old-rel-cl-lstm-3 / checkpoint-76319 /trainer_state.json
abhinavp's picture
Training in progress, epoch 0, checkpoint
82b40b7 verified
{
"best_metric": 4.795548439025879,
"best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/rel-cl/lstm/3/checkpoints/checkpoint-76319",
"epoch": 0.025000278439663435,
"eval_steps": 10,
"global_step": 76319,
"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": 7.5697,
"step": 512
},
{
"epoch": 0.0,
"learning_rate": 4.998322810497896e-05,
"loss": 7.0565,
"step": 1024
},
{
"epoch": 0.0,
"learning_rate": 4.997484215746844e-05,
"loss": 6.9781,
"step": 1536
},
{
"epoch": 0.0,
"learning_rate": 4.996645620995792e-05,
"loss": 6.9507,
"step": 2048
},
{
"epoch": 0.0,
"learning_rate": 4.99580702624474e-05,
"loss": 6.8456,
"step": 2560
},
{
"epoch": 0.0,
"learning_rate": 4.994968431493688e-05,
"loss": 6.7236,
"step": 3072
},
{
"epoch": 0.0,
"learning_rate": 4.994131474623009e-05,
"loss": 6.6208,
"step": 3584
},
{
"epoch": 0.0,
"learning_rate": 4.993292879871958e-05,
"loss": 6.5339,
"step": 4096
},
{
"epoch": 0.0,
"learning_rate": 4.992454285120906e-05,
"loss": 6.4398,
"step": 4608
},
{
"epoch": 0.0,
"learning_rate": 4.991615690369854e-05,
"loss": 6.3912,
"step": 5120
},
{
"epoch": 0.0,
"learning_rate": 4.990777095618801e-05,
"loss": 6.3226,
"step": 5632
},
{
"epoch": 0.0,
"learning_rate": 4.989938500867749e-05,
"loss": 6.2576,
"step": 6144
},
{
"epoch": 0.0,
"learning_rate": 4.989099906116697e-05,
"loss": 6.191,
"step": 6656
},
{
"epoch": 0.0,
"learning_rate": 4.988261311365645e-05,
"loss": 6.1427,
"step": 7168
},
{
"epoch": 0.0,
"learning_rate": 4.987422716614593e-05,
"loss": 6.105,
"step": 7680
},
{
"epoch": 0.0,
"learning_rate": 4.986584121863541e-05,
"loss": 6.0659,
"step": 8192
},
{
"epoch": 0.0,
"learning_rate": 4.985745527112489e-05,
"loss": 6.0125,
"step": 8704
},
{
"epoch": 0.0,
"learning_rate": 4.984906932361437e-05,
"loss": 5.9768,
"step": 9216
},
{
"epoch": 0.0,
"learning_rate": 4.984068337610385e-05,
"loss": 5.9418,
"step": 9728
},
{
"epoch": 0.0,
"learning_rate": 4.983229742859333e-05,
"loss": 5.9049,
"step": 10240
},
{
"epoch": 0.0,
"learning_rate": 4.982391148108281e-05,
"loss": 5.8781,
"step": 10752
},
{
"epoch": 0.0,
"learning_rate": 4.981552553357229e-05,
"loss": 5.8498,
"step": 11264
},
{
"epoch": 0.0,
"learning_rate": 4.9807172343669236e-05,
"loss": 5.8175,
"step": 11776
},
{
"epoch": 0.0,
"learning_rate": 4.9798786396158716e-05,
"loss": 5.7889,
"step": 12288
},
{
"epoch": 0.0,
"learning_rate": 4.9790400448648195e-05,
"loss": 5.7723,
"step": 12800
},
{
"epoch": 0.0,
"learning_rate": 4.9782014501137675e-05,
"loss": 5.7401,
"step": 13312
},
{
"epoch": 0.0,
"learning_rate": 4.9773628553627155e-05,
"loss": 5.7195,
"step": 13824
},
{
"epoch": 0.0,
"learning_rate": 4.9765242606116635e-05,
"loss": 5.6842,
"step": 14336
},
{
"epoch": 0.0,
"learning_rate": 4.9756856658606115e-05,
"loss": 5.6802,
"step": 14848
},
{
"epoch": 0.01,
"learning_rate": 4.9748470711095595e-05,
"loss": 5.6467,
"step": 15360
},
{
"epoch": 0.01,
"learning_rate": 4.974008476358507e-05,
"loss": 5.6311,
"step": 15872
},
{
"epoch": 0.01,
"learning_rate": 4.973169881607455e-05,
"loss": 5.6193,
"step": 16384
},
{
"epoch": 0.01,
"learning_rate": 4.972331286856403e-05,
"loss": 5.602,
"step": 16896
},
{
"epoch": 0.01,
"learning_rate": 4.9714943299857244e-05,
"loss": 5.5709,
"step": 17408
},
{
"epoch": 0.01,
"learning_rate": 4.9706557352346724e-05,
"loss": 5.5577,
"step": 17920
},
{
"epoch": 0.01,
"learning_rate": 4.969818778363994e-05,
"loss": 5.5515,
"step": 18432
},
{
"epoch": 0.01,
"learning_rate": 4.968980183612942e-05,
"loss": 5.5266,
"step": 18944
},
{
"epoch": 0.01,
"learning_rate": 4.96814158886189e-05,
"loss": 5.5154,
"step": 19456
},
{
"epoch": 0.01,
"learning_rate": 4.967302994110837e-05,
"loss": 5.4951,
"step": 19968
},
{
"epoch": 0.01,
"learning_rate": 4.966464399359785e-05,
"loss": 5.4707,
"step": 20480
},
{
"epoch": 0.01,
"learning_rate": 4.965625804608733e-05,
"loss": 5.4599,
"step": 20992
},
{
"epoch": 0.01,
"learning_rate": 4.964788847738054e-05,
"loss": 5.4448,
"step": 21504
},
{
"epoch": 0.01,
"learning_rate": 4.963950252987002e-05,
"loss": 5.4383,
"step": 22016
},
{
"epoch": 0.01,
"learning_rate": 4.96311165823595e-05,
"loss": 5.4263,
"step": 22528
},
{
"epoch": 0.01,
"learning_rate": 4.962273063484898e-05,
"loss": 5.4202,
"step": 23040
},
{
"epoch": 0.01,
"learning_rate": 4.961434468733847e-05,
"loss": 5.3964,
"step": 23552
},
{
"epoch": 0.01,
"learning_rate": 4.960595873982795e-05,
"loss": 5.3959,
"step": 24064
},
{
"epoch": 0.01,
"learning_rate": 4.959757279231743e-05,
"loss": 5.3996,
"step": 24576
},
{
"epoch": 0.01,
"learning_rate": 4.958918684480691e-05,
"loss": 5.3719,
"step": 25088
},
{
"epoch": 0.01,
"learning_rate": 4.958080089729639e-05,
"loss": 5.3628,
"step": 25600
},
{
"epoch": 0.01,
"learning_rate": 4.957241494978587e-05,
"loss": 5.3496,
"step": 26112
},
{
"epoch": 0.01,
"learning_rate": 4.956402900227535e-05,
"loss": 5.359,
"step": 26624
},
{
"epoch": 0.01,
"learning_rate": 4.955565943356856e-05,
"loss": 5.3201,
"step": 27136
},
{
"epoch": 0.01,
"learning_rate": 4.954727348605804e-05,
"loss": 5.3384,
"step": 27648
},
{
"epoch": 0.01,
"learning_rate": 4.953888753854752e-05,
"loss": 5.3088,
"step": 28160
},
{
"epoch": 0.01,
"learning_rate": 4.9530501591037e-05,
"loss": 5.3052,
"step": 28672
},
{
"epoch": 0.01,
"learning_rate": 4.952211564352648e-05,
"loss": 5.2998,
"step": 29184
},
{
"epoch": 0.01,
"learning_rate": 4.951372969601596e-05,
"loss": 5.2898,
"step": 29696
},
{
"epoch": 0.01,
"learning_rate": 4.95053765061129e-05,
"loss": 5.277,
"step": 30208
},
{
"epoch": 0.01,
"learning_rate": 4.949699055860238e-05,
"loss": 5.2662,
"step": 30720
},
{
"epoch": 0.01,
"learning_rate": 4.948860461109186e-05,
"loss": 5.251,
"step": 31232
},
{
"epoch": 0.01,
"learning_rate": 4.948021866358134e-05,
"loss": 5.2581,
"step": 31744
},
{
"epoch": 0.01,
"learning_rate": 4.947183271607082e-05,
"loss": 5.245,
"step": 32256
},
{
"epoch": 0.01,
"learning_rate": 4.94634467685603e-05,
"loss": 5.2357,
"step": 32768
},
{
"epoch": 0.01,
"learning_rate": 4.945506082104978e-05,
"loss": 5.2293,
"step": 33280
},
{
"epoch": 0.01,
"learning_rate": 4.944669125234299e-05,
"loss": 5.2185,
"step": 33792
},
{
"epoch": 0.01,
"learning_rate": 4.943830530483247e-05,
"loss": 5.2008,
"step": 34304
},
{
"epoch": 0.01,
"learning_rate": 4.942991935732195e-05,
"loss": 5.2185,
"step": 34816
},
{
"epoch": 0.01,
"learning_rate": 4.942153340981143e-05,
"loss": 5.1957,
"step": 35328
},
{
"epoch": 0.01,
"learning_rate": 4.9413147462300904e-05,
"loss": 5.1875,
"step": 35840
},
{
"epoch": 0.01,
"learning_rate": 4.940477789359412e-05,
"loss": 5.1983,
"step": 36352
},
{
"epoch": 0.01,
"learning_rate": 4.93963919460836e-05,
"loss": 5.1737,
"step": 36864
},
{
"epoch": 0.01,
"learning_rate": 4.938800599857309e-05,
"loss": 5.1873,
"step": 37376
},
{
"epoch": 0.01,
"learning_rate": 4.937962005106256e-05,
"loss": 5.1776,
"step": 37888
},
{
"epoch": 0.01,
"learning_rate": 4.9371250482355776e-05,
"loss": 5.1662,
"step": 38400
},
{
"epoch": 0.01,
"learning_rate": 4.9362864534845256e-05,
"loss": 5.1671,
"step": 38912
},
{
"epoch": 0.01,
"learning_rate": 4.935447858733473e-05,
"loss": 5.141,
"step": 39424
},
{
"epoch": 0.01,
"learning_rate": 4.934609263982421e-05,
"loss": 5.1373,
"step": 39936
},
{
"epoch": 0.01,
"learning_rate": 4.933770669231369e-05,
"loss": 5.1362,
"step": 40448
},
{
"epoch": 0.01,
"learning_rate": 4.9329337123606905e-05,
"loss": 5.1354,
"step": 40960
},
{
"epoch": 0.01,
"learning_rate": 4.932095117609638e-05,
"loss": 5.1243,
"step": 41472
},
{
"epoch": 0.01,
"learning_rate": 4.931256522858586e-05,
"loss": 5.1112,
"step": 41984
},
{
"epoch": 0.01,
"learning_rate": 4.930417928107534e-05,
"loss": 5.0992,
"step": 42496
},
{
"epoch": 0.01,
"learning_rate": 4.9295793333564825e-05,
"loss": 5.105,
"step": 43008
},
{
"epoch": 0.01,
"learning_rate": 4.9287423764858034e-05,
"loss": 5.1043,
"step": 43520
},
{
"epoch": 0.01,
"learning_rate": 4.927905419615125e-05,
"loss": 5.1007,
"step": 44032
},
{
"epoch": 0.01,
"learning_rate": 4.927066824864073e-05,
"loss": 5.0981,
"step": 44544
},
{
"epoch": 0.01,
"learning_rate": 4.92622823011302e-05,
"loss": 5.0934,
"step": 45056
},
{
"epoch": 0.01,
"learning_rate": 4.925389635361968e-05,
"loss": 5.0793,
"step": 45568
},
{
"epoch": 0.02,
"learning_rate": 4.924551040610916e-05,
"loss": 5.0627,
"step": 46080
},
{
"epoch": 0.02,
"learning_rate": 4.923712445859864e-05,
"loss": 5.0591,
"step": 46592
},
{
"epoch": 0.02,
"learning_rate": 4.922873851108812e-05,
"loss": 5.0652,
"step": 47104
},
{
"epoch": 0.02,
"learning_rate": 4.92203525635776e-05,
"loss": 5.0622,
"step": 47616
},
{
"epoch": 0.02,
"learning_rate": 4.921198299487081e-05,
"loss": 5.0404,
"step": 48128
},
{
"epoch": 0.02,
"learning_rate": 4.920359704736029e-05,
"loss": 5.0438,
"step": 48640
},
{
"epoch": 0.02,
"learning_rate": 4.919521109984978e-05,
"loss": 5.042,
"step": 49152
},
{
"epoch": 0.02,
"learning_rate": 4.918682515233926e-05,
"loss": 5.0424,
"step": 49664
},
{
"epoch": 0.02,
"learning_rate": 4.917843920482874e-05,
"loss": 5.0343,
"step": 50176
},
{
"epoch": 0.02,
"learning_rate": 4.917005325731822e-05,
"loss": 5.034,
"step": 50688
},
{
"epoch": 0.02,
"learning_rate": 4.916168368861143e-05,
"loss": 5.0269,
"step": 51200
},
{
"epoch": 0.02,
"learning_rate": 4.915329774110091e-05,
"loss": 5.0175,
"step": 51712
},
{
"epoch": 0.02,
"learning_rate": 4.914491179359039e-05,
"loss": 5.0164,
"step": 52224
},
{
"epoch": 0.02,
"learning_rate": 4.913652584607987e-05,
"loss": 5.0178,
"step": 52736
},
{
"epoch": 0.02,
"learning_rate": 4.912813989856935e-05,
"loss": 5.0013,
"step": 53248
},
{
"epoch": 0.02,
"learning_rate": 4.9119770329862556e-05,
"loss": 5.0064,
"step": 53760
},
{
"epoch": 0.02,
"learning_rate": 4.9111384382352036e-05,
"loss": 4.9835,
"step": 54272
},
{
"epoch": 0.02,
"learning_rate": 4.9102998434841516e-05,
"loss": 4.9785,
"step": 54784
},
{
"epoch": 0.02,
"learning_rate": 4.909462886613473e-05,
"loss": 4.9778,
"step": 55296
},
{
"epoch": 0.02,
"learning_rate": 4.908624291862421e-05,
"loss": 4.9824,
"step": 55808
},
{
"epoch": 0.02,
"learning_rate": 4.907785697111369e-05,
"loss": 4.9744,
"step": 56320
},
{
"epoch": 0.02,
"learning_rate": 4.906947102360317e-05,
"loss": 4.9516,
"step": 56832
},
{
"epoch": 0.02,
"learning_rate": 4.906108507609265e-05,
"loss": 4.9571,
"step": 57344
},
{
"epoch": 0.02,
"learning_rate": 4.905269912858213e-05,
"loss": 4.9606,
"step": 57856
},
{
"epoch": 0.02,
"learning_rate": 4.904431318107161e-05,
"loss": 4.9677,
"step": 58368
},
{
"epoch": 0.02,
"learning_rate": 4.903592723356109e-05,
"loss": 4.9534,
"step": 58880
},
{
"epoch": 0.02,
"learning_rate": 4.9027541286050565e-05,
"loss": 4.9526,
"step": 59392
},
{
"epoch": 0.02,
"learning_rate": 4.9019155338540045e-05,
"loss": 4.9496,
"step": 59904
},
{
"epoch": 0.02,
"learning_rate": 4.9010769391029525e-05,
"loss": 4.9506,
"step": 60416
},
{
"epoch": 0.02,
"learning_rate": 4.9002383443519005e-05,
"loss": 4.9337,
"step": 60928
},
{
"epoch": 0.02,
"learning_rate": 4.8993997496008485e-05,
"loss": 4.9347,
"step": 61440
},
{
"epoch": 0.02,
"learning_rate": 4.89856279273017e-05,
"loss": 4.9256,
"step": 61952
},
{
"epoch": 0.02,
"learning_rate": 4.897725835859491e-05,
"loss": 4.9299,
"step": 62464
},
{
"epoch": 0.02,
"learning_rate": 4.8968872411084396e-05,
"loss": 4.9112,
"step": 62976
},
{
"epoch": 0.02,
"learning_rate": 4.8960502842377605e-05,
"loss": 4.9247,
"step": 63488
},
{
"epoch": 0.02,
"learning_rate": 4.8952116894867085e-05,
"loss": 4.9142,
"step": 64000
},
{
"epoch": 0.02,
"learning_rate": 4.8943730947356565e-05,
"loss": 4.9212,
"step": 64512
},
{
"epoch": 0.02,
"learning_rate": 4.893534499984604e-05,
"loss": 4.8942,
"step": 65024
},
{
"epoch": 0.02,
"learning_rate": 4.892695905233552e-05,
"loss": 4.9051,
"step": 65536
},
{
"epoch": 0.02,
"learning_rate": 4.8918573104825e-05,
"loss": 4.9138,
"step": 66048
},
{
"epoch": 0.02,
"learning_rate": 4.891018715731448e-05,
"loss": 4.8863,
"step": 66560
},
{
"epoch": 0.02,
"learning_rate": 4.890180120980396e-05,
"loss": 4.905,
"step": 67072
},
{
"epoch": 0.02,
"learning_rate": 4.889343164109717e-05,
"loss": 4.8891,
"step": 67584
},
{
"epoch": 0.02,
"learning_rate": 4.888506207239038e-05,
"loss": 4.8928,
"step": 68096
},
{
"epoch": 0.02,
"learning_rate": 4.887667612487986e-05,
"loss": 4.8827,
"step": 68608
},
{
"epoch": 0.02,
"learning_rate": 4.886829017736934e-05,
"loss": 4.8781,
"step": 69120
},
{
"epoch": 0.02,
"learning_rate": 4.885990422985882e-05,
"loss": 4.8776,
"step": 69632
},
{
"epoch": 0.02,
"learning_rate": 4.88515182823483e-05,
"loss": 4.877,
"step": 70144
},
{
"epoch": 0.02,
"learning_rate": 4.884313233483778e-05,
"loss": 4.8761,
"step": 70656
},
{
"epoch": 0.02,
"learning_rate": 4.883474638732726e-05,
"loss": 4.8701,
"step": 71168
},
{
"epoch": 0.02,
"learning_rate": 4.882636043981674e-05,
"loss": 4.8734,
"step": 71680
},
{
"epoch": 0.02,
"learning_rate": 4.881799087110995e-05,
"loss": 4.8711,
"step": 72192
},
{
"epoch": 0.02,
"learning_rate": 4.880960492359943e-05,
"loss": 4.8518,
"step": 72704
},
{
"epoch": 0.02,
"learning_rate": 4.880121897608891e-05,
"loss": 4.8516,
"step": 73216
},
{
"epoch": 0.02,
"learning_rate": 4.879283302857839e-05,
"loss": 4.8425,
"step": 73728
},
{
"epoch": 0.02,
"learning_rate": 4.878444708106787e-05,
"loss": 4.843,
"step": 74240
},
{
"epoch": 0.02,
"learning_rate": 4.877606113355735e-05,
"loss": 4.8422,
"step": 74752
},
{
"epoch": 0.02,
"learning_rate": 4.876767518604683e-05,
"loss": 4.8361,
"step": 75264
},
{
"epoch": 0.02,
"learning_rate": 4.875928923853632e-05,
"loss": 4.8333,
"step": 75776
},
{
"epoch": 0.02,
"learning_rate": 4.875091966982953e-05,
"loss": 4.8333,
"step": 76288
},
{
"epoch": 0.03,
"eval_loss": 4.795548439025879,
"eval_runtime": 544.0651,
"eval_samples_per_second": 701.37,
"eval_steps_per_second": 21.918,
"step": 76319
}
],
"logging_steps": 512,
"max_steps": 3052726,
"num_train_epochs": 9223372036854775807,
"save_steps": 10,
"total_flos": 3.090533787047635e+16,
"trial_name": null,
"trial_params": null
}