abhinavp's picture
Training in progress, epoch 0, checkpoint
343666b verified
raw
history blame
19 kB
{
"best_metric": 4.76328706741333,
"best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/binding-reconstruction/lstm/3/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": 10.8204,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 4.999161405248948e-05,
"loss": 7.5463,
"step": 512
},
{
"epoch": 0.0,
"learning_rate": 4.998322810497896e-05,
"loss": 7.0572,
"step": 1024
},
{
"epoch": 0.0,
"learning_rate": 4.997484215746844e-05,
"loss": 6.992,
"step": 1536
},
{
"epoch": 0.0,
"learning_rate": 4.996645620995792e-05,
"loss": 6.9466,
"step": 2048
},
{
"epoch": 0.0,
"learning_rate": 4.99580702624474e-05,
"loss": 6.8848,
"step": 2560
},
{
"epoch": 0.0,
"learning_rate": 4.994968431493688e-05,
"loss": 6.7333,
"step": 3072
},
{
"epoch": 0.0,
"learning_rate": 4.994129836742636e-05,
"loss": 6.6274,
"step": 3584
},
{
"epoch": 0.0,
"learning_rate": 4.993291241991584e-05,
"loss": 6.5278,
"step": 4096
},
{
"epoch": 0.0,
"learning_rate": 4.992452647240532e-05,
"loss": 6.4492,
"step": 4608
},
{
"epoch": 0.0,
"learning_rate": 4.99161405248948e-05,
"loss": 6.3879,
"step": 5120
},
{
"epoch": 0.0,
"learning_rate": 4.990775457738428e-05,
"loss": 6.3324,
"step": 5632
},
{
"epoch": 0.0,
"learning_rate": 4.989936862987376e-05,
"loss": 6.2709,
"step": 6144
},
{
"epoch": 0.0,
"learning_rate": 4.989098268236324e-05,
"loss": 6.2013,
"step": 6656
},
{
"epoch": 0.0,
"learning_rate": 4.988259673485272e-05,
"loss": 6.1486,
"step": 7168
},
{
"epoch": 0.0,
"learning_rate": 4.98742107873422e-05,
"loss": 6.086,
"step": 7680
},
{
"epoch": 0.0,
"learning_rate": 4.986582483983168e-05,
"loss": 6.0514,
"step": 8192
},
{
"epoch": 0.0,
"learning_rate": 4.985743889232116e-05,
"loss": 6.0111,
"step": 8704
},
{
"epoch": 0.0,
"learning_rate": 4.984905294481064e-05,
"loss": 5.9594,
"step": 9216
},
{
"epoch": 0.0,
"learning_rate": 4.984066699730012e-05,
"loss": 5.9279,
"step": 9728
},
{
"epoch": 0.0,
"learning_rate": 4.983229742859333e-05,
"loss": 5.8891,
"step": 10240
},
{
"epoch": 0.0,
"learning_rate": 4.982391148108281e-05,
"loss": 5.8571,
"step": 10752
},
{
"epoch": 0.0,
"learning_rate": 4.981552553357229e-05,
"loss": 5.828,
"step": 11264
},
{
"epoch": 0.0,
"learning_rate": 4.980713958606178e-05,
"loss": 5.7935,
"step": 11776
},
{
"epoch": 0.0,
"learning_rate": 4.979875363855125e-05,
"loss": 5.776,
"step": 12288
},
{
"epoch": 0.0,
"learning_rate": 4.979036769104073e-05,
"loss": 5.7382,
"step": 12800
},
{
"epoch": 0.0,
"learning_rate": 4.978198174353021e-05,
"loss": 5.7199,
"step": 13312
},
{
"epoch": 0.0,
"learning_rate": 4.9773612174823426e-05,
"loss": 5.6947,
"step": 13824
},
{
"epoch": 0.0,
"learning_rate": 4.97652262273129e-05,
"loss": 5.6663,
"step": 14336
},
{
"epoch": 0.0,
"learning_rate": 4.975684027980238e-05,
"loss": 5.6387,
"step": 14848
},
{
"epoch": 0.01,
"learning_rate": 4.974845433229186e-05,
"loss": 5.6242,
"step": 15360
},
{
"epoch": 0.01,
"learning_rate": 4.974006838478134e-05,
"loss": 5.612,
"step": 15872
},
{
"epoch": 0.01,
"learning_rate": 4.973168243727082e-05,
"loss": 5.5866,
"step": 16384
},
{
"epoch": 0.01,
"learning_rate": 4.972331286856403e-05,
"loss": 5.5653,
"step": 16896
},
{
"epoch": 0.01,
"learning_rate": 4.9714926921053515e-05,
"loss": 5.5573,
"step": 17408
},
{
"epoch": 0.01,
"learning_rate": 4.9706540973542995e-05,
"loss": 5.5454,
"step": 17920
},
{
"epoch": 0.01,
"learning_rate": 4.9698155026032475e-05,
"loss": 5.5198,
"step": 18432
},
{
"epoch": 0.01,
"learning_rate": 4.9689769078521955e-05,
"loss": 5.4864,
"step": 18944
},
{
"epoch": 0.01,
"learning_rate": 4.9681399509815164e-05,
"loss": 5.4823,
"step": 19456
},
{
"epoch": 0.01,
"learning_rate": 4.9673013562304644e-05,
"loss": 5.4489,
"step": 19968
},
{
"epoch": 0.01,
"learning_rate": 4.9664627614794124e-05,
"loss": 5.4544,
"step": 20480
},
{
"epoch": 0.01,
"learning_rate": 4.9656241667283604e-05,
"loss": 5.4369,
"step": 20992
},
{
"epoch": 0.01,
"learning_rate": 4.964787209857681e-05,
"loss": 5.4347,
"step": 21504
},
{
"epoch": 0.01,
"learning_rate": 4.963948615106629e-05,
"loss": 5.4063,
"step": 22016
},
{
"epoch": 0.01,
"learning_rate": 4.963110020355577e-05,
"loss": 5.4035,
"step": 22528
},
{
"epoch": 0.01,
"learning_rate": 4.962271425604525e-05,
"loss": 5.3896,
"step": 23040
},
{
"epoch": 0.01,
"learning_rate": 4.961432830853473e-05,
"loss": 5.3818,
"step": 23552
},
{
"epoch": 0.01,
"learning_rate": 4.960595873982795e-05,
"loss": 5.3775,
"step": 24064
},
{
"epoch": 0.01,
"learning_rate": 4.959757279231743e-05,
"loss": 5.343,
"step": 24576
},
{
"epoch": 0.01,
"learning_rate": 4.958918684480691e-05,
"loss": 5.3416,
"step": 25088
},
{
"epoch": 0.01,
"learning_rate": 4.958080089729639e-05,
"loss": 5.3449,
"step": 25600
},
{
"epoch": 0.01,
"learning_rate": 4.95724313285896e-05,
"loss": 5.3357,
"step": 26112
},
{
"epoch": 0.01,
"learning_rate": 4.956404538107908e-05,
"loss": 5.3099,
"step": 26624
},
{
"epoch": 0.01,
"learning_rate": 4.955565943356856e-05,
"loss": 5.297,
"step": 27136
},
{
"epoch": 0.01,
"learning_rate": 4.954727348605804e-05,
"loss": 5.2922,
"step": 27648
},
{
"epoch": 0.01,
"learning_rate": 4.953890391735125e-05,
"loss": 5.2822,
"step": 28160
},
{
"epoch": 0.01,
"learning_rate": 4.9530517969840727e-05,
"loss": 5.2933,
"step": 28672
},
{
"epoch": 0.01,
"learning_rate": 4.9522132022330207e-05,
"loss": 5.2534,
"step": 29184
},
{
"epoch": 0.01,
"learning_rate": 4.9513746074819686e-05,
"loss": 5.2642,
"step": 29696
},
{
"epoch": 0.01,
"learning_rate": 4.9505360127309166e-05,
"loss": 5.2538,
"step": 30208
},
{
"epoch": 0.01,
"learning_rate": 4.949699055860238e-05,
"loss": 5.2382,
"step": 30720
},
{
"epoch": 0.01,
"learning_rate": 4.948860461109186e-05,
"loss": 5.2402,
"step": 31232
},
{
"epoch": 0.01,
"learning_rate": 4.948021866358134e-05,
"loss": 5.2133,
"step": 31744
},
{
"epoch": 0.01,
"learning_rate": 4.947183271607082e-05,
"loss": 5.2084,
"step": 32256
},
{
"epoch": 0.01,
"learning_rate": 4.94634467685603e-05,
"loss": 5.2065,
"step": 32768
},
{
"epoch": 0.01,
"learning_rate": 4.945507719985351e-05,
"loss": 5.218,
"step": 33280
},
{
"epoch": 0.01,
"learning_rate": 4.944669125234299e-05,
"loss": 5.1881,
"step": 33792
},
{
"epoch": 0.01,
"learning_rate": 4.94383216836362e-05,
"loss": 5.1832,
"step": 34304
},
{
"epoch": 0.01,
"learning_rate": 4.942993573612568e-05,
"loss": 5.1637,
"step": 34816
},
{
"epoch": 0.01,
"learning_rate": 4.942154978861516e-05,
"loss": 5.1706,
"step": 35328
},
{
"epoch": 0.01,
"learning_rate": 4.941316384110464e-05,
"loss": 5.1721,
"step": 35840
},
{
"epoch": 0.01,
"learning_rate": 4.940477789359412e-05,
"loss": 5.1709,
"step": 36352
},
{
"epoch": 0.01,
"learning_rate": 4.9396408324887336e-05,
"loss": 5.1466,
"step": 36864
},
{
"epoch": 0.01,
"learning_rate": 4.9388022377376816e-05,
"loss": 5.1601,
"step": 37376
},
{
"epoch": 0.01,
"learning_rate": 4.9379636429866296e-05,
"loss": 5.1599,
"step": 37888
},
{
"epoch": 0.01,
"learning_rate": 4.9371250482355776e-05,
"loss": 5.1345,
"step": 38400
},
{
"epoch": 0.01,
"learning_rate": 4.9362864534845256e-05,
"loss": 5.1218,
"step": 38912
},
{
"epoch": 0.01,
"learning_rate": 4.935447858733473e-05,
"loss": 5.114,
"step": 39424
},
{
"epoch": 0.01,
"learning_rate": 4.934609263982421e-05,
"loss": 5.1038,
"step": 39936
},
{
"epoch": 0.01,
"learning_rate": 4.9337723071117425e-05,
"loss": 5.1139,
"step": 40448
},
{
"epoch": 0.01,
"learning_rate": 4.9329337123606905e-05,
"loss": 5.0934,
"step": 40960
},
{
"epoch": 0.01,
"learning_rate": 4.932095117609638e-05,
"loss": 5.1094,
"step": 41472
},
{
"epoch": 0.01,
"learning_rate": 4.931256522858586e-05,
"loss": 5.1001,
"step": 41984
},
{
"epoch": 0.01,
"learning_rate": 4.930417928107534e-05,
"loss": 5.0839,
"step": 42496
},
{
"epoch": 0.01,
"learning_rate": 4.929582609117229e-05,
"loss": 5.0688,
"step": 43008
},
{
"epoch": 0.01,
"learning_rate": 4.928744014366177e-05,
"loss": 5.0705,
"step": 43520
},
{
"epoch": 0.01,
"learning_rate": 4.927905419615125e-05,
"loss": 5.0713,
"step": 44032
},
{
"epoch": 0.01,
"learning_rate": 4.927066824864073e-05,
"loss": 5.0748,
"step": 44544
},
{
"epoch": 0.01,
"learning_rate": 4.92622823011302e-05,
"loss": 5.0549,
"step": 45056
},
{
"epoch": 0.01,
"learning_rate": 4.925389635361968e-05,
"loss": 5.0539,
"step": 45568
},
{
"epoch": 0.02,
"learning_rate": 4.924551040610916e-05,
"loss": 5.0498,
"step": 46080
},
{
"epoch": 0.02,
"learning_rate": 4.923712445859864e-05,
"loss": 5.0342,
"step": 46592
},
{
"epoch": 0.02,
"learning_rate": 4.922873851108812e-05,
"loss": 5.0315,
"step": 47104
},
{
"epoch": 0.02,
"learning_rate": 4.92203525635776e-05,
"loss": 5.03,
"step": 47616
},
{
"epoch": 0.02,
"learning_rate": 4.921196661606708e-05,
"loss": 5.0216,
"step": 48128
},
{
"epoch": 0.02,
"learning_rate": 4.920359704736029e-05,
"loss": 5.0207,
"step": 48640
},
{
"epoch": 0.02,
"learning_rate": 4.919521109984978e-05,
"loss": 5.0093,
"step": 49152
},
{
"epoch": 0.02,
"learning_rate": 4.918682515233926e-05,
"loss": 5.0018,
"step": 49664
},
{
"epoch": 0.02,
"learning_rate": 4.917843920482874e-05,
"loss": 5.0083,
"step": 50176
},
{
"epoch": 0.02,
"learning_rate": 4.917005325731822e-05,
"loss": 5.003,
"step": 50688
},
{
"epoch": 0.02,
"learning_rate": 4.916168368861143e-05,
"loss": 4.9973,
"step": 51200
},
{
"epoch": 0.02,
"learning_rate": 4.915329774110091e-05,
"loss": 4.9909,
"step": 51712
},
{
"epoch": 0.02,
"learning_rate": 4.914491179359039e-05,
"loss": 4.9808,
"step": 52224
},
{
"epoch": 0.02,
"learning_rate": 4.913652584607987e-05,
"loss": 4.9752,
"step": 52736
},
{
"epoch": 0.02,
"learning_rate": 4.912813989856935e-05,
"loss": 4.9725,
"step": 53248
},
{
"epoch": 0.02,
"learning_rate": 4.911975395105883e-05,
"loss": 4.9504,
"step": 53760
},
{
"epoch": 0.02,
"learning_rate": 4.9111384382352036e-05,
"loss": 4.9604,
"step": 54272
},
{
"epoch": 0.02,
"learning_rate": 4.9102998434841516e-05,
"loss": 4.9562,
"step": 54784
},
{
"epoch": 0.02,
"learning_rate": 4.9094612487330996e-05,
"loss": 4.959,
"step": 55296
},
{
"epoch": 0.02,
"learning_rate": 4.9086226539820476e-05,
"loss": 4.9439,
"step": 55808
},
{
"epoch": 0.02,
"learning_rate": 4.907784059230996e-05,
"loss": 4.9504,
"step": 56320
},
{
"epoch": 0.02,
"learning_rate": 4.906945464479944e-05,
"loss": 4.9265,
"step": 56832
},
{
"epoch": 0.02,
"learning_rate": 4.906106869728892e-05,
"loss": 4.9478,
"step": 57344
},
{
"epoch": 0.02,
"learning_rate": 4.9052682749778396e-05,
"loss": 4.9173,
"step": 57856
},
{
"epoch": 0.02,
"learning_rate": 4.9044296802267876e-05,
"loss": 4.9268,
"step": 58368
},
{
"epoch": 0.02,
"learning_rate": 4.903592723356109e-05,
"loss": 4.9273,
"step": 58880
},
{
"epoch": 0.02,
"learning_rate": 4.9027541286050565e-05,
"loss": 4.9219,
"step": 59392
},
{
"epoch": 0.02,
"learning_rate": 4.9019155338540045e-05,
"loss": 4.9183,
"step": 59904
},
{
"epoch": 0.02,
"learning_rate": 4.9010769391029525e-05,
"loss": 4.9056,
"step": 60416
},
{
"epoch": 0.02,
"learning_rate": 4.9002383443519005e-05,
"loss": 4.9103,
"step": 60928
},
{
"epoch": 0.02,
"learning_rate": 4.8993997496008485e-05,
"loss": 4.9138,
"step": 61440
},
{
"epoch": 0.02,
"learning_rate": 4.8985611548497965e-05,
"loss": 4.8967,
"step": 61952
},
{
"epoch": 0.02,
"learning_rate": 4.8977225600987445e-05,
"loss": 4.8921,
"step": 62464
},
{
"epoch": 0.02,
"learning_rate": 4.896885603228066e-05,
"loss": 4.8895,
"step": 62976
},
{
"epoch": 0.02,
"learning_rate": 4.896047008477014e-05,
"loss": 4.9014,
"step": 63488
},
{
"epoch": 0.02,
"learning_rate": 4.895210051606335e-05,
"loss": 4.8686,
"step": 64000
},
{
"epoch": 0.02,
"learning_rate": 4.894371456855283e-05,
"loss": 4.8732,
"step": 64512
},
{
"epoch": 0.02,
"learning_rate": 4.893534499984604e-05,
"loss": 4.8641,
"step": 65024
},
{
"epoch": 0.02,
"learning_rate": 4.892695905233552e-05,
"loss": 4.8732,
"step": 65536
},
{
"epoch": 0.02,
"learning_rate": 4.8918573104825e-05,
"loss": 4.8603,
"step": 66048
},
{
"epoch": 0.02,
"learning_rate": 4.891018715731448e-05,
"loss": 4.8786,
"step": 66560
},
{
"epoch": 0.02,
"learning_rate": 4.890180120980396e-05,
"loss": 4.8672,
"step": 67072
},
{
"epoch": 0.02,
"learning_rate": 4.889341526229344e-05,
"loss": 4.8588,
"step": 67584
},
{
"epoch": 0.02,
"learning_rate": 4.888502931478292e-05,
"loss": 4.8499,
"step": 68096
},
{
"epoch": 0.02,
"learning_rate": 4.88766433672724e-05,
"loss": 4.8562,
"step": 68608
},
{
"epoch": 0.02,
"learning_rate": 4.886825741976188e-05,
"loss": 4.8531,
"step": 69120
},
{
"epoch": 0.02,
"learning_rate": 4.8859871472251365e-05,
"loss": 4.8521,
"step": 69632
},
{
"epoch": 0.02,
"learning_rate": 4.8851485524740845e-05,
"loss": 4.8394,
"step": 70144
},
{
"epoch": 0.02,
"learning_rate": 4.8843099577230325e-05,
"loss": 4.8502,
"step": 70656
},
{
"epoch": 0.02,
"learning_rate": 4.8834730008523534e-05,
"loss": 4.8381,
"step": 71168
},
{
"epoch": 0.02,
"learning_rate": 4.8826344061013014e-05,
"loss": 4.8359,
"step": 71680
},
{
"epoch": 0.02,
"learning_rate": 4.8817958113502494e-05,
"loss": 4.8194,
"step": 72192
},
{
"epoch": 0.02,
"learning_rate": 4.88095885447957e-05,
"loss": 4.8344,
"step": 72704
},
{
"epoch": 0.02,
"learning_rate": 4.880120259728518e-05,
"loss": 4.8183,
"step": 73216
},
{
"epoch": 0.02,
"learning_rate": 4.879281664977466e-05,
"loss": 4.8001,
"step": 73728
},
{
"epoch": 0.02,
"learning_rate": 4.878443070226414e-05,
"loss": 4.8267,
"step": 74240
},
{
"epoch": 0.02,
"learning_rate": 4.877606113355735e-05,
"loss": 4.8134,
"step": 74752
},
{
"epoch": 0.02,
"learning_rate": 4.876767518604683e-05,
"loss": 4.8117,
"step": 75264
},
{
"epoch": 0.02,
"learning_rate": 4.875928923853632e-05,
"loss": 4.8098,
"step": 75776
},
{
"epoch": 0.02,
"learning_rate": 4.87509032910258e-05,
"loss": 4.7957,
"step": 76288
},
{
"epoch": 0.03,
"eval_loss": 4.76328706741333,
"eval_runtime": 292.912,
"eval_samples_per_second": 1302.75,
"eval_steps_per_second": 40.712,
"step": 76320
}
],
"logging_steps": 512,
"max_steps": 3052726,
"num_train_epochs": 9223372036854775807,
"save_steps": 10,
"total_flos": 3.145048926888413e+16,
"trial_name": null,
"trial_params": null
}