rel-cl-lstm-4 / checkpoint-76320 /trainer_state.json
abhinavp's picture
Training in progress, epoch 0, checkpoint
db91844 verified
{
"best_metric": 4.776597499847412,
"best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/rel-cl2/lstm/4/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": 10.82,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 4.999161405248948e-05,
"loss": 7.559,
"step": 512
},
{
"epoch": 0.0,
"learning_rate": 4.998322810497896e-05,
"loss": 7.0554,
"step": 1024
},
{
"epoch": 0.0,
"learning_rate": 4.997484215746844e-05,
"loss": 6.9937,
"step": 1536
},
{
"epoch": 0.0,
"learning_rate": 4.996645620995792e-05,
"loss": 6.9447,
"step": 2048
},
{
"epoch": 0.0,
"learning_rate": 4.99580702624474e-05,
"loss": 6.8936,
"step": 2560
},
{
"epoch": 0.0,
"learning_rate": 4.994968431493688e-05,
"loss": 6.7488,
"step": 3072
},
{
"epoch": 0.0,
"learning_rate": 4.994129836742636e-05,
"loss": 6.6478,
"step": 3584
},
{
"epoch": 0.0,
"learning_rate": 4.993291241991584e-05,
"loss": 6.5524,
"step": 4096
},
{
"epoch": 0.0,
"learning_rate": 4.992452647240532e-05,
"loss": 6.469,
"step": 4608
},
{
"epoch": 0.0,
"learning_rate": 4.99161405248948e-05,
"loss": 6.4073,
"step": 5120
},
{
"epoch": 0.0,
"learning_rate": 4.990775457738428e-05,
"loss": 6.3328,
"step": 5632
},
{
"epoch": 0.0,
"learning_rate": 4.989936862987376e-05,
"loss": 6.2605,
"step": 6144
},
{
"epoch": 0.0,
"learning_rate": 4.989099906116697e-05,
"loss": 6.197,
"step": 6656
},
{
"epoch": 0.0,
"learning_rate": 4.988261311365645e-05,
"loss": 6.1498,
"step": 7168
},
{
"epoch": 0.0,
"learning_rate": 4.987422716614593e-05,
"loss": 6.0945,
"step": 7680
},
{
"epoch": 0.0,
"learning_rate": 4.986584121863541e-05,
"loss": 6.0577,
"step": 8192
},
{
"epoch": 0.0,
"learning_rate": 4.985745527112489e-05,
"loss": 6.022,
"step": 8704
},
{
"epoch": 0.0,
"learning_rate": 4.984906932361437e-05,
"loss": 5.9791,
"step": 9216
},
{
"epoch": 0.0,
"learning_rate": 4.984068337610385e-05,
"loss": 5.9353,
"step": 9728
},
{
"epoch": 0.0,
"learning_rate": 4.983229742859333e-05,
"loss": 5.9076,
"step": 10240
},
{
"epoch": 0.0,
"learning_rate": 4.982391148108281e-05,
"loss": 5.865,
"step": 10752
},
{
"epoch": 0.0,
"learning_rate": 4.981552553357229e-05,
"loss": 5.8413,
"step": 11264
},
{
"epoch": 0.0,
"learning_rate": 4.980713958606178e-05,
"loss": 5.8045,
"step": 11776
},
{
"epoch": 0.0,
"learning_rate": 4.979875363855125e-05,
"loss": 5.7866,
"step": 12288
},
{
"epoch": 0.0,
"learning_rate": 4.979036769104073e-05,
"loss": 5.7486,
"step": 12800
},
{
"epoch": 0.0,
"learning_rate": 4.9781998122333946e-05,
"loss": 5.726,
"step": 13312
},
{
"epoch": 0.0,
"learning_rate": 4.9773612174823426e-05,
"loss": 5.6965,
"step": 13824
},
{
"epoch": 0.0,
"learning_rate": 4.97652262273129e-05,
"loss": 5.6795,
"step": 14336
},
{
"epoch": 0.0,
"learning_rate": 4.975684027980238e-05,
"loss": 5.6523,
"step": 14848
},
{
"epoch": 0.01,
"learning_rate": 4.9748487089899324e-05,
"loss": 5.6403,
"step": 15360
},
{
"epoch": 0.01,
"learning_rate": 4.9740117521192533e-05,
"loss": 5.6224,
"step": 15872
},
{
"epoch": 0.01,
"learning_rate": 4.9731731573682013e-05,
"loss": 5.5978,
"step": 16384
},
{
"epoch": 0.01,
"learning_rate": 4.97233456261715e-05,
"loss": 5.582,
"step": 16896
},
{
"epoch": 0.01,
"learning_rate": 4.971495967866098e-05,
"loss": 5.5683,
"step": 17408
},
{
"epoch": 0.01,
"learning_rate": 4.970657373115046e-05,
"loss": 5.5442,
"step": 17920
},
{
"epoch": 0.01,
"learning_rate": 4.969818778363994e-05,
"loss": 5.5402,
"step": 18432
},
{
"epoch": 0.01,
"learning_rate": 4.968980183612942e-05,
"loss": 5.5074,
"step": 18944
},
{
"epoch": 0.01,
"learning_rate": 4.96814158886189e-05,
"loss": 5.4925,
"step": 19456
},
{
"epoch": 0.01,
"learning_rate": 4.967302994110837e-05,
"loss": 5.4648,
"step": 19968
},
{
"epoch": 0.01,
"learning_rate": 4.966464399359785e-05,
"loss": 5.4631,
"step": 20480
},
{
"epoch": 0.01,
"learning_rate": 4.965625804608733e-05,
"loss": 5.4474,
"step": 20992
},
{
"epoch": 0.01,
"learning_rate": 4.964787209857681e-05,
"loss": 5.4437,
"step": 21504
},
{
"epoch": 0.01,
"learning_rate": 4.963950252987002e-05,
"loss": 5.416,
"step": 22016
},
{
"epoch": 0.01,
"learning_rate": 4.96311165823595e-05,
"loss": 5.4055,
"step": 22528
},
{
"epoch": 0.01,
"learning_rate": 4.962273063484898e-05,
"loss": 5.4031,
"step": 23040
},
{
"epoch": 0.01,
"learning_rate": 4.961434468733847e-05,
"loss": 5.3959,
"step": 23552
},
{
"epoch": 0.01,
"learning_rate": 4.960597511863168e-05,
"loss": 5.3829,
"step": 24064
},
{
"epoch": 0.01,
"learning_rate": 4.9597605549924894e-05,
"loss": 5.3633,
"step": 24576
},
{
"epoch": 0.01,
"learning_rate": 4.9589219602414374e-05,
"loss": 5.3487,
"step": 25088
},
{
"epoch": 0.01,
"learning_rate": 4.958083365490385e-05,
"loss": 5.3653,
"step": 25600
},
{
"epoch": 0.01,
"learning_rate": 4.957244770739333e-05,
"loss": 5.3515,
"step": 26112
},
{
"epoch": 0.01,
"learning_rate": 4.956406175988281e-05,
"loss": 5.3318,
"step": 26624
},
{
"epoch": 0.01,
"learning_rate": 4.955567581237229e-05,
"loss": 5.3092,
"step": 27136
},
{
"epoch": 0.01,
"learning_rate": 4.9547306243665496e-05,
"loss": 5.3194,
"step": 27648
},
{
"epoch": 0.01,
"learning_rate": 4.9538920296154976e-05,
"loss": 5.2973,
"step": 28160
},
{
"epoch": 0.01,
"learning_rate": 4.9530534348644456e-05,
"loss": 5.3025,
"step": 28672
},
{
"epoch": 0.01,
"learning_rate": 4.9522148401133936e-05,
"loss": 5.2727,
"step": 29184
},
{
"epoch": 0.01,
"learning_rate": 4.951376245362342e-05,
"loss": 5.2844,
"step": 29696
},
{
"epoch": 0.01,
"learning_rate": 4.95053765061129e-05,
"loss": 5.2682,
"step": 30208
},
{
"epoch": 0.01,
"learning_rate": 4.949699055860238e-05,
"loss": 5.2502,
"step": 30720
},
{
"epoch": 0.01,
"learning_rate": 4.948860461109186e-05,
"loss": 5.2366,
"step": 31232
},
{
"epoch": 0.01,
"learning_rate": 4.948023504238507e-05,
"loss": 5.2499,
"step": 31744
},
{
"epoch": 0.01,
"learning_rate": 4.947186547367828e-05,
"loss": 5.2068,
"step": 32256
},
{
"epoch": 0.01,
"learning_rate": 4.946347952616776e-05,
"loss": 5.2212,
"step": 32768
},
{
"epoch": 0.01,
"learning_rate": 4.945509357865724e-05,
"loss": 5.2284,
"step": 33280
},
{
"epoch": 0.01,
"learning_rate": 4.944670763114672e-05,
"loss": 5.2113,
"step": 33792
},
{
"epoch": 0.01,
"learning_rate": 4.94383216836362e-05,
"loss": 5.2028,
"step": 34304
},
{
"epoch": 0.01,
"learning_rate": 4.942993573612568e-05,
"loss": 5.1777,
"step": 34816
},
{
"epoch": 0.01,
"learning_rate": 4.942154978861516e-05,
"loss": 5.1749,
"step": 35328
},
{
"epoch": 0.01,
"learning_rate": 4.941316384110464e-05,
"loss": 5.1704,
"step": 35840
},
{
"epoch": 0.01,
"learning_rate": 4.9404794272397856e-05,
"loss": 5.1821,
"step": 36352
},
{
"epoch": 0.01,
"learning_rate": 4.9396408324887336e-05,
"loss": 5.1617,
"step": 36864
},
{
"epoch": 0.01,
"learning_rate": 4.9388022377376816e-05,
"loss": 5.1707,
"step": 37376
},
{
"epoch": 0.01,
"learning_rate": 4.9379652808670025e-05,
"loss": 5.1605,
"step": 37888
},
{
"epoch": 0.01,
"learning_rate": 4.9371266861159505e-05,
"loss": 5.1583,
"step": 38400
},
{
"epoch": 0.01,
"learning_rate": 4.9362880913648985e-05,
"loss": 5.1356,
"step": 38912
},
{
"epoch": 0.01,
"learning_rate": 4.9354511344942194e-05,
"loss": 5.1329,
"step": 39424
},
{
"epoch": 0.01,
"learning_rate": 4.9346125397431674e-05,
"loss": 5.1245,
"step": 39936
},
{
"epoch": 0.01,
"learning_rate": 4.9337739449921154e-05,
"loss": 5.121,
"step": 40448
},
{
"epoch": 0.01,
"learning_rate": 4.9329353502410634e-05,
"loss": 5.119,
"step": 40960
},
{
"epoch": 0.01,
"learning_rate": 4.9320967554900114e-05,
"loss": 5.1014,
"step": 41472
},
{
"epoch": 0.01,
"learning_rate": 4.9312581607389594e-05,
"loss": 5.1045,
"step": 41984
},
{
"epoch": 0.01,
"learning_rate": 4.9304195659879074e-05,
"loss": 5.1023,
"step": 42496
},
{
"epoch": 0.01,
"learning_rate": 4.9295809712368554e-05,
"loss": 5.0775,
"step": 43008
},
{
"epoch": 0.01,
"learning_rate": 4.9287423764858034e-05,
"loss": 5.0852,
"step": 43520
},
{
"epoch": 0.01,
"learning_rate": 4.9279037817347514e-05,
"loss": 5.0834,
"step": 44032
},
{
"epoch": 0.01,
"learning_rate": 4.9270651869836994e-05,
"loss": 5.085,
"step": 44544
},
{
"epoch": 0.01,
"learning_rate": 4.92622823011302e-05,
"loss": 5.066,
"step": 45056
},
{
"epoch": 0.01,
"learning_rate": 4.925389635361968e-05,
"loss": 5.0622,
"step": 45568
},
{
"epoch": 0.02,
"learning_rate": 4.924551040610916e-05,
"loss": 5.0487,
"step": 46080
},
{
"epoch": 0.02,
"learning_rate": 4.923712445859864e-05,
"loss": 5.044,
"step": 46592
},
{
"epoch": 0.02,
"learning_rate": 4.922875488989185e-05,
"loss": 5.045,
"step": 47104
},
{
"epoch": 0.02,
"learning_rate": 4.922036894238133e-05,
"loss": 5.0442,
"step": 47616
},
{
"epoch": 0.02,
"learning_rate": 4.921199937367455e-05,
"loss": 5.0354,
"step": 48128
},
{
"epoch": 0.02,
"learning_rate": 4.920361342616403e-05,
"loss": 5.0387,
"step": 48640
},
{
"epoch": 0.02,
"learning_rate": 4.919522747865351e-05,
"loss": 5.0267,
"step": 49152
},
{
"epoch": 0.02,
"learning_rate": 4.918684153114299e-05,
"loss": 5.0117,
"step": 49664
},
{
"epoch": 0.02,
"learning_rate": 4.917845558363247e-05,
"loss": 5.0086,
"step": 50176
},
{
"epoch": 0.02,
"learning_rate": 4.917006963612195e-05,
"loss": 5.0093,
"step": 50688
},
{
"epoch": 0.02,
"learning_rate": 4.916168368861143e-05,
"loss": 5.0068,
"step": 51200
},
{
"epoch": 0.02,
"learning_rate": 4.915329774110091e-05,
"loss": 5.0104,
"step": 51712
},
{
"epoch": 0.02,
"learning_rate": 4.914491179359039e-05,
"loss": 4.9925,
"step": 52224
},
{
"epoch": 0.02,
"learning_rate": 4.913652584607987e-05,
"loss": 4.9838,
"step": 52736
},
{
"epoch": 0.02,
"learning_rate": 4.912813989856935e-05,
"loss": 4.9888,
"step": 53248
},
{
"epoch": 0.02,
"learning_rate": 4.911975395105883e-05,
"loss": 4.9691,
"step": 53760
},
{
"epoch": 0.02,
"learning_rate": 4.9111400761155765e-05,
"loss": 4.9747,
"step": 54272
},
{
"epoch": 0.02,
"learning_rate": 4.9103014813645245e-05,
"loss": 4.9626,
"step": 54784
},
{
"epoch": 0.02,
"learning_rate": 4.909464524493846e-05,
"loss": 4.964,
"step": 55296
},
{
"epoch": 0.02,
"learning_rate": 4.908625929742794e-05,
"loss": 4.952,
"step": 55808
},
{
"epoch": 0.02,
"learning_rate": 4.907787334991742e-05,
"loss": 4.9532,
"step": 56320
},
{
"epoch": 0.02,
"learning_rate": 4.90694874024069e-05,
"loss": 4.943,
"step": 56832
},
{
"epoch": 0.02,
"learning_rate": 4.906110145489638e-05,
"loss": 4.9474,
"step": 57344
},
{
"epoch": 0.02,
"learning_rate": 4.905271550738586e-05,
"loss": 4.9426,
"step": 57856
},
{
"epoch": 0.02,
"learning_rate": 4.904432955987534e-05,
"loss": 4.9321,
"step": 58368
},
{
"epoch": 0.02,
"learning_rate": 4.903594361236482e-05,
"loss": 4.9421,
"step": 58880
},
{
"epoch": 0.02,
"learning_rate": 4.90275576648543e-05,
"loss": 4.9398,
"step": 59392
},
{
"epoch": 0.02,
"learning_rate": 4.901917171734378e-05,
"loss": 4.9296,
"step": 59904
},
{
"epoch": 0.02,
"learning_rate": 4.901078576983326e-05,
"loss": 4.9159,
"step": 60416
},
{
"epoch": 0.02,
"learning_rate": 4.900241620112647e-05,
"loss": 4.9172,
"step": 60928
},
{
"epoch": 0.02,
"learning_rate": 4.8994046632419686e-05,
"loss": 4.9134,
"step": 61440
},
{
"epoch": 0.02,
"learning_rate": 4.8985660684909166e-05,
"loss": 4.9165,
"step": 61952
},
{
"epoch": 0.02,
"learning_rate": 4.8977274737398646e-05,
"loss": 4.9096,
"step": 62464
},
{
"epoch": 0.02,
"learning_rate": 4.8968888789888125e-05,
"loss": 4.9001,
"step": 62976
},
{
"epoch": 0.02,
"learning_rate": 4.8960502842377605e-05,
"loss": 4.9031,
"step": 63488
},
{
"epoch": 0.02,
"learning_rate": 4.8952116894867085e-05,
"loss": 4.8894,
"step": 64000
},
{
"epoch": 0.02,
"learning_rate": 4.8943730947356565e-05,
"loss": 4.8945,
"step": 64512
},
{
"epoch": 0.02,
"learning_rate": 4.893534499984604e-05,
"loss": 4.8803,
"step": 65024
},
{
"epoch": 0.02,
"learning_rate": 4.8926975431139254e-05,
"loss": 4.8865,
"step": 65536
},
{
"epoch": 0.02,
"learning_rate": 4.8918589483628734e-05,
"loss": 4.8748,
"step": 66048
},
{
"epoch": 0.02,
"learning_rate": 4.8910203536118214e-05,
"loss": 4.8843,
"step": 66560
},
{
"epoch": 0.02,
"learning_rate": 4.890181758860769e-05,
"loss": 4.8833,
"step": 67072
},
{
"epoch": 0.02,
"learning_rate": 4.88934480199009e-05,
"loss": 4.8807,
"step": 67584
},
{
"epoch": 0.02,
"learning_rate": 4.888506207239038e-05,
"loss": 4.8567,
"step": 68096
},
{
"epoch": 0.02,
"learning_rate": 4.887667612487986e-05,
"loss": 4.8631,
"step": 68608
},
{
"epoch": 0.02,
"learning_rate": 4.886829017736934e-05,
"loss": 4.8608,
"step": 69120
},
{
"epoch": 0.02,
"learning_rate": 4.885990422985882e-05,
"loss": 4.864,
"step": 69632
},
{
"epoch": 0.02,
"learning_rate": 4.88515182823483e-05,
"loss": 4.852,
"step": 70144
},
{
"epoch": 0.02,
"learning_rate": 4.884313233483778e-05,
"loss": 4.8563,
"step": 70656
},
{
"epoch": 0.02,
"learning_rate": 4.883476276613099e-05,
"loss": 4.8458,
"step": 71168
},
{
"epoch": 0.02,
"learning_rate": 4.882637681862047e-05,
"loss": 4.8609,
"step": 71680
},
{
"epoch": 0.02,
"learning_rate": 4.881799087110995e-05,
"loss": 4.8292,
"step": 72192
},
{
"epoch": 0.02,
"learning_rate": 4.880960492359943e-05,
"loss": 4.8477,
"step": 72704
},
{
"epoch": 0.02,
"learning_rate": 4.880121897608891e-05,
"loss": 4.8323,
"step": 73216
},
{
"epoch": 0.02,
"learning_rate": 4.879284940738212e-05,
"loss": 4.8283,
"step": 73728
},
{
"epoch": 0.02,
"learning_rate": 4.87844634598716e-05,
"loss": 4.8276,
"step": 74240
},
{
"epoch": 0.02,
"learning_rate": 4.877607751236109e-05,
"loss": 4.813,
"step": 74752
},
{
"epoch": 0.02,
"learning_rate": 4.876769156485057e-05,
"loss": 4.8177,
"step": 75264
},
{
"epoch": 0.02,
"learning_rate": 4.875930561734005e-05,
"loss": 4.8258,
"step": 75776
},
{
"epoch": 0.02,
"learning_rate": 4.875093604863326e-05,
"loss": 4.8127,
"step": 76288
},
{
"epoch": 0.03,
"eval_loss": 4.776597499847412,
"eval_runtime": 334.2731,
"eval_samples_per_second": 1141.555,
"eval_steps_per_second": 35.674,
"step": 76320
}
],
"logging_steps": 512,
"max_steps": 3052726,
"num_train_epochs": 9223372036854775807,
"save_steps": 10,
"total_flos": 3.0937644943415676e+16,
"trial_name": null,
"trial_params": null
}