abhinavp's picture
Training in progress, epoch 0, checkpoint
9a7fed9 verified
raw
history blame
19 kB
{
"best_metric": 4.762792587280273,
"best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/re-irr-sv-agr/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": 10.8198,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 4.999161405248948e-05,
"loss": 7.5444,
"step": 512
},
{
"epoch": 0.0,
"learning_rate": 4.998322810497896e-05,
"loss": 7.0493,
"step": 1024
},
{
"epoch": 0.0,
"learning_rate": 4.997484215746844e-05,
"loss": 6.9839,
"step": 1536
},
{
"epoch": 0.0,
"learning_rate": 4.996645620995792e-05,
"loss": 6.9381,
"step": 2048
},
{
"epoch": 0.0,
"learning_rate": 4.99580702624474e-05,
"loss": 6.879,
"step": 2560
},
{
"epoch": 0.0,
"learning_rate": 4.994968431493688e-05,
"loss": 6.6972,
"step": 3072
},
{
"epoch": 0.0,
"learning_rate": 4.994131474623009e-05,
"loss": 6.5963,
"step": 3584
},
{
"epoch": 0.0,
"learning_rate": 4.993292879871958e-05,
"loss": 6.4992,
"step": 4096
},
{
"epoch": 0.0,
"learning_rate": 4.992454285120906e-05,
"loss": 6.4332,
"step": 4608
},
{
"epoch": 0.0,
"learning_rate": 4.991615690369854e-05,
"loss": 6.375,
"step": 5120
},
{
"epoch": 0.0,
"learning_rate": 4.990777095618801e-05,
"loss": 6.3236,
"step": 5632
},
{
"epoch": 0.0,
"learning_rate": 4.989938500867749e-05,
"loss": 6.2567,
"step": 6144
},
{
"epoch": 0.0,
"learning_rate": 4.989099906116697e-05,
"loss": 6.1862,
"step": 6656
},
{
"epoch": 0.0,
"learning_rate": 4.988261311365645e-05,
"loss": 6.1315,
"step": 7168
},
{
"epoch": 0.0,
"learning_rate": 4.987422716614593e-05,
"loss": 6.0673,
"step": 7680
},
{
"epoch": 0.0,
"learning_rate": 4.986584121863541e-05,
"loss": 6.0263,
"step": 8192
},
{
"epoch": 0.0,
"learning_rate": 4.985745527112489e-05,
"loss": 5.9832,
"step": 8704
},
{
"epoch": 0.0,
"learning_rate": 4.984906932361437e-05,
"loss": 5.9401,
"step": 9216
},
{
"epoch": 0.0,
"learning_rate": 4.984068337610385e-05,
"loss": 5.9022,
"step": 9728
},
{
"epoch": 0.0,
"learning_rate": 4.983229742859333e-05,
"loss": 5.8732,
"step": 10240
},
{
"epoch": 0.0,
"learning_rate": 4.982391148108281e-05,
"loss": 5.831,
"step": 10752
},
{
"epoch": 0.0,
"learning_rate": 4.981552553357229e-05,
"loss": 5.8059,
"step": 11264
},
{
"epoch": 0.0,
"learning_rate": 4.9807155964865506e-05,
"loss": 5.7615,
"step": 11776
},
{
"epoch": 0.0,
"learning_rate": 4.9798770017354986e-05,
"loss": 5.7477,
"step": 12288
},
{
"epoch": 0.0,
"learning_rate": 4.9790384069844466e-05,
"loss": 5.7157,
"step": 12800
},
{
"epoch": 0.0,
"learning_rate": 4.9781998122333946e-05,
"loss": 5.6952,
"step": 13312
},
{
"epoch": 0.0,
"learning_rate": 4.9773612174823426e-05,
"loss": 5.6681,
"step": 13824
},
{
"epoch": 0.0,
"learning_rate": 4.9765242606116635e-05,
"loss": 5.6356,
"step": 14336
},
{
"epoch": 0.0,
"learning_rate": 4.9756856658606115e-05,
"loss": 5.6184,
"step": 14848
},
{
"epoch": 0.01,
"learning_rate": 4.9748470711095595e-05,
"loss": 5.5935,
"step": 15360
},
{
"epoch": 0.01,
"learning_rate": 4.974008476358507e-05,
"loss": 5.5844,
"step": 15872
},
{
"epoch": 0.01,
"learning_rate": 4.973169881607455e-05,
"loss": 5.5577,
"step": 16384
},
{
"epoch": 0.01,
"learning_rate": 4.972331286856403e-05,
"loss": 5.5458,
"step": 16896
},
{
"epoch": 0.01,
"learning_rate": 4.9714926921053515e-05,
"loss": 5.5353,
"step": 17408
},
{
"epoch": 0.01,
"learning_rate": 4.9706557352346724e-05,
"loss": 5.5128,
"step": 17920
},
{
"epoch": 0.01,
"learning_rate": 4.9698171404836204e-05,
"loss": 5.5024,
"step": 18432
},
{
"epoch": 0.01,
"learning_rate": 4.9689785457325684e-05,
"loss": 5.4586,
"step": 18944
},
{
"epoch": 0.01,
"learning_rate": 4.9681399509815164e-05,
"loss": 5.4545,
"step": 19456
},
{
"epoch": 0.01,
"learning_rate": 4.9673013562304644e-05,
"loss": 5.4341,
"step": 19968
},
{
"epoch": 0.01,
"learning_rate": 4.966464399359785e-05,
"loss": 5.4197,
"step": 20480
},
{
"epoch": 0.01,
"learning_rate": 4.965625804608733e-05,
"loss": 5.4159,
"step": 20992
},
{
"epoch": 0.01,
"learning_rate": 4.964787209857681e-05,
"loss": 5.4078,
"step": 21504
},
{
"epoch": 0.01,
"learning_rate": 4.963948615106629e-05,
"loss": 5.3784,
"step": 22016
},
{
"epoch": 0.01,
"learning_rate": 4.96311165823595e-05,
"loss": 5.3794,
"step": 22528
},
{
"epoch": 0.01,
"learning_rate": 4.962273063484898e-05,
"loss": 5.3738,
"step": 23040
},
{
"epoch": 0.01,
"learning_rate": 4.961434468733847e-05,
"loss": 5.3611,
"step": 23552
},
{
"epoch": 0.01,
"learning_rate": 4.960595873982795e-05,
"loss": 5.3548,
"step": 24064
},
{
"epoch": 0.01,
"learning_rate": 4.959757279231743e-05,
"loss": 5.3395,
"step": 24576
},
{
"epoch": 0.01,
"learning_rate": 4.958920322361064e-05,
"loss": 5.3168,
"step": 25088
},
{
"epoch": 0.01,
"learning_rate": 4.958081727610012e-05,
"loss": 5.3194,
"step": 25600
},
{
"epoch": 0.01,
"learning_rate": 4.95724313285896e-05,
"loss": 5.3,
"step": 26112
},
{
"epoch": 0.01,
"learning_rate": 4.956404538107908e-05,
"loss": 5.3012,
"step": 26624
},
{
"epoch": 0.01,
"learning_rate": 4.955565943356856e-05,
"loss": 5.2726,
"step": 27136
},
{
"epoch": 0.01,
"learning_rate": 4.954728986486177e-05,
"loss": 5.2845,
"step": 27648
},
{
"epoch": 0.01,
"learning_rate": 4.953890391735125e-05,
"loss": 5.2577,
"step": 28160
},
{
"epoch": 0.01,
"learning_rate": 4.9530517969840727e-05,
"loss": 5.2741,
"step": 28672
},
{
"epoch": 0.01,
"learning_rate": 4.9522132022330207e-05,
"loss": 5.2461,
"step": 29184
},
{
"epoch": 0.01,
"learning_rate": 4.9513746074819686e-05,
"loss": 5.2331,
"step": 29696
},
{
"epoch": 0.01,
"learning_rate": 4.9505360127309166e-05,
"loss": 5.2281,
"step": 30208
},
{
"epoch": 0.01,
"learning_rate": 4.9496974179798646e-05,
"loss": 5.2223,
"step": 30720
},
{
"epoch": 0.01,
"learning_rate": 4.948858823228813e-05,
"loss": 5.2089,
"step": 31232
},
{
"epoch": 0.01,
"learning_rate": 4.948021866358134e-05,
"loss": 5.2097,
"step": 31744
},
{
"epoch": 0.01,
"learning_rate": 4.947183271607082e-05,
"loss": 5.1824,
"step": 32256
},
{
"epoch": 0.01,
"learning_rate": 4.946346314736403e-05,
"loss": 5.1777,
"step": 32768
},
{
"epoch": 0.01,
"learning_rate": 4.945507719985351e-05,
"loss": 5.1896,
"step": 33280
},
{
"epoch": 0.01,
"learning_rate": 4.944669125234299e-05,
"loss": 5.1812,
"step": 33792
},
{
"epoch": 0.01,
"learning_rate": 4.943830530483247e-05,
"loss": 5.1656,
"step": 34304
},
{
"epoch": 0.01,
"learning_rate": 4.942991935732195e-05,
"loss": 5.1467,
"step": 34816
},
{
"epoch": 0.01,
"learning_rate": 4.942153340981143e-05,
"loss": 5.1466,
"step": 35328
},
{
"epoch": 0.01,
"learning_rate": 4.941316384110464e-05,
"loss": 5.1461,
"step": 35840
},
{
"epoch": 0.01,
"learning_rate": 4.940477789359412e-05,
"loss": 5.1398,
"step": 36352
},
{
"epoch": 0.01,
"learning_rate": 4.93963919460836e-05,
"loss": 5.1359,
"step": 36864
},
{
"epoch": 0.01,
"learning_rate": 4.938800599857309e-05,
"loss": 5.1295,
"step": 37376
},
{
"epoch": 0.01,
"learning_rate": 4.9379636429866296e-05,
"loss": 5.1314,
"step": 37888
},
{
"epoch": 0.01,
"learning_rate": 4.9371250482355776e-05,
"loss": 5.1229,
"step": 38400
},
{
"epoch": 0.01,
"learning_rate": 4.9362864534845256e-05,
"loss": 5.0963,
"step": 38912
},
{
"epoch": 0.01,
"learning_rate": 4.935447858733473e-05,
"loss": 5.0932,
"step": 39424
},
{
"epoch": 0.01,
"learning_rate": 4.934609263982421e-05,
"loss": 5.0903,
"step": 39936
},
{
"epoch": 0.01,
"learning_rate": 4.9337723071117425e-05,
"loss": 5.0773,
"step": 40448
},
{
"epoch": 0.01,
"learning_rate": 4.9329337123606905e-05,
"loss": 5.0854,
"step": 40960
},
{
"epoch": 0.01,
"learning_rate": 4.9320967554900114e-05,
"loss": 5.0695,
"step": 41472
},
{
"epoch": 0.01,
"learning_rate": 4.9312581607389594e-05,
"loss": 5.0863,
"step": 41984
},
{
"epoch": 0.01,
"learning_rate": 4.9304195659879074e-05,
"loss": 5.0594,
"step": 42496
},
{
"epoch": 0.01,
"learning_rate": 4.9295809712368554e-05,
"loss": 5.0441,
"step": 43008
},
{
"epoch": 0.01,
"learning_rate": 4.9287423764858034e-05,
"loss": 5.0498,
"step": 43520
},
{
"epoch": 0.01,
"learning_rate": 4.9279037817347514e-05,
"loss": 5.0495,
"step": 44032
},
{
"epoch": 0.01,
"learning_rate": 4.9270651869836994e-05,
"loss": 5.0454,
"step": 44544
},
{
"epoch": 0.01,
"learning_rate": 4.9262265922326474e-05,
"loss": 5.0322,
"step": 45056
},
{
"epoch": 0.01,
"learning_rate": 4.925389635361968e-05,
"loss": 5.028,
"step": 45568
},
{
"epoch": 0.02,
"learning_rate": 4.924551040610916e-05,
"loss": 5.0201,
"step": 46080
},
{
"epoch": 0.02,
"learning_rate": 4.923712445859864e-05,
"loss": 5.0178,
"step": 46592
},
{
"epoch": 0.02,
"learning_rate": 4.922873851108812e-05,
"loss": 5.0082,
"step": 47104
},
{
"epoch": 0.02,
"learning_rate": 4.92203525635776e-05,
"loss": 5.0043,
"step": 47616
},
{
"epoch": 0.02,
"learning_rate": 4.921196661606708e-05,
"loss": 5.0017,
"step": 48128
},
{
"epoch": 0.02,
"learning_rate": 4.920358066855656e-05,
"loss": 5.0073,
"step": 48640
},
{
"epoch": 0.02,
"learning_rate": 4.919519472104604e-05,
"loss": 4.9857,
"step": 49152
},
{
"epoch": 0.02,
"learning_rate": 4.918682515233926e-05,
"loss": 4.9812,
"step": 49664
},
{
"epoch": 0.02,
"learning_rate": 4.917843920482874e-05,
"loss": 4.974,
"step": 50176
},
{
"epoch": 0.02,
"learning_rate": 4.917005325731822e-05,
"loss": 4.9696,
"step": 50688
},
{
"epoch": 0.02,
"learning_rate": 4.91616673098077e-05,
"loss": 4.9745,
"step": 51200
},
{
"epoch": 0.02,
"learning_rate": 4.915329774110091e-05,
"loss": 4.9612,
"step": 51712
},
{
"epoch": 0.02,
"learning_rate": 4.914491179359039e-05,
"loss": 4.9549,
"step": 52224
},
{
"epoch": 0.02,
"learning_rate": 4.913652584607987e-05,
"loss": 4.9466,
"step": 52736
},
{
"epoch": 0.02,
"learning_rate": 4.912813989856935e-05,
"loss": 4.9508,
"step": 53248
},
{
"epoch": 0.02,
"learning_rate": 4.911975395105883e-05,
"loss": 4.9226,
"step": 53760
},
{
"epoch": 0.02,
"learning_rate": 4.911136800354831e-05,
"loss": 4.9364,
"step": 54272
},
{
"epoch": 0.02,
"learning_rate": 4.910298205603779e-05,
"loss": 4.9256,
"step": 54784
},
{
"epoch": 0.02,
"learning_rate": 4.909459610852727e-05,
"loss": 4.9319,
"step": 55296
},
{
"epoch": 0.02,
"learning_rate": 4.9086226539820476e-05,
"loss": 4.9191,
"step": 55808
},
{
"epoch": 0.02,
"learning_rate": 4.907784059230996e-05,
"loss": 4.926,
"step": 56320
},
{
"epoch": 0.02,
"learning_rate": 4.906945464479944e-05,
"loss": 4.9108,
"step": 56832
},
{
"epoch": 0.02,
"learning_rate": 4.906108507609265e-05,
"loss": 4.9064,
"step": 57344
},
{
"epoch": 0.02,
"learning_rate": 4.905269912858213e-05,
"loss": 4.9152,
"step": 57856
},
{
"epoch": 0.02,
"learning_rate": 4.904431318107161e-05,
"loss": 4.8907,
"step": 58368
},
{
"epoch": 0.02,
"learning_rate": 4.903592723356109e-05,
"loss": 4.9082,
"step": 58880
},
{
"epoch": 0.02,
"learning_rate": 4.9027541286050565e-05,
"loss": 4.8988,
"step": 59392
},
{
"epoch": 0.02,
"learning_rate": 4.901917171734378e-05,
"loss": 4.8949,
"step": 59904
},
{
"epoch": 0.02,
"learning_rate": 4.901078576983326e-05,
"loss": 4.8823,
"step": 60416
},
{
"epoch": 0.02,
"learning_rate": 4.900239982232274e-05,
"loss": 4.8822,
"step": 60928
},
{
"epoch": 0.02,
"learning_rate": 4.8994013874812214e-05,
"loss": 4.8844,
"step": 61440
},
{
"epoch": 0.02,
"learning_rate": 4.89856279273017e-05,
"loss": 4.877,
"step": 61952
},
{
"epoch": 0.02,
"learning_rate": 4.897724197979118e-05,
"loss": 4.8756,
"step": 62464
},
{
"epoch": 0.02,
"learning_rate": 4.896885603228066e-05,
"loss": 4.866,
"step": 62976
},
{
"epoch": 0.02,
"learning_rate": 4.896047008477014e-05,
"loss": 4.8683,
"step": 63488
},
{
"epoch": 0.02,
"learning_rate": 4.895208413725962e-05,
"loss": 4.8476,
"step": 64000
},
{
"epoch": 0.02,
"learning_rate": 4.89436981897491e-05,
"loss": 4.8526,
"step": 64512
},
{
"epoch": 0.02,
"learning_rate": 4.893531224223858e-05,
"loss": 4.834,
"step": 65024
},
{
"epoch": 0.02,
"learning_rate": 4.892694267353179e-05,
"loss": 4.8415,
"step": 65536
},
{
"epoch": 0.02,
"learning_rate": 4.891855672602127e-05,
"loss": 4.8277,
"step": 66048
},
{
"epoch": 0.02,
"learning_rate": 4.891017077851075e-05,
"loss": 4.8489,
"step": 66560
},
{
"epoch": 0.02,
"learning_rate": 4.890178483100023e-05,
"loss": 4.8541,
"step": 67072
},
{
"epoch": 0.02,
"learning_rate": 4.889341526229344e-05,
"loss": 4.8379,
"step": 67584
},
{
"epoch": 0.02,
"learning_rate": 4.888502931478292e-05,
"loss": 4.817,
"step": 68096
},
{
"epoch": 0.02,
"learning_rate": 4.88766433672724e-05,
"loss": 4.8285,
"step": 68608
},
{
"epoch": 0.02,
"learning_rate": 4.886825741976188e-05,
"loss": 4.8264,
"step": 69120
},
{
"epoch": 0.02,
"learning_rate": 4.8859871472251365e-05,
"loss": 4.8203,
"step": 69632
},
{
"epoch": 0.02,
"learning_rate": 4.8851485524740845e-05,
"loss": 4.8159,
"step": 70144
},
{
"epoch": 0.02,
"learning_rate": 4.8843099577230325e-05,
"loss": 4.8175,
"step": 70656
},
{
"epoch": 0.02,
"learning_rate": 4.8834713629719805e-05,
"loss": 4.8132,
"step": 71168
},
{
"epoch": 0.02,
"learning_rate": 4.882632768220928e-05,
"loss": 4.8018,
"step": 71680
},
{
"epoch": 0.02,
"learning_rate": 4.881794173469876e-05,
"loss": 4.8021,
"step": 72192
},
{
"epoch": 0.02,
"learning_rate": 4.8809572165991974e-05,
"loss": 4.803,
"step": 72704
},
{
"epoch": 0.02,
"learning_rate": 4.8801186218481454e-05,
"loss": 4.7911,
"step": 73216
},
{
"epoch": 0.02,
"learning_rate": 4.879280027097093e-05,
"loss": 4.7917,
"step": 73728
},
{
"epoch": 0.02,
"learning_rate": 4.878441432346041e-05,
"loss": 4.7991,
"step": 74240
},
{
"epoch": 0.02,
"learning_rate": 4.877602837594989e-05,
"loss": 4.7873,
"step": 74752
},
{
"epoch": 0.02,
"learning_rate": 4.876764242843937e-05,
"loss": 4.7853,
"step": 75264
},
{
"epoch": 0.02,
"learning_rate": 4.875927285973258e-05,
"loss": 4.7972,
"step": 75776
},
{
"epoch": 0.02,
"learning_rate": 4.875088691222206e-05,
"loss": 4.774,
"step": 76288
},
{
"epoch": 0.03,
"eval_loss": 4.762792587280273,
"eval_runtime": 291.1775,
"eval_samples_per_second": 1310.51,
"eval_steps_per_second": 40.954,
"step": 76320
}
],
"logging_steps": 512,
"max_steps": 3052726,
"num_train_epochs": 9223372036854775807,
"save_steps": 10,
"total_flos": 3.055954929919841e+16,
"trial_name": null,
"trial_params": null
}