abhinavp's picture
Training in progress, epoch 0, checkpoint
157fa71 verified
{
"best_metric": 4.757783889770508,
"best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/binding-case/lstm/0/checkpoints/checkpoint-76320",
"epoch": 0.025000606015738065,
"eval_steps": 10,
"global_step": 76320,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.999998362119627e-05,
"loss": 10.8206,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 4.999161405248948e-05,
"loss": 7.5557,
"step": 512
},
{
"epoch": 0.0,
"learning_rate": 4.998322810497896e-05,
"loss": 7.0612,
"step": 1024
},
{
"epoch": 0.0,
"learning_rate": 4.997484215746844e-05,
"loss": 6.9955,
"step": 1536
},
{
"epoch": 0.0,
"learning_rate": 4.996645620995792e-05,
"loss": 6.949,
"step": 2048
},
{
"epoch": 0.0,
"learning_rate": 4.99580702624474e-05,
"loss": 6.9206,
"step": 2560
},
{
"epoch": 0.0,
"learning_rate": 4.994968431493688e-05,
"loss": 6.7551,
"step": 3072
},
{
"epoch": 0.0,
"learning_rate": 4.994129836742636e-05,
"loss": 6.6436,
"step": 3584
},
{
"epoch": 0.0,
"learning_rate": 4.993291241991584e-05,
"loss": 6.5449,
"step": 4096
},
{
"epoch": 0.0,
"learning_rate": 4.992452647240532e-05,
"loss": 6.4714,
"step": 4608
},
{
"epoch": 0.0,
"learning_rate": 4.991615690369854e-05,
"loss": 6.3994,
"step": 5120
},
{
"epoch": 0.0,
"learning_rate": 4.990777095618801e-05,
"loss": 6.3319,
"step": 5632
},
{
"epoch": 0.0,
"learning_rate": 4.989938500867749e-05,
"loss": 6.2619,
"step": 6144
},
{
"epoch": 0.0,
"learning_rate": 4.989099906116697e-05,
"loss": 6.1949,
"step": 6656
},
{
"epoch": 0.0,
"learning_rate": 4.988261311365645e-05,
"loss": 6.1395,
"step": 7168
},
{
"epoch": 0.0,
"learning_rate": 4.987422716614593e-05,
"loss": 6.0836,
"step": 7680
},
{
"epoch": 0.0,
"learning_rate": 4.986584121863541e-05,
"loss": 6.0391,
"step": 8192
},
{
"epoch": 0.0,
"learning_rate": 4.985745527112489e-05,
"loss": 5.9938,
"step": 8704
},
{
"epoch": 0.0,
"learning_rate": 4.984906932361437e-05,
"loss": 5.9539,
"step": 9216
},
{
"epoch": 0.0,
"learning_rate": 4.984068337610385e-05,
"loss": 5.9146,
"step": 9728
},
{
"epoch": 0.0,
"learning_rate": 4.983229742859333e-05,
"loss": 5.8783,
"step": 10240
},
{
"epoch": 0.0,
"learning_rate": 4.982391148108281e-05,
"loss": 5.8402,
"step": 10752
},
{
"epoch": 0.0,
"learning_rate": 4.981552553357229e-05,
"loss": 5.8076,
"step": 11264
},
{
"epoch": 0.0,
"learning_rate": 4.980713958606178e-05,
"loss": 5.7774,
"step": 11776
},
{
"epoch": 0.0,
"learning_rate": 4.979875363855125e-05,
"loss": 5.7584,
"step": 12288
},
{
"epoch": 0.0,
"learning_rate": 4.979036769104073e-05,
"loss": 5.7174,
"step": 12800
},
{
"epoch": 0.0,
"learning_rate": 4.9781998122333946e-05,
"loss": 5.6961,
"step": 13312
},
{
"epoch": 0.0,
"learning_rate": 4.9773612174823426e-05,
"loss": 5.6694,
"step": 13824
},
{
"epoch": 0.0,
"learning_rate": 4.97652262273129e-05,
"loss": 5.6503,
"step": 14336
},
{
"epoch": 0.0,
"learning_rate": 4.975684027980238e-05,
"loss": 5.623,
"step": 14848
},
{
"epoch": 0.01,
"learning_rate": 4.9748470711095595e-05,
"loss": 5.5988,
"step": 15360
},
{
"epoch": 0.01,
"learning_rate": 4.974008476358507e-05,
"loss": 5.5924,
"step": 15872
},
{
"epoch": 0.01,
"learning_rate": 4.973169881607455e-05,
"loss": 5.566,
"step": 16384
},
{
"epoch": 0.01,
"learning_rate": 4.9723329247367764e-05,
"loss": 5.5574,
"step": 16896
},
{
"epoch": 0.01,
"learning_rate": 4.9714943299857244e-05,
"loss": 5.5424,
"step": 17408
},
{
"epoch": 0.01,
"learning_rate": 4.9706557352346724e-05,
"loss": 5.5176,
"step": 17920
},
{
"epoch": 0.01,
"learning_rate": 4.969818778363994e-05,
"loss": 5.508,
"step": 18432
},
{
"epoch": 0.01,
"learning_rate": 4.968980183612942e-05,
"loss": 5.4698,
"step": 18944
},
{
"epoch": 0.01,
"learning_rate": 4.96814158886189e-05,
"loss": 5.4702,
"step": 19456
},
{
"epoch": 0.01,
"learning_rate": 4.967302994110837e-05,
"loss": 5.4358,
"step": 19968
},
{
"epoch": 0.01,
"learning_rate": 4.966464399359785e-05,
"loss": 5.4358,
"step": 20480
},
{
"epoch": 0.01,
"learning_rate": 4.965625804608733e-05,
"loss": 5.418,
"step": 20992
},
{
"epoch": 0.01,
"learning_rate": 4.964787209857681e-05,
"loss": 5.4185,
"step": 21504
},
{
"epoch": 0.01,
"learning_rate": 4.963948615106629e-05,
"loss": 5.3915,
"step": 22016
},
{
"epoch": 0.01,
"learning_rate": 4.96311165823595e-05,
"loss": 5.3852,
"step": 22528
},
{
"epoch": 0.01,
"learning_rate": 4.962273063484898e-05,
"loss": 5.3764,
"step": 23040
},
{
"epoch": 0.01,
"learning_rate": 4.961434468733847e-05,
"loss": 5.3599,
"step": 23552
},
{
"epoch": 0.01,
"learning_rate": 4.960595873982795e-05,
"loss": 5.3653,
"step": 24064
},
{
"epoch": 0.01,
"learning_rate": 4.959757279231743e-05,
"loss": 5.3357,
"step": 24576
},
{
"epoch": 0.01,
"learning_rate": 4.958920322361064e-05,
"loss": 5.3203,
"step": 25088
},
{
"epoch": 0.01,
"learning_rate": 4.958081727610012e-05,
"loss": 5.3299,
"step": 25600
},
{
"epoch": 0.01,
"learning_rate": 4.957244770739333e-05,
"loss": 5.3239,
"step": 26112
},
{
"epoch": 0.01,
"learning_rate": 4.956407813868654e-05,
"loss": 5.2986,
"step": 26624
},
{
"epoch": 0.01,
"learning_rate": 4.9555692191176016e-05,
"loss": 5.2878,
"step": 27136
},
{
"epoch": 0.01,
"learning_rate": 4.9547306243665496e-05,
"loss": 5.2856,
"step": 27648
},
{
"epoch": 0.01,
"learning_rate": 4.9538920296154976e-05,
"loss": 5.271,
"step": 28160
},
{
"epoch": 0.01,
"learning_rate": 4.9530534348644456e-05,
"loss": 5.2853,
"step": 28672
},
{
"epoch": 0.01,
"learning_rate": 4.9522148401133936e-05,
"loss": 5.2463,
"step": 29184
},
{
"epoch": 0.01,
"learning_rate": 4.951376245362342e-05,
"loss": 5.2454,
"step": 29696
},
{
"epoch": 0.01,
"learning_rate": 4.95053765061129e-05,
"loss": 5.2377,
"step": 30208
},
{
"epoch": 0.01,
"learning_rate": 4.949699055860238e-05,
"loss": 5.2269,
"step": 30720
},
{
"epoch": 0.01,
"learning_rate": 4.948860461109186e-05,
"loss": 5.2277,
"step": 31232
},
{
"epoch": 0.01,
"learning_rate": 4.948021866358134e-05,
"loss": 5.2061,
"step": 31744
},
{
"epoch": 0.01,
"learning_rate": 4.947183271607082e-05,
"loss": 5.2001,
"step": 32256
},
{
"epoch": 0.01,
"learning_rate": 4.94634467685603e-05,
"loss": 5.192,
"step": 32768
},
{
"epoch": 0.01,
"learning_rate": 4.945507719985351e-05,
"loss": 5.2067,
"step": 33280
},
{
"epoch": 0.01,
"learning_rate": 4.944669125234299e-05,
"loss": 5.1779,
"step": 33792
},
{
"epoch": 0.01,
"learning_rate": 4.943830530483247e-05,
"loss": 5.1785,
"step": 34304
},
{
"epoch": 0.01,
"learning_rate": 4.942991935732195e-05,
"loss": 5.1512,
"step": 34816
},
{
"epoch": 0.01,
"learning_rate": 4.942154978861516e-05,
"loss": 5.1598,
"step": 35328
},
{
"epoch": 0.01,
"learning_rate": 4.941316384110464e-05,
"loss": 5.167,
"step": 35840
},
{
"epoch": 0.01,
"learning_rate": 4.940477789359412e-05,
"loss": 5.1598,
"step": 36352
},
{
"epoch": 0.01,
"learning_rate": 4.93963919460836e-05,
"loss": 5.1372,
"step": 36864
},
{
"epoch": 0.01,
"learning_rate": 4.938800599857309e-05,
"loss": 5.1505,
"step": 37376
},
{
"epoch": 0.01,
"learning_rate": 4.937962005106256e-05,
"loss": 5.1458,
"step": 37888
},
{
"epoch": 0.01,
"learning_rate": 4.937123410355204e-05,
"loss": 5.1229,
"step": 38400
},
{
"epoch": 0.01,
"learning_rate": 4.9362864534845256e-05,
"loss": 5.12,
"step": 38912
},
{
"epoch": 0.01,
"learning_rate": 4.935447858733473e-05,
"loss": 5.1027,
"step": 39424
},
{
"epoch": 0.01,
"learning_rate": 4.934609263982421e-05,
"loss": 5.089,
"step": 39936
},
{
"epoch": 0.01,
"learning_rate": 4.933770669231369e-05,
"loss": 5.0878,
"step": 40448
},
{
"epoch": 0.01,
"learning_rate": 4.9329337123606905e-05,
"loss": 5.0921,
"step": 40960
},
{
"epoch": 0.01,
"learning_rate": 4.932095117609638e-05,
"loss": 5.0903,
"step": 41472
},
{
"epoch": 0.01,
"learning_rate": 4.931256522858586e-05,
"loss": 5.096,
"step": 41984
},
{
"epoch": 0.01,
"learning_rate": 4.930417928107534e-05,
"loss": 5.0813,
"step": 42496
},
{
"epoch": 0.01,
"learning_rate": 4.9295809712368554e-05,
"loss": 5.0617,
"step": 43008
},
{
"epoch": 0.01,
"learning_rate": 4.9287423764858034e-05,
"loss": 5.0622,
"step": 43520
},
{
"epoch": 0.01,
"learning_rate": 4.9279037817347514e-05,
"loss": 5.0622,
"step": 44032
},
{
"epoch": 0.01,
"learning_rate": 4.9270651869836994e-05,
"loss": 5.0509,
"step": 44544
},
{
"epoch": 0.01,
"learning_rate": 4.9262265922326474e-05,
"loss": 5.0423,
"step": 45056
},
{
"epoch": 0.01,
"learning_rate": 4.925389635361968e-05,
"loss": 5.0321,
"step": 45568
},
{
"epoch": 0.02,
"learning_rate": 4.92455267849129e-05,
"loss": 5.0325,
"step": 46080
},
{
"epoch": 0.02,
"learning_rate": 4.923714083740238e-05,
"loss": 5.0333,
"step": 46592
},
{
"epoch": 0.02,
"learning_rate": 4.922875488989185e-05,
"loss": 5.0206,
"step": 47104
},
{
"epoch": 0.02,
"learning_rate": 4.922036894238133e-05,
"loss": 5.0233,
"step": 47616
},
{
"epoch": 0.02,
"learning_rate": 4.921198299487081e-05,
"loss": 5.0122,
"step": 48128
},
{
"epoch": 0.02,
"learning_rate": 4.920361342616403e-05,
"loss": 5.0066,
"step": 48640
},
{
"epoch": 0.02,
"learning_rate": 4.919522747865351e-05,
"loss": 4.9943,
"step": 49152
},
{
"epoch": 0.02,
"learning_rate": 4.918684153114299e-05,
"loss": 4.9911,
"step": 49664
},
{
"epoch": 0.02,
"learning_rate": 4.917845558363247e-05,
"loss": 4.9846,
"step": 50176
},
{
"epoch": 0.02,
"learning_rate": 4.917006963612195e-05,
"loss": 4.9879,
"step": 50688
},
{
"epoch": 0.02,
"learning_rate": 4.9161700067415156e-05,
"loss": 4.9896,
"step": 51200
},
{
"epoch": 0.02,
"learning_rate": 4.9153314119904636e-05,
"loss": 4.967,
"step": 51712
},
{
"epoch": 0.02,
"learning_rate": 4.914494455119785e-05,
"loss": 4.9728,
"step": 52224
},
{
"epoch": 0.02,
"learning_rate": 4.9136558603687325e-05,
"loss": 4.9613,
"step": 52736
},
{
"epoch": 0.02,
"learning_rate": 4.9128172656176805e-05,
"loss": 4.9586,
"step": 53248
},
{
"epoch": 0.02,
"learning_rate": 4.9119786708666285e-05,
"loss": 4.9504,
"step": 53760
},
{
"epoch": 0.02,
"learning_rate": 4.9111400761155765e-05,
"loss": 4.9449,
"step": 54272
},
{
"epoch": 0.02,
"learning_rate": 4.9103014813645245e-05,
"loss": 4.9452,
"step": 54784
},
{
"epoch": 0.02,
"learning_rate": 4.909462886613473e-05,
"loss": 4.9367,
"step": 55296
},
{
"epoch": 0.02,
"learning_rate": 4.908624291862421e-05,
"loss": 4.934,
"step": 55808
},
{
"epoch": 0.02,
"learning_rate": 4.907785697111369e-05,
"loss": 4.9274,
"step": 56320
},
{
"epoch": 0.02,
"learning_rate": 4.906947102360317e-05,
"loss": 4.9177,
"step": 56832
},
{
"epoch": 0.02,
"learning_rate": 4.906108507609265e-05,
"loss": 4.9292,
"step": 57344
},
{
"epoch": 0.02,
"learning_rate": 4.905269912858213e-05,
"loss": 4.9178,
"step": 57856
},
{
"epoch": 0.02,
"learning_rate": 4.904432955987534e-05,
"loss": 4.9181,
"step": 58368
},
{
"epoch": 0.02,
"learning_rate": 4.903594361236482e-05,
"loss": 4.9134,
"step": 58880
},
{
"epoch": 0.02,
"learning_rate": 4.90275576648543e-05,
"loss": 4.9144,
"step": 59392
},
{
"epoch": 0.02,
"learning_rate": 4.901917171734378e-05,
"loss": 4.9042,
"step": 59904
},
{
"epoch": 0.02,
"learning_rate": 4.901078576983326e-05,
"loss": 4.8909,
"step": 60416
},
{
"epoch": 0.02,
"learning_rate": 4.900241620112647e-05,
"loss": 4.8986,
"step": 60928
},
{
"epoch": 0.02,
"learning_rate": 4.899403025361595e-05,
"loss": 4.9034,
"step": 61440
},
{
"epoch": 0.02,
"learning_rate": 4.898564430610543e-05,
"loss": 4.8889,
"step": 61952
},
{
"epoch": 0.02,
"learning_rate": 4.897725835859491e-05,
"loss": 4.8826,
"step": 62464
},
{
"epoch": 0.02,
"learning_rate": 4.8968872411084396e-05,
"loss": 4.8807,
"step": 62976
},
{
"epoch": 0.02,
"learning_rate": 4.896048646357387e-05,
"loss": 4.8775,
"step": 63488
},
{
"epoch": 0.02,
"learning_rate": 4.8952116894867085e-05,
"loss": 4.8676,
"step": 64000
},
{
"epoch": 0.02,
"learning_rate": 4.8943747326160294e-05,
"loss": 4.8714,
"step": 64512
},
{
"epoch": 0.02,
"learning_rate": 4.8935361378649774e-05,
"loss": 4.8604,
"step": 65024
},
{
"epoch": 0.02,
"learning_rate": 4.8926975431139254e-05,
"loss": 4.8629,
"step": 65536
},
{
"epoch": 0.02,
"learning_rate": 4.8918589483628734e-05,
"loss": 4.8533,
"step": 66048
},
{
"epoch": 0.02,
"learning_rate": 4.8910203536118214e-05,
"loss": 4.8705,
"step": 66560
},
{
"epoch": 0.02,
"learning_rate": 4.890181758860769e-05,
"loss": 4.8545,
"step": 67072
},
{
"epoch": 0.02,
"learning_rate": 4.889343164109717e-05,
"loss": 4.8449,
"step": 67584
},
{
"epoch": 0.02,
"learning_rate": 4.888506207239038e-05,
"loss": 4.8412,
"step": 68096
},
{
"epoch": 0.02,
"learning_rate": 4.887667612487986e-05,
"loss": 4.8422,
"step": 68608
},
{
"epoch": 0.02,
"learning_rate": 4.886830655617308e-05,
"loss": 4.8429,
"step": 69120
},
{
"epoch": 0.02,
"learning_rate": 4.885992060866256e-05,
"loss": 4.843,
"step": 69632
},
{
"epoch": 0.02,
"learning_rate": 4.885153466115204e-05,
"loss": 4.8318,
"step": 70144
},
{
"epoch": 0.02,
"learning_rate": 4.884314871364151e-05,
"loss": 4.8329,
"step": 70656
},
{
"epoch": 0.02,
"learning_rate": 4.883476276613099e-05,
"loss": 4.828,
"step": 71168
},
{
"epoch": 0.02,
"learning_rate": 4.882637681862047e-05,
"loss": 4.8262,
"step": 71680
},
{
"epoch": 0.02,
"learning_rate": 4.881800724991369e-05,
"loss": 4.8149,
"step": 72192
},
{
"epoch": 0.02,
"learning_rate": 4.880962130240316e-05,
"loss": 4.825,
"step": 72704
},
{
"epoch": 0.02,
"learning_rate": 4.880123535489264e-05,
"loss": 4.813,
"step": 73216
},
{
"epoch": 0.02,
"learning_rate": 4.879284940738212e-05,
"loss": 4.7998,
"step": 73728
},
{
"epoch": 0.02,
"learning_rate": 4.87844634598716e-05,
"loss": 4.8142,
"step": 74240
},
{
"epoch": 0.02,
"learning_rate": 4.877607751236109e-05,
"loss": 4.8032,
"step": 74752
},
{
"epoch": 0.02,
"learning_rate": 4.876769156485057e-05,
"loss": 4.8036,
"step": 75264
},
{
"epoch": 0.02,
"learning_rate": 4.875930561734005e-05,
"loss": 4.8102,
"step": 75776
},
{
"epoch": 0.02,
"learning_rate": 4.875091966982953e-05,
"loss": 4.7895,
"step": 76288
},
{
"epoch": 0.03,
"eval_loss": 4.757783889770508,
"eval_runtime": 371.7114,
"eval_samples_per_second": 1026.579,
"eval_steps_per_second": 32.081,
"step": 76320
}
],
"logging_steps": 512,
"max_steps": 3052726,
"num_train_epochs": 9223372036854775807,
"save_steps": 10,
"total_flos": 3.131296159542617e+16,
"trial_name": null,
"trial_params": null
}