rel-cl-lstm-1 / checkpoint-152640 /trainer_state.json
abhinavp's picture
Training in progress, epoch 1, checkpoint
0c9ab1b verified
{
"best_metric": 4.494524002075195,
"best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/rel-cl2/lstm/1/checkpoints/checkpoint-152640",
"epoch": 1.0250006060157382,
"eval_steps": 10,
"global_step": 152640,
"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": 10.8195,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 4.999161405248948e-05,
"loss": 7.5561,
"step": 512
},
{
"epoch": 0.0,
"learning_rate": 4.998322810497896e-05,
"loss": 7.0599,
"step": 1024
},
{
"epoch": 0.0,
"learning_rate": 4.997484215746844e-05,
"loss": 6.9972,
"step": 1536
},
{
"epoch": 0.0,
"learning_rate": 4.996645620995792e-05,
"loss": 6.9475,
"step": 2048
},
{
"epoch": 0.0,
"learning_rate": 4.99580702624474e-05,
"loss": 6.9188,
"step": 2560
},
{
"epoch": 0.0,
"learning_rate": 4.994968431493688e-05,
"loss": 6.7574,
"step": 3072
},
{
"epoch": 0.0,
"learning_rate": 4.994131474623009e-05,
"loss": 6.6474,
"step": 3584
},
{
"epoch": 0.0,
"learning_rate": 4.993292879871958e-05,
"loss": 6.5534,
"step": 4096
},
{
"epoch": 0.0,
"learning_rate": 4.992454285120906e-05,
"loss": 6.4733,
"step": 4608
},
{
"epoch": 0.0,
"learning_rate": 4.991615690369854e-05,
"loss": 6.4162,
"step": 5120
},
{
"epoch": 0.0,
"learning_rate": 4.990777095618801e-05,
"loss": 6.3506,
"step": 5632
},
{
"epoch": 0.0,
"learning_rate": 4.989938500867749e-05,
"loss": 6.2788,
"step": 6144
},
{
"epoch": 0.0,
"learning_rate": 4.989099906116697e-05,
"loss": 6.2098,
"step": 6656
},
{
"epoch": 0.0,
"learning_rate": 4.988261311365645e-05,
"loss": 6.154,
"step": 7168
},
{
"epoch": 0.0,
"learning_rate": 4.987422716614593e-05,
"loss": 6.0923,
"step": 7680
},
{
"epoch": 0.0,
"learning_rate": 4.986584121863541e-05,
"loss": 6.0503,
"step": 8192
},
{
"epoch": 0.0,
"learning_rate": 4.985745527112489e-05,
"loss": 6.0096,
"step": 8704
},
{
"epoch": 0.0,
"learning_rate": 4.984906932361437e-05,
"loss": 5.963,
"step": 9216
},
{
"epoch": 0.0,
"learning_rate": 4.984068337610385e-05,
"loss": 5.9206,
"step": 9728
},
{
"epoch": 0.0,
"learning_rate": 4.983229742859333e-05,
"loss": 5.8923,
"step": 10240
},
{
"epoch": 0.0,
"learning_rate": 4.982391148108281e-05,
"loss": 5.8495,
"step": 10752
},
{
"epoch": 0.0,
"learning_rate": 4.981552553357229e-05,
"loss": 5.8265,
"step": 11264
},
{
"epoch": 0.0,
"learning_rate": 4.980713958606178e-05,
"loss": 5.788,
"step": 11776
},
{
"epoch": 0.0,
"learning_rate": 4.979875363855125e-05,
"loss": 5.7724,
"step": 12288
},
{
"epoch": 0.0,
"learning_rate": 4.979036769104073e-05,
"loss": 5.7327,
"step": 12800
},
{
"epoch": 0.0,
"learning_rate": 4.978198174353021e-05,
"loss": 5.7102,
"step": 13312
},
{
"epoch": 0.0,
"learning_rate": 4.9773612174823426e-05,
"loss": 5.6808,
"step": 13824
},
{
"epoch": 0.0,
"learning_rate": 4.97652262273129e-05,
"loss": 5.6655,
"step": 14336
},
{
"epoch": 0.0,
"learning_rate": 4.975684027980238e-05,
"loss": 5.6388,
"step": 14848
},
{
"epoch": 0.01,
"learning_rate": 4.974845433229186e-05,
"loss": 5.6265,
"step": 15360
},
{
"epoch": 0.01,
"learning_rate": 4.9740101142388804e-05,
"loss": 5.6087,
"step": 15872
},
{
"epoch": 0.01,
"learning_rate": 4.9731715194878284e-05,
"loss": 5.5848,
"step": 16384
},
{
"epoch": 0.01,
"learning_rate": 4.9723329247367764e-05,
"loss": 5.5705,
"step": 16896
},
{
"epoch": 0.01,
"learning_rate": 4.9714943299857244e-05,
"loss": 5.5555,
"step": 17408
},
{
"epoch": 0.01,
"learning_rate": 4.970657373115046e-05,
"loss": 5.5334,
"step": 17920
},
{
"epoch": 0.01,
"learning_rate": 4.969818778363994e-05,
"loss": 5.5301,
"step": 18432
},
{
"epoch": 0.01,
"learning_rate": 4.968980183612942e-05,
"loss": 5.4959,
"step": 18944
},
{
"epoch": 0.01,
"learning_rate": 4.96814158886189e-05,
"loss": 5.4843,
"step": 19456
},
{
"epoch": 0.01,
"learning_rate": 4.967302994110837e-05,
"loss": 5.4547,
"step": 19968
},
{
"epoch": 0.01,
"learning_rate": 4.966467675120532e-05,
"loss": 5.4546,
"step": 20480
},
{
"epoch": 0.01,
"learning_rate": 4.96562908036948e-05,
"loss": 5.4389,
"step": 20992
},
{
"epoch": 0.01,
"learning_rate": 4.964790485618428e-05,
"loss": 5.4366,
"step": 21504
},
{
"epoch": 0.01,
"learning_rate": 4.963951890867376e-05,
"loss": 5.409,
"step": 22016
},
{
"epoch": 0.01,
"learning_rate": 4.963113296116324e-05,
"loss": 5.3987,
"step": 22528
},
{
"epoch": 0.01,
"learning_rate": 4.962274701365272e-05,
"loss": 5.3976,
"step": 23040
},
{
"epoch": 0.01,
"learning_rate": 4.96143610661422e-05,
"loss": 5.3891,
"step": 23552
},
{
"epoch": 0.01,
"learning_rate": 4.9605991497435414e-05,
"loss": 5.3773,
"step": 24064
},
{
"epoch": 0.01,
"learning_rate": 4.9597605549924894e-05,
"loss": 5.3578,
"step": 24576
},
{
"epoch": 0.01,
"learning_rate": 4.9589219602414374e-05,
"loss": 5.3439,
"step": 25088
},
{
"epoch": 0.01,
"learning_rate": 4.958083365490385e-05,
"loss": 5.3611,
"step": 25600
},
{
"epoch": 0.01,
"learning_rate": 4.957244770739333e-05,
"loss": 5.3472,
"step": 26112
},
{
"epoch": 0.01,
"learning_rate": 4.956406175988281e-05,
"loss": 5.3271,
"step": 26624
},
{
"epoch": 0.01,
"learning_rate": 4.955567581237229e-05,
"loss": 5.3072,
"step": 27136
},
{
"epoch": 0.01,
"learning_rate": 4.954728986486177e-05,
"loss": 5.3151,
"step": 27648
},
{
"epoch": 0.01,
"learning_rate": 4.9538920296154976e-05,
"loss": 5.294,
"step": 28160
},
{
"epoch": 0.01,
"learning_rate": 4.9530534348644456e-05,
"loss": 5.3014,
"step": 28672
},
{
"epoch": 0.01,
"learning_rate": 4.9522148401133936e-05,
"loss": 5.27,
"step": 29184
},
{
"epoch": 0.01,
"learning_rate": 4.951376245362342e-05,
"loss": 5.2815,
"step": 29696
},
{
"epoch": 0.01,
"learning_rate": 4.950539288491663e-05,
"loss": 5.266,
"step": 30208
},
{
"epoch": 0.01,
"learning_rate": 4.949702331620985e-05,
"loss": 5.2482,
"step": 30720
},
{
"epoch": 0.01,
"learning_rate": 4.948863736869932e-05,
"loss": 5.235,
"step": 31232
},
{
"epoch": 0.01,
"learning_rate": 4.94802514211888e-05,
"loss": 5.2471,
"step": 31744
},
{
"epoch": 0.01,
"learning_rate": 4.947186547367828e-05,
"loss": 5.2046,
"step": 32256
},
{
"epoch": 0.01,
"learning_rate": 4.946347952616776e-05,
"loss": 5.2193,
"step": 32768
},
{
"epoch": 0.01,
"learning_rate": 4.945509357865724e-05,
"loss": 5.2267,
"step": 33280
},
{
"epoch": 0.01,
"learning_rate": 4.944670763114672e-05,
"loss": 5.2094,
"step": 33792
},
{
"epoch": 0.01,
"learning_rate": 4.94383216836362e-05,
"loss": 5.2021,
"step": 34304
},
{
"epoch": 0.01,
"learning_rate": 4.942995211492941e-05,
"loss": 5.1768,
"step": 34816
},
{
"epoch": 0.01,
"learning_rate": 4.942156616741889e-05,
"loss": 5.1731,
"step": 35328
},
{
"epoch": 0.01,
"learning_rate": 4.941318021990837e-05,
"loss": 5.171,
"step": 35840
},
{
"epoch": 0.01,
"learning_rate": 4.9404794272397856e-05,
"loss": 5.1818,
"step": 36352
},
{
"epoch": 0.01,
"learning_rate": 4.9396408324887336e-05,
"loss": 5.1619,
"step": 36864
},
{
"epoch": 0.01,
"learning_rate": 4.9388038756180545e-05,
"loss": 5.1702,
"step": 37376
},
{
"epoch": 0.01,
"learning_rate": 4.9379652808670025e-05,
"loss": 5.1592,
"step": 37888
},
{
"epoch": 0.01,
"learning_rate": 4.9371283239963234e-05,
"loss": 5.1568,
"step": 38400
},
{
"epoch": 0.01,
"learning_rate": 4.9362897292452714e-05,
"loss": 5.1342,
"step": 38912
},
{
"epoch": 0.01,
"learning_rate": 4.9354511344942194e-05,
"loss": 5.1321,
"step": 39424
},
{
"epoch": 0.01,
"learning_rate": 4.9346125397431674e-05,
"loss": 5.1243,
"step": 39936
},
{
"epoch": 0.01,
"learning_rate": 4.9337739449921154e-05,
"loss": 5.1194,
"step": 40448
},
{
"epoch": 0.01,
"learning_rate": 4.9329353502410634e-05,
"loss": 5.1185,
"step": 40960
},
{
"epoch": 0.01,
"learning_rate": 4.9320967554900114e-05,
"loss": 5.1013,
"step": 41472
},
{
"epoch": 0.01,
"learning_rate": 4.9312581607389594e-05,
"loss": 5.1026,
"step": 41984
},
{
"epoch": 0.01,
"learning_rate": 4.930421203868281e-05,
"loss": 5.1017,
"step": 42496
},
{
"epoch": 0.01,
"learning_rate": 4.929582609117229e-05,
"loss": 5.0774,
"step": 43008
},
{
"epoch": 0.01,
"learning_rate": 4.928744014366177e-05,
"loss": 5.0845,
"step": 43520
},
{
"epoch": 0.01,
"learning_rate": 4.927905419615125e-05,
"loss": 5.0835,
"step": 44032
},
{
"epoch": 0.01,
"learning_rate": 4.927066824864073e-05,
"loss": 5.0844,
"step": 44544
},
{
"epoch": 0.01,
"learning_rate": 4.92622823011302e-05,
"loss": 5.0655,
"step": 45056
},
{
"epoch": 0.01,
"learning_rate": 4.925389635361968e-05,
"loss": 5.062,
"step": 45568
},
{
"epoch": 0.02,
"learning_rate": 4.924551040610916e-05,
"loss": 5.0485,
"step": 46080
},
{
"epoch": 0.02,
"learning_rate": 4.923712445859864e-05,
"loss": 5.0456,
"step": 46592
},
{
"epoch": 0.02,
"learning_rate": 4.922875488989185e-05,
"loss": 5.0451,
"step": 47104
},
{
"epoch": 0.02,
"learning_rate": 4.922036894238133e-05,
"loss": 5.0429,
"step": 47616
},
{
"epoch": 0.02,
"learning_rate": 4.921198299487081e-05,
"loss": 5.0349,
"step": 48128
},
{
"epoch": 0.02,
"learning_rate": 4.920359704736029e-05,
"loss": 5.037,
"step": 48640
},
{
"epoch": 0.02,
"learning_rate": 4.919522747865351e-05,
"loss": 5.0276,
"step": 49152
},
{
"epoch": 0.02,
"learning_rate": 4.918684153114299e-05,
"loss": 5.0119,
"step": 49664
},
{
"epoch": 0.02,
"learning_rate": 4.917845558363247e-05,
"loss": 5.0076,
"step": 50176
},
{
"epoch": 0.02,
"learning_rate": 4.917006963612195e-05,
"loss": 5.0109,
"step": 50688
},
{
"epoch": 0.02,
"learning_rate": 4.916168368861143e-05,
"loss": 5.0064,
"step": 51200
},
{
"epoch": 0.02,
"learning_rate": 4.9153314119904636e-05,
"loss": 5.0103,
"step": 51712
},
{
"epoch": 0.02,
"learning_rate": 4.9144928172394116e-05,
"loss": 4.9923,
"step": 52224
},
{
"epoch": 0.02,
"learning_rate": 4.9136558603687325e-05,
"loss": 4.9852,
"step": 52736
},
{
"epoch": 0.02,
"learning_rate": 4.9128172656176805e-05,
"loss": 4.9892,
"step": 53248
},
{
"epoch": 0.02,
"learning_rate": 4.9119786708666285e-05,
"loss": 4.9695,
"step": 53760
},
{
"epoch": 0.02,
"learning_rate": 4.9111400761155765e-05,
"loss": 4.975,
"step": 54272
},
{
"epoch": 0.02,
"learning_rate": 4.9103014813645245e-05,
"loss": 4.9649,
"step": 54784
},
{
"epoch": 0.02,
"learning_rate": 4.909462886613473e-05,
"loss": 4.9654,
"step": 55296
},
{
"epoch": 0.02,
"learning_rate": 4.908624291862421e-05,
"loss": 4.9521,
"step": 55808
},
{
"epoch": 0.02,
"learning_rate": 4.907785697111369e-05,
"loss": 4.9555,
"step": 56320
},
{
"epoch": 0.02,
"learning_rate": 4.906947102360317e-05,
"loss": 4.9455,
"step": 56832
},
{
"epoch": 0.02,
"learning_rate": 4.906110145489638e-05,
"loss": 4.9481,
"step": 57344
},
{
"epoch": 0.02,
"learning_rate": 4.905271550738586e-05,
"loss": 4.9428,
"step": 57856
},
{
"epoch": 0.02,
"learning_rate": 4.904432955987534e-05,
"loss": 4.9335,
"step": 58368
},
{
"epoch": 0.02,
"learning_rate": 4.903594361236482e-05,
"loss": 4.9442,
"step": 58880
},
{
"epoch": 0.02,
"learning_rate": 4.902757404365803e-05,
"loss": 4.9418,
"step": 59392
},
{
"epoch": 0.02,
"learning_rate": 4.901918809614751e-05,
"loss": 4.9294,
"step": 59904
},
{
"epoch": 0.02,
"learning_rate": 4.901080214863699e-05,
"loss": 4.9164,
"step": 60416
},
{
"epoch": 0.02,
"learning_rate": 4.900241620112647e-05,
"loss": 4.9174,
"step": 60928
},
{
"epoch": 0.02,
"learning_rate": 4.899403025361595e-05,
"loss": 4.9151,
"step": 61440
},
{
"epoch": 0.02,
"learning_rate": 4.8985660684909166e-05,
"loss": 4.9168,
"step": 61952
},
{
"epoch": 0.02,
"learning_rate": 4.8977274737398646e-05,
"loss": 4.9108,
"step": 62464
},
{
"epoch": 0.02,
"learning_rate": 4.8968888789888125e-05,
"loss": 4.9012,
"step": 62976
},
{
"epoch": 0.02,
"learning_rate": 4.8960502842377605e-05,
"loss": 4.9047,
"step": 63488
},
{
"epoch": 0.02,
"learning_rate": 4.8952116894867085e-05,
"loss": 4.8905,
"step": 64000
},
{
"epoch": 0.02,
"learning_rate": 4.8943730947356565e-05,
"loss": 4.8952,
"step": 64512
},
{
"epoch": 0.02,
"learning_rate": 4.893534499984604e-05,
"loss": 4.8811,
"step": 65024
},
{
"epoch": 0.02,
"learning_rate": 4.892695905233552e-05,
"loss": 4.8873,
"step": 65536
},
{
"epoch": 0.02,
"learning_rate": 4.8918573104825e-05,
"loss": 4.876,
"step": 66048
},
{
"epoch": 0.02,
"learning_rate": 4.8910203536118214e-05,
"loss": 4.8856,
"step": 66560
},
{
"epoch": 0.02,
"learning_rate": 4.890181758860769e-05,
"loss": 4.8846,
"step": 67072
},
{
"epoch": 0.02,
"learning_rate": 4.889343164109717e-05,
"loss": 4.882,
"step": 67584
},
{
"epoch": 0.02,
"learning_rate": 4.888504569358665e-05,
"loss": 4.8584,
"step": 68096
},
{
"epoch": 0.02,
"learning_rate": 4.8876659746076134e-05,
"loss": 4.8639,
"step": 68608
},
{
"epoch": 0.02,
"learning_rate": 4.8868273798565614e-05,
"loss": 4.862,
"step": 69120
},
{
"epoch": 0.02,
"learning_rate": 4.8859887851055094e-05,
"loss": 4.8639,
"step": 69632
},
{
"epoch": 0.02,
"learning_rate": 4.88515182823483e-05,
"loss": 4.852,
"step": 70144
},
{
"epoch": 0.02,
"learning_rate": 4.884313233483778e-05,
"loss": 4.8572,
"step": 70656
},
{
"epoch": 0.02,
"learning_rate": 4.883474638732726e-05,
"loss": 4.8458,
"step": 71168
},
{
"epoch": 0.02,
"learning_rate": 4.882636043981674e-05,
"loss": 4.861,
"step": 71680
},
{
"epoch": 0.02,
"learning_rate": 4.881797449230622e-05,
"loss": 4.8289,
"step": 72192
},
{
"epoch": 0.02,
"learning_rate": 4.880960492359943e-05,
"loss": 4.8484,
"step": 72704
},
{
"epoch": 0.02,
"learning_rate": 4.880121897608891e-05,
"loss": 4.8358,
"step": 73216
},
{
"epoch": 0.02,
"learning_rate": 4.879283302857839e-05,
"loss": 4.8287,
"step": 73728
},
{
"epoch": 0.02,
"learning_rate": 4.878444708106787e-05,
"loss": 4.8282,
"step": 74240
},
{
"epoch": 0.02,
"learning_rate": 4.877607751236109e-05,
"loss": 4.813,
"step": 74752
},
{
"epoch": 0.02,
"learning_rate": 4.876769156485057e-05,
"loss": 4.8183,
"step": 75264
},
{
"epoch": 0.02,
"learning_rate": 4.875932199614378e-05,
"loss": 4.8269,
"step": 75776
},
{
"epoch": 0.02,
"learning_rate": 4.875093604863326e-05,
"loss": 4.8123,
"step": 76288
},
{
"epoch": 0.03,
"eval_loss": 4.779831886291504,
"eval_runtime": 288.1178,
"eval_samples_per_second": 1324.427,
"eval_steps_per_second": 41.389,
"step": 76320
},
{
"epoch": 1.0,
"learning_rate": 4.874255010112274e-05,
"loss": 4.8111,
"step": 76800
},
{
"epoch": 1.0,
"learning_rate": 4.873416415361222e-05,
"loss": 4.8048,
"step": 77312
},
{
"epoch": 1.0,
"learning_rate": 4.87257782061017e-05,
"loss": 4.8175,
"step": 77824
},
{
"epoch": 1.0,
"learning_rate": 4.871739225859118e-05,
"loss": 4.8068,
"step": 78336
},
{
"epoch": 1.0,
"learning_rate": 4.8709006311080657e-05,
"loss": 4.8127,
"step": 78848
},
{
"epoch": 1.0,
"learning_rate": 4.8700636742373866e-05,
"loss": 4.7888,
"step": 79360
},
{
"epoch": 1.0,
"learning_rate": 4.8692250794863346e-05,
"loss": 4.796,
"step": 79872
},
{
"epoch": 1.0,
"learning_rate": 4.8683864847352826e-05,
"loss": 4.7801,
"step": 80384
},
{
"epoch": 1.0,
"learning_rate": 4.8675478899842306e-05,
"loss": 4.7839,
"step": 80896
},
{
"epoch": 1.0,
"learning_rate": 4.8667092952331785e-05,
"loss": 4.788,
"step": 81408
},
{
"epoch": 1.0,
"learning_rate": 4.865870700482127e-05,
"loss": 4.7842,
"step": 81920
},
{
"epoch": 1.0,
"learning_rate": 4.865033743611448e-05,
"loss": 4.7876,
"step": 82432
},
{
"epoch": 1.0,
"learning_rate": 4.864196786740769e-05,
"loss": 4.7664,
"step": 82944
},
{
"epoch": 1.0,
"learning_rate": 4.863358191989717e-05,
"loss": 4.7803,
"step": 83456
},
{
"epoch": 1.0,
"learning_rate": 4.862519597238665e-05,
"loss": 4.7618,
"step": 83968
},
{
"epoch": 1.0,
"learning_rate": 4.861682640367986e-05,
"loss": 4.7667,
"step": 84480
},
{
"epoch": 1.0,
"learning_rate": 4.860844045616934e-05,
"loss": 4.7686,
"step": 84992
},
{
"epoch": 1.0,
"learning_rate": 4.860005450865882e-05,
"loss": 4.7593,
"step": 85504
},
{
"epoch": 1.0,
"learning_rate": 4.85916685611483e-05,
"loss": 4.7627,
"step": 86016
},
{
"epoch": 1.0,
"learning_rate": 4.858328261363778e-05,
"loss": 4.775,
"step": 86528
},
{
"epoch": 1.0,
"learning_rate": 4.857489666612726e-05,
"loss": 4.7505,
"step": 87040
},
{
"epoch": 1.0,
"learning_rate": 4.856651071861674e-05,
"loss": 4.7575,
"step": 87552
},
{
"epoch": 1.0,
"learning_rate": 4.8558124771106226e-05,
"loss": 4.7503,
"step": 88064
},
{
"epoch": 1.0,
"learning_rate": 4.85497388235957e-05,
"loss": 4.7591,
"step": 88576
},
{
"epoch": 1.0,
"learning_rate": 4.854135287608518e-05,
"loss": 4.748,
"step": 89088
},
{
"epoch": 1.0,
"learning_rate": 4.853296692857466e-05,
"loss": 4.7478,
"step": 89600
},
{
"epoch": 1.0,
"learning_rate": 4.852458098106414e-05,
"loss": 4.7309,
"step": 90112
},
{
"epoch": 1.0,
"learning_rate": 4.851619503355362e-05,
"loss": 4.7403,
"step": 90624
},
{
"epoch": 1.0,
"learning_rate": 4.85078090860431e-05,
"loss": 4.7271,
"step": 91136
},
{
"epoch": 1.01,
"learning_rate": 4.849942313853258e-05,
"loss": 4.738,
"step": 91648
},
{
"epoch": 1.01,
"learning_rate": 4.849105356982579e-05,
"loss": 4.7333,
"step": 92160
},
{
"epoch": 1.01,
"learning_rate": 4.848266762231527e-05,
"loss": 4.7318,
"step": 92672
},
{
"epoch": 1.01,
"learning_rate": 4.847428167480475e-05,
"loss": 4.7342,
"step": 93184
},
{
"epoch": 1.01,
"learning_rate": 4.846589572729423e-05,
"loss": 4.725,
"step": 93696
},
{
"epoch": 1.01,
"learning_rate": 4.8457526158587444e-05,
"loss": 4.7303,
"step": 94208
},
{
"epoch": 1.01,
"learning_rate": 4.8449140211076924e-05,
"loss": 4.7232,
"step": 94720
},
{
"epoch": 1.01,
"learning_rate": 4.8440754263566404e-05,
"loss": 4.7129,
"step": 95232
},
{
"epoch": 1.01,
"learning_rate": 4.8432368316055884e-05,
"loss": 4.7066,
"step": 95744
},
{
"epoch": 1.01,
"learning_rate": 4.8423982368545363e-05,
"loss": 4.6953,
"step": 96256
},
{
"epoch": 1.01,
"learning_rate": 4.8415596421034843e-05,
"loss": 4.7051,
"step": 96768
},
{
"epoch": 1.01,
"learning_rate": 4.840721047352432e-05,
"loss": 4.7039,
"step": 97280
},
{
"epoch": 1.01,
"learning_rate": 4.839884090481753e-05,
"loss": 4.7105,
"step": 97792
},
{
"epoch": 1.01,
"learning_rate": 4.839045495730701e-05,
"loss": 4.6933,
"step": 98304
},
{
"epoch": 1.01,
"learning_rate": 4.838206900979649e-05,
"loss": 4.6952,
"step": 98816
},
{
"epoch": 1.01,
"learning_rate": 4.837368306228597e-05,
"loss": 4.6984,
"step": 99328
},
{
"epoch": 1.01,
"learning_rate": 4.836529711477545e-05,
"loss": 4.6911,
"step": 99840
},
{
"epoch": 1.01,
"learning_rate": 4.835691116726493e-05,
"loss": 4.6973,
"step": 100352
},
{
"epoch": 1.01,
"learning_rate": 4.834852521975441e-05,
"loss": 4.6833,
"step": 100864
},
{
"epoch": 1.01,
"learning_rate": 4.834013927224389e-05,
"loss": 4.675,
"step": 101376
},
{
"epoch": 1.01,
"learning_rate": 4.833176970353711e-05,
"loss": 4.7029,
"step": 101888
},
{
"epoch": 1.01,
"learning_rate": 4.832338375602659e-05,
"loss": 4.6967,
"step": 102400
},
{
"epoch": 1.01,
"learning_rate": 4.831499780851606e-05,
"loss": 4.6819,
"step": 102912
},
{
"epoch": 1.01,
"learning_rate": 4.830661186100554e-05,
"loss": 4.678,
"step": 103424
},
{
"epoch": 1.01,
"learning_rate": 4.829822591349502e-05,
"loss": 4.6853,
"step": 103936
},
{
"epoch": 1.01,
"learning_rate": 4.828985634478824e-05,
"loss": 4.6695,
"step": 104448
},
{
"epoch": 1.01,
"learning_rate": 4.828147039727771e-05,
"loss": 4.687,
"step": 104960
},
{
"epoch": 1.01,
"learning_rate": 4.827308444976719e-05,
"loss": 4.6578,
"step": 105472
},
{
"epoch": 1.01,
"learning_rate": 4.826469850225667e-05,
"loss": 4.682,
"step": 105984
},
{
"epoch": 1.01,
"learning_rate": 4.825631255474615e-05,
"loss": 4.6739,
"step": 106496
},
{
"epoch": 1.01,
"learning_rate": 4.824792660723563e-05,
"loss": 4.6568,
"step": 107008
},
{
"epoch": 1.01,
"learning_rate": 4.823954065972511e-05,
"loss": 4.6549,
"step": 107520
},
{
"epoch": 1.01,
"learning_rate": 4.82311547122146e-05,
"loss": 4.6691,
"step": 108032
},
{
"epoch": 1.01,
"learning_rate": 4.8222801522311535e-05,
"loss": 4.6315,
"step": 108544
},
{
"epoch": 1.01,
"learning_rate": 4.8214415574801015e-05,
"loss": 4.654,
"step": 109056
},
{
"epoch": 1.01,
"learning_rate": 4.8206029627290495e-05,
"loss": 4.6645,
"step": 109568
},
{
"epoch": 1.01,
"learning_rate": 4.8197643679779975e-05,
"loss": 4.656,
"step": 110080
},
{
"epoch": 1.01,
"learning_rate": 4.8189274111073184e-05,
"loss": 4.6474,
"step": 110592
},
{
"epoch": 1.01,
"learning_rate": 4.8180888163562664e-05,
"loss": 4.6334,
"step": 111104
},
{
"epoch": 1.01,
"learning_rate": 4.8172502216052144e-05,
"loss": 4.639,
"step": 111616
},
{
"epoch": 1.01,
"learning_rate": 4.8164116268541624e-05,
"loss": 4.6418,
"step": 112128
},
{
"epoch": 1.01,
"learning_rate": 4.815574669983483e-05,
"loss": 4.6485,
"step": 112640
},
{
"epoch": 1.01,
"learning_rate": 4.814736075232432e-05,
"loss": 4.6363,
"step": 113152
},
{
"epoch": 1.01,
"learning_rate": 4.81389748048138e-05,
"loss": 4.6456,
"step": 113664
},
{
"epoch": 1.01,
"learning_rate": 4.813058885730328e-05,
"loss": 4.649,
"step": 114176
},
{
"epoch": 1.01,
"learning_rate": 4.812220290979276e-05,
"loss": 4.6462,
"step": 114688
},
{
"epoch": 1.01,
"learning_rate": 4.811381696228224e-05,
"loss": 4.6283,
"step": 115200
},
{
"epoch": 1.01,
"learning_rate": 4.810543101477172e-05,
"loss": 4.6338,
"step": 115712
},
{
"epoch": 1.01,
"learning_rate": 4.80970450672612e-05,
"loss": 4.6326,
"step": 116224
},
{
"epoch": 1.01,
"learning_rate": 4.808867549855441e-05,
"loss": 4.625,
"step": 116736
},
{
"epoch": 1.01,
"learning_rate": 4.808028955104389e-05,
"loss": 4.6308,
"step": 117248
},
{
"epoch": 1.01,
"learning_rate": 4.807190360353337e-05,
"loss": 4.6205,
"step": 117760
},
{
"epoch": 1.01,
"learning_rate": 4.806351765602285e-05,
"loss": 4.6291,
"step": 118272
},
{
"epoch": 1.01,
"learning_rate": 4.805514808731606e-05,
"loss": 4.6264,
"step": 118784
},
{
"epoch": 1.01,
"learning_rate": 4.804676213980554e-05,
"loss": 4.6066,
"step": 119296
},
{
"epoch": 1.01,
"learning_rate": 4.803837619229502e-05,
"loss": 4.6191,
"step": 119808
},
{
"epoch": 1.01,
"learning_rate": 4.8029990244784504e-05,
"loss": 4.6188,
"step": 120320
},
{
"epoch": 1.01,
"learning_rate": 4.8021604297273984e-05,
"loss": 4.6249,
"step": 120832
},
{
"epoch": 1.01,
"learning_rate": 4.801323472856719e-05,
"loss": 4.6154,
"step": 121344
},
{
"epoch": 1.01,
"learning_rate": 4.800484878105667e-05,
"loss": 4.6108,
"step": 121856
},
{
"epoch": 1.02,
"learning_rate": 4.799646283354615e-05,
"loss": 4.6036,
"step": 122368
},
{
"epoch": 1.02,
"learning_rate": 4.798807688603563e-05,
"loss": 4.5998,
"step": 122880
},
{
"epoch": 1.02,
"learning_rate": 4.797970731732884e-05,
"loss": 4.604,
"step": 123392
},
{
"epoch": 1.02,
"learning_rate": 4.797132136981832e-05,
"loss": 4.6104,
"step": 123904
},
{
"epoch": 1.02,
"learning_rate": 4.79629354223078e-05,
"loss": 4.6054,
"step": 124416
},
{
"epoch": 1.02,
"learning_rate": 4.795454947479728e-05,
"loss": 4.6106,
"step": 124928
},
{
"epoch": 1.02,
"learning_rate": 4.794616352728676e-05,
"loss": 4.6005,
"step": 125440
},
{
"epoch": 1.02,
"learning_rate": 4.793779395857997e-05,
"loss": 4.595,
"step": 125952
},
{
"epoch": 1.02,
"learning_rate": 4.792940801106946e-05,
"loss": 4.5925,
"step": 126464
},
{
"epoch": 1.02,
"learning_rate": 4.792102206355894e-05,
"loss": 4.5972,
"step": 126976
},
{
"epoch": 1.02,
"learning_rate": 4.791263611604842e-05,
"loss": 4.6017,
"step": 127488
},
{
"epoch": 1.02,
"learning_rate": 4.79042501685379e-05,
"loss": 4.6028,
"step": 128000
},
{
"epoch": 1.02,
"learning_rate": 4.7895880599831107e-05,
"loss": 4.5901,
"step": 128512
},
{
"epoch": 1.02,
"learning_rate": 4.7887494652320587e-05,
"loss": 4.5824,
"step": 129024
},
{
"epoch": 1.02,
"learning_rate": 4.7879108704810067e-05,
"loss": 4.5922,
"step": 129536
},
{
"epoch": 1.02,
"learning_rate": 4.7870722757299546e-05,
"loss": 4.577,
"step": 130048
},
{
"epoch": 1.02,
"learning_rate": 4.786233680978902e-05,
"loss": 4.584,
"step": 130560
},
{
"epoch": 1.02,
"learning_rate": 4.7853967241082236e-05,
"loss": 4.5802,
"step": 131072
},
{
"epoch": 1.02,
"learning_rate": 4.7845581293571715e-05,
"loss": 4.5804,
"step": 131584
},
{
"epoch": 1.02,
"learning_rate": 4.7837195346061195e-05,
"loss": 4.5706,
"step": 132096
},
{
"epoch": 1.02,
"learning_rate": 4.7828809398550675e-05,
"loss": 4.5814,
"step": 132608
},
{
"epoch": 1.02,
"learning_rate": 4.7820423451040155e-05,
"loss": 4.5675,
"step": 133120
},
{
"epoch": 1.02,
"learning_rate": 4.781205388233337e-05,
"loss": 4.5722,
"step": 133632
},
{
"epoch": 1.02,
"learning_rate": 4.7803667934822844e-05,
"loss": 4.5762,
"step": 134144
},
{
"epoch": 1.02,
"learning_rate": 4.7795281987312324e-05,
"loss": 4.5642,
"step": 134656
},
{
"epoch": 1.02,
"learning_rate": 4.7786896039801804e-05,
"loss": 4.5851,
"step": 135168
},
{
"epoch": 1.02,
"learning_rate": 4.777852647109502e-05,
"loss": 4.5813,
"step": 135680
},
{
"epoch": 1.02,
"learning_rate": 4.777014052358449e-05,
"loss": 4.5737,
"step": 136192
},
{
"epoch": 1.02,
"learning_rate": 4.776175457607397e-05,
"loss": 4.5632,
"step": 136704
},
{
"epoch": 1.02,
"learning_rate": 4.775336862856345e-05,
"loss": 4.5608,
"step": 137216
},
{
"epoch": 1.02,
"learning_rate": 4.774499905985666e-05,
"loss": 4.5673,
"step": 137728
},
{
"epoch": 1.02,
"learning_rate": 4.773661311234615e-05,
"loss": 4.5617,
"step": 138240
},
{
"epoch": 1.02,
"learning_rate": 4.772822716483563e-05,
"loss": 4.5727,
"step": 138752
},
{
"epoch": 1.02,
"learning_rate": 4.771984121732511e-05,
"loss": 4.5525,
"step": 139264
},
{
"epoch": 1.02,
"learning_rate": 4.771145526981459e-05,
"loss": 4.5608,
"step": 139776
},
{
"epoch": 1.02,
"learning_rate": 4.770306932230407e-05,
"loss": 4.5584,
"step": 140288
},
{
"epoch": 1.02,
"learning_rate": 4.769468337479355e-05,
"loss": 4.5635,
"step": 140800
},
{
"epoch": 1.02,
"learning_rate": 4.768629742728303e-05,
"loss": 4.546,
"step": 141312
},
{
"epoch": 1.02,
"learning_rate": 4.767792785857624e-05,
"loss": 4.5561,
"step": 141824
},
{
"epoch": 1.02,
"learning_rate": 4.766954191106572e-05,
"loss": 4.5529,
"step": 142336
},
{
"epoch": 1.02,
"learning_rate": 4.76611559635552e-05,
"loss": 4.555,
"step": 142848
},
{
"epoch": 1.02,
"learning_rate": 4.765277001604468e-05,
"loss": 4.5589,
"step": 143360
},
{
"epoch": 1.02,
"learning_rate": 4.764438406853416e-05,
"loss": 4.5623,
"step": 143872
},
{
"epoch": 1.02,
"learning_rate": 4.763599812102364e-05,
"loss": 4.5398,
"step": 144384
},
{
"epoch": 1.02,
"learning_rate": 4.762761217351312e-05,
"loss": 4.5484,
"step": 144896
},
{
"epoch": 1.02,
"learning_rate": 4.76192262260026e-05,
"loss": 4.5526,
"step": 145408
},
{
"epoch": 1.02,
"learning_rate": 4.7610856657295813e-05,
"loss": 4.5491,
"step": 145920
},
{
"epoch": 1.02,
"learning_rate": 4.7602470709785293e-05,
"loss": 4.5443,
"step": 146432
},
{
"epoch": 1.02,
"learning_rate": 4.75941011410785e-05,
"loss": 4.5479,
"step": 146944
},
{
"epoch": 1.02,
"learning_rate": 4.758571519356798e-05,
"loss": 4.5424,
"step": 147456
},
{
"epoch": 1.02,
"learning_rate": 4.757732924605746e-05,
"loss": 4.5595,
"step": 147968
},
{
"epoch": 1.02,
"learning_rate": 4.756894329854694e-05,
"loss": 4.5306,
"step": 148480
},
{
"epoch": 1.02,
"learning_rate": 4.756055735103642e-05,
"loss": 4.5488,
"step": 148992
},
{
"epoch": 1.02,
"learning_rate": 4.75521714035259e-05,
"loss": 4.5361,
"step": 149504
},
{
"epoch": 1.02,
"learning_rate": 4.7543785456015375e-05,
"loss": 4.5327,
"step": 150016
},
{
"epoch": 1.02,
"learning_rate": 4.753541588730859e-05,
"loss": 4.5343,
"step": 150528
},
{
"epoch": 1.02,
"learning_rate": 4.752702993979807e-05,
"loss": 4.5265,
"step": 151040
},
{
"epoch": 1.02,
"learning_rate": 4.751864399228755e-05,
"loss": 4.5283,
"step": 151552
},
{
"epoch": 1.02,
"learning_rate": 4.751025804477703e-05,
"loss": 4.534,
"step": 152064
},
{
"epoch": 1.02,
"learning_rate": 4.750187209726651e-05,
"loss": 4.5274,
"step": 152576
},
{
"epoch": 1.03,
"eval_loss": 4.494524002075195,
"eval_runtime": 288.449,
"eval_samples_per_second": 1322.906,
"eval_steps_per_second": 41.342,
"step": 152640
}
],
"logging_steps": 512,
"max_steps": 3052726,
"num_train_epochs": 9223372036854775807,
"save_steps": 10,
"total_flos": 6.187528988683136e+16,
"trial_name": null,
"trial_params": null
}