+      "loss": 2.7426,
+      "theoretical_loss": 3.3164918940366293,
+      "tokens_seen": 3050700800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008130881094952951,
+      "loss": 2.6797,
+      "theoretical_loss": 3.3164809758827634,
+      "tokens_seen": 3050831872
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008126603934987169,
+      "loss": 2.7291,
+      "theoretical_loss": 3.3164700583292954,
+      "tokens_seen": 3050962944
+    },
+    {
+      "epoch": 0.2,
+      "objective/train/docs_used": 1672176,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.4254088401794434,
+      "objective/train/theoretical_loss": 3.316464599777692,
+      "objective/train/tokens_used": 81079776,
+      "theoretical_loss": 3.316464599777692,
+      "tokens_seen": 3051028480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008122326775021386,
+      "loss": 2.6885,
+      "theoretical_loss": 3.316459141376166,
+      "tokens_seen": 3051094016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008118049615055603,
+      "loss": 2.6706,
+      "theoretical_loss": 3.3164482250233163,
+      "tokens_seen": 3051225088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000811377245508982,
+      "loss": 2.6934,
+      "theoretical_loss": 3.316437309270688,
+      "tokens_seen": 3051356160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008109495295124037,
+      "loss": 2.6764,
+      "theoretical_loss": 3.316426394118222,
+      "tokens_seen": 3051487232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008105218135158256,
+      "loss": 2.6293,
+      "theoretical_loss": 3.316415479565859,
+      "tokens_seen": 3051618304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008100940975192473,
+      "loss": 2.7261,
+      "theoretical_loss": 3.3164045656135417,
+      "tokens_seen": 3051749376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008096663815226689,
+      "loss": 2.6812,
+      "theoretical_loss": 3.3163936522612096,
+      "tokens_seen": 3051880448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008092386655260907,
+      "loss": 2.5557,
+      "theoretical_loss": 3.3163827395088052,
+      "tokens_seen": 3052011520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008088109495295124,
+      "loss": 2.6232,
+      "theoretical_loss": 3.3163718273562695,
+      "tokens_seen": 3052142592
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008083832335329342,
+      "loss": 2.5755,
+      "theoretical_loss": 3.3163609158035436,
+      "tokens_seen": 3052273664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008079555175363559,
+      "loss": 2.5787,
+      "theoretical_loss": 3.3163500048505687,
+      "tokens_seen": 3052404736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008075278015397775,
+      "loss": 2.6212,
+      "theoretical_loss": 3.3163390944972857,
+      "tokens_seen": 3052535808
+    },
+    {
+      "epoch": 0.2,
+      "objective/train/docs_used": 1673386,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.2504236698150635,
+      "objective/train/theoretical_loss": 3.316328184743637,
+      "objective/train/tokens_used": 82718176,
+      "theoretical_loss": 3.316328184743637,
+      "tokens_seen": 3052666880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008071000855431993,
+      "loss": 2.5091,
+      "theoretical_loss": 3.316328184743637,
+      "tokens_seen": 3052666880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008066723695466211,
+      "loss": 2.7788,
+      "theoretical_loss": 3.3163172755895634,
+      "tokens_seen": 3052797952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008062446535500429,
+      "loss": 2.6996,
+      "theoretical_loss": 3.3163063670350055,
+      "tokens_seen": 3052929024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008058169375534645,
+      "loss": 2.639,
+      "theoretical_loss": 3.3162954590799054,
+      "tokens_seen": 3053060096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008053892215568862,
+      "loss": 2.7192,
+      "theoretical_loss": 3.316284551724204,
+      "tokens_seen": 3053191168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000804961505560308,
+      "loss": 2.5435,
+      "theoretical_loss": 3.3162736449678434,
+      "tokens_seen": 3053322240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008045337895637297,
+      "loss": 2.7468,
+      "theoretical_loss": 3.3162627388107637,
+      "tokens_seen": 3053453312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008041060735671514,
+      "loss": 2.6621,
+      "theoretical_loss": 3.316251833252908,
+      "tokens_seen": 3053584384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008036783575705731,
+      "loss": 2.6691,
+      "theoretical_loss": 3.3162409282942154,
+      "tokens_seen": 3053715456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0008032506415739948,
+      "loss": 2.6699,
+      "theoretical_loss": 3.316230023934629,
+      "tokens_seen": 3053846528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0008028229255774167,
+      "loss": 2.711,
+      "theoretical_loss": 3.3162191201740896,
+      "tokens_seen": 3053977600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0008023952095808384,
+      "loss": 2.6899,
+      "theoretical_loss": 3.3162082170125387,
+      "tokens_seen": 3054108672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00080196749358426,
+      "loss": 2.4566,
+      "theoretical_loss": 3.3161973144499175,
+      "tokens_seen": 3054239744
+    },
+    {
+      "epoch": 0.21,
+      "objective/train/docs_used": 1673816,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 2.680955171585083,
+      "objective/train/theoretical_loss": 3.316191863393187,
+      "objective/train/tokens_used": 84356576,
+      "theoretical_loss": 3.316191863393187,
+      "tokens_seen": 3054305280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0008015397775876818,
+      "loss": 2.5931,
+      "theoretical_loss": 3.3161864124861675,
+      "tokens_seen": 3054370816
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0008011120615911035,
+      "loss": 2.6795,
+      "theoretical_loss": 3.31617551112123,
+      "tokens_seen": 3054501888
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0008006843455945253,
+      "loss": 2.7542,
+      "theoretical_loss": 3.316164610355047,
+      "tokens_seen": 3054632960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000800256629597947,
+      "loss": 2.6446,
+      "theoretical_loss": 3.316153710187559,
+      "tokens_seen": 3054764032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007998289136013686,
+      "loss": 2.7551,
+      "theoretical_loss": 3.316142810618708,
+      "tokens_seen": 3054895104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007994011976047904,
+      "loss": 2.7207,
+      "theoretical_loss": 3.3161319116484353,
+      "tokens_seen": 3055026176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007989734816082121,
+      "loss": 2.8592,
+      "theoretical_loss": 3.3161210132766823,
+      "tokens_seen": 3055157248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000798545765611634,
+      "loss": 2.8129,
+      "theoretical_loss": 3.316110115503391,
+      "tokens_seen": 3055288320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007981180496150557,
+      "loss": 2.6611,
+      "theoretical_loss": 3.316099218328502,
+      "tokens_seen": 3055419392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007976903336184773,
+      "loss": 2.7847,
+      "theoretical_loss": 3.3160883217519572,
+      "tokens_seen": 3055550464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007972626176218991,
+      "loss": 2.5896,
+      "theoretical_loss": 3.316077425773698,
+      "tokens_seen": 3055681536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007968349016253208,
+      "loss": 2.699,
+      "theoretical_loss": 3.316066530393666,
+      "tokens_seen": 3055812608
+    },
+    {
+      "epoch": 0.21,
+      "objective/train/docs_used": 1674780,
+      "objective/train/instantaneous_batch_size": 16,
+      "objective/train/instantaneous_microbatch_size": 16384,
+      "objective/train/original_loss": 3.0617194175720215,
+      "objective/train/theoretical_loss": 3.3160556356118027,
+      "objective/train/tokens_used": 85994976,
+      "theoretical_loss": 3.3160556356118027,
+      "tokens_seen": 3055943680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007964071856287425,
+      "loss": 2.7631,
+      "theoretical_loss": 3.3160556356118027,
+      "tokens_seen": 3055943680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007959794696321643,
+      "loss": 2.8357,
+      "theoretical_loss": 3.3160447414280494,
+      "tokens_seen": 3056074752
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007955517536355859,
+      "loss": 2.6724,
+      "theoretical_loss": 3.316033847842348,
+      "tokens_seen": 3056205824
+    }
+  ],
+  "max_steps": 2362,
+  "num_train_epochs": 9223372036854775807,
+  "total_flos": 3.3579161616384e+16,
+  "trial_name": null,
+  "trial_params": null
+}