full_deepseek-1.3-tc / trainer_state.json
Yhhhhhhhhh's picture
Upload folder using huggingface_hub
4f17f74 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.9921414538310414,
"eval_steps": 50000,
"global_step": 1016,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.03929273084479371,
"grad_norm": 4.250463008880615,
"learning_rate": 1.6129032258064516e-06,
"loss": 0.8414,
"step": 10
},
{
"epoch": 0.07858546168958742,
"grad_norm": 2.718027114868164,
"learning_rate": 3.225806451612903e-06,
"loss": 0.8388,
"step": 20
},
{
"epoch": 0.11787819253438114,
"grad_norm": 1.6809030771255493,
"learning_rate": 4.838709677419355e-06,
"loss": 0.6738,
"step": 30
},
{
"epoch": 0.15717092337917485,
"grad_norm": 1.230251431465149,
"learning_rate": 4.998970106077018e-06,
"loss": 0.5541,
"step": 40
},
{
"epoch": 0.19646365422396855,
"grad_norm": 1.1485263109207153,
"learning_rate": 4.99541106832608e-06,
"loss": 0.5261,
"step": 50
},
{
"epoch": 0.2357563850687623,
"grad_norm": 1.0606619119644165,
"learning_rate": 4.989313791265896e-06,
"loss": 0.6025,
"step": 60
},
{
"epoch": 0.275049115913556,
"grad_norm": 1.3169667720794678,
"learning_rate": 4.9806844768198724e-06,
"loss": 0.571,
"step": 70
},
{
"epoch": 0.3143418467583497,
"grad_norm": 1.87986159324646,
"learning_rate": 4.969531902405652e-06,
"loss": 0.5795,
"step": 80
},
{
"epoch": 0.35363457760314343,
"grad_norm": 1.7818858623504639,
"learning_rate": 4.955867412007052e-06,
"loss": 0.5766,
"step": 90
},
{
"epoch": 0.3929273084479371,
"grad_norm": 1.3598214387893677,
"learning_rate": 4.939704904635388e-06,
"loss": 0.5421,
"step": 100
},
{
"epoch": 0.43222003929273084,
"grad_norm": 1.1836585998535156,
"learning_rate": 4.921060820191909e-06,
"loss": 0.5133,
"step": 110
},
{
"epoch": 0.4715127701375246,
"grad_norm": 1.4962817430496216,
"learning_rate": 4.8999541227457514e-06,
"loss": 0.5643,
"step": 120
},
{
"epoch": 0.5108055009823183,
"grad_norm": 1.0972118377685547,
"learning_rate": 4.8764062812443875e-06,
"loss": 0.5117,
"step": 130
},
{
"epoch": 0.550098231827112,
"grad_norm": 1.249541997909546,
"learning_rate": 4.8504412476762105e-06,
"loss": 0.5394,
"step": 140
},
{
"epoch": 0.5893909626719057,
"grad_norm": 1.2363935708999634,
"learning_rate": 4.822085432707465e-06,
"loss": 0.5244,
"step": 150
},
{
"epoch": 0.6286836935166994,
"grad_norm": 0.9700586199760437,
"learning_rate": 4.791367678818299e-06,
"loss": 0.4974,
"step": 160
},
{
"epoch": 0.6679764243614931,
"grad_norm": 1.324613332748413,
"learning_rate": 4.758319230965267e-06,
"loss": 0.51,
"step": 170
},
{
"epoch": 0.7072691552062869,
"grad_norm": 1.7756155729293823,
"learning_rate": 4.72297370480012e-06,
"loss": 0.4972,
"step": 180
},
{
"epoch": 0.7465618860510805,
"grad_norm": 1.4743613004684448,
"learning_rate": 4.685367052477218e-06,
"loss": 0.4932,
"step": 190
},
{
"epoch": 0.7858546168958742,
"grad_norm": 1.4711652994155884,
"learning_rate": 4.645537526084331e-06,
"loss": 0.472,
"step": 200
},
{
"epoch": 0.825147347740668,
"grad_norm": 1.2896119356155396,
"learning_rate": 4.603525638734049e-06,
"loss": 0.5046,
"step": 210
},
{
"epoch": 0.8644400785854617,
"grad_norm": 1.195989966392517,
"learning_rate": 4.559374123355337e-06,
"loss": 0.5002,
"step": 220
},
{
"epoch": 0.9037328094302554,
"grad_norm": 1.3485041856765747,
"learning_rate": 4.5131278892272e-06,
"loss": 0.4411,
"step": 230
},
{
"epoch": 0.9430255402750491,
"grad_norm": 1.3075335025787354,
"learning_rate": 4.46483397629863e-06,
"loss": 0.4711,
"step": 240
},
{
"epoch": 0.9823182711198428,
"grad_norm": 1.2377111911773682,
"learning_rate": 4.414541507341323e-06,
"loss": 0.4635,
"step": 250
},
{
"epoch": 1.0216110019646365,
"grad_norm": 1.4121922254562378,
"learning_rate": 4.362301637983815e-06,
"loss": 0.4179,
"step": 260
},
{
"epoch": 1.0609037328094302,
"grad_norm": 1.6095746755599976,
"learning_rate": 4.308167504677893e-06,
"loss": 0.4041,
"step": 270
},
{
"epoch": 1.1001964636542239,
"grad_norm": 1.8479869365692139,
"learning_rate": 4.2521941706501625e-06,
"loss": 0.3847,
"step": 280
},
{
"epoch": 1.1394891944990178,
"grad_norm": 1.4423387050628662,
"learning_rate": 4.194438569893784e-06,
"loss": 0.3876,
"step": 290
},
{
"epoch": 1.1787819253438114,
"grad_norm": 1.4060430526733398,
"learning_rate": 4.134959449257335e-06,
"loss": 0.3743,
"step": 300
},
{
"epoch": 1.218074656188605,
"grad_norm": 1.1776665449142456,
"learning_rate": 4.0738173086896995e-06,
"loss": 0.3689,
"step": 310
},
{
"epoch": 1.2573673870333988,
"grad_norm": 1.248273491859436,
"learning_rate": 4.011074339701772e-06,
"loss": 0.3813,
"step": 320
},
{
"epoch": 1.2966601178781925,
"grad_norm": 1.229260802268982,
"learning_rate": 3.946794362107564e-06,
"loss": 0.3786,
"step": 330
},
{
"epoch": 1.3359528487229864,
"grad_norm": 1.7943792343139648,
"learning_rate": 3.8810427591090635e-06,
"loss": 0.4117,
"step": 340
},
{
"epoch": 1.37524557956778,
"grad_norm": 1.9054116010665894,
"learning_rate": 3.813886410790879e-06,
"loss": 0.3861,
"step": 350
},
{
"epoch": 1.4145383104125737,
"grad_norm": 1.1942715644836426,
"learning_rate": 3.7453936260922983e-06,
"loss": 0.3636,
"step": 360
},
{
"epoch": 1.4538310412573674,
"grad_norm": 1.5429688692092896,
"learning_rate": 3.675634073325981e-06,
"loss": 0.4099,
"step": 370
},
{
"epoch": 1.493123772102161,
"grad_norm": 1.4392579793930054,
"learning_rate": 3.6046787093139415e-06,
"loss": 0.381,
"step": 380
},
{
"epoch": 1.5324165029469548,
"grad_norm": 1.8450443744659424,
"learning_rate": 3.5325997072129066e-06,
"loss": 0.3769,
"step": 390
},
{
"epoch": 1.5717092337917484,
"grad_norm": 1.5081740617752075,
"learning_rate": 3.4594703831024723e-06,
"loss": 0.3923,
"step": 400
},
{
"epoch": 1.611001964636542,
"grad_norm": 1.912083387374878,
"learning_rate": 3.385365121410706e-06,
"loss": 0.3653,
"step": 410
},
{
"epoch": 1.650294695481336,
"grad_norm": 1.4762853384017944,
"learning_rate": 3.310359299253082e-06,
"loss": 0.3544,
"step": 420
},
{
"epoch": 1.6895874263261297,
"grad_norm": 1.6862307786941528,
"learning_rate": 3.234529209761676e-06,
"loss": 0.3878,
"step": 430
},
{
"epoch": 1.7288801571709234,
"grad_norm": 1.4678353071212769,
"learning_rate": 3.157951984482635e-06,
"loss": 0.3846,
"step": 440
},
{
"epoch": 1.768172888015717,
"grad_norm": 1.1692405939102173,
"learning_rate": 3.080705514920836e-06,
"loss": 0.4512,
"step": 450
},
{
"epoch": 1.807465618860511,
"grad_norm": 1.5150586366653442,
"learning_rate": 3.0028683733115417e-06,
"loss": 0.3761,
"step": 460
},
{
"epoch": 1.8467583497053046,
"grad_norm": 1.101525902748108,
"learning_rate": 2.9245197326996515e-06,
"loss": 0.3673,
"step": 470
},
{
"epoch": 1.8860510805500983,
"grad_norm": 1.644962191581726,
"learning_rate": 2.845739286407821e-06,
"loss": 0.3874,
"step": 480
},
{
"epoch": 1.925343811394892,
"grad_norm": 1.2484012842178345,
"learning_rate": 2.7666071669753807e-06,
"loss": 0.3562,
"step": 490
},
{
"epoch": 1.9646365422396856,
"grad_norm": 1.538386583328247,
"learning_rate": 2.687203864650497e-06,
"loss": 0.3739,
"step": 500
},
{
"epoch": 2.0039292730844793,
"grad_norm": 1.4169279336929321,
"learning_rate": 2.6076101455184867e-06,
"loss": 0.3554,
"step": 510
},
{
"epoch": 2.043222003929273,
"grad_norm": 1.270408034324646,
"learning_rate": 2.527906969349559e-06,
"loss": 0.323,
"step": 520
},
{
"epoch": 2.0825147347740667,
"grad_norm": 1.7707313299179077,
"learning_rate": 2.44817540724955e-06,
"loss": 0.2903,
"step": 530
},
{
"epoch": 2.1218074656188604,
"grad_norm": 2.075082540512085,
"learning_rate": 2.3684965591974084e-06,
"loss": 0.306,
"step": 540
},
{
"epoch": 2.161100196463654,
"grad_norm": 1.2306314706802368,
"learning_rate": 2.288951471553316e-06,
"loss": 0.3889,
"step": 550
},
{
"epoch": 2.2003929273084477,
"grad_norm": 1.4437307119369507,
"learning_rate": 2.2096210546213397e-06,
"loss": 0.2868,
"step": 560
},
{
"epoch": 2.239685658153242,
"grad_norm": 1.1352663040161133,
"learning_rate": 2.130586000350486e-06,
"loss": 0.2873,
"step": 570
},
{
"epoch": 2.2789783889980355,
"grad_norm": 2.1770286560058594,
"learning_rate": 2.0519267002578517e-06,
"loss": 0.2845,
"step": 580
},
{
"epoch": 2.318271119842829,
"grad_norm": 2.0281052589416504,
"learning_rate": 1.9737231636573595e-06,
"loss": 0.3272,
"step": 590
},
{
"epoch": 2.357563850687623,
"grad_norm": 1.668318510055542,
"learning_rate": 1.8960549362772618e-06,
"loss": 0.2833,
"step": 600
},
{
"epoch": 2.3968565815324165,
"grad_norm": 2.1316215991973877,
"learning_rate": 1.8190010193491869e-06,
"loss": 0.3254,
"step": 610
},
{
"epoch": 2.43614931237721,
"grad_norm": 1.4150116443634033,
"learning_rate": 1.7426397892510244e-06,
"loss": 0.3161,
"step": 620
},
{
"epoch": 2.475442043222004,
"grad_norm": 1.6361192464828491,
"learning_rate": 1.6670489177853854e-06,
"loss": 0.3081,
"step": 630
},
{
"epoch": 2.5147347740667976,
"grad_norm": 1.5381343364715576,
"learning_rate": 1.5923052931747408e-06,
"loss": 0.2698,
"step": 640
},
{
"epoch": 2.5540275049115913,
"grad_norm": 1.8007198572158813,
"learning_rate": 1.5184849418535763e-06,
"loss": 0.3066,
"step": 650
},
{
"epoch": 2.593320235756385,
"grad_norm": 1.7028734683990479,
"learning_rate": 1.4456629511371346e-06,
"loss": 0.2608,
"step": 660
},
{
"epoch": 2.6326129666011786,
"grad_norm": 1.3481054306030273,
"learning_rate": 1.3739133928453884e-06,
"loss": 0.2841,
"step": 670
},
{
"epoch": 2.6719056974459727,
"grad_norm": 1.6009360551834106,
"learning_rate": 1.3033092479599453e-06,
"loss": 0.2789,
"step": 680
},
{
"epoch": 2.711198428290766,
"grad_norm": 1.5498707294464111,
"learning_rate": 1.2339223323905026e-06,
"loss": 0.2586,
"step": 690
},
{
"epoch": 2.75049115913556,
"grad_norm": 1.4760583639144897,
"learning_rate": 1.1658232239263815e-06,
"loss": 0.2768,
"step": 700
},
{
"epoch": 2.7897838899803538,
"grad_norm": 1.7059721946716309,
"learning_rate": 1.099081190447418e-06,
"loss": 0.2831,
"step": 710
},
{
"epoch": 2.8290766208251474,
"grad_norm": 1.3393672704696655,
"learning_rate": 1.0337641194672608e-06,
"loss": 0.2779,
"step": 720
},
{
"epoch": 2.868369351669941,
"grad_norm": 1.9578804969787598,
"learning_rate": 9.699384490807113e-07,
"loss": 0.3024,
"step": 730
},
{
"epoch": 2.907662082514735,
"grad_norm": 1.727266788482666,
"learning_rate": 9.076691003853666e-07,
"loss": 0.2706,
"step": 740
},
{
"epoch": 2.9469548133595285,
"grad_norm": 1.587507963180542,
"learning_rate": 8.470194114462965e-07,
"loss": 0.2732,
"step": 750
},
{
"epoch": 2.986247544204322,
"grad_norm": 1.7859814167022705,
"learning_rate": 7.880510728709176e-07,
"loss": 0.3005,
"step": 760
},
{
"epoch": 3.025540275049116,
"grad_norm": 1.4465070962905884,
"learning_rate": 7.30824065059603e-07,
"loss": 0.2967,
"step": 770
},
{
"epoch": 3.0648330058939095,
"grad_norm": 2.266911506652832,
"learning_rate": 6.753965971958487e-07,
"loss": 0.252,
"step": 780
},
{
"epoch": 3.104125736738703,
"grad_norm": 1.5692194700241089,
"learning_rate": 6.218250480380611e-07,
"loss": 0.2569,
"step": 790
},
{
"epoch": 3.143418467583497,
"grad_norm": 1.6480083465576172,
"learning_rate": 5.701639085731786e-07,
"loss": 0.2773,
"step": 800
},
{
"epoch": 3.182711198428291,
"grad_norm": 1.6557282209396362,
"learning_rate": 5.204657265904664e-07,
"loss": 0.2404,
"step": 810
},
{
"epoch": 3.2220039292730847,
"grad_norm": 1.824144959449768,
"learning_rate": 4.7278105323186244e-07,
"loss": 0.2641,
"step": 820
},
{
"epoch": 3.2612966601178783,
"grad_norm": 1.5101735591888428,
"learning_rate": 4.2715839157323384e-07,
"loss": 0.2366,
"step": 830
},
{
"epoch": 3.300589390962672,
"grad_norm": 1.9978082180023193,
"learning_rate": 3.8364414728885414e-07,
"loss": 0.224,
"step": 840
},
{
"epoch": 3.3398821218074657,
"grad_norm": 1.7423945665359497,
"learning_rate": 3.4228258144927587e-07,
"loss": 0.2223,
"step": 850
},
{
"epoch": 3.3791748526522594,
"grad_norm": 1.030571460723877,
"learning_rate": 3.031157655006187e-07,
"loss": 0.2585,
"step": 860
},
{
"epoch": 3.418467583497053,
"grad_norm": 1.687889575958252,
"learning_rate": 2.6618353847105703e-07,
"loss": 0.2527,
"step": 870
},
{
"epoch": 3.4577603143418467,
"grad_norm": 1.345308542251587,
"learning_rate": 2.315234664480448e-07,
"loss": 0.2442,
"step": 880
},
{
"epoch": 3.4970530451866404,
"grad_norm": 1.6126112937927246,
"learning_rate": 1.9917080436748615e-07,
"loss": 0.2248,
"step": 890
},
{
"epoch": 3.536345776031434,
"grad_norm": 1.585175633430481,
"learning_rate": 1.6915846015372855e-07,
"loss": 0.2251,
"step": 900
},
{
"epoch": 3.5756385068762278,
"grad_norm": 1.6584875583648682,
"learning_rate": 1.4151696124684504e-07,
"loss": 0.232,
"step": 910
},
{
"epoch": 3.614931237721022,
"grad_norm": 1.8172000646591187,
"learning_rate": 1.1627442355125834e-07,
"loss": 0.2409,
"step": 920
},
{
"epoch": 3.654223968565815,
"grad_norm": 2.262449026107788,
"learning_rate": 9.345652283728828e-08,
"loss": 0.2335,
"step": 930
},
{
"epoch": 3.6935166994106092,
"grad_norm": 1.6031078100204468,
"learning_rate": 7.308646862471319e-08,
"loss": 0.2436,
"step": 940
},
{
"epoch": 3.732809430255403,
"grad_norm": 2.1290814876556396,
"learning_rate": 5.5184980574911617e-08,
"loss": 0.2531,
"step": 950
},
{
"epoch": 3.7721021611001966,
"grad_norm": 1.8082916736602783,
"learning_rate": 3.9770267415590866e-08,
"loss": 0.2311,
"step": 960
},
{
"epoch": 3.8113948919449903,
"grad_norm": 2.1202828884124756,
"learning_rate": 2.6858008419548e-08,
"loss": 0.2407,
"step": 970
},
{
"epoch": 3.850687622789784,
"grad_norm": 2.0655148029327393,
"learning_rate": 1.6461337456297187e-08,
"loss": 0.2375,
"step": 980
},
{
"epoch": 3.8899803536345776,
"grad_norm": 1.8078986406326294,
"learning_rate": 8.590829632785513e-09,
"loss": 0.2385,
"step": 990
},
{
"epoch": 3.9292730844793713,
"grad_norm": 2.0520102977752686,
"learning_rate": 3.2544905367876134e-09,
"loss": 0.2567,
"step": 1000
},
{
"epoch": 3.968565815324165,
"grad_norm": 1.6944043636322021,
"learning_rate": 4.577480939194079e-10,
"loss": 0.3128,
"step": 1010
},
{
"epoch": 3.9921414538310414,
"step": 1016,
"total_flos": 1.8674965727831654e+17,
"train_loss": 0.3693922672450073,
"train_runtime": 2946.7744,
"train_samples_per_second": 5.518,
"train_steps_per_second": 0.345
}
],
"logging_steps": 10,
"max_steps": 1016,
"num_input_tokens_seen": 0,
"num_train_epochs": 4,
"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": 1.8674965727831654e+17,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}