{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 5.0,
"eval_steps": 500,
"global_step": 54900,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.04553734061930783,
"grad_norm": 18.591150283813477,
"learning_rate": 2.9726775956284154e-05,
"loss": 2.5583,
"step": 500
},
{
"epoch": 0.09107468123861566,
"grad_norm": 15.338476181030273,
"learning_rate": 2.9453551912568308e-05,
"loss": 1.6841,
"step": 1000
},
{
"epoch": 0.1366120218579235,
"grad_norm": 14.975069999694824,
"learning_rate": 2.9180327868852458e-05,
"loss": 1.5423,
"step": 1500
},
{
"epoch": 0.18214936247723132,
"grad_norm": 17.511743545532227,
"learning_rate": 2.890710382513661e-05,
"loss": 1.4113,
"step": 2000
},
{
"epoch": 0.22768670309653916,
"grad_norm": 18.18260955810547,
"learning_rate": 2.8633879781420765e-05,
"loss": 1.3533,
"step": 2500
},
{
"epoch": 0.273224043715847,
"grad_norm": 16.508344650268555,
"learning_rate": 2.836065573770492e-05,
"loss": 1.299,
"step": 3000
},
{
"epoch": 0.31876138433515483,
"grad_norm": 25.282175064086914,
"learning_rate": 2.8087431693989072e-05,
"loss": 1.282,
"step": 3500
},
{
"epoch": 0.36429872495446264,
"grad_norm": 24.085981369018555,
"learning_rate": 2.7814207650273226e-05,
"loss": 1.2146,
"step": 4000
},
{
"epoch": 0.4098360655737705,
"grad_norm": 12.455412864685059,
"learning_rate": 2.754098360655738e-05,
"loss": 1.2332,
"step": 4500
},
{
"epoch": 0.4553734061930783,
"grad_norm": 10.222314834594727,
"learning_rate": 2.7267759562841533e-05,
"loss": 1.1607,
"step": 5000
},
{
"epoch": 0.5009107468123861,
"grad_norm": 10.9562406539917,
"learning_rate": 2.6994535519125683e-05,
"loss": 1.1293,
"step": 5500
},
{
"epoch": 0.546448087431694,
"grad_norm": 16.84743309020996,
"learning_rate": 2.6721311475409837e-05,
"loss": 1.129,
"step": 6000
},
{
"epoch": 0.5919854280510018,
"grad_norm": 21.29459571838379,
"learning_rate": 2.644808743169399e-05,
"loss": 1.1002,
"step": 6500
},
{
"epoch": 0.6375227686703097,
"grad_norm": 25.021621704101562,
"learning_rate": 2.6174863387978144e-05,
"loss": 1.0397,
"step": 7000
},
{
"epoch": 0.6830601092896175,
"grad_norm": 11.044733047485352,
"learning_rate": 2.5901639344262294e-05,
"loss": 1.0806,
"step": 7500
},
{
"epoch": 0.7285974499089253,
"grad_norm": 10.266705513000488,
"learning_rate": 2.5628415300546447e-05,
"loss": 1.047,
"step": 8000
},
{
"epoch": 0.7741347905282332,
"grad_norm": 13.432859420776367,
"learning_rate": 2.53551912568306e-05,
"loss": 1.0566,
"step": 8500
},
{
"epoch": 0.819672131147541,
"grad_norm": 25.581703186035156,
"learning_rate": 2.5081967213114754e-05,
"loss": 1.0277,
"step": 9000
},
{
"epoch": 0.8652094717668488,
"grad_norm": 14.602388381958008,
"learning_rate": 2.4808743169398908e-05,
"loss": 1.0154,
"step": 9500
},
{
"epoch": 0.9107468123861566,
"grad_norm": 17.17100715637207,
"learning_rate": 2.453551912568306e-05,
"loss": 1.0149,
"step": 10000
},
{
"epoch": 0.9562841530054644,
"grad_norm": 10.567970275878906,
"learning_rate": 2.4262295081967215e-05,
"loss": 1.007,
"step": 10500
},
{
"epoch": 1.0018214936247722,
"grad_norm": 22.679931640625,
"learning_rate": 2.398907103825137e-05,
"loss": 0.995,
"step": 11000
},
{
"epoch": 1.04735883424408,
"grad_norm": 12.259163856506348,
"learning_rate": 2.371584699453552e-05,
"loss": 0.7177,
"step": 11500
},
{
"epoch": 1.092896174863388,
"grad_norm": 18.350513458251953,
"learning_rate": 2.3442622950819672e-05,
"loss": 0.7203,
"step": 12000
},
{
"epoch": 1.138433515482696,
"grad_norm": 21.86613655090332,
"learning_rate": 2.3169398907103826e-05,
"loss": 0.7585,
"step": 12500
},
{
"epoch": 1.1839708561020037,
"grad_norm": 18.372119903564453,
"learning_rate": 2.289617486338798e-05,
"loss": 0.7324,
"step": 13000
},
{
"epoch": 1.2295081967213115,
"grad_norm": 25.320674896240234,
"learning_rate": 2.262295081967213e-05,
"loss": 0.7241,
"step": 13500
},
{
"epoch": 1.2750455373406193,
"grad_norm": 10.435434341430664,
"learning_rate": 2.2349726775956283e-05,
"loss": 0.7322,
"step": 14000
},
{
"epoch": 1.3205828779599271,
"grad_norm": 29.3476505279541,
"learning_rate": 2.2076502732240437e-05,
"loss": 0.745,
"step": 14500
},
{
"epoch": 1.366120218579235,
"grad_norm": 8.349024772644043,
"learning_rate": 2.180327868852459e-05,
"loss": 0.7543,
"step": 15000
},
{
"epoch": 1.4116575591985427,
"grad_norm": 10.698665618896484,
"learning_rate": 2.153005464480874e-05,
"loss": 0.7475,
"step": 15500
},
{
"epoch": 1.4571948998178508,
"grad_norm": 10.810667037963867,
"learning_rate": 2.1256830601092897e-05,
"loss": 0.7336,
"step": 16000
},
{
"epoch": 1.5027322404371586,
"grad_norm": 14.816246032714844,
"learning_rate": 2.098360655737705e-05,
"loss": 0.7393,
"step": 16500
},
{
"epoch": 1.5482695810564664,
"grad_norm": 11.557518005371094,
"learning_rate": 2.0710382513661204e-05,
"loss": 0.7333,
"step": 17000
},
{
"epoch": 1.5938069216757742,
"grad_norm": 16.669546127319336,
"learning_rate": 2.0437158469945358e-05,
"loss": 0.7255,
"step": 17500
},
{
"epoch": 1.639344262295082,
"grad_norm": 14.28069019317627,
"learning_rate": 2.0163934426229508e-05,
"loss": 0.7369,
"step": 18000
},
{
"epoch": 1.6848816029143898,
"grad_norm": 22.36986541748047,
"learning_rate": 1.9890710382513662e-05,
"loss": 0.7287,
"step": 18500
},
{
"epoch": 1.7304189435336976,
"grad_norm": 11.109138488769531,
"learning_rate": 1.9617486338797815e-05,
"loss": 0.7406,
"step": 19000
},
{
"epoch": 1.7759562841530054,
"grad_norm": 16.205177307128906,
"learning_rate": 1.934426229508197e-05,
"loss": 0.734,
"step": 19500
},
{
"epoch": 1.8214936247723132,
"grad_norm": 16.029556274414062,
"learning_rate": 1.907103825136612e-05,
"loss": 0.7261,
"step": 20000
},
{
"epoch": 1.867030965391621,
"grad_norm": 17.520740509033203,
"learning_rate": 1.8797814207650273e-05,
"loss": 0.7289,
"step": 20500
},
{
"epoch": 1.9125683060109289,
"grad_norm": 31.18975257873535,
"learning_rate": 1.8524590163934426e-05,
"loss": 0.7012,
"step": 21000
},
{
"epoch": 1.9581056466302367,
"grad_norm": 11.743003845214844,
"learning_rate": 1.825136612021858e-05,
"loss": 0.7185,
"step": 21500
},
{
"epoch": 2.0036429872495445,
"grad_norm": 13.794853210449219,
"learning_rate": 1.7978142076502733e-05,
"loss": 0.6816,
"step": 22000
},
{
"epoch": 2.0491803278688523,
"grad_norm": 28.844125747680664,
"learning_rate": 1.7704918032786887e-05,
"loss": 0.4469,
"step": 22500
},
{
"epoch": 2.09471766848816,
"grad_norm": 5.327408790588379,
"learning_rate": 1.743169398907104e-05,
"loss": 0.4577,
"step": 23000
},
{
"epoch": 2.1402550091074684,
"grad_norm": 29.065576553344727,
"learning_rate": 1.7158469945355194e-05,
"loss": 0.4617,
"step": 23500
},
{
"epoch": 2.185792349726776,
"grad_norm": 7.028495788574219,
"learning_rate": 1.6885245901639344e-05,
"loss": 0.4542,
"step": 24000
},
{
"epoch": 2.231329690346084,
"grad_norm": 17.614421844482422,
"learning_rate": 1.6612021857923497e-05,
"loss": 0.4809,
"step": 24500
},
{
"epoch": 2.276867030965392,
"grad_norm": 16.79141616821289,
"learning_rate": 1.633879781420765e-05,
"loss": 0.4717,
"step": 25000
},
{
"epoch": 2.3224043715846996,
"grad_norm": 29.323766708374023,
"learning_rate": 1.6065573770491805e-05,
"loss": 0.4758,
"step": 25500
},
{
"epoch": 2.3679417122040074,
"grad_norm": 19.425329208374023,
"learning_rate": 1.5792349726775955e-05,
"loss": 0.4679,
"step": 26000
},
{
"epoch": 2.413479052823315,
"grad_norm": 5.9673380851745605,
"learning_rate": 1.5519125683060108e-05,
"loss": 0.4885,
"step": 26500
},
{
"epoch": 2.459016393442623,
"grad_norm": 12.733803749084473,
"learning_rate": 1.5245901639344264e-05,
"loss": 0.4811,
"step": 27000
},
{
"epoch": 2.504553734061931,
"grad_norm": 22.212753295898438,
"learning_rate": 1.4972677595628415e-05,
"loss": 0.4613,
"step": 27500
},
{
"epoch": 2.5500910746812386,
"grad_norm": 18.736804962158203,
"learning_rate": 1.4699453551912569e-05,
"loss": 0.4786,
"step": 28000
},
{
"epoch": 2.5956284153005464,
"grad_norm": 20.595645904541016,
"learning_rate": 1.442622950819672e-05,
"loss": 0.4997,
"step": 28500
},
{
"epoch": 2.6411657559198543,
"grad_norm": 18.30236053466797,
"learning_rate": 1.4153005464480874e-05,
"loss": 0.4792,
"step": 29000
},
{
"epoch": 2.686703096539162,
"grad_norm": 19.671579360961914,
"learning_rate": 1.3879781420765026e-05,
"loss": 0.4556,
"step": 29500
},
{
"epoch": 2.73224043715847,
"grad_norm": 14.210960388183594,
"learning_rate": 1.3606557377049181e-05,
"loss": 0.4737,
"step": 30000
},
{
"epoch": 2.7777777777777777,
"grad_norm": 15.706062316894531,
"learning_rate": 1.3333333333333333e-05,
"loss": 0.4773,
"step": 30500
},
{
"epoch": 2.8233151183970855,
"grad_norm": 2.883192300796509,
"learning_rate": 1.3060109289617487e-05,
"loss": 0.4734,
"step": 31000
},
{
"epoch": 2.8688524590163933,
"grad_norm": 12.449463844299316,
"learning_rate": 1.2786885245901639e-05,
"loss": 0.4579,
"step": 31500
},
{
"epoch": 2.9143897996357016,
"grad_norm": 19.76999855041504,
"learning_rate": 1.2513661202185792e-05,
"loss": 0.4772,
"step": 32000
},
{
"epoch": 2.959927140255009,
"grad_norm": 38.29131317138672,
"learning_rate": 1.2240437158469946e-05,
"loss": 0.4689,
"step": 32500
},
{
"epoch": 3.0054644808743167,
"grad_norm": 4.822471618652344,
"learning_rate": 1.19672131147541e-05,
"loss": 0.4374,
"step": 33000
},
{
"epoch": 3.051001821493625,
"grad_norm": 21.936952590942383,
"learning_rate": 1.1693989071038253e-05,
"loss": 0.2893,
"step": 33500
},
{
"epoch": 3.096539162112933,
"grad_norm": 25.08475685119629,
"learning_rate": 1.1420765027322405e-05,
"loss": 0.2888,
"step": 34000
},
{
"epoch": 3.1420765027322406,
"grad_norm": 13.904535293579102,
"learning_rate": 1.1147540983606558e-05,
"loss": 0.2915,
"step": 34500
},
{
"epoch": 3.1876138433515484,
"grad_norm": 3.9972009658813477,
"learning_rate": 1.087431693989071e-05,
"loss": 0.3081,
"step": 35000
},
{
"epoch": 3.233151183970856,
"grad_norm": 30.3763370513916,
"learning_rate": 1.0601092896174864e-05,
"loss": 0.3158,
"step": 35500
},
{
"epoch": 3.278688524590164,
"grad_norm": 4.742646217346191,
"learning_rate": 1.0327868852459017e-05,
"loss": 0.2869,
"step": 36000
},
{
"epoch": 3.324225865209472,
"grad_norm": 42.35346984863281,
"learning_rate": 1.005464480874317e-05,
"loss": 0.2932,
"step": 36500
},
{
"epoch": 3.3697632058287796,
"grad_norm": 13.786494255065918,
"learning_rate": 9.781420765027323e-06,
"loss": 0.2976,
"step": 37000
},
{
"epoch": 3.4153005464480874,
"grad_norm": 11.841999053955078,
"learning_rate": 9.508196721311476e-06,
"loss": 0.2938,
"step": 37500
},
{
"epoch": 3.4608378870673953,
"grad_norm": 29.609792709350586,
"learning_rate": 9.234972677595628e-06,
"loss": 0.3006,
"step": 38000
},
{
"epoch": 3.506375227686703,
"grad_norm": 18.663509368896484,
"learning_rate": 8.961748633879782e-06,
"loss": 0.3039,
"step": 38500
},
{
"epoch": 3.551912568306011,
"grad_norm": 1.889535903930664,
"learning_rate": 8.688524590163935e-06,
"loss": 0.3082,
"step": 39000
},
{
"epoch": 3.5974499089253187,
"grad_norm": 19.884906768798828,
"learning_rate": 8.415300546448089e-06,
"loss": 0.3085,
"step": 39500
},
{
"epoch": 3.6429872495446265,
"grad_norm": 12.94946002960205,
"learning_rate": 8.14207650273224e-06,
"loss": 0.3086,
"step": 40000
},
{
"epoch": 3.6885245901639343,
"grad_norm": 27.61246681213379,
"learning_rate": 7.868852459016394e-06,
"loss": 0.3109,
"step": 40500
},
{
"epoch": 3.734061930783242,
"grad_norm": 14.3275146484375,
"learning_rate": 7.595628415300546e-06,
"loss": 0.3025,
"step": 41000
},
{
"epoch": 3.7795992714025504,
"grad_norm": 6.992326259613037,
"learning_rate": 7.3224043715847e-06,
"loss": 0.3108,
"step": 41500
},
{
"epoch": 3.8251366120218577,
"grad_norm": 5.084273815155029,
"learning_rate": 7.049180327868853e-06,
"loss": 0.3012,
"step": 42000
},
{
"epoch": 3.870673952641166,
"grad_norm": 27.192960739135742,
"learning_rate": 6.775956284153006e-06,
"loss": 0.2996,
"step": 42500
},
{
"epoch": 3.9162112932604733,
"grad_norm": 42.71589279174805,
"learning_rate": 6.502732240437159e-06,
"loss": 0.31,
"step": 43000
},
{
"epoch": 3.9617486338797816,
"grad_norm": 11.759929656982422,
"learning_rate": 6.229508196721312e-06,
"loss": 0.2953,
"step": 43500
},
{
"epoch": 4.007285974499089,
"grad_norm": 30.232032775878906,
"learning_rate": 5.956284153005465e-06,
"loss": 0.2612,
"step": 44000
},
{
"epoch": 4.052823315118397,
"grad_norm": 62.14134979248047,
"learning_rate": 5.683060109289618e-06,
"loss": 0.1843,
"step": 44500
},
{
"epoch": 4.098360655737705,
"grad_norm": 7.216005802154541,
"learning_rate": 5.409836065573771e-06,
"loss": 0.1859,
"step": 45000
},
{
"epoch": 4.143897996357013,
"grad_norm": 17.130714416503906,
"learning_rate": 5.136612021857924e-06,
"loss": 0.2106,
"step": 45500
},
{
"epoch": 4.18943533697632,
"grad_norm": 4.9142985343933105,
"learning_rate": 4.863387978142076e-06,
"loss": 0.1932,
"step": 46000
},
{
"epoch": 4.2349726775956285,
"grad_norm": 13.510531425476074,
"learning_rate": 4.59016393442623e-06,
"loss": 0.2081,
"step": 46500
},
{
"epoch": 4.280510018214937,
"grad_norm": 5.9509172439575195,
"learning_rate": 4.316939890710383e-06,
"loss": 0.1861,
"step": 47000
},
{
"epoch": 4.326047358834244,
"grad_norm": 7.964664936065674,
"learning_rate": 4.043715846994535e-06,
"loss": 0.1863,
"step": 47500
},
{
"epoch": 4.371584699453552,
"grad_norm": 25.88454246520996,
"learning_rate": 3.770491803278689e-06,
"loss": 0.1938,
"step": 48000
},
{
"epoch": 4.41712204007286,
"grad_norm": 2.440138339996338,
"learning_rate": 3.4972677595628415e-06,
"loss": 0.1963,
"step": 48500
},
{
"epoch": 4.462659380692168,
"grad_norm": 6.778371334075928,
"learning_rate": 3.2240437158469947e-06,
"loss": 0.1829,
"step": 49000
},
{
"epoch": 4.508196721311475,
"grad_norm": 31.79254722595215,
"learning_rate": 2.9508196721311474e-06,
"loss": 0.1897,
"step": 49500
},
{
"epoch": 4.553734061930784,
"grad_norm": 42.17744064331055,
"learning_rate": 2.6775956284153005e-06,
"loss": 0.1921,
"step": 50000
},
{
"epoch": 4.599271402550091,
"grad_norm": 0.5351423621177673,
"learning_rate": 2.4043715846994536e-06,
"loss": 0.2084,
"step": 50500
},
{
"epoch": 4.644808743169399,
"grad_norm": 19.15989112854004,
"learning_rate": 2.1311475409836063e-06,
"loss": 0.1898,
"step": 51000
},
{
"epoch": 4.6903460837887065,
"grad_norm": 3.171184539794922,
"learning_rate": 1.8579234972677596e-06,
"loss": 0.1928,
"step": 51500
},
{
"epoch": 4.735883424408015,
"grad_norm": 61.36581802368164,
"learning_rate": 1.5846994535519126e-06,
"loss": 0.198,
"step": 52000
},
{
"epoch": 4.781420765027322,
"grad_norm": 6.246685028076172,
"learning_rate": 1.3114754098360657e-06,
"loss": 0.1794,
"step": 52500
},
{
"epoch": 4.82695810564663,
"grad_norm": 10.183932304382324,
"learning_rate": 1.0382513661202186e-06,
"loss": 0.1845,
"step": 53000
},
{
"epoch": 4.872495446265938,
"grad_norm": 29.7720947265625,
"learning_rate": 7.650273224043716e-07,
"loss": 0.1889,
"step": 53500
},
{
"epoch": 4.918032786885246,
"grad_norm": 9.350275039672852,
"learning_rate": 4.918032786885246e-07,
"loss": 0.1941,
"step": 54000
},
{
"epoch": 4.963570127504553,
"grad_norm": 16.810739517211914,
"learning_rate": 2.185792349726776e-07,
"loss": 0.2008,
"step": 54500
},
{
"epoch": 5.0,
"step": 54900,
"total_flos": 1.2910084832623104e+17,
"train_loss": 0.07113273273183131,
"train_runtime": 1918.8519,
"train_samples_per_second": 343.315,
"train_steps_per_second": 28.611
}
],
"logging_steps": 500,
"max_steps": 54900,
"num_input_tokens_seen": 0,
"num_train_epochs": 5,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 1.2910084832623104e+17,
"train_batch_size": 12,
"trial_name": null,
"trial_params": null
}