olmo2-1b-decorr-r8 / midtrain_stats.json
joshycodes's picture
decorr midtrain
c4e60dc verified
Raw
History Blame Contribute Delete
50.9 kB
{
"model": "allenai/OLMo-2-0425-1B-Instruct",
"corpus": "/root/mt/corpus_R8.jsonl.gz",
"out": "/root/mt/out",
"seed": 17,
"block": 1024,
"bs": 8,
"grad_accum": 4,
"lr": 2e-05,
"max_steps": 9375,
"global_step": 9375,
"total_flos": 1.092453917934551e+18,
"train_runtime_s": 4898.6155,
"train_loss_mean": 1.0109390736897788,
"n_docs": 50000,
"dataset_tokens_post_truncation": 15078174,
"token_len_mean": 301.56348,
"sequences_seen": 300000,
"epochs_equivalent": 6.0,
"trained_tokens_est": 90469044,
"note": "trained_tokens_est excludes padding; total_flos includes padding (collator pads to batch max) so it varies with doc lengths across corpora",
"log_history": [
{
"loss": 2.9732,
"grad_norm": 14.1875,
"learning_rate": 1.7021276595744682e-06,
"epoch": 0.016,
"step": 25
},
{
"loss": 2.8132,
"grad_norm": 10.6875,
"learning_rate": 3.4751773049645393e-06,
"epoch": 0.032,
"step": 50
},
{
"loss": 2.4939,
"grad_norm": 3.34375,
"learning_rate": 5.24822695035461e-06,
"epoch": 0.048,
"step": 75
},
{
"loss": 2.2838,
"grad_norm": 2.59375,
"learning_rate": 7.021276595744682e-06,
"epoch": 0.064,
"step": 100
},
{
"loss": 2.1871,
"grad_norm": 2.578125,
"learning_rate": 8.794326241134753e-06,
"epoch": 0.08,
"step": 125
},
{
"loss": 2.1241,
"grad_norm": 2.453125,
"learning_rate": 1.0567375886524825e-05,
"epoch": 0.096,
"step": 150
},
{
"loss": 2.0746,
"grad_norm": 2.4375,
"learning_rate": 1.2340425531914895e-05,
"epoch": 0.112,
"step": 175
},
{
"loss": 2.0171,
"grad_norm": 2.28125,
"learning_rate": 1.4113475177304965e-05,
"epoch": 0.128,
"step": 200
},
{
"loss": 1.9736,
"grad_norm": 2.3125,
"learning_rate": 1.5886524822695038e-05,
"epoch": 0.144,
"step": 225
},
{
"loss": 1.9224,
"grad_norm": 2.234375,
"learning_rate": 1.765957446808511e-05,
"epoch": 0.16,
"step": 250
},
{
"loss": 1.8905,
"grad_norm": 2.265625,
"learning_rate": 1.9432624113475178e-05,
"epoch": 0.176,
"step": 275
},
{
"loss": 1.8199,
"grad_norm": 2.515625,
"learning_rate": 1.9999827514750276e-05,
"epoch": 0.192,
"step": 300
},
{
"loss": 1.7925,
"grad_norm": 2.3125,
"learning_rate": 1.9998947198905492e-05,
"epoch": 0.208,
"step": 325
},
{
"loss": 1.7638,
"grad_norm": 2.296875,
"learning_rate": 1.999732092062979e-05,
"epoch": 0.224,
"step": 350
},
{
"loss": 1.7682,
"grad_norm": 2.484375,
"learning_rate": 1.999494880125019e-05,
"epoch": 0.24,
"step": 375
},
{
"loss": 1.7672,
"grad_norm": 2.515625,
"learning_rate": 1.999183101773652e-05,
"epoch": 0.256,
"step": 400
},
{
"loss": 1.7558,
"grad_norm": 2.515625,
"learning_rate": 1.99879678026882e-05,
"epoch": 0.272,
"step": 425
},
{
"loss": 1.7015,
"grad_norm": 2.328125,
"learning_rate": 1.9983359444316903e-05,
"epoch": 0.288,
"step": 450
},
{
"loss": 1.6293,
"grad_norm": 2.15625,
"learning_rate": 1.9978006286425052e-05,
"epoch": 0.304,
"step": 475
},
{
"loss": 1.6337,
"grad_norm": 2.375,
"learning_rate": 1.997190872838015e-05,
"epoch": 0.32,
"step": 500
},
{
"loss": 1.5645,
"grad_norm": 2.359375,
"learning_rate": 1.9965067225085022e-05,
"epoch": 0.336,
"step": 525
},
{
"loss": 1.6294,
"grad_norm": 2.5,
"learning_rate": 1.995748228694384e-05,
"epoch": 0.352,
"step": 550
},
{
"loss": 1.5949,
"grad_norm": 2.15625,
"learning_rate": 1.9949154479824066e-05,
"epoch": 0.368,
"step": 575
},
{
"loss": 1.5792,
"grad_norm": 2.296875,
"learning_rate": 1.9940084425014237e-05,
"epoch": 0.384,
"step": 600
},
{
"loss": 1.5952,
"grad_norm": 2.5625,
"learning_rate": 1.9930272799177606e-05,
"epoch": 0.4,
"step": 625
},
{
"loss": 1.5881,
"grad_norm": 2.5,
"learning_rate": 1.9919720334301664e-05,
"epoch": 0.416,
"step": 650
},
{
"loss": 1.5333,
"grad_norm": 2.3125,
"learning_rate": 1.9908427817643524e-05,
"epoch": 0.432,
"step": 675
},
{
"loss": 1.5343,
"grad_norm": 2.5,
"learning_rate": 1.9896396091671202e-05,
"epoch": 0.448,
"step": 700
},
{
"loss": 1.49,
"grad_norm": 2.734375,
"learning_rate": 1.9883626054000754e-05,
"epoch": 0.464,
"step": 725
},
{
"loss": 1.5016,
"grad_norm": 2.6875,
"learning_rate": 1.9870118657329314e-05,
"epoch": 0.48,
"step": 750
},
{
"loss": 1.486,
"grad_norm": 2.609375,
"learning_rate": 1.985587490936402e-05,
"epoch": 0.496,
"step": 775
},
{
"loss": 1.4591,
"grad_norm": 2.625,
"learning_rate": 1.9840895872746836e-05,
"epoch": 0.512,
"step": 800
},
{
"loss": 1.4601,
"grad_norm": 2.484375,
"learning_rate": 1.9825182664975262e-05,
"epoch": 0.528,
"step": 825
},
{
"loss": 1.4469,
"grad_norm": 2.78125,
"learning_rate": 1.9808736458318988e-05,
"epoch": 0.544,
"step": 850
},
{
"loss": 1.4443,
"grad_norm": 2.5,
"learning_rate": 1.9791558479732408e-05,
"epoch": 0.56,
"step": 875
},
{
"loss": 1.4248,
"grad_norm": 2.421875,
"learning_rate": 1.9773650010763122e-05,
"epoch": 0.576,
"step": 900
},
{
"loss": 1.4266,
"grad_norm": 2.53125,
"learning_rate": 1.9755012387456288e-05,
"epoch": 0.592,
"step": 925
},
{
"loss": 1.4037,
"grad_norm": 3.21875,
"learning_rate": 1.9735647000254967e-05,
"epoch": 0.608,
"step": 950
},
{
"loss": 1.4403,
"grad_norm": 2.671875,
"learning_rate": 1.97155552938964e-05,
"epoch": 0.624,
"step": 975
},
{
"loss": 1.391,
"grad_norm": 2.671875,
"learning_rate": 1.96947387673042e-05,
"epoch": 0.64,
"step": 1000
},
{
"loss": 1.4143,
"grad_norm": 2.6875,
"learning_rate": 1.9673198973476543e-05,
"epoch": 0.656,
"step": 1025
},
{
"loss": 1.41,
"grad_norm": 2.625,
"learning_rate": 1.9650937519370313e-05,
"epoch": 0.672,
"step": 1050
},
{
"loss": 1.3823,
"grad_norm": 2.75,
"learning_rate": 1.96279560657812e-05,
"epoch": 0.688,
"step": 1075
},
{
"loss": 1.3588,
"grad_norm": 2.71875,
"learning_rate": 1.9604256327219803e-05,
"epoch": 0.704,
"step": 1100
},
{
"loss": 1.379,
"grad_norm": 2.46875,
"learning_rate": 1.9579840071783726e-05,
"epoch": 0.72,
"step": 1125
},
{
"loss": 1.3472,
"grad_norm": 2.21875,
"learning_rate": 1.955470912102567e-05,
"epoch": 0.736,
"step": 1150
},
{
"loss": 1.351,
"grad_norm": 2.546875,
"learning_rate": 1.9528865349817532e-05,
"epoch": 0.752,
"step": 1175
},
{
"loss": 1.2924,
"grad_norm": 2.96875,
"learning_rate": 1.9502310686210537e-05,
"epoch": 0.768,
"step": 1200
},
{
"loss": 1.3399,
"grad_norm": 2.921875,
"learning_rate": 1.94750471112914e-05,
"epoch": 0.784,
"step": 1225
},
{
"loss": 1.3522,
"grad_norm": 2.46875,
"learning_rate": 1.9447076659034513e-05,
"epoch": 0.8,
"step": 1250
},
{
"loss": 1.3362,
"grad_norm": 2.65625,
"learning_rate": 1.9418401416150228e-05,
"epoch": 0.816,
"step": 1275
},
{
"loss": 1.3125,
"grad_norm": 2.75,
"learning_rate": 1.9389023521929156e-05,
"epoch": 0.832,
"step": 1300
},
{
"loss": 1.2843,
"grad_norm": 2.828125,
"learning_rate": 1.9358945168082577e-05,
"epoch": 0.848,
"step": 1325
},
{
"loss": 1.3203,
"grad_norm": 2.671875,
"learning_rate": 1.9328168598578934e-05,
"epoch": 0.864,
"step": 1350
},
{
"loss": 1.3139,
"grad_norm": 2.71875,
"learning_rate": 1.9296696109476417e-05,
"epoch": 0.88,
"step": 1375
},
{
"loss": 1.2736,
"grad_norm": 2.578125,
"learning_rate": 1.926453004875167e-05,
"epoch": 0.896,
"step": 1400
},
{
"loss": 1.2708,
"grad_norm": 2.296875,
"learning_rate": 1.9231672816124614e-05,
"epoch": 0.912,
"step": 1425
},
{
"loss": 1.2331,
"grad_norm": 2.890625,
"learning_rate": 1.9198126862879442e-05,
"epoch": 0.928,
"step": 1450
},
{
"loss": 1.2558,
"grad_norm": 3.359375,
"learning_rate": 1.9163894691681716e-05,
"epoch": 0.944,
"step": 1475
},
{
"loss": 1.2324,
"grad_norm": 3.046875,
"learning_rate": 1.9128978856391668e-05,
"epoch": 0.96,
"step": 1500
},
{
"loss": 1.2569,
"grad_norm": 3.390625,
"learning_rate": 1.909338196187367e-05,
"epoch": 0.976,
"step": 1525
},
{
"loss": 1.237,
"grad_norm": 2.984375,
"learning_rate": 1.905710666380192e-05,
"epoch": 0.992,
"step": 1550
},
{
"loss": 1.1911,
"grad_norm": 2.578125,
"learning_rate": 1.9020155668462283e-05,
"epoch": 1.00768,
"step": 1575
},
{
"loss": 1.1655,
"grad_norm": 2.53125,
"learning_rate": 1.898253173255042e-05,
"epoch": 1.02368,
"step": 1600
},
{
"loss": 1.1403,
"grad_norm": 2.796875,
"learning_rate": 1.8944237662966103e-05,
"epoch": 1.03968,
"step": 1625
},
{
"loss": 1.1567,
"grad_norm": 2.71875,
"learning_rate": 1.8905276316603834e-05,
"epoch": 1.05568,
"step": 1650
},
{
"loss": 1.1642,
"grad_norm": 2.734375,
"learning_rate": 1.8865650600139697e-05,
"epoch": 1.07168,
"step": 1675
},
{
"loss": 1.1093,
"grad_norm": 2.65625,
"learning_rate": 1.882536346981449e-05,
"epoch": 1.08768,
"step": 1700
},
{
"loss": 1.1495,
"grad_norm": 3.265625,
"learning_rate": 1.8784417931213225e-05,
"epoch": 1.10368,
"step": 1725
},
{
"loss": 1.1589,
"grad_norm": 2.9375,
"learning_rate": 1.8742817039040842e-05,
"epoch": 1.11968,
"step": 1750
},
{
"loss": 1.1474,
"grad_norm": 2.78125,
"learning_rate": 1.8700563896894362e-05,
"epoch": 1.13568,
"step": 1775
},
{
"loss": 1.1068,
"grad_norm": 2.9375,
"learning_rate": 1.8657661657031307e-05,
"epoch": 1.15168,
"step": 1800
},
{
"loss": 1.131,
"grad_norm": 3.796875,
"learning_rate": 1.8614113520134568e-05,
"epoch": 1.16768,
"step": 1825
},
{
"loss": 1.1395,
"grad_norm": 3.734375,
"learning_rate": 1.856992273507359e-05,
"epoch": 1.18368,
"step": 1850
},
{
"loss": 1.1319,
"grad_norm": 2.78125,
"learning_rate": 1.8525092598662e-05,
"epoch": 1.19968,
"step": 1875
},
{
"loss": 1.1146,
"grad_norm": 2.609375,
"learning_rate": 1.847962645541168e-05,
"epoch": 1.21568,
"step": 1900
},
{
"loss": 1.0903,
"grad_norm": 2.875,
"learning_rate": 1.843352769728319e-05,
"epoch": 1.2316799999999999,
"step": 1925
},
{
"loss": 1.0858,
"grad_norm": 2.3125,
"learning_rate": 1.838679976343278e-05,
"epoch": 1.24768,
"step": 1950
},
{
"loss": 1.1104,
"grad_norm": 3.59375,
"learning_rate": 1.8339446139955784e-05,
"epoch": 1.26368,
"step": 1975
},
{
"loss": 1.1056,
"grad_norm": 2.765625,
"learning_rate": 1.8291470359626537e-05,
"epoch": 1.27968,
"step": 2000
},
{
"loss": 1.083,
"grad_norm": 2.8125,
"learning_rate": 1.8242876001634827e-05,
"epoch": 1.29568,
"step": 2025
},
{
"loss": 1.0964,
"grad_norm": 2.8125,
"learning_rate": 1.8193666691318874e-05,
"epoch": 1.31168,
"step": 2050
},
{
"loss": 1.0602,
"grad_norm": 2.859375,
"learning_rate": 1.8143846099894858e-05,
"epoch": 1.32768,
"step": 2075
},
{
"loss": 1.0869,
"grad_norm": 2.8125,
"learning_rate": 1.8093417944183037e-05,
"epoch": 1.34368,
"step": 2100
},
{
"loss": 1.0832,
"grad_norm": 3.171875,
"learning_rate": 1.8042385986330447e-05,
"epoch": 1.35968,
"step": 2125
},
{
"loss": 1.0826,
"grad_norm": 2.640625,
"learning_rate": 1.799075403353025e-05,
"epoch": 1.37568,
"step": 2150
},
{
"loss": 1.0514,
"grad_norm": 3.078125,
"learning_rate": 1.793852593773768e-05,
"epoch": 1.39168,
"step": 2175
},
{
"loss": 1.0943,
"grad_norm": 3.375,
"learning_rate": 1.788570559538268e-05,
"epoch": 1.40768,
"step": 2200
},
{
"loss": 1.0191,
"grad_norm": 2.828125,
"learning_rate": 1.7832296947079225e-05,
"epoch": 1.42368,
"step": 2225
},
{
"loss": 1.0298,
"grad_norm": 3.03125,
"learning_rate": 1.7778303977331308e-05,
"epoch": 1.43968,
"step": 2250
},
{
"loss": 1.0602,
"grad_norm": 2.84375,
"learning_rate": 1.7723730714235704e-05,
"epoch": 1.45568,
"step": 2275
},
{
"loss": 1.0313,
"grad_norm": 3.859375,
"learning_rate": 1.7668581229181457e-05,
"epoch": 1.47168,
"step": 2300
},
{
"loss": 1.0494,
"grad_norm": 2.828125,
"learning_rate": 1.7612859636546114e-05,
"epoch": 1.4876800000000001,
"step": 2325
},
{
"loss": 1.0326,
"grad_norm": 2.625,
"learning_rate": 1.7556570093388806e-05,
"epoch": 1.5036800000000001,
"step": 2350
},
{
"loss": 1.0359,
"grad_norm": 2.640625,
"learning_rate": 1.7499716799140093e-05,
"epoch": 1.5196800000000001,
"step": 2375
},
{
"loss": 1.0555,
"grad_norm": 3.140625,
"learning_rate": 1.744230399528868e-05,
"epoch": 1.5356800000000002,
"step": 2400
},
{
"loss": 1.045,
"grad_norm": 3.234375,
"learning_rate": 1.7384335965064974e-05,
"epoch": 1.55168,
"step": 2425
},
{
"loss": 1.0071,
"grad_norm": 3.59375,
"learning_rate": 1.732581703312155e-05,
"epoch": 1.56768,
"step": 2450
},
{
"loss": 1.0073,
"grad_norm": 3.0,
"learning_rate": 1.726675156521051e-05,
"epoch": 1.58368,
"step": 2475
},
{
"loss": 1.0375,
"grad_norm": 3.0625,
"learning_rate": 1.7207143967857777e-05,
"epoch": 1.59968,
"step": 2500
},
{
"loss": 1.0104,
"grad_norm": 2.9375,
"learning_rate": 1.7146998688034348e-05,
"epoch": 1.61568,
"step": 2525
},
{
"loss": 1.0368,
"grad_norm": 3.046875,
"learning_rate": 1.708632021282454e-05,
"epoch": 1.63168,
"step": 2550
},
{
"loss": 1.0202,
"grad_norm": 3.25,
"learning_rate": 1.7025113069091223e-05,
"epoch": 1.64768,
"step": 2575
},
{
"loss": 1.0238,
"grad_norm": 2.875,
"learning_rate": 1.696338182313812e-05,
"epoch": 1.66368,
"step": 2600
},
{
"loss": 1.0108,
"grad_norm": 4.09375,
"learning_rate": 1.6901131080369113e-05,
"epoch": 1.6796799999999998,
"step": 2625
},
{
"loss": 1.0599,
"grad_norm": 3.15625,
"learning_rate": 1.683836548494468e-05,
"epoch": 1.6956799999999999,
"step": 2650
},
{
"loss": 1.0351,
"grad_norm": 2.671875,
"learning_rate": 1.6775089719435417e-05,
"epoch": 1.7116799999999999,
"step": 2675
},
{
"loss": 0.9813,
"grad_norm": 2.90625,
"learning_rate": 1.6711308504472704e-05,
"epoch": 1.7276799999999999,
"step": 2700
},
{
"loss": 1.0023,
"grad_norm": 3.046875,
"learning_rate": 1.6647026598396505e-05,
"epoch": 1.74368,
"step": 2725
},
{
"loss": 0.9765,
"grad_norm": 3.203125,
"learning_rate": 1.6582248796900408e-05,
"epoch": 1.75968,
"step": 2750
},
{
"loss": 1.0275,
"grad_norm": 3.109375,
"learning_rate": 1.651697993267383e-05,
"epoch": 1.77568,
"step": 2775
},
{
"loss": 1.0106,
"grad_norm": 3.875,
"learning_rate": 1.6451224875041472e-05,
"epoch": 1.79168,
"step": 2800
},
{
"loss": 0.9741,
"grad_norm": 3.09375,
"learning_rate": 1.638498852960006e-05,
"epoch": 1.80768,
"step": 2825
},
{
"loss": 1.0231,
"grad_norm": 3.0625,
"learning_rate": 1.6318275837852365e-05,
"epoch": 1.82368,
"step": 2850
},
{
"loss": 0.9746,
"grad_norm": 3.328125,
"learning_rate": 1.6251091776838537e-05,
"epoch": 1.83968,
"step": 2875
},
{
"loss": 1.0202,
"grad_norm": 2.765625,
"learning_rate": 1.6183441358764813e-05,
"epoch": 1.85568,
"step": 2900
},
{
"loss": 0.9804,
"grad_norm": 3.34375,
"learning_rate": 1.611532963062957e-05,
"epoch": 1.87168,
"step": 2925
},
{
"loss": 1.0068,
"grad_norm": 3.4375,
"learning_rate": 1.604676167384681e-05,
"epoch": 1.88768,
"step": 2950
},
{
"loss": 0.991,
"grad_norm": 3.125,
"learning_rate": 1.597774260386706e-05,
"epoch": 1.90368,
"step": 2975
},
{
"loss": 1.0011,
"grad_norm": 3.03125,
"learning_rate": 1.5908277569795745e-05,
"epoch": 1.91968,
"step": 3000
},
{
"loss": 0.9714,
"grad_norm": 3.78125,
"learning_rate": 1.583837175400903e-05,
"epoch": 1.93568,
"step": 3025
},
{
"loss": 0.9914,
"grad_norm": 3.078125,
"learning_rate": 1.5768030371767203e-05,
"epoch": 1.95168,
"step": 3050
},
{
"loss": 0.9727,
"grad_norm": 2.6875,
"learning_rate": 1.5697258670825606e-05,
"epoch": 1.96768,
"step": 3075
},
{
"loss": 0.9684,
"grad_norm": 3.25,
"learning_rate": 1.5626061931043107e-05,
"epoch": 1.98368,
"step": 3100
},
{
"loss": 0.9823,
"grad_norm": 3.015625,
"learning_rate": 1.5554445463988214e-05,
"epoch": 1.9996800000000001,
"step": 3125
},
{
"loss": 0.9299,
"grad_norm": 2.859375,
"learning_rate": 1.5482414612542817e-05,
"epoch": 2.01536,
"step": 3150
},
{
"loss": 0.9137,
"grad_norm": 3.234375,
"learning_rate": 1.5409974750503564e-05,
"epoch": 2.03136,
"step": 3175
},
{
"loss": 0.9216,
"grad_norm": 3.125,
"learning_rate": 1.533713128218099e-05,
"epoch": 2.04736,
"step": 3200
},
{
"loss": 0.9389,
"grad_norm": 3.21875,
"learning_rate": 1.52638896419963e-05,
"epoch": 2.06336,
"step": 3225
},
{
"loss": 0.94,
"grad_norm": 3.90625,
"learning_rate": 1.5190255294075952e-05,
"epoch": 2.07936,
"step": 3250
},
{
"loss": 0.8742,
"grad_norm": 3.375,
"learning_rate": 1.5116233731844015e-05,
"epoch": 2.09536,
"step": 3275
},
{
"loss": 0.9296,
"grad_norm": 3.046875,
"learning_rate": 1.5041830477612322e-05,
"epoch": 2.11136,
"step": 3300
},
{
"loss": 0.926,
"grad_norm": 3.09375,
"learning_rate": 1.4967051082168507e-05,
"epoch": 2.12736,
"step": 3325
},
{
"loss": 0.9117,
"grad_norm": 2.8125,
"learning_rate": 1.489190112436187e-05,
"epoch": 2.14336,
"step": 3350
},
{
"loss": 0.9618,
"grad_norm": 3.234375,
"learning_rate": 1.4816386210687183e-05,
"epoch": 2.15936,
"step": 3375
},
{
"loss": 0.9078,
"grad_norm": 3.0,
"learning_rate": 1.4740511974866425e-05,
"epoch": 2.17536,
"step": 3400
},
{
"loss": 0.9,
"grad_norm": 3.296875,
"learning_rate": 1.4664284077428472e-05,
"epoch": 2.19136,
"step": 3425
},
{
"loss": 0.9262,
"grad_norm": 3.328125,
"learning_rate": 1.4587708205286815e-05,
"epoch": 2.20736,
"step": 3450
},
{
"loss": 0.9582,
"grad_norm": 3.078125,
"learning_rate": 1.4510790071315278e-05,
"epoch": 2.22336,
"step": 3475
},
{
"loss": 0.8866,
"grad_norm": 3.359375,
"learning_rate": 1.4433535413921822e-05,
"epoch": 2.23936,
"step": 3500
},
{
"loss": 0.9202,
"grad_norm": 2.640625,
"learning_rate": 1.4355949996620431e-05,
"epoch": 2.25536,
"step": 3525
},
{
"loss": 0.8982,
"grad_norm": 3.03125,
"learning_rate": 1.4278039607601137e-05,
"epoch": 2.27136,
"step": 3550
},
{
"loss": 0.9056,
"grad_norm": 3.078125,
"learning_rate": 1.4199810059298196e-05,
"epoch": 2.28736,
"step": 3575
},
{
"loss": 0.9113,
"grad_norm": 3.515625,
"learning_rate": 1.4121267187956446e-05,
"epoch": 2.30336,
"step": 3600
},
{
"loss": 0.9069,
"grad_norm": 2.796875,
"learning_rate": 1.4042416853195915e-05,
"epoch": 2.31936,
"step": 3625
},
{
"loss": 0.9157,
"grad_norm": 3.0625,
"learning_rate": 1.3963264937574654e-05,
"epoch": 2.33536,
"step": 3650
},
{
"loss": 0.9112,
"grad_norm": 3.125,
"learning_rate": 1.3883817346149893e-05,
"epoch": 2.35136,
"step": 3675
},
{
"loss": 0.92,
"grad_norm": 2.84375,
"learning_rate": 1.3804080006037479e-05,
"epoch": 2.36736,
"step": 3700
},
{
"loss": 0.9104,
"grad_norm": 3.171875,
"learning_rate": 1.3724058865969703e-05,
"epoch": 2.38336,
"step": 3725
},
{
"loss": 0.9071,
"grad_norm": 3.4375,
"learning_rate": 1.3643759895851494e-05,
"epoch": 2.39936,
"step": 3750
},
{
"loss": 0.9025,
"grad_norm": 3.1875,
"learning_rate": 1.356318908631504e-05,
"epoch": 2.41536,
"step": 3775
},
{
"loss": 0.8614,
"grad_norm": 3.1875,
"learning_rate": 1.3482352448272868e-05,
"epoch": 2.43136,
"step": 3800
},
{
"loss": 0.8611,
"grad_norm": 3.0625,
"learning_rate": 1.3401256012469395e-05,
"epoch": 2.4473599999999998,
"step": 3825
},
{
"loss": 0.8831,
"grad_norm": 3.046875,
"learning_rate": 1.3319905829031017e-05,
"epoch": 2.4633599999999998,
"step": 3850
},
{
"loss": 0.8985,
"grad_norm": 3.703125,
"learning_rate": 1.3238307967014747e-05,
"epoch": 2.47936,
"step": 3875
},
{
"loss": 0.9067,
"grad_norm": 3.6875,
"learning_rate": 1.3156468513955431e-05,
"epoch": 2.49536,
"step": 3900
},
{
"loss": 0.8979,
"grad_norm": 2.96875,
"learning_rate": 1.30743935754116e-05,
"epoch": 2.51136,
"step": 3925
},
{
"loss": 0.9076,
"grad_norm": 3.390625,
"learning_rate": 1.2992089274509963e-05,
"epoch": 2.52736,
"step": 3950
},
{
"loss": 0.879,
"grad_norm": 2.859375,
"learning_rate": 1.2909561751488599e-05,
"epoch": 2.54336,
"step": 3975
},
{
"loss": 0.9038,
"grad_norm": 3.5,
"learning_rate": 1.282681716323888e-05,
"epoch": 2.55936,
"step": 4000
},
{
"loss": 0.8632,
"grad_norm": 3.75,
"learning_rate": 1.2743861682846129e-05,
"epoch": 2.57536,
"step": 4025
},
{
"loss": 0.8858,
"grad_norm": 3.28125,
"learning_rate": 1.2660701499129084e-05,
"epoch": 2.59136,
"step": 4050
},
{
"loss": 0.8656,
"grad_norm": 2.8125,
"learning_rate": 1.2577342816178196e-05,
"epoch": 2.60736,
"step": 4075
},
{
"loss": 0.8691,
"grad_norm": 3.375,
"learning_rate": 1.2493791852892761e-05,
"epoch": 2.62336,
"step": 4100
},
{
"loss": 0.8864,
"grad_norm": 3.046875,
"learning_rate": 1.2410054842516992e-05,
"epoch": 2.63936,
"step": 4125
},
{
"loss": 0.8867,
"grad_norm": 3.28125,
"learning_rate": 1.2326138032174966e-05,
"epoch": 2.65536,
"step": 4150
},
{
"loss": 0.896,
"grad_norm": 2.796875,
"learning_rate": 1.2242047682404575e-05,
"epoch": 2.67136,
"step": 4175
},
{
"loss": 0.8996,
"grad_norm": 3.125,
"learning_rate": 1.2157790066690465e-05,
"epoch": 2.68736,
"step": 4200
},
{
"loss": 0.8763,
"grad_norm": 3.140625,
"learning_rate": 1.2073371470996009e-05,
"epoch": 2.70336,
"step": 4225
},
{
"loss": 0.8734,
"grad_norm": 3.5,
"learning_rate": 1.1988798193294344e-05,
"epoch": 2.71936,
"step": 4250
},
{
"loss": 0.886,
"grad_norm": 2.78125,
"learning_rate": 1.190407654309852e-05,
"epoch": 2.73536,
"step": 4275
},
{
"loss": 0.9159,
"grad_norm": 3.21875,
"learning_rate": 1.181921284099078e-05,
"epoch": 2.75136,
"step": 4300
},
{
"loss": 0.8778,
"grad_norm": 3.109375,
"learning_rate": 1.1734213418151022e-05,
"epoch": 2.76736,
"step": 4325
},
{
"loss": 0.89,
"grad_norm": 3.140625,
"learning_rate": 1.1649084615884472e-05,
"epoch": 2.78336,
"step": 4350
},
{
"loss": 0.8476,
"grad_norm": 3.5,
"learning_rate": 1.1563832785148583e-05,
"epoch": 2.79936,
"step": 4375
},
{
"loss": 0.8796,
"grad_norm": 3.28125,
"learning_rate": 1.1478464286079246e-05,
"epoch": 2.81536,
"step": 4400
},
{
"loss": 0.8856,
"grad_norm": 2.75,
"learning_rate": 1.1392985487516275e-05,
"epoch": 2.83136,
"step": 4425
},
{
"loss": 0.8882,
"grad_norm": 3.46875,
"learning_rate": 1.1307402766528293e-05,
"epoch": 2.84736,
"step": 4450
},
{
"loss": 0.8566,
"grad_norm": 3.375,
"learning_rate": 1.1221722507936949e-05,
"epoch": 2.86336,
"step": 4475
},
{
"loss": 0.8796,
"grad_norm": 3.09375,
"learning_rate": 1.1135951103840604e-05,
"epoch": 2.87936,
"step": 4500
},
{
"loss": 0.8752,
"grad_norm": 3.171875,
"learning_rate": 1.1050094953137444e-05,
"epoch": 2.89536,
"step": 4525
},
{
"loss": 0.8761,
"grad_norm": 3.1875,
"learning_rate": 1.0964160461048097e-05,
"epoch": 2.91136,
"step": 4550
},
{
"loss": 0.8575,
"grad_norm": 3.015625,
"learning_rate": 1.0878154038637789e-05,
"epoch": 2.92736,
"step": 4575
},
{
"loss": 0.8811,
"grad_norm": 3.234375,
"learning_rate": 1.0792082102338031e-05,
"epoch": 2.94336,
"step": 4600
},
{
"loss": 0.8709,
"grad_norm": 3.171875,
"learning_rate": 1.0705951073467948e-05,
"epoch": 2.95936,
"step": 4625
},
{
"loss": 0.8601,
"grad_norm": 2.90625,
"learning_rate": 1.0619767377755203e-05,
"epoch": 2.9753600000000002,
"step": 4650
},
{
"loss": 0.873,
"grad_norm": 3.1875,
"learning_rate": 1.0533537444856637e-05,
"epoch": 2.9913600000000002,
"step": 4675
},
{
"loss": 0.8749,
"grad_norm": 2.984375,
"learning_rate": 1.0447267707878552e-05,
"epoch": 3.00704,
"step": 4700
},
{
"loss": 0.8454,
"grad_norm": 3.265625,
"learning_rate": 1.036096460289682e-05,
"epoch": 3.02304,
"step": 4725
},
{
"loss": 0.8675,
"grad_norm": 3.296875,
"learning_rate": 1.0274634568476687e-05,
"epoch": 3.03904,
"step": 4750
},
{
"loss": 0.8341,
"grad_norm": 3.96875,
"learning_rate": 1.0188284045192448e-05,
"epoch": 3.05504,
"step": 4775
},
{
"loss": 0.8566,
"grad_norm": 3.09375,
"learning_rate": 1.0101919475146966e-05,
"epoch": 3.07104,
"step": 4800
},
{
"loss": 0.8482,
"grad_norm": 3.203125,
"learning_rate": 1.0015547301491028e-05,
"epoch": 3.08704,
"step": 4825
},
{
"loss": 0.8621,
"grad_norm": 3.140625,
"learning_rate": 9.929173967942693e-06,
"epoch": 3.10304,
"step": 4850
},
{
"loss": 0.8643,
"grad_norm": 3.234375,
"learning_rate": 9.842805918306557e-06,
"epoch": 3.11904,
"step": 4875
},
{
"loss": 0.8685,
"grad_norm": 3.734375,
"learning_rate": 9.756449595993005e-06,
"epoch": 3.13504,
"step": 4900
},
{
"loss": 0.8062,
"grad_norm": 3.109375,
"learning_rate": 9.67011144353752e-06,
"epoch": 3.15104,
"step": 4925
},
{
"loss": 0.8389,
"grad_norm": 3.046875,
"learning_rate": 9.583797902120036e-06,
"epoch": 3.16704,
"step": 4950
},
{
"loss": 0.8696,
"grad_norm": 3.375,
"learning_rate": 9.49751541108441e-06,
"epoch": 3.18304,
"step": 4975
},
{
"loss": 0.8514,
"grad_norm": 3.25,
"learning_rate": 9.411270407458008e-06,
"epoch": 3.19904,
"step": 5000
},
{
"loss": 0.8348,
"grad_norm": 2.984375,
"learning_rate": 9.325069325471483e-06,
"epoch": 3.21504,
"step": 5025
},
{
"loss": 0.8586,
"grad_norm": 3.59375,
"learning_rate": 9.238918596078747e-06,
"epoch": 3.23104,
"step": 5050
},
{
"loss": 0.8489,
"grad_norm": 2.890625,
"learning_rate": 9.152824646477218e-06,
"epoch": 3.24704,
"step": 5075
},
{
"loss": 0.8381,
"grad_norm": 3.359375,
"learning_rate": 9.066793899628293e-06,
"epoch": 3.26304,
"step": 5100
},
{
"loss": 0.8601,
"grad_norm": 3.234375,
"learning_rate": 8.980832773778194e-06,
"epoch": 3.27904,
"step": 5125
},
{
"loss": 0.8571,
"grad_norm": 3.0,
"learning_rate": 8.894947681979127e-06,
"epoch": 3.29504,
"step": 5150
},
{
"loss": 0.8373,
"grad_norm": 2.859375,
"learning_rate": 8.809145031610848e-06,
"epoch": 3.31104,
"step": 5175
},
{
"loss": 0.8695,
"grad_norm": 3.125,
"learning_rate": 8.723431223902641e-06,
"epoch": 3.32704,
"step": 5200
},
{
"loss": 0.8284,
"grad_norm": 3.078125,
"learning_rate": 8.637812653455771e-06,
"epoch": 3.3430400000000002,
"step": 5225
},
{
"loss": 0.8587,
"grad_norm": 3.578125,
"learning_rate": 8.552295707766406e-06,
"epoch": 3.3590400000000002,
"step": 5250
},
{
"loss": 0.827,
"grad_norm": 3.28125,
"learning_rate": 8.46688676674909e-06,
"epoch": 3.37504,
"step": 5275
},
{
"loss": 0.8279,
"grad_norm": 3.28125,
"learning_rate": 8.381592202260784e-06,
"epoch": 3.39104,
"step": 5300
},
{
"loss": 0.8452,
"grad_norm": 2.703125,
"learning_rate": 8.296418377625489e-06,
"epoch": 3.40704,
"step": 5325
},
{
"loss": 0.8531,
"grad_norm": 3.28125,
"learning_rate": 8.211371647159509e-06,
"epoch": 3.42304,
"step": 5350
},
{
"loss": 0.8326,
"grad_norm": 3.421875,
"learning_rate": 8.126458355697411e-06,
"epoch": 3.43904,
"step": 5375
},
{
"loss": 0.8611,
"grad_norm": 3.421875,
"learning_rate": 8.041684838118665e-06,
"epoch": 3.45504,
"step": 5400
},
{
"loss": 0.8159,
"grad_norm": 3.046875,
"learning_rate": 7.957057418875036e-06,
"epoch": 3.47104,
"step": 5425
},
{
"loss": 0.8669,
"grad_norm": 3.25,
"learning_rate": 7.87258241151875e-06,
"epoch": 3.48704,
"step": 5450
},
{
"loss": 0.8501,
"grad_norm": 3.265625,
"learning_rate": 7.78826611823149e-06,
"epoch": 3.50304,
"step": 5475
},
{
"loss": 0.834,
"grad_norm": 3.234375,
"learning_rate": 7.704114829354204e-06,
"epoch": 3.51904,
"step": 5500
},
{
"loss": 0.8511,
"grad_norm": 3.40625,
"learning_rate": 7.62013482291786e-06,
"epoch": 3.53504,
"step": 5525
},
{
"loss": 0.8386,
"grad_norm": 2.984375,
"learning_rate": 7.536332364175031e-06,
"epoch": 3.55104,
"step": 5550
},
{
"loss": 0.8765,
"grad_norm": 3.296875,
"learning_rate": 7.452713705132515e-06,
"epoch": 3.56704,
"step": 5575
},
{
"loss": 0.8525,
"grad_norm": 3.0,
"learning_rate": 7.369285084084898e-06,
"epoch": 3.58304,
"step": 5600
},
{
"loss": 0.8313,
"grad_norm": 3.265625,
"learning_rate": 7.286052725149149e-06,
"epoch": 3.59904,
"step": 5625
},
{
"loss": 0.8599,
"grad_norm": 2.875,
"learning_rate": 7.203022837800286e-06,
"epoch": 3.61504,
"step": 5650
},
{
"loss": 0.8342,
"grad_norm": 3.09375,
"learning_rate": 7.120201616408109e-06,
"epoch": 3.63104,
"step": 5675
},
{
"loss": 0.8704,
"grad_norm": 3.46875,
"learning_rate": 7.037595239775093e-06,
"epoch": 3.64704,
"step": 5700
},
{
"loss": 0.8358,
"grad_norm": 3.5625,
"learning_rate": 6.955209870675403e-06,
"epoch": 3.66304,
"step": 5725
},
{
"loss": 0.8066,
"grad_norm": 3.3125,
"learning_rate": 6.873051655395147e-06,
"epoch": 3.67904,
"step": 5750
},
{
"loss": 0.8621,
"grad_norm": 3.125,
"learning_rate": 6.791126723273813e-06,
"epoch": 3.69504,
"step": 5775
},
{
"loss": 0.8633,
"grad_norm": 3.484375,
"learning_rate": 6.709441186247028e-06,
"epoch": 3.71104,
"step": 5800
},
{
"loss": 0.8438,
"grad_norm": 3.234375,
"learning_rate": 6.628001138390547e-06,
"epoch": 3.72704,
"step": 5825
},
{
"loss": 0.8394,
"grad_norm": 3.140625,
"learning_rate": 6.546812655465637e-06,
"epoch": 3.74304,
"step": 5850
},
{
"loss": 0.8576,
"grad_norm": 3.546875,
"learning_rate": 6.465881794465787e-06,
"epoch": 3.75904,
"step": 5875
},
{
"loss": 0.8151,
"grad_norm": 3.109375,
"learning_rate": 6.3852145931648345e-06,
"epoch": 3.7750399999999997,
"step": 5900
},
{
"loss": 0.8627,
"grad_norm": 3.046875,
"learning_rate": 6.3048170696665245e-06,
"epoch": 3.7910399999999997,
"step": 5925
},
{
"loss": 0.8287,
"grad_norm": 2.71875,
"learning_rate": 6.224695221955528e-06,
"epoch": 3.8070399999999998,
"step": 5950
},
{
"loss": 0.8832,
"grad_norm": 3.28125,
"learning_rate": 6.144855027449986e-06,
"epoch": 3.8230399999999998,
"step": 5975
},
{
"loss": 0.8686,
"grad_norm": 4.0625,
"learning_rate": 6.065302442555543e-06,
"epoch": 3.83904,
"step": 6000
},
{
"loss": 0.8542,
"grad_norm": 3.53125,
"learning_rate": 5.986043402220991e-06,
"epoch": 3.85504,
"step": 6025
},
{
"loss": 0.8432,
"grad_norm": 3.28125,
"learning_rate": 5.9070838194954995e-06,
"epoch": 3.87104,
"step": 6050
},
{
"loss": 0.8515,
"grad_norm": 3.71875,
"learning_rate": 5.828429585087467e-06,
"epoch": 3.88704,
"step": 6075
},
{
"loss": 0.829,
"grad_norm": 3.09375,
"learning_rate": 5.7500865669250625e-06,
"epoch": 3.90304,
"step": 6100
},
{
"loss": 0.8172,
"grad_norm": 3.09375,
"learning_rate": 5.672060609718444e-06,
"epoch": 3.91904,
"step": 6125
},
{
"loss": 0.8301,
"grad_norm": 3.09375,
"learning_rate": 5.5943575345237235e-06,
"epoch": 3.93504,
"step": 6150
},
{
"loss": 0.8396,
"grad_norm": 2.890625,
"learning_rate": 5.516983138308689e-06,
"epoch": 3.95104,
"step": 6175
},
{
"loss": 0.8633,
"grad_norm": 3.15625,
"learning_rate": 5.439943193520343e-06,
"epoch": 3.96704,
"step": 6200
},
{
"loss": 0.8266,
"grad_norm": 3.0625,
"learning_rate": 5.363243447654225e-06,
"epoch": 3.98304,
"step": 6225
},
{
"loss": 0.8667,
"grad_norm": 3.203125,
"learning_rate": 5.2868896228256504e-06,
"epoch": 3.99904,
"step": 6250
},
{
"loss": 0.8279,
"grad_norm": 2.84375,
"learning_rate": 5.2108874153427995e-06,
"epoch": 4.01472,
"step": 6275
},
{
"loss": 0.8266,
"grad_norm": 3.234375,
"learning_rate": 5.1352424952817715e-06,
"epoch": 4.03072,
"step": 6300
},
{
"loss": 0.845,
"grad_norm": 3.375,
"learning_rate": 5.059960506063548e-06,
"epoch": 4.04672,
"step": 6325
},
{
"loss": 0.8284,
"grad_norm": 3.734375,
"learning_rate": 4.985047064032987e-06,
"epoch": 4.06272,
"step": 6350
},
{
"loss": 0.8448,
"grad_norm": 2.921875,
"learning_rate": 4.910507758039815e-06,
"epoch": 4.07872,
"step": 6375
},
{
"loss": 0.8555,
"grad_norm": 3.578125,
"learning_rate": 4.836348149021675e-06,
"epoch": 4.09472,
"step": 6400
},
{
"loss": 0.8053,
"grad_norm": 2.921875,
"learning_rate": 4.7625737695892715e-06,
"epoch": 4.11072,
"step": 6425
},
{
"loss": 0.7963,
"grad_norm": 3.0625,
"learning_rate": 4.689190123613587e-06,
"epoch": 4.12672,
"step": 6450
},
{
"loss": 0.8185,
"grad_norm": 3.3125,
"learning_rate": 4.6162026858153e-06,
"epoch": 4.14272,
"step": 6475
},
{
"loss": 0.8525,
"grad_norm": 3.125,
"learning_rate": 4.543616901356326e-06,
"epoch": 4.15872,
"step": 6500
},
{
"loss": 0.8708,
"grad_norm": 2.875,
"learning_rate": 4.471438185433602e-06,
"epoch": 4.17472,
"step": 6525
},
{
"loss": 0.8474,
"grad_norm": 2.875,
"learning_rate": 4.3996719228750824e-06,
"epoch": 4.19072,
"step": 6550
},
{
"loss": 0.8385,
"grad_norm": 3.4375,
"learning_rate": 4.328323467738011e-06,
"epoch": 4.20672,
"step": 6575
},
{
"loss": 0.851,
"grad_norm": 2.859375,
"learning_rate": 4.257398142909483e-06,
"epoch": 4.22272,
"step": 6600
},
{
"loss": 0.8429,
"grad_norm": 3.34375,
"learning_rate": 4.186901239709342e-06,
"epoch": 4.23872,
"step": 6625
},
{
"loss": 0.8465,
"grad_norm": 3.5,
"learning_rate": 4.116838017495426e-06,
"epoch": 4.25472,
"step": 6650
},
{
"loss": 0.8196,
"grad_norm": 3.34375,
"learning_rate": 4.047213703271188e-06,
"epoch": 4.27072,
"step": 6675
},
{
"loss": 0.8327,
"grad_norm": 2.6875,
"learning_rate": 3.9780334912957495e-06,
"epoch": 4.28672,
"step": 6700
},
{
"loss": 0.839,
"grad_norm": 3.09375,
"learning_rate": 3.909302542696384e-06,
"epoch": 4.30272,
"step": 6725
},
{
"loss": 0.8177,
"grad_norm": 3.359375,
"learning_rate": 3.841025985083475e-06,
"epoch": 4.31872,
"step": 6750
},
{
"loss": 0.8426,
"grad_norm": 3.390625,
"learning_rate": 3.7732089121679805e-06,
"epoch": 4.33472,
"step": 6775
},
{
"loss": 0.8315,
"grad_norm": 3.3125,
"learning_rate": 3.705856383381414e-06,
"epoch": 4.35072,
"step": 6800
},
{
"loss": 0.832,
"grad_norm": 2.96875,
"learning_rate": 3.6389734234983955e-06,
"epoch": 4.36672,
"step": 6825
},
{
"loss": 0.84,
"grad_norm": 3.125,
"learning_rate": 3.5725650222617745e-06,
"epoch": 4.38272,
"step": 6850
},
{
"loss": 0.8274,
"grad_norm": 3.28125,
"learning_rate": 3.506636134010395e-06,
"epoch": 4.39872,
"step": 6875
},
{
"loss": 0.84,
"grad_norm": 3.15625,
"learning_rate": 3.4411916773094546e-06,
"epoch": 4.41472,
"step": 6900
},
{
"loss": 0.8248,
"grad_norm": 3.3125,
"learning_rate": 3.3762365345835758e-06,
"epoch": 4.43072,
"step": 6925
},
{
"loss": 0.8523,
"grad_norm": 3.921875,
"learning_rate": 3.3117755517525563e-06,
"epoch": 4.44672,
"step": 6950
},
{
"loss": 0.8459,
"grad_norm": 3.59375,
"learning_rate": 3.247813537869837e-06,
"epoch": 4.46272,
"step": 6975
},
{
"loss": 0.8096,
"grad_norm": 3.765625,
"learning_rate": 3.184355264763731e-06,
"epoch": 4.47872,
"step": 7000
},
{
"loss": 0.8147,
"grad_norm": 2.921875,
"learning_rate": 3.121405466681433e-06,
"epoch": 4.49472,
"step": 7025
},
{
"loss": 0.8227,
"grad_norm": 3.21875,
"learning_rate": 3.0589688399358088e-06,
"epoch": 4.51072,
"step": 7050
},
{
"loss": 0.8486,
"grad_norm": 3.046875,
"learning_rate": 2.997050042555042e-06,
"epoch": 4.52672,
"step": 7075
},
{
"loss": 0.8253,
"grad_norm": 2.890625,
"learning_rate": 2.935653693935122e-06,
"epoch": 4.54272,
"step": 7100
},
{
"loss": 0.8226,
"grad_norm": 2.984375,
"learning_rate": 2.8747843744952287e-06,
"epoch": 4.55872,
"step": 7125
},
{
"loss": 0.815,
"grad_norm": 3.171875,
"learning_rate": 2.814446625335997e-06,
"epoch": 4.57472,
"step": 7150
},
{
"loss": 0.852,
"grad_norm": 3.15625,
"learning_rate": 2.7546449479007407e-06,
"epoch": 4.59072,
"step": 7175
},
{
"loss": 0.8407,
"grad_norm": 3.0625,
"learning_rate": 2.6953838036396317e-06,
"epoch": 4.60672,
"step": 7200
},
{
"loss": 0.8247,
"grad_norm": 3.078125,
"learning_rate": 2.6366676136768486e-06,
"epoch": 4.62272,
"step": 7225
},
{
"loss": 0.8355,
"grad_norm": 3.1875,
"learning_rate": 2.5785007584807475e-06,
"epoch": 4.63872,
"step": 7250
},
{
"loss": 0.8578,
"grad_norm": 3.5625,
"learning_rate": 2.5208875775370613e-06,
"epoch": 4.65472,
"step": 7275
},
{
"loss": 0.8669,
"grad_norm": 2.765625,
"learning_rate": 2.463832369025149e-06,
"epoch": 4.67072,
"step": 7300
},
{
"loss": 0.8225,
"grad_norm": 2.796875,
"learning_rate": 2.4073393894973497e-06,
"epoch": 4.68672,
"step": 7325
},
{
"loss": 0.856,
"grad_norm": 2.90625,
"learning_rate": 2.351412853561412e-06,
"epoch": 4.70272,
"step": 7350
},
{
"loss": 0.8247,
"grad_norm": 3.109375,
"learning_rate": 2.2960569335660686e-06,
"epoch": 4.71872,
"step": 7375
},
{
"loss": 0.866,
"grad_norm": 3.265625,
"learning_rate": 2.2412757592897694e-06,
"epoch": 4.73472,
"step": 7400
},
{
"loss": 0.8328,
"grad_norm": 3.09375,
"learning_rate": 2.1870734176325746e-06,
"epoch": 4.75072,
"step": 7425
},
{
"loss": 0.8494,
"grad_norm": 3.234375,
"learning_rate": 2.133453952311264e-06,
"epoch": 4.76672,
"step": 7450
},
{
"loss": 0.8513,
"grad_norm": 3.59375,
"learning_rate": 2.080421363557651e-06,
"epoch": 4.78272,
"step": 7475
},
{
"loss": 0.8259,
"grad_norm": 3.078125,
"learning_rate": 2.027979607820152e-06,
"epoch": 4.79872,
"step": 7500
},
{
"loss": 0.82,
"grad_norm": 3.0625,
"learning_rate": 1.9761325974686206e-06,
"epoch": 4.81472,
"step": 7525
},
{
"loss": 0.8382,
"grad_norm": 3.40625,
"learning_rate": 1.9248842005024716e-06,
"epoch": 4.83072,
"step": 7550
},
{
"loss": 0.8031,
"grad_norm": 3.609375,
"learning_rate": 1.874238240262103e-06,
"epoch": 4.84672,
"step": 7575
},
{
"loss": 0.8392,
"grad_norm": 3.203125,
"learning_rate": 1.8241984951436665e-06,
"epoch": 4.86272,
"step": 7600
},
{
"loss": 0.8595,
"grad_norm": 3.484375,
"learning_rate": 1.774768698317182e-06,
"epoch": 4.8787199999999995,
"step": 7625
},
{
"loss": 0.8151,
"grad_norm": 2.953125,
"learning_rate": 1.7259525374480258e-06,
"epoch": 4.8947199999999995,
"step": 7650
},
{
"loss": 0.855,
"grad_norm": 2.875,
"learning_rate": 1.677753654421821e-06,
"epoch": 4.9107199999999995,
"step": 7675
},
{
"loss": 0.8447,
"grad_norm": 3.484375,
"learning_rate": 1.6301756450727313e-06,
"epoch": 4.9267199999999995,
"step": 7700
},
{
"loss": 0.8401,
"grad_norm": 3.15625,
"learning_rate": 1.583222058915199e-06,
"epoch": 4.94272,
"step": 7725
},
{
"loss": 0.8119,
"grad_norm": 3.015625,
"learning_rate": 1.5368963988791451e-06,
"epoch": 4.95872,
"step": 7750
},
{
"loss": 0.7993,
"grad_norm": 3.328125,
"learning_rate": 1.4912021210486205e-06,
"epoch": 4.97472,
"step": 7775
},
{
"loss": 0.8639,
"grad_norm": 3.140625,
"learning_rate": 1.4461426344039797e-06,
"epoch": 4.99072,
"step": 7800
},
{
"loss": 0.8207,
"grad_norm": 2.75,
"learning_rate": 1.4017213005675568e-06,
"epoch": 5.0064,
"step": 7825
},
{
"loss": 0.8249,
"grad_norm": 3.21875,
"learning_rate": 1.3579414335528673e-06,
"epoch": 5.0224,
"step": 7850
},
{
"loss": 0.8443,
"grad_norm": 3.46875,
"learning_rate": 1.314806299517375e-06,
"epoch": 5.0384,
"step": 7875
},
{
"loss": 0.8236,
"grad_norm": 3.328125,
"learning_rate": 1.272319116518822e-06,
"epoch": 5.0544,
"step": 7900
},
{
"loss": 0.8655,
"grad_norm": 3.375,
"learning_rate": 1.2304830542751466e-06,
"epoch": 5.0704,
"step": 7925
},
{
"loss": 0.8309,
"grad_norm": 3.140625,
"learning_rate": 1.189301233928013e-06,
"epoch": 5.0864,
"step": 7950
},
{
"loss": 0.8442,
"grad_norm": 2.890625,
"learning_rate": 1.148776727809956e-06,
"epoch": 5.1024,
"step": 7975
},
{
"loss": 0.8459,
"grad_norm": 2.96875,
"learning_rate": 1.1089125592151817e-06,
"epoch": 5.1184,
"step": 8000
},
{
"loss": 0.8337,
"grad_norm": 3.171875,
"learning_rate": 1.0697117021740066e-06,
"epoch": 5.1344,
"step": 8025
},
{
"loss": 0.8385,
"grad_norm": 2.734375,
"learning_rate": 1.0311770812309873e-06,
"epoch": 5.1504,
"step": 8050
},
{
"loss": 0.8464,
"grad_norm": 3.53125,
"learning_rate": 9.933115712267393e-07,
"epoch": 5.1664,
"step": 8075
},
{
"loss": 0.8581,
"grad_norm": 3.4375,
"learning_rate": 9.561179970834588e-07,
"epoch": 5.1824,
"step": 8100
},
{
"loss": 0.8361,
"grad_norm": 3.125,
"learning_rate": 9.195991335941756e-07,
"epoch": 5.1984,
"step": 8125
},
{
"loss": 0.8178,
"grad_norm": 3.4375,
"learning_rate": 8.837577052157387e-07,
"epoch": 5.2144,
"step": 8150
},
{
"loss": 0.8338,
"grad_norm": 3.15625,
"learning_rate": 8.485963858655621e-07,
"epoch": 5.2304,
"step": 8175
},
{
"loss": 0.8407,
"grad_norm": 2.84375,
"learning_rate": 8.141177987221394e-07,
"epoch": 5.2464,
"step": 8200
},
{
"loss": 0.8289,
"grad_norm": 2.90625,
"learning_rate": 7.803245160293471e-07,
"epoch": 5.2624,
"step": 8225
},
{
"loss": 0.8319,
"grad_norm": 3.15625,
"learning_rate": 7.47219058904537e-07,
"epoch": 5.2783999999999995,
"step": 8250
},
{
"loss": 0.8479,
"grad_norm": 3.125,
"learning_rate": 7.1480389715046e-07,
"epoch": 5.2943999999999996,
"step": 8275
},
{
"loss": 0.8574,
"grad_norm": 3.1875,
"learning_rate": 6.830814490709992e-07,
"epoch": 5.3104,
"step": 8300
},
{
"loss": 0.8146,
"grad_norm": 2.96875,
"learning_rate": 6.520540812907627e-07,
"epoch": 5.3264,
"step": 8325
},
{
"loss": 0.8304,
"grad_norm": 3.234375,
"learning_rate": 6.217241085785186e-07,
"epoch": 5.3424,
"step": 8350
},
{
"loss": 0.8534,
"grad_norm": 2.921875,
"learning_rate": 5.920937936745064e-07,
"epoch": 5.3584,
"step": 8375
},
{
"loss": 0.8395,
"grad_norm": 2.875,
"learning_rate": 5.631653471216248e-07,
"epoch": 5.3744,
"step": 8400
},
{
"loss": 0.8414,
"grad_norm": 3.265625,
"learning_rate": 5.349409271005168e-07,
"epoch": 5.3904,
"step": 8425
},
{
"loss": 0.836,
"grad_norm": 2.625,
"learning_rate": 5.074226392685688e-07,
"epoch": 5.4064,
"step": 8450
},
{
"loss": 0.8273,
"grad_norm": 3.359375,
"learning_rate": 4.806125366028025e-07,
"epoch": 5.4224,
"step": 8475
},
{
"loss": 0.8205,
"grad_norm": 3.5625,
"learning_rate": 4.5451261924673083e-07,
"epoch": 5.4384,
"step": 8500
},
{
"loss": 0.8038,
"grad_norm": 3.265625,
"learning_rate": 4.291248343611287e-07,
"epoch": 5.4544,
"step": 8525
},
{
"loss": 0.8326,
"grad_norm": 2.96875,
"learning_rate": 4.0445107597877673e-07,
"epoch": 5.4704,
"step": 8550
},
{
"loss": 0.8312,
"grad_norm": 3.453125,
"learning_rate": 3.804931848631466e-07,
"epoch": 5.4864,
"step": 8575
},
{
"loss": 0.8174,
"grad_norm": 2.734375,
"learning_rate": 3.572529483710818e-07,
"epoch": 5.5024,
"step": 8600
},
{
"loss": 0.8211,
"grad_norm": 2.859375,
"learning_rate": 3.3473210031944813e-07,
"epoch": 5.5184,
"step": 8625
},
{
"loss": 0.8266,
"grad_norm": 3.515625,
"learning_rate": 3.129323208557888e-07,
"epoch": 5.5344,
"step": 8650
},
{
"loss": 0.874,
"grad_norm": 3.375,
"learning_rate": 2.918552363329752e-07,
"epoch": 5.5504,
"step": 8675
},
{
"loss": 0.8305,
"grad_norm": 3.046875,
"learning_rate": 2.7150241918787345e-07,
"epoch": 5.5664,
"step": 8700
},
{
"loss": 0.8113,
"grad_norm": 3.140625,
"learning_rate": 2.5187538782403653e-07,
"epoch": 5.5824,
"step": 8725
},
{
"loss": 0.8302,
"grad_norm": 3.3125,
"learning_rate": 2.3297560649842567e-07,
"epoch": 5.5984,
"step": 8750
},
{
"loss": 0.8244,
"grad_norm": 3.03125,
"learning_rate": 2.1480448521216667e-07,
"epoch": 5.6144,
"step": 8775
},
{
"loss": 0.8486,
"grad_norm": 4.3125,
"learning_rate": 1.97363379605362e-07,
"epoch": 5.6304,
"step": 8800
},
{
"loss": 0.8252,
"grad_norm": 2.890625,
"learning_rate": 1.806535908559548e-07,
"epoch": 5.6464,
"step": 8825
},
{
"loss": 0.8284,
"grad_norm": 3.25,
"learning_rate": 1.6467636558265222e-07,
"epoch": 5.6624,
"step": 8850
},
{
"loss": 0.8403,
"grad_norm": 3.390625,
"learning_rate": 1.494328957519242e-07,
"epoch": 5.6784,
"step": 8875
},
{
"loss": 0.835,
"grad_norm": 3.234375,
"learning_rate": 1.3492431858908027e-07,
"epoch": 5.6944,
"step": 8900
},
{
"loss": 0.8642,
"grad_norm": 3.390625,
"learning_rate": 1.2115171649342504e-07,
"epoch": 5.7104,
"step": 8925
},
{
"loss": 0.8183,
"grad_norm": 3.609375,
"learning_rate": 1.0811611695750513e-07,
"epoch": 5.7264,
"step": 8950
},
{
"loss": 0.8225,
"grad_norm": 3.0625,
"learning_rate": 9.581849249045817e-08,
"epoch": 5.7424,
"step": 8975
},
{
"loss": 0.8629,
"grad_norm": 3.171875,
"learning_rate": 8.425976054545981e-08,
"epoch": 5.7584,
"step": 9000
},
{
"loss": 0.8439,
"grad_norm": 3.375,
"learning_rate": 7.344078345127292e-08,
"epoch": 5.7744,
"step": 9025
},
{
"loss": 0.8171,
"grad_norm": 2.78125,
"learning_rate": 6.336236834791898e-08,
"epoch": 5.7904,
"step": 9050
},
{
"loss": 0.8495,
"grad_norm": 3.265625,
"learning_rate": 5.4025267126459615e-08,
"epoch": 5.8064,
"step": 9075
},
{
"loss": 0.8467,
"grad_norm": 2.828125,
"learning_rate": 4.54301763729037e-08,
"epoch": 5.8224,
"step": 9100
},
{
"loss": 0.8243,
"grad_norm": 3.0625,
"learning_rate": 3.757773731624115e-08,
"epoch": 5.8384,
"step": 9125
},
{
"loss": 0.8087,
"grad_norm": 2.609375,
"learning_rate": 3.046853578059783e-08,
"epoch": 5.8544,
"step": 9150
},
{
"loss": 0.8313,
"grad_norm": 3.0625,
"learning_rate": 2.4103102141539437e-08,
"epoch": 5.8704,
"step": 9175
},
{
"loss": 0.8222,
"grad_norm": 3.203125,
"learning_rate": 1.8481911286496458e-08,
"epoch": 5.8864,
"step": 9200
},
{
"loss": 0.8346,
"grad_norm": 3.5,
"learning_rate": 1.3605382579339188e-08,
"epoch": 5.9024,
"step": 9225
},
{
"loss": 0.801,
"grad_norm": 3.328125,
"learning_rate": 9.473879829091648e-09,
"epoch": 5.9184,
"step": 9250
},
{
"loss": 0.8376,
"grad_norm": 2.546875,
"learning_rate": 6.087711262785512e-09,
"epoch": 5.9344,
"step": 9275
},
{
"loss": 0.8218,
"grad_norm": 3.125,
"learning_rate": 3.4471295024707253e-09,
"epoch": 5.9504,
"step": 9300
},
{
"loss": 0.8291,
"grad_norm": 3.125,
"learning_rate": 1.5523315463672472e-09,
"epoch": 5.9664,
"step": 9325
},
{
"loss": 0.8432,
"grad_norm": 3.234375,
"learning_rate": 4.034587541656976e-10,
"epoch": 5.9824,
"step": 9350
},
{
"loss": 0.8457,
"grad_norm": 3.28125,
"learning_rate": 5.968364835684526e-13,
"epoch": 5.9984,
"step": 9375
},
{
"train_runtime": 4898.6155,
"train_samples_per_second": 61.242,
"train_steps_per_second": 1.914,
"total_flos": 1.092453917934551e+18,
"train_loss": 1.0109390736897788,
"epoch": 5.9984,
"step": 9375
}
]
}