npi-only-lstm-3 / checkpoint-76320 /trainer_state.json
abhinavp's picture
Training in progress, epoch 0, checkpoint
1ac83a1 verified
raw
history blame
19 kB
{
"best_metric": 4.750964641571045,
"best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/npi-only/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": 10.8208,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 4.999161405248948e-05,
"loss": 7.5525,
"step": 512
},
{
"epoch": 0.0,
"learning_rate": 4.998322810497896e-05,
"loss": 7.0564,
"step": 1024
},
{
"epoch": 0.0,
"learning_rate": 4.997484215746844e-05,
"loss": 6.9917,
"step": 1536
},
{
"epoch": 0.0,
"learning_rate": 4.996645620995792e-05,
"loss": 6.9456,
"step": 2048
},
{
"epoch": 0.0,
"learning_rate": 4.99580702624474e-05,
"loss": 6.8947,
"step": 2560
},
{
"epoch": 0.0,
"learning_rate": 4.994968431493688e-05,
"loss": 6.7148,
"step": 3072
},
{
"epoch": 0.0,
"learning_rate": 4.994129836742636e-05,
"loss": 6.608,
"step": 3584
},
{
"epoch": 0.0,
"learning_rate": 4.993291241991584e-05,
"loss": 6.5116,
"step": 4096
},
{
"epoch": 0.0,
"learning_rate": 4.992452647240532e-05,
"loss": 6.4526,
"step": 4608
},
{
"epoch": 0.0,
"learning_rate": 4.99161405248948e-05,
"loss": 6.4085,
"step": 5120
},
{
"epoch": 0.0,
"learning_rate": 4.990775457738428e-05,
"loss": 6.3554,
"step": 5632
},
{
"epoch": 0.0,
"learning_rate": 4.989936862987376e-05,
"loss": 6.2919,
"step": 6144
},
{
"epoch": 0.0,
"learning_rate": 4.989098268236324e-05,
"loss": 6.2369,
"step": 6656
},
{
"epoch": 0.0,
"learning_rate": 4.988259673485272e-05,
"loss": 6.1912,
"step": 7168
},
{
"epoch": 0.0,
"learning_rate": 4.98742107873422e-05,
"loss": 6.1306,
"step": 7680
},
{
"epoch": 0.0,
"learning_rate": 4.986582483983168e-05,
"loss": 6.0968,
"step": 8192
},
{
"epoch": 0.0,
"learning_rate": 4.985743889232116e-05,
"loss": 6.0456,
"step": 8704
},
{
"epoch": 0.0,
"learning_rate": 4.984906932361437e-05,
"loss": 5.9932,
"step": 9216
},
{
"epoch": 0.0,
"learning_rate": 4.984068337610385e-05,
"loss": 5.9602,
"step": 9728
},
{
"epoch": 0.0,
"learning_rate": 4.983229742859333e-05,
"loss": 5.9203,
"step": 10240
},
{
"epoch": 0.0,
"learning_rate": 4.9823927859886547e-05,
"loss": 5.8789,
"step": 10752
},
{
"epoch": 0.0,
"learning_rate": 4.9815541912376026e-05,
"loss": 5.8514,
"step": 11264
},
{
"epoch": 0.0,
"learning_rate": 4.9807155964865506e-05,
"loss": 5.8072,
"step": 11776
},
{
"epoch": 0.0,
"learning_rate": 4.9798770017354986e-05,
"loss": 5.786,
"step": 12288
},
{
"epoch": 0.0,
"learning_rate": 4.9790384069844466e-05,
"loss": 5.7572,
"step": 12800
},
{
"epoch": 0.0,
"learning_rate": 4.9781998122333946e-05,
"loss": 5.7303,
"step": 13312
},
{
"epoch": 0.0,
"learning_rate": 4.9773612174823426e-05,
"loss": 5.6995,
"step": 13824
},
{
"epoch": 0.0,
"learning_rate": 4.97652262273129e-05,
"loss": 5.6648,
"step": 14336
},
{
"epoch": 0.0,
"learning_rate": 4.975684027980238e-05,
"loss": 5.652,
"step": 14848
},
{
"epoch": 0.01,
"learning_rate": 4.974845433229186e-05,
"loss": 5.6281,
"step": 15360
},
{
"epoch": 0.01,
"learning_rate": 4.974008476358507e-05,
"loss": 5.6148,
"step": 15872
},
{
"epoch": 0.01,
"learning_rate": 4.973169881607455e-05,
"loss": 5.5876,
"step": 16384
},
{
"epoch": 0.01,
"learning_rate": 4.972331286856403e-05,
"loss": 5.5684,
"step": 16896
},
{
"epoch": 0.01,
"learning_rate": 4.9714926921053515e-05,
"loss": 5.5576,
"step": 17408
},
{
"epoch": 0.01,
"learning_rate": 4.9706540973542995e-05,
"loss": 5.5397,
"step": 17920
},
{
"epoch": 0.01,
"learning_rate": 4.9698171404836204e-05,
"loss": 5.5201,
"step": 18432
},
{
"epoch": 0.01,
"learning_rate": 4.9689785457325684e-05,
"loss": 5.4885,
"step": 18944
},
{
"epoch": 0.01,
"learning_rate": 4.9681399509815164e-05,
"loss": 5.4832,
"step": 19456
},
{
"epoch": 0.01,
"learning_rate": 4.9673013562304644e-05,
"loss": 5.4514,
"step": 19968
},
{
"epoch": 0.01,
"learning_rate": 4.966464399359785e-05,
"loss": 5.4533,
"step": 20480
},
{
"epoch": 0.01,
"learning_rate": 4.965625804608733e-05,
"loss": 5.427,
"step": 20992
},
{
"epoch": 0.01,
"learning_rate": 4.964787209857681e-05,
"loss": 5.4288,
"step": 21504
},
{
"epoch": 0.01,
"learning_rate": 4.963948615106629e-05,
"loss": 5.3997,
"step": 22016
},
{
"epoch": 0.01,
"learning_rate": 4.963110020355577e-05,
"loss": 5.3924,
"step": 22528
},
{
"epoch": 0.01,
"learning_rate": 4.962273063484898e-05,
"loss": 5.3859,
"step": 23040
},
{
"epoch": 0.01,
"learning_rate": 4.961434468733847e-05,
"loss": 5.3776,
"step": 23552
},
{
"epoch": 0.01,
"learning_rate": 4.960597511863168e-05,
"loss": 5.3604,
"step": 24064
},
{
"epoch": 0.01,
"learning_rate": 4.959758917112116e-05,
"loss": 5.3383,
"step": 24576
},
{
"epoch": 0.01,
"learning_rate": 4.958920322361064e-05,
"loss": 5.3357,
"step": 25088
},
{
"epoch": 0.01,
"learning_rate": 4.958081727610012e-05,
"loss": 5.3381,
"step": 25600
},
{
"epoch": 0.01,
"learning_rate": 4.95724313285896e-05,
"loss": 5.3259,
"step": 26112
},
{
"epoch": 0.01,
"learning_rate": 4.956404538107908e-05,
"loss": 5.3046,
"step": 26624
},
{
"epoch": 0.01,
"learning_rate": 4.955565943356856e-05,
"loss": 5.2941,
"step": 27136
},
{
"epoch": 0.01,
"learning_rate": 4.954727348605804e-05,
"loss": 5.2901,
"step": 27648
},
{
"epoch": 0.01,
"learning_rate": 4.953890391735125e-05,
"loss": 5.2767,
"step": 28160
},
{
"epoch": 0.01,
"learning_rate": 4.9530517969840727e-05,
"loss": 5.2868,
"step": 28672
},
{
"epoch": 0.01,
"learning_rate": 4.9522132022330207e-05,
"loss": 5.2426,
"step": 29184
},
{
"epoch": 0.01,
"learning_rate": 4.9513746074819686e-05,
"loss": 5.2558,
"step": 29696
},
{
"epoch": 0.01,
"learning_rate": 4.9505360127309166e-05,
"loss": 5.2416,
"step": 30208
},
{
"epoch": 0.01,
"learning_rate": 4.9496974179798646e-05,
"loss": 5.2263,
"step": 30720
},
{
"epoch": 0.01,
"learning_rate": 4.948860461109186e-05,
"loss": 5.234,
"step": 31232
},
{
"epoch": 0.01,
"learning_rate": 4.948021866358134e-05,
"loss": 5.212,
"step": 31744
},
{
"epoch": 0.01,
"learning_rate": 4.947183271607082e-05,
"loss": 5.2018,
"step": 32256
},
{
"epoch": 0.01,
"learning_rate": 4.94634467685603e-05,
"loss": 5.1982,
"step": 32768
},
{
"epoch": 0.01,
"learning_rate": 4.945507719985351e-05,
"loss": 5.209,
"step": 33280
},
{
"epoch": 0.01,
"learning_rate": 4.944669125234299e-05,
"loss": 5.1833,
"step": 33792
},
{
"epoch": 0.01,
"learning_rate": 4.94383216836362e-05,
"loss": 5.1782,
"step": 34304
},
{
"epoch": 0.01,
"learning_rate": 4.942993573612568e-05,
"loss": 5.1637,
"step": 34816
},
{
"epoch": 0.01,
"learning_rate": 4.942154978861516e-05,
"loss": 5.1557,
"step": 35328
},
{
"epoch": 0.01,
"learning_rate": 4.941316384110464e-05,
"loss": 5.161,
"step": 35840
},
{
"epoch": 0.01,
"learning_rate": 4.940477789359412e-05,
"loss": 5.165,
"step": 36352
},
{
"epoch": 0.01,
"learning_rate": 4.93963919460836e-05,
"loss": 5.1376,
"step": 36864
},
{
"epoch": 0.01,
"learning_rate": 4.938800599857309e-05,
"loss": 5.148,
"step": 37376
},
{
"epoch": 0.01,
"learning_rate": 4.937962005106256e-05,
"loss": 5.1488,
"step": 37888
},
{
"epoch": 0.01,
"learning_rate": 4.937123410355204e-05,
"loss": 5.1286,
"step": 38400
},
{
"epoch": 0.01,
"learning_rate": 4.9362864534845256e-05,
"loss": 5.1109,
"step": 38912
},
{
"epoch": 0.01,
"learning_rate": 4.935447858733473e-05,
"loss": 5.1119,
"step": 39424
},
{
"epoch": 0.01,
"learning_rate": 4.934609263982421e-05,
"loss": 5.1045,
"step": 39936
},
{
"epoch": 0.01,
"learning_rate": 4.933770669231369e-05,
"loss": 5.0941,
"step": 40448
},
{
"epoch": 0.01,
"learning_rate": 4.932932074480317e-05,
"loss": 5.0918,
"step": 40960
},
{
"epoch": 0.01,
"learning_rate": 4.932095117609638e-05,
"loss": 5.0943,
"step": 41472
},
{
"epoch": 0.01,
"learning_rate": 4.931256522858586e-05,
"loss": 5.0933,
"step": 41984
},
{
"epoch": 0.01,
"learning_rate": 4.930417928107534e-05,
"loss": 5.0756,
"step": 42496
},
{
"epoch": 0.01,
"learning_rate": 4.9295809712368554e-05,
"loss": 5.0548,
"step": 43008
},
{
"epoch": 0.01,
"learning_rate": 4.9287423764858034e-05,
"loss": 5.0617,
"step": 43520
},
{
"epoch": 0.01,
"learning_rate": 4.9279037817347514e-05,
"loss": 5.0651,
"step": 44032
},
{
"epoch": 0.01,
"learning_rate": 4.9270651869836994e-05,
"loss": 5.0534,
"step": 44544
},
{
"epoch": 0.01,
"learning_rate": 4.9262265922326474e-05,
"loss": 5.0504,
"step": 45056
},
{
"epoch": 0.01,
"learning_rate": 4.9253879974815954e-05,
"loss": 5.0318,
"step": 45568
},
{
"epoch": 0.02,
"learning_rate": 4.9245494027305433e-05,
"loss": 5.0316,
"step": 46080
},
{
"epoch": 0.02,
"learning_rate": 4.9237108079794913e-05,
"loss": 5.0262,
"step": 46592
},
{
"epoch": 0.02,
"learning_rate": 4.922873851108812e-05,
"loss": 5.0245,
"step": 47104
},
{
"epoch": 0.02,
"learning_rate": 4.92203525635776e-05,
"loss": 5.0253,
"step": 47616
},
{
"epoch": 0.02,
"learning_rate": 4.921196661606708e-05,
"loss": 5.0075,
"step": 48128
},
{
"epoch": 0.02,
"learning_rate": 4.920358066855656e-05,
"loss": 5.0243,
"step": 48640
},
{
"epoch": 0.02,
"learning_rate": 4.919521109984978e-05,
"loss": 5.0082,
"step": 49152
},
{
"epoch": 0.02,
"learning_rate": 4.918682515233926e-05,
"loss": 4.989,
"step": 49664
},
{
"epoch": 0.02,
"learning_rate": 4.917843920482874e-05,
"loss": 4.9839,
"step": 50176
},
{
"epoch": 0.02,
"learning_rate": 4.917005325731822e-05,
"loss": 4.9849,
"step": 50688
},
{
"epoch": 0.02,
"learning_rate": 4.91616673098077e-05,
"loss": 4.983,
"step": 51200
},
{
"epoch": 0.02,
"learning_rate": 4.915328136229718e-05,
"loss": 4.9812,
"step": 51712
},
{
"epoch": 0.02,
"learning_rate": 4.914489541478666e-05,
"loss": 4.9736,
"step": 52224
},
{
"epoch": 0.02,
"learning_rate": 4.913650946727614e-05,
"loss": 4.9657,
"step": 52736
},
{
"epoch": 0.02,
"learning_rate": 4.912813989856935e-05,
"loss": 4.9618,
"step": 53248
},
{
"epoch": 0.02,
"learning_rate": 4.911975395105883e-05,
"loss": 4.9411,
"step": 53760
},
{
"epoch": 0.02,
"learning_rate": 4.911136800354831e-05,
"loss": 4.945,
"step": 54272
},
{
"epoch": 0.02,
"learning_rate": 4.9102998434841516e-05,
"loss": 4.9434,
"step": 54784
},
{
"epoch": 0.02,
"learning_rate": 4.9094612487330996e-05,
"loss": 4.9392,
"step": 55296
},
{
"epoch": 0.02,
"learning_rate": 4.9086226539820476e-05,
"loss": 4.9336,
"step": 55808
},
{
"epoch": 0.02,
"learning_rate": 4.907785697111369e-05,
"loss": 4.9397,
"step": 56320
},
{
"epoch": 0.02,
"learning_rate": 4.906947102360317e-05,
"loss": 4.9143,
"step": 56832
},
{
"epoch": 0.02,
"learning_rate": 4.906108507609265e-05,
"loss": 4.9181,
"step": 57344
},
{
"epoch": 0.02,
"learning_rate": 4.905269912858213e-05,
"loss": 4.9116,
"step": 57856
},
{
"epoch": 0.02,
"learning_rate": 4.904431318107161e-05,
"loss": 4.9176,
"step": 58368
},
{
"epoch": 0.02,
"learning_rate": 4.903592723356109e-05,
"loss": 4.922,
"step": 58880
},
{
"epoch": 0.02,
"learning_rate": 4.9027541286050565e-05,
"loss": 4.9052,
"step": 59392
},
{
"epoch": 0.02,
"learning_rate": 4.9019155338540045e-05,
"loss": 4.9018,
"step": 59904
},
{
"epoch": 0.02,
"learning_rate": 4.9010769391029525e-05,
"loss": 4.8923,
"step": 60416
},
{
"epoch": 0.02,
"learning_rate": 4.900239982232274e-05,
"loss": 4.9004,
"step": 60928
},
{
"epoch": 0.02,
"learning_rate": 4.8994013874812214e-05,
"loss": 4.894,
"step": 61440
},
{
"epoch": 0.02,
"learning_rate": 4.89856279273017e-05,
"loss": 4.8885,
"step": 61952
},
{
"epoch": 0.02,
"learning_rate": 4.897724197979118e-05,
"loss": 4.8824,
"step": 62464
},
{
"epoch": 0.02,
"learning_rate": 4.896885603228066e-05,
"loss": 4.8746,
"step": 62976
},
{
"epoch": 0.02,
"learning_rate": 4.8960502842377605e-05,
"loss": 4.8702,
"step": 63488
},
{
"epoch": 0.02,
"learning_rate": 4.8952133273670815e-05,
"loss": 4.859,
"step": 64000
},
{
"epoch": 0.02,
"learning_rate": 4.8943747326160294e-05,
"loss": 4.8638,
"step": 64512
},
{
"epoch": 0.02,
"learning_rate": 4.8935361378649774e-05,
"loss": 4.8584,
"step": 65024
},
{
"epoch": 0.02,
"learning_rate": 4.8926975431139254e-05,
"loss": 4.8544,
"step": 65536
},
{
"epoch": 0.02,
"learning_rate": 4.8918589483628734e-05,
"loss": 4.8448,
"step": 66048
},
{
"epoch": 0.02,
"learning_rate": 4.8910219914921943e-05,
"loss": 4.8607,
"step": 66560
},
{
"epoch": 0.02,
"learning_rate": 4.8901833967411423e-05,
"loss": 4.8604,
"step": 67072
},
{
"epoch": 0.02,
"learning_rate": 4.88934480199009e-05,
"loss": 4.8495,
"step": 67584
},
{
"epoch": 0.02,
"learning_rate": 4.888506207239038e-05,
"loss": 4.8376,
"step": 68096
},
{
"epoch": 0.02,
"learning_rate": 4.887667612487986e-05,
"loss": 4.8368,
"step": 68608
},
{
"epoch": 0.02,
"learning_rate": 4.886829017736934e-05,
"loss": 4.8398,
"step": 69120
},
{
"epoch": 0.02,
"learning_rate": 4.885990422985882e-05,
"loss": 4.8364,
"step": 69632
},
{
"epoch": 0.02,
"learning_rate": 4.88515182823483e-05,
"loss": 4.8313,
"step": 70144
},
{
"epoch": 0.02,
"learning_rate": 4.884313233483778e-05,
"loss": 4.8279,
"step": 70656
},
{
"epoch": 0.02,
"learning_rate": 4.883474638732726e-05,
"loss": 4.8247,
"step": 71168
},
{
"epoch": 0.02,
"learning_rate": 4.882636043981674e-05,
"loss": 4.8173,
"step": 71680
},
{
"epoch": 0.02,
"learning_rate": 4.881797449230622e-05,
"loss": 4.8127,
"step": 72192
},
{
"epoch": 0.02,
"learning_rate": 4.88095885447957e-05,
"loss": 4.8167,
"step": 72704
},
{
"epoch": 0.02,
"learning_rate": 4.880121897608891e-05,
"loss": 4.8047,
"step": 73216
},
{
"epoch": 0.02,
"learning_rate": 4.879284940738212e-05,
"loss": 4.7993,
"step": 73728
},
{
"epoch": 0.02,
"learning_rate": 4.87844634598716e-05,
"loss": 4.8073,
"step": 74240
},
{
"epoch": 0.02,
"learning_rate": 4.877607751236109e-05,
"loss": 4.7931,
"step": 74752
},
{
"epoch": 0.02,
"learning_rate": 4.876769156485057e-05,
"loss": 4.8016,
"step": 75264
},
{
"epoch": 0.02,
"learning_rate": 4.875930561734005e-05,
"loss": 4.7919,
"step": 75776
},
{
"epoch": 0.02,
"learning_rate": 4.875091966982953e-05,
"loss": 4.7826,
"step": 76288
},
{
"epoch": 0.03,
"eval_loss": 4.750964641571045,
"eval_runtime": 293.5866,
"eval_samples_per_second": 1299.756,
"eval_steps_per_second": 40.618,
"step": 76320
}
],
"logging_steps": 512,
"max_steps": 3052726,
"num_train_epochs": 9223372036854775807,
"save_steps": 10,
"total_flos": 3.159246072248334e+16,
"trial_name": null,
"trial_params": null
}