30706 lines
620 KiB
JSON
30706 lines
620 KiB
JSON
{
|
|
"best_metric": 3.841752052307129,
|
|
"best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/full2/transformer/0/checkpoints/checkpoint-992160",
|
|
"epoch": 1.0250006060157382,
|
|
"eval_steps": 10,
|
|
"global_step": 2594880,
|
|
"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": 11.0453,
|
|
"step": 1
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.999161405248948e-05,
|
|
"loss": 6.8203,
|
|
"step": 512
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.998322810497896e-05,
|
|
"loss": 6.1803,
|
|
"step": 1024
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.997484215746844e-05,
|
|
"loss": 5.976,
|
|
"step": 1536
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.996645620995792e-05,
|
|
"loss": 5.8136,
|
|
"step": 2048
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.99580702624474e-05,
|
|
"loss": 5.7055,
|
|
"step": 2560
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.994968431493688e-05,
|
|
"loss": 5.6058,
|
|
"step": 3072
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.994129836742636e-05,
|
|
"loss": 5.5376,
|
|
"step": 3584
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.993291241991584e-05,
|
|
"loss": 5.4512,
|
|
"step": 4096
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.992452647240532e-05,
|
|
"loss": 5.403,
|
|
"step": 4608
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.99161405248948e-05,
|
|
"loss": 5.3536,
|
|
"step": 5120
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.990775457738428e-05,
|
|
"loss": 5.33,
|
|
"step": 5632
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.989938500867749e-05,
|
|
"loss": 5.2746,
|
|
"step": 6144
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.989099906116697e-05,
|
|
"loss": 5.211,
|
|
"step": 6656
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.988261311365645e-05,
|
|
"loss": 5.1872,
|
|
"step": 7168
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.987422716614593e-05,
|
|
"loss": 5.1491,
|
|
"step": 7680
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.986584121863541e-05,
|
|
"loss": 5.1193,
|
|
"step": 8192
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.985745527112489e-05,
|
|
"loss": 5.0928,
|
|
"step": 8704
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.984906932361437e-05,
|
|
"loss": 5.0671,
|
|
"step": 9216
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.984068337610385e-05,
|
|
"loss": 5.0268,
|
|
"step": 9728
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.983229742859333e-05,
|
|
"loss": 5.0218,
|
|
"step": 10240
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9823927859886547e-05,
|
|
"loss": 4.986,
|
|
"step": 10752
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9815541912376026e-05,
|
|
"loss": 4.9659,
|
|
"step": 11264
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9807155964865506e-05,
|
|
"loss": 4.9377,
|
|
"step": 11776
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9798770017354986e-05,
|
|
"loss": 4.9353,
|
|
"step": 12288
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9790400448648195e-05,
|
|
"loss": 4.9084,
|
|
"step": 12800
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9782014501137675e-05,
|
|
"loss": 4.8814,
|
|
"step": 13312
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9773628553627155e-05,
|
|
"loss": 4.8641,
|
|
"step": 13824
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9765242606116635e-05,
|
|
"loss": 4.8465,
|
|
"step": 14336
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9756873037409844e-05,
|
|
"loss": 4.8346,
|
|
"step": 14848
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9748487089899324e-05,
|
|
"loss": 4.8112,
|
|
"step": 15360
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9740117521192533e-05,
|
|
"loss": 4.8046,
|
|
"step": 15872
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9731731573682013e-05,
|
|
"loss": 4.7911,
|
|
"step": 16384
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.97233456261715e-05,
|
|
"loss": 4.7771,
|
|
"step": 16896
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.971495967866098e-05,
|
|
"loss": 4.7744,
|
|
"step": 17408
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.970657373115046e-05,
|
|
"loss": 4.7607,
|
|
"step": 17920
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.969818778363994e-05,
|
|
"loss": 4.7448,
|
|
"step": 18432
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.968980183612942e-05,
|
|
"loss": 4.7186,
|
|
"step": 18944
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.96814158886189e-05,
|
|
"loss": 4.714,
|
|
"step": 19456
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.967304631991211e-05,
|
|
"loss": 4.6948,
|
|
"step": 19968
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.966466037240159e-05,
|
|
"loss": 4.6941,
|
|
"step": 20480
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.965627442489107e-05,
|
|
"loss": 4.6788,
|
|
"step": 20992
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.964788847738054e-05,
|
|
"loss": 4.6731,
|
|
"step": 21504
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.963950252987002e-05,
|
|
"loss": 4.656,
|
|
"step": 22016
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.963113296116324e-05,
|
|
"loss": 4.6628,
|
|
"step": 22528
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.962274701365272e-05,
|
|
"loss": 4.6451,
|
|
"step": 23040
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.96143610661422e-05,
|
|
"loss": 4.6473,
|
|
"step": 23552
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.960597511863168e-05,
|
|
"loss": 4.6369,
|
|
"step": 24064
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.959758917112116e-05,
|
|
"loss": 4.609,
|
|
"step": 24576
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.95892359812181e-05,
|
|
"loss": 4.602,
|
|
"step": 25088
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.958085003370758e-05,
|
|
"loss": 4.616,
|
|
"step": 25600
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.957246408619706e-05,
|
|
"loss": 4.6008,
|
|
"step": 26112
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.956407813868654e-05,
|
|
"loss": 4.5844,
|
|
"step": 26624
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9555692191176016e-05,
|
|
"loss": 4.5682,
|
|
"step": 27136
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9547306243665496e-05,
|
|
"loss": 4.5748,
|
|
"step": 27648
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9538920296154976e-05,
|
|
"loss": 4.5575,
|
|
"step": 28160
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9530534348644456e-05,
|
|
"loss": 4.5697,
|
|
"step": 28672
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.952216477993767e-05,
|
|
"loss": 4.5357,
|
|
"step": 29184
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.951377883242715e-05,
|
|
"loss": 4.5472,
|
|
"step": 29696
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.950539288491663e-05,
|
|
"loss": 4.5447,
|
|
"step": 30208
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.949700693740611e-05,
|
|
"loss": 4.5208,
|
|
"step": 30720
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.948863736869932e-05,
|
|
"loss": 4.5379,
|
|
"step": 31232
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.94802514211888e-05,
|
|
"loss": 4.5087,
|
|
"step": 31744
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.947186547367828e-05,
|
|
"loss": 4.5105,
|
|
"step": 32256
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.946347952616776e-05,
|
|
"loss": 4.4967,
|
|
"step": 32768
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.945510995746097e-05,
|
|
"loss": 4.513,
|
|
"step": 33280
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.944672400995045e-05,
|
|
"loss": 4.486,
|
|
"step": 33792
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.943833806243993e-05,
|
|
"loss": 4.4883,
|
|
"step": 34304
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.942995211492941e-05,
|
|
"loss": 4.4747,
|
|
"step": 34816
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.942156616741889e-05,
|
|
"loss": 4.4701,
|
|
"step": 35328
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9413196598712105e-05,
|
|
"loss": 4.4835,
|
|
"step": 35840
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.940482703000532e-05,
|
|
"loss": 4.4759,
|
|
"step": 36352
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9396441082494794e-05,
|
|
"loss": 4.4659,
|
|
"step": 36864
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9388055134984274e-05,
|
|
"loss": 4.4653,
|
|
"step": 37376
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9379669187473754e-05,
|
|
"loss": 4.4719,
|
|
"step": 37888
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9371283239963234e-05,
|
|
"loss": 4.446,
|
|
"step": 38400
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.936291367125644e-05,
|
|
"loss": 4.4601,
|
|
"step": 38912
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.935452772374592e-05,
|
|
"loss": 4.4346,
|
|
"step": 39424
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.93461417762354e-05,
|
|
"loss": 4.4338,
|
|
"step": 39936
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.933775582872488e-05,
|
|
"loss": 4.4317,
|
|
"step": 40448
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.932936988121436e-05,
|
|
"loss": 4.4345,
|
|
"step": 40960
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.932098393370384e-05,
|
|
"loss": 4.4255,
|
|
"step": 41472
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.931259798619332e-05,
|
|
"loss": 4.4334,
|
|
"step": 41984
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.930421203868281e-05,
|
|
"loss": 4.418,
|
|
"step": 42496
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.929584246997602e-05,
|
|
"loss": 4.3969,
|
|
"step": 43008
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.92874565224655e-05,
|
|
"loss": 4.3995,
|
|
"step": 43520
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.927907057495498e-05,
|
|
"loss": 4.411,
|
|
"step": 44032
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.927070100624819e-05,
|
|
"loss": 4.3954,
|
|
"step": 44544
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.92623314375414e-05,
|
|
"loss": 4.3944,
|
|
"step": 45056
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.925394549003088e-05,
|
|
"loss": 4.3981,
|
|
"step": 45568
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.924555954252036e-05,
|
|
"loss": 4.3813,
|
|
"step": 46080
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.923717359500984e-05,
|
|
"loss": 4.3868,
|
|
"step": 46592
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.9228804026303046e-05,
|
|
"loss": 4.3807,
|
|
"step": 47104
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.922041807879253e-05,
|
|
"loss": 4.3858,
|
|
"step": 47616
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.921203213128201e-05,
|
|
"loss": 4.3832,
|
|
"step": 48128
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.920364618377149e-05,
|
|
"loss": 4.3773,
|
|
"step": 48640
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.919526023626097e-05,
|
|
"loss": 4.3577,
|
|
"step": 49152
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.918687428875045e-05,
|
|
"loss": 4.3646,
|
|
"step": 49664
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.917848834123993e-05,
|
|
"loss": 4.3568,
|
|
"step": 50176
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.917010239372941e-05,
|
|
"loss": 4.3604,
|
|
"step": 50688
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.916171644621889e-05,
|
|
"loss": 4.3648,
|
|
"step": 51200
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.915333049870837e-05,
|
|
"loss": 4.3437,
|
|
"step": 51712
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.914494455119785e-05,
|
|
"loss": 4.3483,
|
|
"step": 52224
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.913657498249106e-05,
|
|
"loss": 4.3495,
|
|
"step": 52736
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.912818903498054e-05,
|
|
"loss": 4.3432,
|
|
"step": 53248
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.911980308747002e-05,
|
|
"loss": 4.3316,
|
|
"step": 53760
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.91114171399595e-05,
|
|
"loss": 4.335,
|
|
"step": 54272
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.910303119244898e-05,
|
|
"loss": 4.322,
|
|
"step": 54784
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.909464524493846e-05,
|
|
"loss": 4.3218,
|
|
"step": 55296
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.908627567623168e-05,
|
|
"loss": 4.3195,
|
|
"step": 55808
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.907788972872115e-05,
|
|
"loss": 4.3196,
|
|
"step": 56320
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.906950378121063e-05,
|
|
"loss": 4.314,
|
|
"step": 56832
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.906111783370011e-05,
|
|
"loss": 4.3183,
|
|
"step": 57344
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.905273188618959e-05,
|
|
"loss": 4.3097,
|
|
"step": 57856
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.904434593867907e-05,
|
|
"loss": 4.3127,
|
|
"step": 58368
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.903595999116855e-05,
|
|
"loss": 4.3213,
|
|
"step": 58880
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.902759042246176e-05,
|
|
"loss": 4.3152,
|
|
"step": 59392
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.901920447495124e-05,
|
|
"loss": 4.3094,
|
|
"step": 59904
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.901081852744072e-05,
|
|
"loss": 4.2991,
|
|
"step": 60416
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.90024325799302e-05,
|
|
"loss": 4.2987,
|
|
"step": 60928
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8994046632419686e-05,
|
|
"loss": 4.2997,
|
|
"step": 61440
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8985660684909166e-05,
|
|
"loss": 4.2858,
|
|
"step": 61952
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8977291116202375e-05,
|
|
"loss": 4.2887,
|
|
"step": 62464
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8968905168691855e-05,
|
|
"loss": 4.2851,
|
|
"step": 62976
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8960519221181335e-05,
|
|
"loss": 4.2893,
|
|
"step": 63488
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8952133273670815e-05,
|
|
"loss": 4.2801,
|
|
"step": 64000
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8943747326160294e-05,
|
|
"loss": 4.2895,
|
|
"step": 64512
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8935377757453504e-05,
|
|
"loss": 4.2816,
|
|
"step": 65024
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8926991809942984e-05,
|
|
"loss": 4.2735,
|
|
"step": 65536
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8918605862432463e-05,
|
|
"loss": 4.2684,
|
|
"step": 66048
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8910219914921943e-05,
|
|
"loss": 4.2846,
|
|
"step": 66560
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8901833967411423e-05,
|
|
"loss": 4.2742,
|
|
"step": 67072
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.88934480199009e-05,
|
|
"loss": 4.2679,
|
|
"step": 67584
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.888506207239038e-05,
|
|
"loss": 4.2691,
|
|
"step": 68096
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.88766925036836e-05,
|
|
"loss": 4.2702,
|
|
"step": 68608
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.886830655617308e-05,
|
|
"loss": 4.2672,
|
|
"step": 69120
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.885992060866256e-05,
|
|
"loss": 4.2696,
|
|
"step": 69632
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.885155103995577e-05,
|
|
"loss": 4.2588,
|
|
"step": 70144
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.884316509244525e-05,
|
|
"loss": 4.2571,
|
|
"step": 70656
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.883477914493473e-05,
|
|
"loss": 4.2581,
|
|
"step": 71168
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.882639319742421e-05,
|
|
"loss": 4.25,
|
|
"step": 71680
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.881800724991369e-05,
|
|
"loss": 4.2452,
|
|
"step": 72192
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.880962130240316e-05,
|
|
"loss": 4.245,
|
|
"step": 72704
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.880123535489264e-05,
|
|
"loss": 4.2489,
|
|
"step": 73216
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.879286578618586e-05,
|
|
"loss": 4.2394,
|
|
"step": 73728
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.878447983867534e-05,
|
|
"loss": 4.2489,
|
|
"step": 74240
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.877609389116482e-05,
|
|
"loss": 4.2412,
|
|
"step": 74752
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.87677079436543e-05,
|
|
"loss": 4.2424,
|
|
"step": 75264
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.875933837494751e-05,
|
|
"loss": 4.2462,
|
|
"step": 75776
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8750952427436986e-05,
|
|
"loss": 4.2278,
|
|
"step": 76288
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 4.198204517364502,
|
|
"eval_runtime": 302.7571,
|
|
"eval_samples_per_second": 1260.387,
|
|
"eval_steps_per_second": 39.388,
|
|
"step": 76320
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8742566479926466e-05,
|
|
"loss": 4.2153,
|
|
"step": 76800
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8734180532415946e-05,
|
|
"loss": 4.2138,
|
|
"step": 77312
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.872581096370916e-05,
|
|
"loss": 4.238,
|
|
"step": 77824
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8717425016198635e-05,
|
|
"loss": 4.2213,
|
|
"step": 78336
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8709039068688115e-05,
|
|
"loss": 4.2241,
|
|
"step": 78848
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8700653121177595e-05,
|
|
"loss": 4.2182,
|
|
"step": 79360
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8692267173667075e-05,
|
|
"loss": 4.2109,
|
|
"step": 79872
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8683881226156555e-05,
|
|
"loss": 4.202,
|
|
"step": 80384
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.867551165744977e-05,
|
|
"loss": 4.2111,
|
|
"step": 80896
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.866712570993925e-05,
|
|
"loss": 4.2031,
|
|
"step": 81408
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.865873976242873e-05,
|
|
"loss": 4.2263,
|
|
"step": 81920
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.865035381491821e-05,
|
|
"loss": 4.2114,
|
|
"step": 82432
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.864198424621142e-05,
|
|
"loss": 4.1925,
|
|
"step": 82944
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.86335982987009e-05,
|
|
"loss": 4.1953,
|
|
"step": 83456
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.862521235119038e-05,
|
|
"loss": 4.1971,
|
|
"step": 83968
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.861682640367986e-05,
|
|
"loss": 4.1843,
|
|
"step": 84480
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.860844045616934e-05,
|
|
"loss": 4.1969,
|
|
"step": 84992
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.860005450865882e-05,
|
|
"loss": 4.1913,
|
|
"step": 85504
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.85916685611483e-05,
|
|
"loss": 4.1805,
|
|
"step": 86016
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.858328261363778e-05,
|
|
"loss": 4.2049,
|
|
"step": 86528
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.857489666612726e-05,
|
|
"loss": 4.1852,
|
|
"step": 87040
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.856651071861674e-05,
|
|
"loss": 4.186,
|
|
"step": 87552
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8558141149909955e-05,
|
|
"loss": 4.1768,
|
|
"step": 88064
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8549755202399435e-05,
|
|
"loss": 4.1956,
|
|
"step": 88576
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8541369254888915e-05,
|
|
"loss": 4.1737,
|
|
"step": 89088
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8532983307378395e-05,
|
|
"loss": 4.1744,
|
|
"step": 89600
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8524597359867875e-05,
|
|
"loss": 4.1703,
|
|
"step": 90112
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8516227791161084e-05,
|
|
"loss": 4.1688,
|
|
"step": 90624
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8507841843650564e-05,
|
|
"loss": 4.1673,
|
|
"step": 91136
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8499455896140044e-05,
|
|
"loss": 4.166,
|
|
"step": 91648
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8491069948629524e-05,
|
|
"loss": 4.1693,
|
|
"step": 92160
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8482684001119e-05,
|
|
"loss": 4.1671,
|
|
"step": 92672
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.847429805360848e-05,
|
|
"loss": 4.1652,
|
|
"step": 93184
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8465912106097964e-05,
|
|
"loss": 4.1701,
|
|
"step": 93696
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.845754253739117e-05,
|
|
"loss": 4.1644,
|
|
"step": 94208
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.844915658988065e-05,
|
|
"loss": 4.1653,
|
|
"step": 94720
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.844077064237013e-05,
|
|
"loss": 4.1492,
|
|
"step": 95232
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.843238469485961e-05,
|
|
"loss": 4.1523,
|
|
"step": 95744
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.842399874734909e-05,
|
|
"loss": 4.1433,
|
|
"step": 96256
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.841561279983857e-05,
|
|
"loss": 4.1509,
|
|
"step": 96768
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.840724323113178e-05,
|
|
"loss": 4.1423,
|
|
"step": 97280
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.839885728362126e-05,
|
|
"loss": 4.1517,
|
|
"step": 97792
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.839047133611074e-05,
|
|
"loss": 4.1339,
|
|
"step": 98304
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.838208538860022e-05,
|
|
"loss": 4.1537,
|
|
"step": 98816
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.83736994410897e-05,
|
|
"loss": 4.1457,
|
|
"step": 99328
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.836531349357918e-05,
|
|
"loss": 4.1453,
|
|
"step": 99840
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.83569439248724e-05,
|
|
"loss": 4.1428,
|
|
"step": 100352
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.834855797736188e-05,
|
|
"loss": 4.1279,
|
|
"step": 100864
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.834017202985136e-05,
|
|
"loss": 4.13,
|
|
"step": 101376
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.833178608234084e-05,
|
|
"loss": 4.1439,
|
|
"step": 101888
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.832340013483032e-05,
|
|
"loss": 4.1345,
|
|
"step": 102400
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.83150141873198e-05,
|
|
"loss": 4.1282,
|
|
"step": 102912
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.830662823980928e-05,
|
|
"loss": 4.1195,
|
|
"step": 103424
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8298258671102486e-05,
|
|
"loss": 4.1293,
|
|
"step": 103936
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8289872723591966e-05,
|
|
"loss": 4.1114,
|
|
"step": 104448
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8281486776081446e-05,
|
|
"loss": 4.137,
|
|
"step": 104960
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8273100828570926e-05,
|
|
"loss": 4.1076,
|
|
"step": 105472
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8264714881060406e-05,
|
|
"loss": 4.127,
|
|
"step": 105984
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8256328933549886e-05,
|
|
"loss": 4.1235,
|
|
"step": 106496
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8247942986039366e-05,
|
|
"loss": 4.1103,
|
|
"step": 107008
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.823957341733258e-05,
|
|
"loss": 4.1243,
|
|
"step": 107520
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.823118746982206e-05,
|
|
"loss": 4.1121,
|
|
"step": 108032
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8222801522311535e-05,
|
|
"loss": 4.1111,
|
|
"step": 108544
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8214415574801015e-05,
|
|
"loss": 4.0977,
|
|
"step": 109056
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8206029627290495e-05,
|
|
"loss": 4.1234,
|
|
"step": 109568
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8197643679779975e-05,
|
|
"loss": 4.0986,
|
|
"step": 110080
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8189257732269455e-05,
|
|
"loss": 4.107,
|
|
"step": 110592
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8180888163562664e-05,
|
|
"loss": 4.0888,
|
|
"step": 111104
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8172502216052144e-05,
|
|
"loss": 4.1007,
|
|
"step": 111616
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8164116268541624e-05,
|
|
"loss": 4.1112,
|
|
"step": 112128
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8155730321031104e-05,
|
|
"loss": 4.1088,
|
|
"step": 112640
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.814736075232432e-05,
|
|
"loss": 4.1031,
|
|
"step": 113152
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.81389748048138e-05,
|
|
"loss": 4.1086,
|
|
"step": 113664
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.813058885730328e-05,
|
|
"loss": 4.1114,
|
|
"step": 114176
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.812220290979276e-05,
|
|
"loss": 4.0973,
|
|
"step": 114688
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.811381696228224e-05,
|
|
"loss": 4.1069,
|
|
"step": 115200
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.810544739357545e-05,
|
|
"loss": 4.0937,
|
|
"step": 115712
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.809706144606493e-05,
|
|
"loss": 4.0938,
|
|
"step": 116224
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.808867549855441e-05,
|
|
"loss": 4.0892,
|
|
"step": 116736
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.808028955104389e-05,
|
|
"loss": 4.0989,
|
|
"step": 117248
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.80719199823371e-05,
|
|
"loss": 4.0985,
|
|
"step": 117760
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.806353403482658e-05,
|
|
"loss": 4.1027,
|
|
"step": 118272
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.805514808731606e-05,
|
|
"loss": 4.0917,
|
|
"step": 118784
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.804676213980554e-05,
|
|
"loss": 4.072,
|
|
"step": 119296
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.803839257109875e-05,
|
|
"loss": 4.0766,
|
|
"step": 119808
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.803000662358823e-05,
|
|
"loss": 4.0888,
|
|
"step": 120320
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.802162067607771e-05,
|
|
"loss": 4.0822,
|
|
"step": 120832
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.801323472856719e-05,
|
|
"loss": 4.0807,
|
|
"step": 121344
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.80048651598604e-05,
|
|
"loss": 4.0884,
|
|
"step": 121856
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.799647921234988e-05,
|
|
"loss": 4.0742,
|
|
"step": 122368
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.798809326483936e-05,
|
|
"loss": 4.0768,
|
|
"step": 122880
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.797970731732884e-05,
|
|
"loss": 4.0781,
|
|
"step": 123392
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.797133774862205e-05,
|
|
"loss": 4.0878,
|
|
"step": 123904
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.796295180111153e-05,
|
|
"loss": 4.0841,
|
|
"step": 124416
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.795456585360101e-05,
|
|
"loss": 4.0889,
|
|
"step": 124928
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.794617990609049e-05,
|
|
"loss": 4.0642,
|
|
"step": 125440
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.793781033738371e-05,
|
|
"loss": 4.073,
|
|
"step": 125952
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.792942438987319e-05,
|
|
"loss": 4.0714,
|
|
"step": 126464
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.792103844236267e-05,
|
|
"loss": 4.0711,
|
|
"step": 126976
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.791265249485215e-05,
|
|
"loss": 4.0835,
|
|
"step": 127488
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.790426654734163e-05,
|
|
"loss": 4.0634,
|
|
"step": 128000
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7895896978634836e-05,
|
|
"loss": 4.0669,
|
|
"step": 128512
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7887511031124316e-05,
|
|
"loss": 4.074,
|
|
"step": 129024
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7879125083613796e-05,
|
|
"loss": 4.0708,
|
|
"step": 129536
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7870739136103276e-05,
|
|
"loss": 4.0543,
|
|
"step": 130048
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7862369567396485e-05,
|
|
"loss": 4.0656,
|
|
"step": 130560
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7853983619885965e-05,
|
|
"loss": 4.0573,
|
|
"step": 131072
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7845597672375445e-05,
|
|
"loss": 4.0569,
|
|
"step": 131584
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7837211724864925e-05,
|
|
"loss": 4.0536,
|
|
"step": 132096
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.782882577735441e-05,
|
|
"loss": 4.0531,
|
|
"step": 132608
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.782045620864762e-05,
|
|
"loss": 4.0534,
|
|
"step": 133120
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.78120702611371e-05,
|
|
"loss": 4.0642,
|
|
"step": 133632
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.780368431362658e-05,
|
|
"loss": 4.0494,
|
|
"step": 134144
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.779529836611606e-05,
|
|
"loss": 4.0583,
|
|
"step": 134656
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.778692879740927e-05,
|
|
"loss": 4.0687,
|
|
"step": 135168
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.777854284989875e-05,
|
|
"loss": 4.0629,
|
|
"step": 135680
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.777017328119196e-05,
|
|
"loss": 4.0586,
|
|
"step": 136192
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.776178733368144e-05,
|
|
"loss": 4.0525,
|
|
"step": 136704
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.775340138617092e-05,
|
|
"loss": 4.0492,
|
|
"step": 137216
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.77450154386604e-05,
|
|
"loss": 4.0526,
|
|
"step": 137728
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.773662949114988e-05,
|
|
"loss": 4.0443,
|
|
"step": 138240
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7728259922443094e-05,
|
|
"loss": 4.0453,
|
|
"step": 138752
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7719873974932574e-05,
|
|
"loss": 4.046,
|
|
"step": 139264
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7711488027422054e-05,
|
|
"loss": 4.0504,
|
|
"step": 139776
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7703102079911534e-05,
|
|
"loss": 4.047,
|
|
"step": 140288
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7694716132401014e-05,
|
|
"loss": 4.0485,
|
|
"step": 140800
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7686330184890494e-05,
|
|
"loss": 4.0481,
|
|
"step": 141312
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.767794423737997e-05,
|
|
"loss": 4.0367,
|
|
"step": 141824
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.766957466867318e-05,
|
|
"loss": 4.0365,
|
|
"step": 142336
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.766118872116266e-05,
|
|
"loss": 4.0551,
|
|
"step": 142848
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7652802773652136e-05,
|
|
"loss": 4.0489,
|
|
"step": 143360
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7644416826141616e-05,
|
|
"loss": 4.044,
|
|
"step": 143872
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7636030878631096e-05,
|
|
"loss": 4.0391,
|
|
"step": 144384
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.762766130992431e-05,
|
|
"loss": 4.0439,
|
|
"step": 144896
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.761927536241379e-05,
|
|
"loss": 4.0417,
|
|
"step": 145408
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.761088941490327e-05,
|
|
"loss": 4.0507,
|
|
"step": 145920
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.760250346739275e-05,
|
|
"loss": 4.0414,
|
|
"step": 146432
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.759413389868596e-05,
|
|
"loss": 4.0368,
|
|
"step": 146944
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.758574795117544e-05,
|
|
"loss": 4.0389,
|
|
"step": 147456
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.757736200366492e-05,
|
|
"loss": 4.0347,
|
|
"step": 147968
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.75689760561544e-05,
|
|
"loss": 4.0279,
|
|
"step": 148480
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.756060648744761e-05,
|
|
"loss": 4.0322,
|
|
"step": 148992
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.755222053993709e-05,
|
|
"loss": 4.0382,
|
|
"step": 149504
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.754383459242657e-05,
|
|
"loss": 4.0225,
|
|
"step": 150016
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.753544864491605e-05,
|
|
"loss": 4.0408,
|
|
"step": 150528
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7527079076209266e-05,
|
|
"loss": 4.0322,
|
|
"step": 151040
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7518693128698745e-05,
|
|
"loss": 4.0297,
|
|
"step": 151552
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7510307181188225e-05,
|
|
"loss": 4.0355,
|
|
"step": 152064
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7501921233677705e-05,
|
|
"loss": 4.022,
|
|
"step": 152576
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 4.027729034423828,
|
|
"eval_runtime": 302.5198,
|
|
"eval_samples_per_second": 1261.375,
|
|
"eval_steps_per_second": 39.419,
|
|
"step": 152640
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7493535286167185e-05,
|
|
"loss": 4.0091,
|
|
"step": 153088
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7485149338656665e-05,
|
|
"loss": 4.0107,
|
|
"step": 153600
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7476763391146145e-05,
|
|
"loss": 4.0363,
|
|
"step": 154112
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7468377443635625e-05,
|
|
"loss": 4.0207,
|
|
"step": 154624
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7459991496125105e-05,
|
|
"loss": 4.0299,
|
|
"step": 155136
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7451605548614585e-05,
|
|
"loss": 4.0155,
|
|
"step": 155648
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7443219601104065e-05,
|
|
"loss": 4.0168,
|
|
"step": 156160
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7434833653593545e-05,
|
|
"loss": 4.0049,
|
|
"step": 156672
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7426447706083025e-05,
|
|
"loss": 4.0167,
|
|
"step": 157184
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7418061758572505e-05,
|
|
"loss": 4.0108,
|
|
"step": 157696
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7409675811061985e-05,
|
|
"loss": 4.0314,
|
|
"step": 158208
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7401289863551465e-05,
|
|
"loss": 4.0211,
|
|
"step": 158720
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.739293667364841e-05,
|
|
"loss": 3.9997,
|
|
"step": 159232
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.738455072613789e-05,
|
|
"loss": 4.0054,
|
|
"step": 159744
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.737616477862737e-05,
|
|
"loss": 4.0103,
|
|
"step": 160256
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.736777883111685e-05,
|
|
"loss": 3.9944,
|
|
"step": 160768
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.735939288360632e-05,
|
|
"loss": 4.0083,
|
|
"step": 161280
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.73510069360958e-05,
|
|
"loss": 4.0086,
|
|
"step": 161792
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.734262098858528e-05,
|
|
"loss": 3.9988,
|
|
"step": 162304
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.733423504107476e-05,
|
|
"loss": 4.0262,
|
|
"step": 162816
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.732586547236797e-05,
|
|
"loss": 4.0019,
|
|
"step": 163328
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.731747952485746e-05,
|
|
"loss": 4.0083,
|
|
"step": 163840
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.730909357734694e-05,
|
|
"loss": 3.999,
|
|
"step": 164352
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.730070762983642e-05,
|
|
"loss": 4.0161,
|
|
"step": 164864
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.729233806112963e-05,
|
|
"loss": 3.9915,
|
|
"step": 165376
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.728395211361911e-05,
|
|
"loss": 4.0005,
|
|
"step": 165888
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.727556616610859e-05,
|
|
"loss": 3.9929,
|
|
"step": 166400
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.726718021859807e-05,
|
|
"loss": 3.998,
|
|
"step": 166912
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7258810649891277e-05,
|
|
"loss": 3.9925,
|
|
"step": 167424
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7250424702380757e-05,
|
|
"loss": 3.9929,
|
|
"step": 167936
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7242038754870237e-05,
|
|
"loss": 3.9991,
|
|
"step": 168448
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7233652807359716e-05,
|
|
"loss": 3.997,
|
|
"step": 168960
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7225283238652926e-05,
|
|
"loss": 3.9921,
|
|
"step": 169472
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.721689729114241e-05,
|
|
"loss": 4.0027,
|
|
"step": 169984
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.720851134363189e-05,
|
|
"loss": 3.9928,
|
|
"step": 170496
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.720012539612137e-05,
|
|
"loss": 3.9981,
|
|
"step": 171008
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.719175582741458e-05,
|
|
"loss": 3.9835,
|
|
"step": 171520
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.718336987990406e-05,
|
|
"loss": 3.9878,
|
|
"step": 172032
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.717498393239354e-05,
|
|
"loss": 3.9749,
|
|
"step": 172544
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.716659798488302e-05,
|
|
"loss": 3.9898,
|
|
"step": 173056
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.715822841617623e-05,
|
|
"loss": 3.9765,
|
|
"step": 173568
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.714984246866571e-05,
|
|
"loss": 3.9905,
|
|
"step": 174080
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.714145652115519e-05,
|
|
"loss": 3.9759,
|
|
"step": 174592
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.713307057364467e-05,
|
|
"loss": 3.9871,
|
|
"step": 175104
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.712470100493788e-05,
|
|
"loss": 3.9899,
|
|
"step": 175616
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7116315057427366e-05,
|
|
"loss": 3.9842,
|
|
"step": 176128
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7107929109916846e-05,
|
|
"loss": 3.9856,
|
|
"step": 176640
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7099543162406326e-05,
|
|
"loss": 3.9708,
|
|
"step": 177152
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7091173593699535e-05,
|
|
"loss": 3.9721,
|
|
"step": 177664
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7082787646189015e-05,
|
|
"loss": 3.9872,
|
|
"step": 178176
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7074401698678495e-05,
|
|
"loss": 3.9778,
|
|
"step": 178688
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7066015751167975e-05,
|
|
"loss": 3.9775,
|
|
"step": 179200
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7057646182461184e-05,
|
|
"loss": 3.9644,
|
|
"step": 179712
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.704927661375439e-05,
|
|
"loss": 3.9781,
|
|
"step": 180224
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.704089066624387e-05,
|
|
"loss": 3.9544,
|
|
"step": 180736
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.703250471873335e-05,
|
|
"loss": 3.9878,
|
|
"step": 181248
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.702411877122283e-05,
|
|
"loss": 3.9558,
|
|
"step": 181760
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.701573282371232e-05,
|
|
"loss": 3.9763,
|
|
"step": 182272
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.70073468762018e-05,
|
|
"loss": 3.9773,
|
|
"step": 182784
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.699896092869128e-05,
|
|
"loss": 3.9599,
|
|
"step": 183296
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.699057498118076e-05,
|
|
"loss": 3.9784,
|
|
"step": 183808
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.698220541247397e-05,
|
|
"loss": 3.9628,
|
|
"step": 184320
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.697381946496345e-05,
|
|
"loss": 3.964,
|
|
"step": 184832
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.696543351745293e-05,
|
|
"loss": 3.952,
|
|
"step": 185344
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.695704756994241e-05,
|
|
"loss": 3.9766,
|
|
"step": 185856
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.694867800123562e-05,
|
|
"loss": 3.9557,
|
|
"step": 186368
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.69402920537251e-05,
|
|
"loss": 3.9609,
|
|
"step": 186880
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.693190610621458e-05,
|
|
"loss": 3.9473,
|
|
"step": 187392
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.692352015870406e-05,
|
|
"loss": 3.9563,
|
|
"step": 187904
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.6915150589997267e-05,
|
|
"loss": 3.9664,
|
|
"step": 188416
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.690676464248675e-05,
|
|
"loss": 3.9695,
|
|
"step": 188928
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.689837869497623e-05,
|
|
"loss": 3.9581,
|
|
"step": 189440
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.688999274746571e-05,
|
|
"loss": 3.9721,
|
|
"step": 189952
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.688162317875892e-05,
|
|
"loss": 3.9701,
|
|
"step": 190464
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.68732372312484e-05,
|
|
"loss": 3.9617,
|
|
"step": 190976
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.686485128373788e-05,
|
|
"loss": 3.9624,
|
|
"step": 191488
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.685646533622736e-05,
|
|
"loss": 3.9597,
|
|
"step": 192000
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.684809576752057e-05,
|
|
"loss": 3.9542,
|
|
"step": 192512
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.683970982001005e-05,
|
|
"loss": 3.954,
|
|
"step": 193024
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.683132387249953e-05,
|
|
"loss": 3.96,
|
|
"step": 193536
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.682293792498901e-05,
|
|
"loss": 3.9627,
|
|
"step": 194048
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.681456835628222e-05,
|
|
"loss": 3.9699,
|
|
"step": 194560
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.680618240877171e-05,
|
|
"loss": 3.9544,
|
|
"step": 195072
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.679779646126119e-05,
|
|
"loss": 3.9353,
|
|
"step": 195584
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.678941051375067e-05,
|
|
"loss": 3.9439,
|
|
"step": 196096
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.6781040945043876e-05,
|
|
"loss": 3.9539,
|
|
"step": 196608
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.6772654997533356e-05,
|
|
"loss": 3.9505,
|
|
"step": 197120
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.6764269050022836e-05,
|
|
"loss": 3.9479,
|
|
"step": 197632
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.6755883102512316e-05,
|
|
"loss": 3.9543,
|
|
"step": 198144
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6747513533805525e-05,
|
|
"loss": 3.9489,
|
|
"step": 198656
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6739127586295005e-05,
|
|
"loss": 3.9393,
|
|
"step": 199168
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6730741638784485e-05,
|
|
"loss": 3.9497,
|
|
"step": 199680
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6722355691273965e-05,
|
|
"loss": 3.9589,
|
|
"step": 200192
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6713986122567174e-05,
|
|
"loss": 3.9531,
|
|
"step": 200704
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.670560017505666e-05,
|
|
"loss": 3.9583,
|
|
"step": 201216
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.669721422754614e-05,
|
|
"loss": 3.9415,
|
|
"step": 201728
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.668882828003562e-05,
|
|
"loss": 3.9446,
|
|
"step": 202240
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.668045871132883e-05,
|
|
"loss": 3.9413,
|
|
"step": 202752
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.667207276381831e-05,
|
|
"loss": 3.9457,
|
|
"step": 203264
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.666368681630779e-05,
|
|
"loss": 3.9545,
|
|
"step": 203776
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.665530086879727e-05,
|
|
"loss": 3.9385,
|
|
"step": 204288
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.664693130009048e-05,
|
|
"loss": 3.9397,
|
|
"step": 204800
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.663854535257996e-05,
|
|
"loss": 3.9476,
|
|
"step": 205312
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.663015940506944e-05,
|
|
"loss": 3.9488,
|
|
"step": 205824
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.662177345755892e-05,
|
|
"loss": 3.9306,
|
|
"step": 206336
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6613420267655863e-05,
|
|
"loss": 3.9381,
|
|
"step": 206848
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.660503432014534e-05,
|
|
"loss": 3.9392,
|
|
"step": 207360
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.659664837263482e-05,
|
|
"loss": 3.9339,
|
|
"step": 207872
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.65882624251243e-05,
|
|
"loss": 3.9298,
|
|
"step": 208384
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.657987647761378e-05,
|
|
"loss": 3.9321,
|
|
"step": 208896
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.657149053010326e-05,
|
|
"loss": 3.9358,
|
|
"step": 209408
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.656310458259274e-05,
|
|
"loss": 3.938,
|
|
"step": 209920
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.655471863508222e-05,
|
|
"loss": 3.929,
|
|
"step": 210432
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.654634906637543e-05,
|
|
"loss": 3.9371,
|
|
"step": 210944
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.653796311886491e-05,
|
|
"loss": 3.9474,
|
|
"step": 211456
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.652957717135439e-05,
|
|
"loss": 3.949,
|
|
"step": 211968
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.65212076026476e-05,
|
|
"loss": 3.9373,
|
|
"step": 212480
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.651282165513708e-05,
|
|
"loss": 3.9367,
|
|
"step": 212992
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.650443570762657e-05,
|
|
"loss": 3.9314,
|
|
"step": 213504
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.649604976011605e-05,
|
|
"loss": 3.9355,
|
|
"step": 214016
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.648766381260553e-05,
|
|
"loss": 3.9252,
|
|
"step": 214528
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.647929424389874e-05,
|
|
"loss": 3.9278,
|
|
"step": 215040
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.647090829638822e-05,
|
|
"loss": 3.9285,
|
|
"step": 215552
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.64625223488777e-05,
|
|
"loss": 3.938,
|
|
"step": 216064
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.645413640136718e-05,
|
|
"loss": 3.9302,
|
|
"step": 216576
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.644575045385665e-05,
|
|
"loss": 3.9327,
|
|
"step": 217088
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.643736450634613e-05,
|
|
"loss": 3.9322,
|
|
"step": 217600
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.642897855883561e-05,
|
|
"loss": 3.9229,
|
|
"step": 218112
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.642059261132509e-05,
|
|
"loss": 3.9205,
|
|
"step": 218624
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6412223042618306e-05,
|
|
"loss": 3.9408,
|
|
"step": 219136
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6403837095107786e-05,
|
|
"loss": 3.9383,
|
|
"step": 219648
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6395451147597266e-05,
|
|
"loss": 3.9303,
|
|
"step": 220160
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6387081578890475e-05,
|
|
"loss": 3.9268,
|
|
"step": 220672
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.637871201018369e-05,
|
|
"loss": 3.9271,
|
|
"step": 221184
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.637032606267317e-05,
|
|
"loss": 3.9306,
|
|
"step": 221696
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.636194011516265e-05,
|
|
"loss": 3.9418,
|
|
"step": 222208
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6353554167652124e-05,
|
|
"loss": 3.929,
|
|
"step": 222720
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6345168220141604e-05,
|
|
"loss": 3.9228,
|
|
"step": 223232
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6336782272631084e-05,
|
|
"loss": 3.9264,
|
|
"step": 223744
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6328396325120564e-05,
|
|
"loss": 3.9232,
|
|
"step": 224256
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6320010377610043e-05,
|
|
"loss": 3.9204,
|
|
"step": 224768
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.631164080890326e-05,
|
|
"loss": 3.9203,
|
|
"step": 225280
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.630325486139274e-05,
|
|
"loss": 3.9275,
|
|
"step": 225792
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.629486891388222e-05,
|
|
"loss": 3.9163,
|
|
"step": 226304
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.62864829663717e-05,
|
|
"loss": 3.9289,
|
|
"step": 226816
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.627811339766491e-05,
|
|
"loss": 3.9262,
|
|
"step": 227328
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.626972745015439e-05,
|
|
"loss": 3.9209,
|
|
"step": 227840
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.626134150264387e-05,
|
|
"loss": 3.9269,
|
|
"step": 228352
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.625295555513335e-05,
|
|
"loss": 3.9114,
|
|
"step": 228864
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.95308780670166,
|
|
"eval_runtime": 304.6541,
|
|
"eval_samples_per_second": 1252.539,
|
|
"eval_steps_per_second": 39.143,
|
|
"step": 228960
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.624458598642656e-05,
|
|
"loss": 3.915,
|
|
"step": 229376
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.623620003891604e-05,
|
|
"loss": 3.9042,
|
|
"step": 229888
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.622781409140552e-05,
|
|
"loss": 3.9287,
|
|
"step": 230400
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.6219428143895e-05,
|
|
"loss": 3.9135,
|
|
"step": 230912
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.621105857518821e-05,
|
|
"loss": 3.9283,
|
|
"step": 231424
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.620267262767769e-05,
|
|
"loss": 3.9081,
|
|
"step": 231936
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.619428668016717e-05,
|
|
"loss": 3.9171,
|
|
"step": 232448
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.618591711146038e-05,
|
|
"loss": 3.8989,
|
|
"step": 232960
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.617753116394986e-05,
|
|
"loss": 3.9165,
|
|
"step": 233472
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.616914521643934e-05,
|
|
"loss": 3.9041,
|
|
"step": 233984
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.616077564773255e-05,
|
|
"loss": 3.9276,
|
|
"step": 234496
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.615238970022203e-05,
|
|
"loss": 3.9171,
|
|
"step": 235008
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.614402013151524e-05,
|
|
"loss": 3.8957,
|
|
"step": 235520
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.613563418400472e-05,
|
|
"loss": 3.9043,
|
|
"step": 236032
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.61272482364942e-05,
|
|
"loss": 3.9104,
|
|
"step": 236544
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.611886228898368e-05,
|
|
"loss": 3.8943,
|
|
"step": 237056
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.611047634147316e-05,
|
|
"loss": 3.9046,
|
|
"step": 237568
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.610209039396265e-05,
|
|
"loss": 3.9067,
|
|
"step": 238080
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.609370444645213e-05,
|
|
"loss": 3.8982,
|
|
"step": 238592
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.6085318498941607e-05,
|
|
"loss": 3.9294,
|
|
"step": 239104
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.6076932551431087e-05,
|
|
"loss": 3.902,
|
|
"step": 239616
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.6068546603920566e-05,
|
|
"loss": 3.9048,
|
|
"step": 240128
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.6060177035213776e-05,
|
|
"loss": 3.9026,
|
|
"step": 240640
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.6051791087703256e-05,
|
|
"loss": 3.9106,
|
|
"step": 241152
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.6043405140192735e-05,
|
|
"loss": 3.8958,
|
|
"step": 241664
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.6035019192682215e-05,
|
|
"loss": 3.9016,
|
|
"step": 242176
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.6026633245171695e-05,
|
|
"loss": 3.8971,
|
|
"step": 242688
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.6018263676464904e-05,
|
|
"loss": 3.9002,
|
|
"step": 243200
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.6009877728954384e-05,
|
|
"loss": 3.8941,
|
|
"step": 243712
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.6001491781443864e-05,
|
|
"loss": 3.8956,
|
|
"step": 244224
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5993105833933344e-05,
|
|
"loss": 3.8977,
|
|
"step": 244736
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.598471988642283e-05,
|
|
"loss": 3.9027,
|
|
"step": 245248
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.597633393891231e-05,
|
|
"loss": 3.9027,
|
|
"step": 245760
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5967947991401784e-05,
|
|
"loss": 3.9028,
|
|
"step": 246272
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5959578422695e-05,
|
|
"loss": 3.8961,
|
|
"step": 246784
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.595119247518448e-05,
|
|
"loss": 3.9025,
|
|
"step": 247296
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.594280652767395e-05,
|
|
"loss": 3.8846,
|
|
"step": 247808
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.593442058016343e-05,
|
|
"loss": 3.899,
|
|
"step": 248320
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.592603463265291e-05,
|
|
"loss": 3.8816,
|
|
"step": 248832
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.591764868514239e-05,
|
|
"loss": 3.8918,
|
|
"step": 249344
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.590926273763187e-05,
|
|
"loss": 3.8831,
|
|
"step": 249856
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.590087679012135e-05,
|
|
"loss": 3.9,
|
|
"step": 250368
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.58925236002183e-05,
|
|
"loss": 3.8821,
|
|
"step": 250880
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5884137652707785e-05,
|
|
"loss": 3.8901,
|
|
"step": 251392
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.587575170519726e-05,
|
|
"loss": 3.9007,
|
|
"step": 251904
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.586736575768674e-05,
|
|
"loss": 3.893,
|
|
"step": 252416
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.585897981017622e-05,
|
|
"loss": 3.8952,
|
|
"step": 252928
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.585061024146943e-05,
|
|
"loss": 3.8799,
|
|
"step": 253440
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.584222429395891e-05,
|
|
"loss": 3.8795,
|
|
"step": 253952
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.583383834644839e-05,
|
|
"loss": 3.8934,
|
|
"step": 254464
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.582545239893787e-05,
|
|
"loss": 3.8872,
|
|
"step": 254976
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.581706645142735e-05,
|
|
"loss": 3.8842,
|
|
"step": 255488
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.580868050391683e-05,
|
|
"loss": 3.8778,
|
|
"step": 256000
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5800310935210036e-05,
|
|
"loss": 3.8886,
|
|
"step": 256512
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.579192498769952e-05,
|
|
"loss": 3.8676,
|
|
"step": 257024
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5783539040189e-05,
|
|
"loss": 3.8898,
|
|
"step": 257536
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.577515309267848e-05,
|
|
"loss": 3.8708,
|
|
"step": 258048
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.576676714516796e-05,
|
|
"loss": 3.882,
|
|
"step": 258560
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.575838119765744e-05,
|
|
"loss": 3.892,
|
|
"step": 259072
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.574999525014692e-05,
|
|
"loss": 3.8722,
|
|
"step": 259584
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.57416093026364e-05,
|
|
"loss": 3.8855,
|
|
"step": 260096
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.573323973392961e-05,
|
|
"loss": 3.8773,
|
|
"step": 260608
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.572485378641909e-05,
|
|
"loss": 3.8731,
|
|
"step": 261120
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.57164842177123e-05,
|
|
"loss": 3.8693,
|
|
"step": 261632
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.570809827020178e-05,
|
|
"loss": 3.8849,
|
|
"step": 262144
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.569971232269126e-05,
|
|
"loss": 3.8712,
|
|
"step": 262656
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.569132637518074e-05,
|
|
"loss": 3.8743,
|
|
"step": 263168
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.568294042767022e-05,
|
|
"loss": 3.8621,
|
|
"step": 263680
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.567455448015971e-05,
|
|
"loss": 3.867,
|
|
"step": 264192
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.566616853264919e-05,
|
|
"loss": 3.8787,
|
|
"step": 264704
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5657798963942396e-05,
|
|
"loss": 3.8837,
|
|
"step": 265216
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5649413016431876e-05,
|
|
"loss": 3.8725,
|
|
"step": 265728
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5641027068921356e-05,
|
|
"loss": 3.8872,
|
|
"step": 266240
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5632641121410836e-05,
|
|
"loss": 3.8814,
|
|
"step": 266752
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5624255173900316e-05,
|
|
"loss": 3.8741,
|
|
"step": 267264
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.561586922638979e-05,
|
|
"loss": 3.8793,
|
|
"step": 267776
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.560748327887927e-05,
|
|
"loss": 3.8739,
|
|
"step": 268288
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5599113710172485e-05,
|
|
"loss": 3.8727,
|
|
"step": 268800
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5590727762661965e-05,
|
|
"loss": 3.8701,
|
|
"step": 269312
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5582341815151445e-05,
|
|
"loss": 3.8744,
|
|
"step": 269824
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5573955867640925e-05,
|
|
"loss": 3.8806,
|
|
"step": 270336
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5565569920130405e-05,
|
|
"loss": 3.8784,
|
|
"step": 270848
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5557200351423614e-05,
|
|
"loss": 3.873,
|
|
"step": 271360
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5548814403913094e-05,
|
|
"loss": 3.8545,
|
|
"step": 271872
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5540428456402574e-05,
|
|
"loss": 3.8604,
|
|
"step": 272384
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5532042508892054e-05,
|
|
"loss": 3.8699,
|
|
"step": 272896
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5523656561381534e-05,
|
|
"loss": 3.8693,
|
|
"step": 273408
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5515270613871014e-05,
|
|
"loss": 3.8658,
|
|
"step": 273920
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.5506884666360494e-05,
|
|
"loss": 3.8705,
|
|
"step": 274432
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.54985150976537e-05,
|
|
"loss": 3.867,
|
|
"step": 274944
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.549012915014318e-05,
|
|
"loss": 3.8598,
|
|
"step": 275456
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.548174320263266e-05,
|
|
"loss": 3.8654,
|
|
"step": 275968
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.547335725512214e-05,
|
|
"loss": 3.8724,
|
|
"step": 276480
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.546497130761162e-05,
|
|
"loss": 3.8707,
|
|
"step": 276992
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.545660173890484e-05,
|
|
"loss": 3.8805,
|
|
"step": 277504
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.544821579139432e-05,
|
|
"loss": 3.8625,
|
|
"step": 278016
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.54398298438838e-05,
|
|
"loss": 3.8603,
|
|
"step": 278528
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.543144389637328e-05,
|
|
"loss": 3.8639,
|
|
"step": 279040
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.542305794886276e-05,
|
|
"loss": 3.8637,
|
|
"step": 279552
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.541467200135224e-05,
|
|
"loss": 3.8751,
|
|
"step": 280064
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.540630243264545e-05,
|
|
"loss": 3.858,
|
|
"step": 280576
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.539791648513493e-05,
|
|
"loss": 3.8618,
|
|
"step": 281088
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.538953053762441e-05,
|
|
"loss": 3.8658,
|
|
"step": 281600
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.538114459011389e-05,
|
|
"loss": 3.8699,
|
|
"step": 282112
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.537275864260337e-05,
|
|
"loss": 3.8519,
|
|
"step": 282624
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.5364389073896576e-05,
|
|
"loss": 3.8596,
|
|
"step": 283136
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.535600312638606e-05,
|
|
"loss": 3.861,
|
|
"step": 283648
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.534761717887554e-05,
|
|
"loss": 3.8543,
|
|
"step": 284160
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.533923123136502e-05,
|
|
"loss": 3.8479,
|
|
"step": 284672
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.53308452838545e-05,
|
|
"loss": 3.8571,
|
|
"step": 285184
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.5322459336343976e-05,
|
|
"loss": 3.8591,
|
|
"step": 285696
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.531408976763719e-05,
|
|
"loss": 3.8586,
|
|
"step": 286208
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.530570382012667e-05,
|
|
"loss": 3.8529,
|
|
"step": 286720
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.529731787261615e-05,
|
|
"loss": 3.8563,
|
|
"step": 287232
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.5288931925105625e-05,
|
|
"loss": 3.8716,
|
|
"step": 287744
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.5280545977595105e-05,
|
|
"loss": 3.8697,
|
|
"step": 288256
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.527217640888832e-05,
|
|
"loss": 3.8632,
|
|
"step": 288768
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.52637904613778e-05,
|
|
"loss": 3.8587,
|
|
"step": 289280
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.525540451386728e-05,
|
|
"loss": 3.8559,
|
|
"step": 289792
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.524701856635676e-05,
|
|
"loss": 3.8548,
|
|
"step": 290304
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.523863261884624e-05,
|
|
"loss": 3.8483,
|
|
"step": 290816
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.523026305013945e-05,
|
|
"loss": 3.8519,
|
|
"step": 291328
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.522187710262893e-05,
|
|
"loss": 3.852,
|
|
"step": 291840
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.521349115511841e-05,
|
|
"loss": 3.8612,
|
|
"step": 292352
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.520510520760789e-05,
|
|
"loss": 3.8535,
|
|
"step": 292864
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.51967356389011e-05,
|
|
"loss": 3.8604,
|
|
"step": 293376
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.518834969139058e-05,
|
|
"loss": 3.8522,
|
|
"step": 293888
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.517996374388006e-05,
|
|
"loss": 3.8507,
|
|
"step": 294400
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.517157779636954e-05,
|
|
"loss": 3.8452,
|
|
"step": 294912
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.516319184885902e-05,
|
|
"loss": 3.8666,
|
|
"step": 295424
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.5154822280152234e-05,
|
|
"loss": 3.8648,
|
|
"step": 295936
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.5146436332641714e-05,
|
|
"loss": 3.8551,
|
|
"step": 296448
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.5138050385131194e-05,
|
|
"loss": 3.8512,
|
|
"step": 296960
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.5129664437620674e-05,
|
|
"loss": 3.856,
|
|
"step": 297472
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.5121278490110154e-05,
|
|
"loss": 3.8536,
|
|
"step": 297984
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.5112892542599634e-05,
|
|
"loss": 3.8672,
|
|
"step": 298496
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.5104506595089114e-05,
|
|
"loss": 3.8496,
|
|
"step": 299008
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.509613702638232e-05,
|
|
"loss": 3.8537,
|
|
"step": 299520
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.50877510788718e-05,
|
|
"loss": 3.8513,
|
|
"step": 300032
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.507936513136128e-05,
|
|
"loss": 3.8557,
|
|
"step": 300544
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.507097918385076e-05,
|
|
"loss": 3.8441,
|
|
"step": 301056
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.506259323634024e-05,
|
|
"loss": 3.8465,
|
|
"step": 301568
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.505422366763345e-05,
|
|
"loss": 3.8545,
|
|
"step": 302080
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.504583772012293e-05,
|
|
"loss": 3.8464,
|
|
"step": 302592
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.503745177261242e-05,
|
|
"loss": 3.8544,
|
|
"step": 303104
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.50290658251019e-05,
|
|
"loss": 3.8558,
|
|
"step": 303616
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.502067987759138e-05,
|
|
"loss": 3.8488,
|
|
"step": 304128
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.501231030888459e-05,
|
|
"loss": 3.8515,
|
|
"step": 304640
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.500392436137407e-05,
|
|
"loss": 3.8457,
|
|
"step": 305152
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 3.9113528728485107,
|
|
"eval_runtime": 304.0304,
|
|
"eval_samples_per_second": 1255.108,
|
|
"eval_steps_per_second": 39.223,
|
|
"step": 305280
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.499553841386355e-05,
|
|
"loss": 3.835,
|
|
"step": 305664
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.498715246635303e-05,
|
|
"loss": 3.8334,
|
|
"step": 306176
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.497876651884251e-05,
|
|
"loss": 3.855,
|
|
"step": 306688
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.497038057133199e-05,
|
|
"loss": 3.8469,
|
|
"step": 307200
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.496199462382146e-05,
|
|
"loss": 3.8579,
|
|
"step": 307712
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.495360867631094e-05,
|
|
"loss": 3.8369,
|
|
"step": 308224
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.494522272880042e-05,
|
|
"loss": 3.8458,
|
|
"step": 308736
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.49368367812899e-05,
|
|
"loss": 3.8284,
|
|
"step": 309248
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.4928467212583116e-05,
|
|
"loss": 3.8483,
|
|
"step": 309760
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.4920081265072596e-05,
|
|
"loss": 3.8335,
|
|
"step": 310272
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.4911695317562076e-05,
|
|
"loss": 3.8516,
|
|
"step": 310784
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.4903309370051556e-05,
|
|
"loss": 3.849,
|
|
"step": 311296
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.4894939801344765e-05,
|
|
"loss": 3.8236,
|
|
"step": 311808
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.4886553853834245e-05,
|
|
"loss": 3.8379,
|
|
"step": 312320
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.4878167906323725e-05,
|
|
"loss": 3.8383,
|
|
"step": 312832
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.4869781958813205e-05,
|
|
"loss": 3.8259,
|
|
"step": 313344
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.4861412390106414e-05,
|
|
"loss": 3.8361,
|
|
"step": 313856
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.4853026442595894e-05,
|
|
"loss": 3.8371,
|
|
"step": 314368
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.4844640495085374e-05,
|
|
"loss": 3.8334,
|
|
"step": 314880
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.4836254547574854e-05,
|
|
"loss": 3.8543,
|
|
"step": 315392
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.482788497886807e-05,
|
|
"loss": 3.8338,
|
|
"step": 315904
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.481949903135755e-05,
|
|
"loss": 3.8351,
|
|
"step": 316416
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.481111308384703e-05,
|
|
"loss": 3.8378,
|
|
"step": 316928
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.480272713633651e-05,
|
|
"loss": 3.8412,
|
|
"step": 317440
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.479435756762972e-05,
|
|
"loss": 3.8288,
|
|
"step": 317952
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.47859716201192e-05,
|
|
"loss": 3.8402,
|
|
"step": 318464
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.477758567260868e-05,
|
|
"loss": 3.8259,
|
|
"step": 318976
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.476919972509816e-05,
|
|
"loss": 3.8298,
|
|
"step": 319488
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.476083015639137e-05,
|
|
"loss": 3.8273,
|
|
"step": 320000
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.475244420888085e-05,
|
|
"loss": 3.8292,
|
|
"step": 320512
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.474405826137033e-05,
|
|
"loss": 3.8315,
|
|
"step": 321024
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.473567231385981e-05,
|
|
"loss": 3.8365,
|
|
"step": 321536
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4727302745153024e-05,
|
|
"loss": 3.8369,
|
|
"step": 322048
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4718916797642504e-05,
|
|
"loss": 3.8362,
|
|
"step": 322560
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4710530850131984e-05,
|
|
"loss": 3.8281,
|
|
"step": 323072
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4702144902621464e-05,
|
|
"loss": 3.8381,
|
|
"step": 323584
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.469377533391467e-05,
|
|
"loss": 3.8242,
|
|
"step": 324096
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.468538938640415e-05,
|
|
"loss": 3.8267,
|
|
"step": 324608
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.467700343889363e-05,
|
|
"loss": 3.8179,
|
|
"step": 325120
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.466861749138311e-05,
|
|
"loss": 3.8251,
|
|
"step": 325632
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.466024792267632e-05,
|
|
"loss": 3.8208,
|
|
"step": 326144
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.46518619751658e-05,
|
|
"loss": 3.8341,
|
|
"step": 326656
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.464347602765528e-05,
|
|
"loss": 3.8167,
|
|
"step": 327168
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.463509008014476e-05,
|
|
"loss": 3.8231,
|
|
"step": 327680
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.462670413263425e-05,
|
|
"loss": 3.8397,
|
|
"step": 328192
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.461833456392746e-05,
|
|
"loss": 3.8293,
|
|
"step": 328704
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.460994861641694e-05,
|
|
"loss": 3.8285,
|
|
"step": 329216
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.460156266890642e-05,
|
|
"loss": 3.8197,
|
|
"step": 329728
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.45931767213959e-05,
|
|
"loss": 3.8095,
|
|
"step": 330240
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4584807152689106e-05,
|
|
"loss": 3.8317,
|
|
"step": 330752
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4576421205178586e-05,
|
|
"loss": 3.8243,
|
|
"step": 331264
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4568035257668066e-05,
|
|
"loss": 3.819,
|
|
"step": 331776
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4559649310157546e-05,
|
|
"loss": 3.8164,
|
|
"step": 332288
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4551279741450755e-05,
|
|
"loss": 3.8259,
|
|
"step": 332800
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4542893793940235e-05,
|
|
"loss": 3.8029,
|
|
"step": 333312
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4534507846429715e-05,
|
|
"loss": 3.8282,
|
|
"step": 333824
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.45261218989192e-05,
|
|
"loss": 3.8056,
|
|
"step": 334336
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.451775233021241e-05,
|
|
"loss": 3.8206,
|
|
"step": 334848
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.450936638270189e-05,
|
|
"loss": 3.8253,
|
|
"step": 335360
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.450098043519137e-05,
|
|
"loss": 3.8134,
|
|
"step": 335872
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.449259448768085e-05,
|
|
"loss": 3.8176,
|
|
"step": 336384
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.448420854017033e-05,
|
|
"loss": 3.8168,
|
|
"step": 336896
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.447583897146354e-05,
|
|
"loss": 3.8084,
|
|
"step": 337408
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.446745302395302e-05,
|
|
"loss": 3.815,
|
|
"step": 337920
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.44590670764425e-05,
|
|
"loss": 3.8175,
|
|
"step": 338432
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.445068112893198e-05,
|
|
"loss": 3.812,
|
|
"step": 338944
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.444231156022519e-05,
|
|
"loss": 3.8119,
|
|
"step": 339456
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.443392561271467e-05,
|
|
"loss": 3.8028,
|
|
"step": 339968
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4425539665204156e-05,
|
|
"loss": 3.8024,
|
|
"step": 340480
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4417153717693636e-05,
|
|
"loss": 3.816,
|
|
"step": 340992
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4408784148986845e-05,
|
|
"loss": 3.8252,
|
|
"step": 341504
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4400398201476325e-05,
|
|
"loss": 3.808,
|
|
"step": 342016
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4392012253965805e-05,
|
|
"loss": 3.8275,
|
|
"step": 342528
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4383626306455285e-05,
|
|
"loss": 3.8235,
|
|
"step": 343040
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4375240358944764e-05,
|
|
"loss": 3.8112,
|
|
"step": 343552
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4366870790237974e-05,
|
|
"loss": 3.8146,
|
|
"step": 344064
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4358484842727454e-05,
|
|
"loss": 3.8205,
|
|
"step": 344576
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4350098895216933e-05,
|
|
"loss": 3.8117,
|
|
"step": 345088
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.4341712947706413e-05,
|
|
"loss": 3.8096,
|
|
"step": 345600
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.433334337899962e-05,
|
|
"loss": 3.8166,
|
|
"step": 346112
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.43249574314891e-05,
|
|
"loss": 3.8161,
|
|
"step": 346624
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.431657148397859e-05,
|
|
"loss": 3.8193,
|
|
"step": 347136
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.430818553646807e-05,
|
|
"loss": 3.8107,
|
|
"step": 347648
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.429981596776128e-05,
|
|
"loss": 3.7951,
|
|
"step": 348160
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.429143002025076e-05,
|
|
"loss": 3.7979,
|
|
"step": 348672
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.428304407274024e-05,
|
|
"loss": 3.8096,
|
|
"step": 349184
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.427465812522971e-05,
|
|
"loss": 3.8106,
|
|
"step": 349696
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.426628855652293e-05,
|
|
"loss": 3.8077,
|
|
"step": 350208
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.425790260901241e-05,
|
|
"loss": 3.8102,
|
|
"step": 350720
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.424951666150189e-05,
|
|
"loss": 3.8051,
|
|
"step": 351232
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.424113071399136e-05,
|
|
"loss": 3.802,
|
|
"step": 351744
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.4232761145284576e-05,
|
|
"loss": 3.8071,
|
|
"step": 352256
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.4224375197774056e-05,
|
|
"loss": 3.8079,
|
|
"step": 352768
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.421598925026354e-05,
|
|
"loss": 3.8154,
|
|
"step": 353280
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.4207603302753016e-05,
|
|
"loss": 3.8166,
|
|
"step": 353792
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.419923373404623e-05,
|
|
"loss": 3.8104,
|
|
"step": 354304
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.419084778653571e-05,
|
|
"loss": 3.7991,
|
|
"step": 354816
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.4182461839025185e-05,
|
|
"loss": 3.8107,
|
|
"step": 355328
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.4174075891514665e-05,
|
|
"loss": 3.8024,
|
|
"step": 355840
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.416570632280788e-05,
|
|
"loss": 3.8161,
|
|
"step": 356352
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.415732037529736e-05,
|
|
"loss": 3.803,
|
|
"step": 356864
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.4148934427786834e-05,
|
|
"loss": 3.7983,
|
|
"step": 357376
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.4140548480276314e-05,
|
|
"loss": 3.8141,
|
|
"step": 357888
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.413217891156953e-05,
|
|
"loss": 3.8097,
|
|
"step": 358400
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.412379296405901e-05,
|
|
"loss": 3.8001,
|
|
"step": 358912
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.411540701654849e-05,
|
|
"loss": 3.8043,
|
|
"step": 359424
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.410702106903797e-05,
|
|
"loss": 3.8048,
|
|
"step": 359936
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.4098651500331186e-05,
|
|
"loss": 3.7885,
|
|
"step": 360448
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.409026555282066e-05,
|
|
"loss": 3.7919,
|
|
"step": 360960
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.408187960531014e-05,
|
|
"loss": 3.8042,
|
|
"step": 361472
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.407349365779962e-05,
|
|
"loss": 3.7999,
|
|
"step": 361984
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.4065124089092835e-05,
|
|
"loss": 3.8018,
|
|
"step": 362496
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.405673814158231e-05,
|
|
"loss": 3.7961,
|
|
"step": 363008
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.404835219407179e-05,
|
|
"loss": 3.7967,
|
|
"step": 363520
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.403996624656127e-05,
|
|
"loss": 3.8159,
|
|
"step": 364032
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.4031596677854484e-05,
|
|
"loss": 3.8152,
|
|
"step": 364544
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.4023210730343963e-05,
|
|
"loss": 3.8079,
|
|
"step": 365056
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.4014824782833443e-05,
|
|
"loss": 3.8029,
|
|
"step": 365568
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.4006438835322923e-05,
|
|
"loss": 3.796,
|
|
"step": 366080
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.399806926661613e-05,
|
|
"loss": 3.799,
|
|
"step": 366592
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.398968331910561e-05,
|
|
"loss": 3.7937,
|
|
"step": 367104
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.398129737159509e-05,
|
|
"loss": 3.7995,
|
|
"step": 367616
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.397291142408457e-05,
|
|
"loss": 3.7963,
|
|
"step": 368128
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.396454185537778e-05,
|
|
"loss": 3.8055,
|
|
"step": 368640
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.395615590786726e-05,
|
|
"loss": 3.7993,
|
|
"step": 369152
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.394776996035674e-05,
|
|
"loss": 3.8023,
|
|
"step": 369664
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.393938401284622e-05,
|
|
"loss": 3.7984,
|
|
"step": 370176
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.39309980653357e-05,
|
|
"loss": 3.7931,
|
|
"step": 370688
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.392262849662892e-05,
|
|
"loss": 3.7927,
|
|
"step": 371200
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.39142425491184e-05,
|
|
"loss": 3.8059,
|
|
"step": 371712
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.390585660160788e-05,
|
|
"loss": 3.8094,
|
|
"step": 372224
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.389747065409736e-05,
|
|
"loss": 3.8024,
|
|
"step": 372736
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.3889101085390566e-05,
|
|
"loss": 3.7935,
|
|
"step": 373248
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.3880715137880046e-05,
|
|
"loss": 3.8035,
|
|
"step": 373760
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.3872329190369526e-05,
|
|
"loss": 3.7955,
|
|
"step": 374272
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.3863943242859006e-05,
|
|
"loss": 3.8105,
|
|
"step": 374784
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.3855573674152215e-05,
|
|
"loss": 3.7968,
|
|
"step": 375296
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.3847187726641695e-05,
|
|
"loss": 3.7977,
|
|
"step": 375808
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.3838801779131175e-05,
|
|
"loss": 3.7959,
|
|
"step": 376320
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.3830415831620655e-05,
|
|
"loss": 3.8041,
|
|
"step": 376832
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.382204626291387e-05,
|
|
"loss": 3.7876,
|
|
"step": 377344
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.381366031540335e-05,
|
|
"loss": 3.7929,
|
|
"step": 377856
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.380527436789283e-05,
|
|
"loss": 3.7995,
|
|
"step": 378368
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.379688842038231e-05,
|
|
"loss": 3.7924,
|
|
"step": 378880
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.378851885167552e-05,
|
|
"loss": 3.8002,
|
|
"step": 379392
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.3780132904165e-05,
|
|
"loss": 3.8013,
|
|
"step": 379904
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.377174695665448e-05,
|
|
"loss": 3.7984,
|
|
"step": 380416
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.376336100914396e-05,
|
|
"loss": 3.7933,
|
|
"step": 380928
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.375499144043717e-05,
|
|
"loss": 3.7937,
|
|
"step": 381440
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.8858225345611572,
|
|
"eval_runtime": 335.7172,
|
|
"eval_samples_per_second": 1136.644,
|
|
"eval_steps_per_second": 35.521,
|
|
"step": 381600
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.374660549292665e-05,
|
|
"loss": 3.7837,
|
|
"step": 381952
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.373821954541613e-05,
|
|
"loss": 3.7775,
|
|
"step": 382464
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.372983359790561e-05,
|
|
"loss": 3.8026,
|
|
"step": 382976
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.3721447650395095e-05,
|
|
"loss": 3.795,
|
|
"step": 383488
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.3713061702884575e-05,
|
|
"loss": 3.8023,
|
|
"step": 384000
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.3704675755374055e-05,
|
|
"loss": 3.7857,
|
|
"step": 384512
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.3696289807863535e-05,
|
|
"loss": 3.7887,
|
|
"step": 385024
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.3687903860353015e-05,
|
|
"loss": 3.7811,
|
|
"step": 385536
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.3679534291646224e-05,
|
|
"loss": 3.79,
|
|
"step": 386048
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.3671148344135704e-05,
|
|
"loss": 3.7867,
|
|
"step": 386560
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.3662762396625184e-05,
|
|
"loss": 3.7974,
|
|
"step": 387072
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.3654376449114664e-05,
|
|
"loss": 3.7977,
|
|
"step": 387584
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.364600688040787e-05,
|
|
"loss": 3.7709,
|
|
"step": 388096
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.363762093289735e-05,
|
|
"loss": 3.7877,
|
|
"step": 388608
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.362923498538683e-05,
|
|
"loss": 3.7878,
|
|
"step": 389120
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.362084903787631e-05,
|
|
"loss": 3.7752,
|
|
"step": 389632
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.361246309036579e-05,
|
|
"loss": 3.7824,
|
|
"step": 390144
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.360407714285527e-05,
|
|
"loss": 3.7844,
|
|
"step": 390656
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.359569119534475e-05,
|
|
"loss": 3.783,
|
|
"step": 391168
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.358730524783423e-05,
|
|
"loss": 3.7977,
|
|
"step": 391680
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.357893567912744e-05,
|
|
"loss": 3.783,
|
|
"step": 392192
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.357054973161692e-05,
|
|
"loss": 3.7887,
|
|
"step": 392704
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.35621637841064e-05,
|
|
"loss": 3.7829,
|
|
"step": 393216
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.355377783659588e-05,
|
|
"loss": 3.7872,
|
|
"step": 393728
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.354540826788909e-05,
|
|
"loss": 3.7771,
|
|
"step": 394240
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.353702232037857e-05,
|
|
"loss": 3.7848,
|
|
"step": 394752
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.352863637286805e-05,
|
|
"loss": 3.7808,
|
|
"step": 395264
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.352025042535753e-05,
|
|
"loss": 3.7765,
|
|
"step": 395776
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.351188085665075e-05,
|
|
"loss": 3.7773,
|
|
"step": 396288
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.350349490914023e-05,
|
|
"loss": 3.7777,
|
|
"step": 396800
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.349510896162971e-05,
|
|
"loss": 3.7801,
|
|
"step": 397312
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3486723014119187e-05,
|
|
"loss": 3.7829,
|
|
"step": 397824
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3478353445412396e-05,
|
|
"loss": 3.7883,
|
|
"step": 398336
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3469967497901876e-05,
|
|
"loss": 3.7862,
|
|
"step": 398848
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3461581550391356e-05,
|
|
"loss": 3.7782,
|
|
"step": 399360
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3453195602880836e-05,
|
|
"loss": 3.7851,
|
|
"step": 399872
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3444826034174045e-05,
|
|
"loss": 3.7766,
|
|
"step": 400384
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3436440086663525e-05,
|
|
"loss": 3.7783,
|
|
"step": 400896
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3428054139153005e-05,
|
|
"loss": 3.7662,
|
|
"step": 401408
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3419668191642485e-05,
|
|
"loss": 3.773,
|
|
"step": 401920
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.34112986229357e-05,
|
|
"loss": 3.7749,
|
|
"step": 402432
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.340291267542518e-05,
|
|
"loss": 3.7817,
|
|
"step": 402944
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.339452672791466e-05,
|
|
"loss": 3.7709,
|
|
"step": 403456
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.338614078040414e-05,
|
|
"loss": 3.7731,
|
|
"step": 403968
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.337777121169735e-05,
|
|
"loss": 3.7885,
|
|
"step": 404480
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.336938526418683e-05,
|
|
"loss": 3.7794,
|
|
"step": 404992
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.336099931667631e-05,
|
|
"loss": 3.7801,
|
|
"step": 405504
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.335261336916579e-05,
|
|
"loss": 3.7703,
|
|
"step": 406016
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3344243800459e-05,
|
|
"loss": 3.759,
|
|
"step": 406528
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.333585785294848e-05,
|
|
"loss": 3.7812,
|
|
"step": 407040
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.332747190543796e-05,
|
|
"loss": 3.7767,
|
|
"step": 407552
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.331908595792744e-05,
|
|
"loss": 3.7702,
|
|
"step": 408064
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3310716389220654e-05,
|
|
"loss": 3.7703,
|
|
"step": 408576
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3302330441710134e-05,
|
|
"loss": 3.7787,
|
|
"step": 409088
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3293944494199614e-05,
|
|
"loss": 3.7499,
|
|
"step": 409600
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3285558546689094e-05,
|
|
"loss": 3.7767,
|
|
"step": 410112
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.32771889779823e-05,
|
|
"loss": 3.7621,
|
|
"step": 410624
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.326880303047178e-05,
|
|
"loss": 3.7706,
|
|
"step": 411136
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.326041708296126e-05,
|
|
"loss": 3.7786,
|
|
"step": 411648
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.325203113545074e-05,
|
|
"loss": 3.7674,
|
|
"step": 412160
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.324366156674395e-05,
|
|
"loss": 3.766,
|
|
"step": 412672
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.323527561923343e-05,
|
|
"loss": 3.7745,
|
|
"step": 413184
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.322688967172291e-05,
|
|
"loss": 3.7622,
|
|
"step": 413696
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.321850372421239e-05,
|
|
"loss": 3.7619,
|
|
"step": 414208
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.321013415550561e-05,
|
|
"loss": 3.7716,
|
|
"step": 414720
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.320174820799509e-05,
|
|
"loss": 3.7675,
|
|
"step": 415232
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.319336226048457e-05,
|
|
"loss": 3.7604,
|
|
"step": 415744
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.318497631297405e-05,
|
|
"loss": 3.7566,
|
|
"step": 416256
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.317660674426726e-05,
|
|
"loss": 3.7554,
|
|
"step": 416768
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.316822079675674e-05,
|
|
"loss": 3.7663,
|
|
"step": 417280
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3159834849246217e-05,
|
|
"loss": 3.7754,
|
|
"step": 417792
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3151448901735697e-05,
|
|
"loss": 3.7615,
|
|
"step": 418304
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3143079333028906e-05,
|
|
"loss": 3.7809,
|
|
"step": 418816
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3134693385518386e-05,
|
|
"loss": 3.776,
|
|
"step": 419328
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3126307438007866e-05,
|
|
"loss": 3.7674,
|
|
"step": 419840
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3117921490497346e-05,
|
|
"loss": 3.7603,
|
|
"step": 420352
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.310955192179056e-05,
|
|
"loss": 3.7742,
|
|
"step": 420864
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.310116597428004e-05,
|
|
"loss": 3.7667,
|
|
"step": 421376
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.309278002676952e-05,
|
|
"loss": 3.7604,
|
|
"step": 421888
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.308441045806273e-05,
|
|
"loss": 3.7735,
|
|
"step": 422400
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.307602451055221e-05,
|
|
"loss": 3.7672,
|
|
"step": 422912
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.306763856304169e-05,
|
|
"loss": 3.7737,
|
|
"step": 423424
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.305925261553117e-05,
|
|
"loss": 3.765,
|
|
"step": 423936
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.305088304682438e-05,
|
|
"loss": 3.752,
|
|
"step": 424448
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.304249709931386e-05,
|
|
"loss": 3.7496,
|
|
"step": 424960
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.303411115180334e-05,
|
|
"loss": 3.7608,
|
|
"step": 425472
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.302572520429282e-05,
|
|
"loss": 3.7653,
|
|
"step": 425984
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3017355635586035e-05,
|
|
"loss": 3.7609,
|
|
"step": 426496
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.3008969688075515e-05,
|
|
"loss": 3.7614,
|
|
"step": 427008
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.3000583740564995e-05,
|
|
"loss": 3.7608,
|
|
"step": 427520
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2992197793054475e-05,
|
|
"loss": 3.7524,
|
|
"step": 428032
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2983828224347684e-05,
|
|
"loss": 3.7635,
|
|
"step": 428544
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2975442276837164e-05,
|
|
"loss": 3.7571,
|
|
"step": 429056
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2967056329326644e-05,
|
|
"loss": 3.772,
|
|
"step": 429568
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2958670381816124e-05,
|
|
"loss": 3.7726,
|
|
"step": 430080
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.295030081310933e-05,
|
|
"loss": 3.7637,
|
|
"step": 430592
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.294191486559881e-05,
|
|
"loss": 3.7557,
|
|
"step": 431104
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.293352891808829e-05,
|
|
"loss": 3.7603,
|
|
"step": 431616
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.292514297057777e-05,
|
|
"loss": 3.7576,
|
|
"step": 432128
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.291677340187099e-05,
|
|
"loss": 3.7726,
|
|
"step": 432640
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.290838745436047e-05,
|
|
"loss": 3.7556,
|
|
"step": 433152
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.290000150684995e-05,
|
|
"loss": 3.7532,
|
|
"step": 433664
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.289161555933943e-05,
|
|
"loss": 3.767,
|
|
"step": 434176
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.288324599063264e-05,
|
|
"loss": 3.7625,
|
|
"step": 434688
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.287486004312212e-05,
|
|
"loss": 3.7539,
|
|
"step": 435200
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.28664740956116e-05,
|
|
"loss": 3.7599,
|
|
"step": 435712
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.285808814810108e-05,
|
|
"loss": 3.7568,
|
|
"step": 436224
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.284971857939429e-05,
|
|
"loss": 3.7529,
|
|
"step": 436736
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.284133263188377e-05,
|
|
"loss": 3.7402,
|
|
"step": 437248
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2832946684373247e-05,
|
|
"loss": 3.7591,
|
|
"step": 437760
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2824560736862727e-05,
|
|
"loss": 3.7574,
|
|
"step": 438272
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.281619116815594e-05,
|
|
"loss": 3.7545,
|
|
"step": 438784
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.280780522064542e-05,
|
|
"loss": 3.7542,
|
|
"step": 439296
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.27994192731349e-05,
|
|
"loss": 3.7494,
|
|
"step": 439808
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.279103332562438e-05,
|
|
"loss": 3.7699,
|
|
"step": 440320
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.278266375691759e-05,
|
|
"loss": 3.7698,
|
|
"step": 440832
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.277427780940707e-05,
|
|
"loss": 3.7639,
|
|
"step": 441344
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.276589186189655e-05,
|
|
"loss": 3.761,
|
|
"step": 441856
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.275750591438603e-05,
|
|
"loss": 3.7524,
|
|
"step": 442368
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.274913634567924e-05,
|
|
"loss": 3.7527,
|
|
"step": 442880
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.274075039816872e-05,
|
|
"loss": 3.7487,
|
|
"step": 443392
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.27323644506582e-05,
|
|
"loss": 3.7528,
|
|
"step": 443904
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.272397850314768e-05,
|
|
"loss": 3.7556,
|
|
"step": 444416
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2715608934440896e-05,
|
|
"loss": 3.7577,
|
|
"step": 444928
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2707222986930376e-05,
|
|
"loss": 3.7584,
|
|
"step": 445440
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2698837039419856e-05,
|
|
"loss": 3.7535,
|
|
"step": 445952
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2690451091909336e-05,
|
|
"loss": 3.7565,
|
|
"step": 446464
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2682081523202545e-05,
|
|
"loss": 3.7516,
|
|
"step": 446976
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2673695575692025e-05,
|
|
"loss": 3.7515,
|
|
"step": 447488
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2665309628181505e-05,
|
|
"loss": 3.7557,
|
|
"step": 448000
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2656923680670985e-05,
|
|
"loss": 3.7635,
|
|
"step": 448512
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2648554111964194e-05,
|
|
"loss": 3.7639,
|
|
"step": 449024
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2640168164453674e-05,
|
|
"loss": 3.7474,
|
|
"step": 449536
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2631782216943154e-05,
|
|
"loss": 3.7611,
|
|
"step": 450048
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.2623396269432634e-05,
|
|
"loss": 3.7544,
|
|
"step": 450560
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.261502670072584e-05,
|
|
"loss": 3.7629,
|
|
"step": 451072
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.260664075321533e-05,
|
|
"loss": 3.7536,
|
|
"step": 451584
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.259825480570481e-05,
|
|
"loss": 3.7577,
|
|
"step": 452096
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.258986885819429e-05,
|
|
"loss": 3.753,
|
|
"step": 452608
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.25814992894875e-05,
|
|
"loss": 3.7587,
|
|
"step": 453120
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.257311334197698e-05,
|
|
"loss": 3.7454,
|
|
"step": 453632
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.256472739446646e-05,
|
|
"loss": 3.7514,
|
|
"step": 454144
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.255634144695594e-05,
|
|
"loss": 3.753,
|
|
"step": 454656
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.254797187824915e-05,
|
|
"loss": 3.7551,
|
|
"step": 455168
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.253958593073863e-05,
|
|
"loss": 3.7563,
|
|
"step": 455680
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.253119998322811e-05,
|
|
"loss": 3.7515,
|
|
"step": 456192
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.252281403571758e-05,
|
|
"loss": 3.763,
|
|
"step": 456704
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.25144444670108e-05,
|
|
"loss": 3.7472,
|
|
"step": 457216
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.250605851950028e-05,
|
|
"loss": 3.754,
|
|
"step": 457728
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.8698978424072266,
|
|
"eval_runtime": 303.666,
|
|
"eval_samples_per_second": 1256.614,
|
|
"eval_steps_per_second": 39.27,
|
|
"step": 457920
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.249767257198976e-05,
|
|
"loss": 3.7504,
|
|
"step": 458240
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.2489286624479237e-05,
|
|
"loss": 3.7343,
|
|
"step": 458752
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.248091705577245e-05,
|
|
"loss": 3.7598,
|
|
"step": 459264
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.247253110826193e-05,
|
|
"loss": 3.753,
|
|
"step": 459776
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.246414516075141e-05,
|
|
"loss": 3.7619,
|
|
"step": 460288
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.2455759213240885e-05,
|
|
"loss": 3.7449,
|
|
"step": 460800
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.24473896445341e-05,
|
|
"loss": 3.7421,
|
|
"step": 461312
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.243900369702358e-05,
|
|
"loss": 3.7441,
|
|
"step": 461824
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.2430617749513054e-05,
|
|
"loss": 3.7453,
|
|
"step": 462336
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.2422231802002534e-05,
|
|
"loss": 3.7456,
|
|
"step": 462848
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.241386223329575e-05,
|
|
"loss": 3.7548,
|
|
"step": 463360
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.240547628578524e-05,
|
|
"loss": 3.7579,
|
|
"step": 463872
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.2397106717078446e-05,
|
|
"loss": 3.7322,
|
|
"step": 464384
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.2388720769567926e-05,
|
|
"loss": 3.7423,
|
|
"step": 464896
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.2380334822057406e-05,
|
|
"loss": 3.7468,
|
|
"step": 465408
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.2371948874546886e-05,
|
|
"loss": 3.7336,
|
|
"step": 465920
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.236356292703636e-05,
|
|
"loss": 3.7403,
|
|
"step": 466432
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.235517697952584e-05,
|
|
"loss": 3.7416,
|
|
"step": 466944
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.234679103201532e-05,
|
|
"loss": 3.7418,
|
|
"step": 467456
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.23384050845048e-05,
|
|
"loss": 3.7568,
|
|
"step": 467968
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.233003551579801e-05,
|
|
"loss": 3.7473,
|
|
"step": 468480
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.232164956828749e-05,
|
|
"loss": 3.7453,
|
|
"step": 468992
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.2313263620776975e-05,
|
|
"loss": 3.7422,
|
|
"step": 469504
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.2304877673266455e-05,
|
|
"loss": 3.7477,
|
|
"step": 470016
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.2296508104559664e-05,
|
|
"loss": 3.7395,
|
|
"step": 470528
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.2288122157049144e-05,
|
|
"loss": 3.7444,
|
|
"step": 471040
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.2279736209538624e-05,
|
|
"loss": 3.7366,
|
|
"step": 471552
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.2271350262028104e-05,
|
|
"loss": 3.7364,
|
|
"step": 472064
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.226298069332131e-05,
|
|
"loss": 3.7388,
|
|
"step": 472576
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.225459474581079e-05,
|
|
"loss": 3.7372,
|
|
"step": 473088
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.224620879830027e-05,
|
|
"loss": 3.7407,
|
|
"step": 473600
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.223783922959348e-05,
|
|
"loss": 3.7423,
|
|
"step": 474112
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.222945328208296e-05,
|
|
"loss": 3.7485,
|
|
"step": 474624
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.222106733457244e-05,
|
|
"loss": 3.7474,
|
|
"step": 475136
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.221268138706193e-05,
|
|
"loss": 3.7366,
|
|
"step": 475648
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.220431181835514e-05,
|
|
"loss": 3.7442,
|
|
"step": 476160
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.219592587084462e-05,
|
|
"loss": 3.7408,
|
|
"step": 476672
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.21875399233341e-05,
|
|
"loss": 3.7367,
|
|
"step": 477184
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.217915397582358e-05,
|
|
"loss": 3.7241,
|
|
"step": 477696
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.2170784407116787e-05,
|
|
"loss": 3.7335,
|
|
"step": 478208
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.2162398459606267e-05,
|
|
"loss": 3.7349,
|
|
"step": 478720
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.2154012512095746e-05,
|
|
"loss": 3.7393,
|
|
"step": 479232
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.2145626564585226e-05,
|
|
"loss": 3.7318,
|
|
"step": 479744
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.2137256995878436e-05,
|
|
"loss": 3.732,
|
|
"step": 480256
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.2128871048367915e-05,
|
|
"loss": 3.7531,
|
|
"step": 480768
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.2120485100857395e-05,
|
|
"loss": 3.7367,
|
|
"step": 481280
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.211209915334688e-05,
|
|
"loss": 3.741,
|
|
"step": 481792
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.210372958464009e-05,
|
|
"loss": 3.7258,
|
|
"step": 482304
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.209534363712957e-05,
|
|
"loss": 3.7199,
|
|
"step": 482816
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.208695768961905e-05,
|
|
"loss": 3.7451,
|
|
"step": 483328
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.207857174210853e-05,
|
|
"loss": 3.7333,
|
|
"step": 483840
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.207020217340174e-05,
|
|
"loss": 3.7318,
|
|
"step": 484352
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.206181622589122e-05,
|
|
"loss": 3.7297,
|
|
"step": 484864
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.20534302783807e-05,
|
|
"loss": 3.7362,
|
|
"step": 485376
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.204504433087018e-05,
|
|
"loss": 3.7127,
|
|
"step": 485888
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.203667476216339e-05,
|
|
"loss": 3.7348,
|
|
"step": 486400
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.202828881465287e-05,
|
|
"loss": 3.7262,
|
|
"step": 486912
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.201990286714235e-05,
|
|
"loss": 3.724,
|
|
"step": 487424
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.2011516919631836e-05,
|
|
"loss": 3.7438,
|
|
"step": 487936
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.2003147350925045e-05,
|
|
"loss": 3.7296,
|
|
"step": 488448
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.1994761403414525e-05,
|
|
"loss": 3.7264,
|
|
"step": 488960
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.1986375455904005e-05,
|
|
"loss": 3.7361,
|
|
"step": 489472
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.1977989508393485e-05,
|
|
"loss": 3.7223,
|
|
"step": 489984
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.1969619939686694e-05,
|
|
"loss": 3.7244,
|
|
"step": 490496
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.1961233992176174e-05,
|
|
"loss": 3.7308,
|
|
"step": 491008
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.1952848044665654e-05,
|
|
"loss": 3.7273,
|
|
"step": 491520
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.1944462097155134e-05,
|
|
"loss": 3.7277,
|
|
"step": 492032
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.193609252844834e-05,
|
|
"loss": 3.7207,
|
|
"step": 492544
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.192770658093782e-05,
|
|
"loss": 3.7119,
|
|
"step": 493056
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.19193206334273e-05,
|
|
"loss": 3.7249,
|
|
"step": 493568
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.191093468591679e-05,
|
|
"loss": 3.7368,
|
|
"step": 494080
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.190256511721e-05,
|
|
"loss": 3.7213,
|
|
"step": 494592
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.189417916969948e-05,
|
|
"loss": 3.7443,
|
|
"step": 495104
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.188579322218896e-05,
|
|
"loss": 3.7366,
|
|
"step": 495616
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.187740727467844e-05,
|
|
"loss": 3.7349,
|
|
"step": 496128
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.186903770597165e-05,
|
|
"loss": 3.7197,
|
|
"step": 496640
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.186065175846113e-05,
|
|
"loss": 3.7377,
|
|
"step": 497152
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.185226581095061e-05,
|
|
"loss": 3.7288,
|
|
"step": 497664
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.184387986344009e-05,
|
|
"loss": 3.7204,
|
|
"step": 498176
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.1835526673537026e-05,
|
|
"loss": 3.737,
|
|
"step": 498688
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.182714072602651e-05,
|
|
"loss": 3.7286,
|
|
"step": 499200
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.181875477851599e-05,
|
|
"loss": 3.7356,
|
|
"step": 499712
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.181036883100547e-05,
|
|
"loss": 3.7277,
|
|
"step": 500224
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.180198288349495e-05,
|
|
"loss": 3.7117,
|
|
"step": 500736
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.179359693598443e-05,
|
|
"loss": 3.7109,
|
|
"step": 501248
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.178522736727764e-05,
|
|
"loss": 3.719,
|
|
"step": 501760
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.177684141976712e-05,
|
|
"loss": 3.7281,
|
|
"step": 502272
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.17684554722566e-05,
|
|
"loss": 3.725,
|
|
"step": 502784
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.176006952474608e-05,
|
|
"loss": 3.722,
|
|
"step": 503296
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.175169995603929e-05,
|
|
"loss": 3.7255,
|
|
"step": 503808
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.174331400852877e-05,
|
|
"loss": 3.716,
|
|
"step": 504320
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.173492806101825e-05,
|
|
"loss": 3.7231,
|
|
"step": 504832
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.172654211350773e-05,
|
|
"loss": 3.718,
|
|
"step": 505344
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.171815616599721e-05,
|
|
"loss": 3.7394,
|
|
"step": 505856
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.170977021848669e-05,
|
|
"loss": 3.7373,
|
|
"step": 506368
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.170138427097618e-05,
|
|
"loss": 3.7257,
|
|
"step": 506880
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.169299832346566e-05,
|
|
"loss": 3.7186,
|
|
"step": 507392
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1684628754758866e-05,
|
|
"loss": 3.7263,
|
|
"step": 507904
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1676242807248346e-05,
|
|
"loss": 3.7162,
|
|
"step": 508416
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1667856859737826e-05,
|
|
"loss": 3.7376,
|
|
"step": 508928
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1659470912227306e-05,
|
|
"loss": 3.7184,
|
|
"step": 509440
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1651101343520515e-05,
|
|
"loss": 3.7193,
|
|
"step": 509952
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1642715396009995e-05,
|
|
"loss": 3.7278,
|
|
"step": 510464
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1634329448499475e-05,
|
|
"loss": 3.7289,
|
|
"step": 510976
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1625943500988955e-05,
|
|
"loss": 3.7151,
|
|
"step": 511488
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1617573932282164e-05,
|
|
"loss": 3.724,
|
|
"step": 512000
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1609187984771644e-05,
|
|
"loss": 3.7162,
|
|
"step": 512512
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.160080203726113e-05,
|
|
"loss": 3.7139,
|
|
"step": 513024
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.159241608975061e-05,
|
|
"loss": 3.7047,
|
|
"step": 513536
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.158404652104382e-05,
|
|
"loss": 3.7216,
|
|
"step": 514048
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.15756605735333e-05,
|
|
"loss": 3.7233,
|
|
"step": 514560
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.156727462602278e-05,
|
|
"loss": 3.7162,
|
|
"step": 515072
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.155888867851226e-05,
|
|
"loss": 3.7184,
|
|
"step": 515584
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.155051910980547e-05,
|
|
"loss": 3.7162,
|
|
"step": 516096
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.154213316229495e-05,
|
|
"loss": 3.7303,
|
|
"step": 516608
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.153374721478443e-05,
|
|
"loss": 3.7357,
|
|
"step": 517120
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.152537764607764e-05,
|
|
"loss": 3.7293,
|
|
"step": 517632
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.151699169856712e-05,
|
|
"loss": 3.7228,
|
|
"step": 518144
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.15086057510566e-05,
|
|
"loss": 3.7137,
|
|
"step": 518656
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1500219803546084e-05,
|
|
"loss": 3.7186,
|
|
"step": 519168
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.149183385603556e-05,
|
|
"loss": 3.7084,
|
|
"step": 519680
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.148344790852504e-05,
|
|
"loss": 3.7169,
|
|
"step": 520192
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.147506196101452e-05,
|
|
"loss": 3.72,
|
|
"step": 520704
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1466676013504e-05,
|
|
"loss": 3.7192,
|
|
"step": 521216
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1458306444797206e-05,
|
|
"loss": 3.7235,
|
|
"step": 521728
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1449920497286686e-05,
|
|
"loss": 3.7168,
|
|
"step": 522240
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1441534549776166e-05,
|
|
"loss": 3.7263,
|
|
"step": 522752
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1433148602265646e-05,
|
|
"loss": 3.7144,
|
|
"step": 523264
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1424779033558855e-05,
|
|
"loss": 3.7185,
|
|
"step": 523776
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1416393086048335e-05,
|
|
"loss": 3.721,
|
|
"step": 524288
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.140800713853782e-05,
|
|
"loss": 3.7227,
|
|
"step": 524800
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.13996211910273e-05,
|
|
"loss": 3.7306,
|
|
"step": 525312
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.139125162232051e-05,
|
|
"loss": 3.7117,
|
|
"step": 525824
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.138286567480999e-05,
|
|
"loss": 3.7264,
|
|
"step": 526336
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.137447972729947e-05,
|
|
"loss": 3.7186,
|
|
"step": 526848
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.136611015859268e-05,
|
|
"loss": 3.7265,
|
|
"step": 527360
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.135772421108216e-05,
|
|
"loss": 3.717,
|
|
"step": 527872
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.134933826357164e-05,
|
|
"loss": 3.7238,
|
|
"step": 528384
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.134095231606112e-05,
|
|
"loss": 3.714,
|
|
"step": 528896
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.133258274735433e-05,
|
|
"loss": 3.7257,
|
|
"step": 529408
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.132419679984381e-05,
|
|
"loss": 3.708,
|
|
"step": 529920
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.131581085233329e-05,
|
|
"loss": 3.7156,
|
|
"step": 530432
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1307424904822776e-05,
|
|
"loss": 3.7169,
|
|
"step": 530944
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1299055336115985e-05,
|
|
"loss": 3.7217,
|
|
"step": 531456
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1290669388605465e-05,
|
|
"loss": 3.7164,
|
|
"step": 531968
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1282283441094945e-05,
|
|
"loss": 3.7192,
|
|
"step": 532480
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1273897493584425e-05,
|
|
"loss": 3.7256,
|
|
"step": 532992
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1265527924877634e-05,
|
|
"loss": 3.711,
|
|
"step": 533504
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.1257141977367114e-05,
|
|
"loss": 3.7241,
|
|
"step": 534016
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 3.8603413105010986,
|
|
"eval_runtime": 303.9807,
|
|
"eval_samples_per_second": 1255.313,
|
|
"eval_steps_per_second": 39.229,
|
|
"step": 534240
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.1248756029856594e-05,
|
|
"loss": 3.71,
|
|
"step": 534528
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.1240370082346074e-05,
|
|
"loss": 3.7017,
|
|
"step": 535040
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.1231984134835553e-05,
|
|
"loss": 3.7192,
|
|
"step": 535552
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.1223598187325033e-05,
|
|
"loss": 3.7179,
|
|
"step": 536064
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.1215212239814513e-05,
|
|
"loss": 3.7239,
|
|
"step": 536576
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.120682629230399e-05,
|
|
"loss": 3.7116,
|
|
"step": 537088
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.119844034479347e-05,
|
|
"loss": 3.7106,
|
|
"step": 537600
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.119005439728296e-05,
|
|
"loss": 3.7115,
|
|
"step": 538112
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.118168482857617e-05,
|
|
"loss": 3.7116,
|
|
"step": 538624
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.117329888106565e-05,
|
|
"loss": 3.7077,
|
|
"step": 539136
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.116492931235886e-05,
|
|
"loss": 3.7196,
|
|
"step": 539648
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.115654336484834e-05,
|
|
"loss": 3.7193,
|
|
"step": 540160
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.114815741733782e-05,
|
|
"loss": 3.7059,
|
|
"step": 540672
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.11397714698273e-05,
|
|
"loss": 3.7032,
|
|
"step": 541184
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.113138552231678e-05,
|
|
"loss": 3.7121,
|
|
"step": 541696
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.112299957480626e-05,
|
|
"loss": 3.6947,
|
|
"step": 542208
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.111461362729574e-05,
|
|
"loss": 3.708,
|
|
"step": 542720
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.110622767978521e-05,
|
|
"loss": 3.7055,
|
|
"step": 543232
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.109785811107843e-05,
|
|
"loss": 3.7114,
|
|
"step": 543744
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.108947216356791e-05,
|
|
"loss": 3.7179,
|
|
"step": 544256
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.1081086216057394e-05,
|
|
"loss": 3.7133,
|
|
"step": 544768
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.107270026854687e-05,
|
|
"loss": 3.7089,
|
|
"step": 545280
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.106433069984008e-05,
|
|
"loss": 3.7079,
|
|
"step": 545792
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.105594475232956e-05,
|
|
"loss": 3.7117,
|
|
"step": 546304
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.104755880481904e-05,
|
|
"loss": 3.704,
|
|
"step": 546816
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.1039172857308516e-05,
|
|
"loss": 3.7131,
|
|
"step": 547328
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.103080328860173e-05,
|
|
"loss": 3.7052,
|
|
"step": 547840
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.102241734109121e-05,
|
|
"loss": 3.6996,
|
|
"step": 548352
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.1014031393580685e-05,
|
|
"loss": 3.7025,
|
|
"step": 548864
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.1005645446070165e-05,
|
|
"loss": 3.7081,
|
|
"step": 549376
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.099727587736338e-05,
|
|
"loss": 3.7022,
|
|
"step": 549888
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.098888992985286e-05,
|
|
"loss": 3.7113,
|
|
"step": 550400
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.098050398234234e-05,
|
|
"loss": 3.7177,
|
|
"step": 550912
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0972134413635556e-05,
|
|
"loss": 3.7091,
|
|
"step": 551424
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0963748466125036e-05,
|
|
"loss": 3.6992,
|
|
"step": 551936
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0955362518614516e-05,
|
|
"loss": 3.7101,
|
|
"step": 552448
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.094697657110399e-05,
|
|
"loss": 3.7092,
|
|
"step": 552960
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.093859062359347e-05,
|
|
"loss": 3.7025,
|
|
"step": 553472
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.093020467608295e-05,
|
|
"loss": 3.6929,
|
|
"step": 553984
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.092181872857243e-05,
|
|
"loss": 3.6959,
|
|
"step": 554496
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.091344915986564e-05,
|
|
"loss": 3.7024,
|
|
"step": 555008
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.090506321235512e-05,
|
|
"loss": 3.7063,
|
|
"step": 555520
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.08966772648446e-05,
|
|
"loss": 3.6957,
|
|
"step": 556032
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0888291317334085e-05,
|
|
"loss": 3.7004,
|
|
"step": 556544
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0879905369823565e-05,
|
|
"loss": 3.7163,
|
|
"step": 557056
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0871519422313045e-05,
|
|
"loss": 3.7026,
|
|
"step": 557568
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0863133474802525e-05,
|
|
"loss": 3.7097,
|
|
"step": 558080
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0854763906095734e-05,
|
|
"loss": 3.6936,
|
|
"step": 558592
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0846377958585214e-05,
|
|
"loss": 3.6831,
|
|
"step": 559104
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0837992011074694e-05,
|
|
"loss": 3.7143,
|
|
"step": 559616
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.08296224423679e-05,
|
|
"loss": 3.6995,
|
|
"step": 560128
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.082123649485738e-05,
|
|
"loss": 3.6974,
|
|
"step": 560640
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.081285054734686e-05,
|
|
"loss": 3.7034,
|
|
"step": 561152
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.080446459983634e-05,
|
|
"loss": 3.696,
|
|
"step": 561664
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.079607865232582e-05,
|
|
"loss": 3.6876,
|
|
"step": 562176
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.078770908361904e-05,
|
|
"loss": 3.6954,
|
|
"step": 562688
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.077932313610852e-05,
|
|
"loss": 3.693,
|
|
"step": 563200
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0770937188598e-05,
|
|
"loss": 3.6953,
|
|
"step": 563712
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.076255124108748e-05,
|
|
"loss": 3.7069,
|
|
"step": 564224
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.075416529357696e-05,
|
|
"loss": 3.6982,
|
|
"step": 564736
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.074577934606644e-05,
|
|
"loss": 3.6919,
|
|
"step": 565248
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.073739339855592e-05,
|
|
"loss": 3.7052,
|
|
"step": 565760
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.07290074510454e-05,
|
|
"loss": 3.6874,
|
|
"step": 566272
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.072065426114234e-05,
|
|
"loss": 3.6928,
|
|
"step": 566784
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.071226831363182e-05,
|
|
"loss": 3.6969,
|
|
"step": 567296
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.07038823661213e-05,
|
|
"loss": 3.6966,
|
|
"step": 567808
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0695496418610777e-05,
|
|
"loss": 3.6912,
|
|
"step": 568320
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0687110471100257e-05,
|
|
"loss": 3.6916,
|
|
"step": 568832
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0678724523589736e-05,
|
|
"loss": 3.6758,
|
|
"step": 569344
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.067033857607922e-05,
|
|
"loss": 3.6929,
|
|
"step": 569856
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.06619526285687e-05,
|
|
"loss": 3.7051,
|
|
"step": 570368
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.065359943866564e-05,
|
|
"loss": 3.6883,
|
|
"step": 570880
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.064521349115512e-05,
|
|
"loss": 3.71,
|
|
"step": 571392
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.06368275436446e-05,
|
|
"loss": 3.7003,
|
|
"step": 571904
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.062844159613408e-05,
|
|
"loss": 3.7051,
|
|
"step": 572416
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.062007202742729e-05,
|
|
"loss": 3.6898,
|
|
"step": 572928
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.061168607991677e-05,
|
|
"loss": 3.7036,
|
|
"step": 573440
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.060330013240625e-05,
|
|
"loss": 3.6945,
|
|
"step": 573952
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.059491418489573e-05,
|
|
"loss": 3.6898,
|
|
"step": 574464
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0586544616188946e-05,
|
|
"loss": 3.7064,
|
|
"step": 574976
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0578158668678426e-05,
|
|
"loss": 3.6942,
|
|
"step": 575488
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0569772721167906e-05,
|
|
"loss": 3.7025,
|
|
"step": 576000
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0561386773657386e-05,
|
|
"loss": 3.6985,
|
|
"step": 576512
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0553000826146866e-05,
|
|
"loss": 3.6742,
|
|
"step": 577024
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0544614878636346e-05,
|
|
"loss": 3.6859,
|
|
"step": 577536
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0536245309929555e-05,
|
|
"loss": 3.6847,
|
|
"step": 578048
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0527859362419035e-05,
|
|
"loss": 3.6967,
|
|
"step": 578560
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0519473414908515e-05,
|
|
"loss": 3.6948,
|
|
"step": 579072
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.0511087467397995e-05,
|
|
"loss": 3.6907,
|
|
"step": 579584
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0502717898691204e-05,
|
|
"loss": 3.6919,
|
|
"step": 580096
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0494331951180684e-05,
|
|
"loss": 3.6856,
|
|
"step": 580608
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0485946003670164e-05,
|
|
"loss": 3.6892,
|
|
"step": 581120
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0477560056159644e-05,
|
|
"loss": 3.6869,
|
|
"step": 581632
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0469174108649124e-05,
|
|
"loss": 3.7053,
|
|
"step": 582144
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.046080453994234e-05,
|
|
"loss": 3.7105,
|
|
"step": 582656
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.045241859243182e-05,
|
|
"loss": 3.6896,
|
|
"step": 583168
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.044403264492129e-05,
|
|
"loss": 3.6877,
|
|
"step": 583680
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.043564669741077e-05,
|
|
"loss": 3.6945,
|
|
"step": 584192
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.042726074990025e-05,
|
|
"loss": 3.6852,
|
|
"step": 584704
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.041889118119347e-05,
|
|
"loss": 3.7039,
|
|
"step": 585216
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.041050523368294e-05,
|
|
"loss": 3.6878,
|
|
"step": 585728
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.040211928617242e-05,
|
|
"loss": 3.6914,
|
|
"step": 586240
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.03937333386619e-05,
|
|
"loss": 3.6955,
|
|
"step": 586752
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.038534739115138e-05,
|
|
"loss": 3.6964,
|
|
"step": 587264
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.03769778224446e-05,
|
|
"loss": 3.6844,
|
|
"step": 587776
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.036859187493408e-05,
|
|
"loss": 3.6927,
|
|
"step": 588288
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.036020592742356e-05,
|
|
"loss": 3.6892,
|
|
"step": 588800
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.035181997991304e-05,
|
|
"loss": 3.6807,
|
|
"step": 589312
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0343450411206246e-05,
|
|
"loss": 3.676,
|
|
"step": 589824
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0335064463695726e-05,
|
|
"loss": 3.6878,
|
|
"step": 590336
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0326678516185206e-05,
|
|
"loss": 3.6909,
|
|
"step": 590848
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0318292568674686e-05,
|
|
"loss": 3.6839,
|
|
"step": 591360
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0309906621164166e-05,
|
|
"loss": 3.6899,
|
|
"step": 591872
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0301520673653646e-05,
|
|
"loss": 3.6838,
|
|
"step": 592384
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0293151104946855e-05,
|
|
"loss": 3.7001,
|
|
"step": 592896
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0284765157436335e-05,
|
|
"loss": 3.7025,
|
|
"step": 593408
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0276379209925815e-05,
|
|
"loss": 3.6987,
|
|
"step": 593920
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.02679932624153e-05,
|
|
"loss": 3.6915,
|
|
"step": 594432
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.025960731490478e-05,
|
|
"loss": 3.6805,
|
|
"step": 594944
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.025122136739426e-05,
|
|
"loss": 3.6908,
|
|
"step": 595456
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.024283541988374e-05,
|
|
"loss": 3.6795,
|
|
"step": 595968
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.023444947237322e-05,
|
|
"loss": 3.6871,
|
|
"step": 596480
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.022607990366643e-05,
|
|
"loss": 3.6846,
|
|
"step": 596992
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.021769395615591e-05,
|
|
"loss": 3.6889,
|
|
"step": 597504
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.020930800864539e-05,
|
|
"loss": 3.6893,
|
|
"step": 598016
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.020092206113487e-05,
|
|
"loss": 3.6884,
|
|
"step": 598528
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.019255249242808e-05,
|
|
"loss": 3.6951,
|
|
"step": 599040
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.018416654491756e-05,
|
|
"loss": 3.6842,
|
|
"step": 599552
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.017579697621077e-05,
|
|
"loss": 3.6879,
|
|
"step": 600064
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0167411028700256e-05,
|
|
"loss": 3.6923,
|
|
"step": 600576
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0159025081189736e-05,
|
|
"loss": 3.6889,
|
|
"step": 601088
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0150639133679216e-05,
|
|
"loss": 3.7007,
|
|
"step": 601600
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0142253186168696e-05,
|
|
"loss": 3.6823,
|
|
"step": 602112
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0133867238658175e-05,
|
|
"loss": 3.696,
|
|
"step": 602624
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0125481291147655e-05,
|
|
"loss": 3.6896,
|
|
"step": 603136
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.011709534363713e-05,
|
|
"loss": 3.6921,
|
|
"step": 603648
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0108725774930344e-05,
|
|
"loss": 3.687,
|
|
"step": 604160
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0100339827419824e-05,
|
|
"loss": 3.6919,
|
|
"step": 604672
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0091953879909304e-05,
|
|
"loss": 3.6846,
|
|
"step": 605184
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0083584311202513e-05,
|
|
"loss": 3.6962,
|
|
"step": 605696
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.0075198363691993e-05,
|
|
"loss": 3.6763,
|
|
"step": 606208
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.006681241618147e-05,
|
|
"loss": 3.6857,
|
|
"step": 606720
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.005842646867095e-05,
|
|
"loss": 3.6914,
|
|
"step": 607232
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.005004052116043e-05,
|
|
"loss": 3.6901,
|
|
"step": 607744
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.004165457364991e-05,
|
|
"loss": 3.6848,
|
|
"step": 608256
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.003326862613939e-05,
|
|
"loss": 3.6855,
|
|
"step": 608768
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.00248990574326e-05,
|
|
"loss": 3.6983,
|
|
"step": 609280
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.001651310992208e-05,
|
|
"loss": 3.6807,
|
|
"step": 609792
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.000812716241156e-05,
|
|
"loss": 3.6953,
|
|
"step": 610304
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.8519527912139893,
|
|
"eval_runtime": 304.1849,
|
|
"eval_samples_per_second": 1254.47,
|
|
"eval_steps_per_second": 39.203,
|
|
"step": 610560
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.999974121490104e-05,
|
|
"loss": 3.6788,
|
|
"step": 610816
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.999135526739052e-05,
|
|
"loss": 3.677,
|
|
"step": 611328
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.998298569868373e-05,
|
|
"loss": 3.6865,
|
|
"step": 611840
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.997459975117321e-05,
|
|
"loss": 3.6884,
|
|
"step": 612352
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.996621380366269e-05,
|
|
"loss": 3.6978,
|
|
"step": 612864
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.995782785615218e-05,
|
|
"loss": 3.6786,
|
|
"step": 613376
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.994944190864166e-05,
|
|
"loss": 3.685,
|
|
"step": 613888
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.994107233993487e-05,
|
|
"loss": 3.6819,
|
|
"step": 614400
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.993268639242435e-05,
|
|
"loss": 3.6773,
|
|
"step": 614912
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.992430044491383e-05,
|
|
"loss": 3.6783,
|
|
"step": 615424
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.991591449740331e-05,
|
|
"loss": 3.6894,
|
|
"step": 615936
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.990752854989279e-05,
|
|
"loss": 3.6907,
|
|
"step": 616448
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.9899158981185996e-05,
|
|
"loss": 3.6725,
|
|
"step": 616960
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.9890773033675476e-05,
|
|
"loss": 3.6716,
|
|
"step": 617472
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.9882387086164956e-05,
|
|
"loss": 3.6839,
|
|
"step": 617984
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.9874001138654436e-05,
|
|
"loss": 3.6655,
|
|
"step": 618496
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.9865615191143916e-05,
|
|
"loss": 3.6789,
|
|
"step": 619008
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.9857229243633396e-05,
|
|
"loss": 3.6749,
|
|
"step": 619520
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.9848843296122876e-05,
|
|
"loss": 3.6827,
|
|
"step": 620032
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.9840457348612356e-05,
|
|
"loss": 3.6841,
|
|
"step": 620544
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.983208777990557e-05,
|
|
"loss": 3.6875,
|
|
"step": 621056
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.982370183239505e-05,
|
|
"loss": 3.6816,
|
|
"step": 621568
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.981531588488453e-05,
|
|
"loss": 3.6785,
|
|
"step": 622080
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.980692993737401e-05,
|
|
"loss": 3.6781,
|
|
"step": 622592
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.979856036866722e-05,
|
|
"loss": 3.6788,
|
|
"step": 623104
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.97901744211567e-05,
|
|
"loss": 3.6822,
|
|
"step": 623616
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.978178847364618e-05,
|
|
"loss": 3.6756,
|
|
"step": 624128
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.977341890493939e-05,
|
|
"loss": 3.6714,
|
|
"step": 624640
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.976503295742887e-05,
|
|
"loss": 3.6766,
|
|
"step": 625152
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.975664700991835e-05,
|
|
"loss": 3.6742,
|
|
"step": 625664
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.974826106240783e-05,
|
|
"loss": 3.6704,
|
|
"step": 626176
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.973987511489731e-05,
|
|
"loss": 3.6857,
|
|
"step": 626688
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.973148916738679e-05,
|
|
"loss": 3.6835,
|
|
"step": 627200
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9723119598680005e-05,
|
|
"loss": 3.6788,
|
|
"step": 627712
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9714733651169485e-05,
|
|
"loss": 3.6718,
|
|
"step": 628224
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9706347703658965e-05,
|
|
"loss": 3.6855,
|
|
"step": 628736
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.969796175614844e-05,
|
|
"loss": 3.6781,
|
|
"step": 629248
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.968957580863792e-05,
|
|
"loss": 3.6758,
|
|
"step": 629760
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.96811898611274e-05,
|
|
"loss": 3.6602,
|
|
"step": 630272
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9672820292420614e-05,
|
|
"loss": 3.6675,
|
|
"step": 630784
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.966443434491009e-05,
|
|
"loss": 3.6709,
|
|
"step": 631296
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.965604839739957e-05,
|
|
"loss": 3.6797,
|
|
"step": 631808
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.964766244988905e-05,
|
|
"loss": 3.6677,
|
|
"step": 632320
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.963929288118226e-05,
|
|
"loss": 3.6709,
|
|
"step": 632832
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.963090693367174e-05,
|
|
"loss": 3.6862,
|
|
"step": 633344
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.962252098616122e-05,
|
|
"loss": 3.6743,
|
|
"step": 633856
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.96141350386507e-05,
|
|
"loss": 3.683,
|
|
"step": 634368
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.960574909114018e-05,
|
|
"loss": 3.6656,
|
|
"step": 634880
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.959736314362966e-05,
|
|
"loss": 3.65,
|
|
"step": 635392
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.958899357492287e-05,
|
|
"loss": 3.6868,
|
|
"step": 635904
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.958060762741235e-05,
|
|
"loss": 3.6729,
|
|
"step": 636416
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.957222167990183e-05,
|
|
"loss": 3.6714,
|
|
"step": 636928
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.956383573239131e-05,
|
|
"loss": 3.6722,
|
|
"step": 637440
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.955544978488079e-05,
|
|
"loss": 3.6711,
|
|
"step": 637952
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9547080216174e-05,
|
|
"loss": 3.6568,
|
|
"step": 638464
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.953869426866349e-05,
|
|
"loss": 3.6638,
|
|
"step": 638976
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.953030832115297e-05,
|
|
"loss": 3.6657,
|
|
"step": 639488
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.952192237364245e-05,
|
|
"loss": 3.6676,
|
|
"step": 640000
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9513552804935656e-05,
|
|
"loss": 3.6822,
|
|
"step": 640512
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9505166857425136e-05,
|
|
"loss": 3.6678,
|
|
"step": 641024
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9496780909914616e-05,
|
|
"loss": 3.6646,
|
|
"step": 641536
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9488394962404096e-05,
|
|
"loss": 3.6805,
|
|
"step": 642048
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9480009014893576e-05,
|
|
"loss": 3.656,
|
|
"step": 642560
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9471639446186785e-05,
|
|
"loss": 3.6622,
|
|
"step": 643072
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9463253498676265e-05,
|
|
"loss": 3.6688,
|
|
"step": 643584
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9454867551165745e-05,
|
|
"loss": 3.6673,
|
|
"step": 644096
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9446481603655225e-05,
|
|
"loss": 3.6633,
|
|
"step": 644608
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.943811203494844e-05,
|
|
"loss": 3.6654,
|
|
"step": 645120
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.942972608743792e-05,
|
|
"loss": 3.6516,
|
|
"step": 645632
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.94213401399274e-05,
|
|
"loss": 3.6629,
|
|
"step": 646144
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.941295419241688e-05,
|
|
"loss": 3.6771,
|
|
"step": 646656
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.940458462371009e-05,
|
|
"loss": 3.6624,
|
|
"step": 647168
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.939619867619957e-05,
|
|
"loss": 3.6813,
|
|
"step": 647680
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.938781272868905e-05,
|
|
"loss": 3.6745,
|
|
"step": 648192
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.937942678117853e-05,
|
|
"loss": 3.6806,
|
|
"step": 648704
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.937105721247174e-05,
|
|
"loss": 3.6597,
|
|
"step": 649216
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.936267126496122e-05,
|
|
"loss": 3.6742,
|
|
"step": 649728
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.93542853174507e-05,
|
|
"loss": 3.6716,
|
|
"step": 650240
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.934589936994018e-05,
|
|
"loss": 3.657,
|
|
"step": 650752
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9337529801233395e-05,
|
|
"loss": 3.6797,
|
|
"step": 651264
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9329143853722875e-05,
|
|
"loss": 3.669,
|
|
"step": 651776
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9320757906212355e-05,
|
|
"loss": 3.6708,
|
|
"step": 652288
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9312371958701835e-05,
|
|
"loss": 3.6726,
|
|
"step": 652800
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9303986011191315e-05,
|
|
"loss": 3.651,
|
|
"step": 653312
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9295600063680794e-05,
|
|
"loss": 3.6545,
|
|
"step": 653824
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9287230494974004e-05,
|
|
"loss": 3.6601,
|
|
"step": 654336
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9278844547463484e-05,
|
|
"loss": 3.6668,
|
|
"step": 654848
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9270458599952964e-05,
|
|
"loss": 3.6644,
|
|
"step": 655360
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.9262072652442443e-05,
|
|
"loss": 3.6607,
|
|
"step": 655872
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.925370308373565e-05,
|
|
"loss": 3.663,
|
|
"step": 656384
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.924531713622513e-05,
|
|
"loss": 3.663,
|
|
"step": 656896
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.923693118871461e-05,
|
|
"loss": 3.6602,
|
|
"step": 657408
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.922854524120409e-05,
|
|
"loss": 3.6595,
|
|
"step": 657920
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.922015929369357e-05,
|
|
"loss": 3.6804,
|
|
"step": 658432
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.921177334618305e-05,
|
|
"loss": 3.6777,
|
|
"step": 658944
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.920340377747627e-05,
|
|
"loss": 3.6674,
|
|
"step": 659456
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.919501782996575e-05,
|
|
"loss": 3.6558,
|
|
"step": 659968
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.918663188245522e-05,
|
|
"loss": 3.6672,
|
|
"step": 660480
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.91782459349447e-05,
|
|
"loss": 3.6579,
|
|
"step": 660992
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.916987636623792e-05,
|
|
"loss": 3.6767,
|
|
"step": 661504
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.916149041872739e-05,
|
|
"loss": 3.6617,
|
|
"step": 662016
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.915310447121687e-05,
|
|
"loss": 3.6628,
|
|
"step": 662528
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.914471852370635e-05,
|
|
"loss": 3.6661,
|
|
"step": 663040
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.9136348954999566e-05,
|
|
"loss": 3.6724,
|
|
"step": 663552
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.9127963007489046e-05,
|
|
"loss": 3.656,
|
|
"step": 664064
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.9119577059978526e-05,
|
|
"loss": 3.6644,
|
|
"step": 664576
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.9111191112468006e-05,
|
|
"loss": 3.6582,
|
|
"step": 665088
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.9102805164957486e-05,
|
|
"loss": 3.6568,
|
|
"step": 665600
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.9094435596250695e-05,
|
|
"loss": 3.6486,
|
|
"step": 666112
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.9086049648740175e-05,
|
|
"loss": 3.6608,
|
|
"step": 666624
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.9077663701229655e-05,
|
|
"loss": 3.6598,
|
|
"step": 667136
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.9069277753719135e-05,
|
|
"loss": 3.6621,
|
|
"step": 667648
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.9060891806208615e-05,
|
|
"loss": 3.6605,
|
|
"step": 668160
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.9052505858698095e-05,
|
|
"loss": 3.6563,
|
|
"step": 668672
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.9044136289991304e-05,
|
|
"loss": 3.6702,
|
|
"step": 669184
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.9035750342480784e-05,
|
|
"loss": 3.6767,
|
|
"step": 669696
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.9027364394970264e-05,
|
|
"loss": 3.6649,
|
|
"step": 670208
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.901897844745975e-05,
|
|
"loss": 3.671,
|
|
"step": 670720
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.901059249994923e-05,
|
|
"loss": 3.6475,
|
|
"step": 671232
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.900222293124244e-05,
|
|
"loss": 3.6654,
|
|
"step": 671744
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.899383698373192e-05,
|
|
"loss": 3.6571,
|
|
"step": 672256
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.89854510362214e-05,
|
|
"loss": 3.6562,
|
|
"step": 672768
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.897706508871088e-05,
|
|
"loss": 3.6576,
|
|
"step": 673280
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.896867914120036e-05,
|
|
"loss": 3.6599,
|
|
"step": 673792
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.896029319368984e-05,
|
|
"loss": 3.6626,
|
|
"step": 674304
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.895192362498305e-05,
|
|
"loss": 3.6609,
|
|
"step": 674816
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.894353767747253e-05,
|
|
"loss": 3.6696,
|
|
"step": 675328
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.893515172996201e-05,
|
|
"loss": 3.6571,
|
|
"step": 675840
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.892676578245149e-05,
|
|
"loss": 3.6568,
|
|
"step": 676352
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.8918396213744704e-05,
|
|
"loss": 3.6666,
|
|
"step": 676864
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.8910010266234184e-05,
|
|
"loss": 3.6626,
|
|
"step": 677376
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.8901624318723664e-05,
|
|
"loss": 3.6749,
|
|
"step": 677888
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.8893238371213144e-05,
|
|
"loss": 3.6547,
|
|
"step": 678400
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.888486880250635e-05,
|
|
"loss": 3.6724,
|
|
"step": 678912
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.887648285499583e-05,
|
|
"loss": 3.6615,
|
|
"step": 679424
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.886809690748531e-05,
|
|
"loss": 3.6659,
|
|
"step": 679936
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.885971095997479e-05,
|
|
"loss": 3.6621,
|
|
"step": 680448
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.885132501246427e-05,
|
|
"loss": 3.6618,
|
|
"step": 680960
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.884293906495375e-05,
|
|
"loss": 3.6558,
|
|
"step": 681472
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.883456949624696e-05,
|
|
"loss": 3.6726,
|
|
"step": 681984
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.882618354873644e-05,
|
|
"loss": 3.6515,
|
|
"step": 682496
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.881779760122592e-05,
|
|
"loss": 3.6589,
|
|
"step": 683008
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.88094116537154e-05,
|
|
"loss": 3.665,
|
|
"step": 683520
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.880104208500862e-05,
|
|
"loss": 3.6689,
|
|
"step": 684032
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.87926561374981e-05,
|
|
"loss": 3.6536,
|
|
"step": 684544
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.878427018998758e-05,
|
|
"loss": 3.66,
|
|
"step": 685056
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.877588424247705e-05,
|
|
"loss": 3.6666,
|
|
"step": 685568
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.876749829496653e-05,
|
|
"loss": 3.6608,
|
|
"step": 686080
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.875911234745601e-05,
|
|
"loss": 3.6648,
|
|
"step": 686592
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 3.846325397491455,
|
|
"eval_runtime": 303.4315,
|
|
"eval_samples_per_second": 1257.585,
|
|
"eval_steps_per_second": 39.3,
|
|
"step": 686880
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.875072639994549e-05,
|
|
"loss": 3.6399,
|
|
"step": 687104
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.874234045243497e-05,
|
|
"loss": 3.6537,
|
|
"step": 687616
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.873395450492445e-05,
|
|
"loss": 3.6578,
|
|
"step": 688128
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.872556855741393e-05,
|
|
"loss": 3.663,
|
|
"step": 688640
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.871718260990341e-05,
|
|
"loss": 3.6694,
|
|
"step": 689152
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.8708813041196626e-05,
|
|
"loss": 3.6532,
|
|
"step": 689664
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.8700427093686106e-05,
|
|
"loss": 3.6576,
|
|
"step": 690176
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.8692041146175586e-05,
|
|
"loss": 3.6581,
|
|
"step": 690688
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.8683655198665066e-05,
|
|
"loss": 3.6478,
|
|
"step": 691200
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.8675269251154546e-05,
|
|
"loss": 3.6538,
|
|
"step": 691712
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.8666883303644026e-05,
|
|
"loss": 3.6611,
|
|
"step": 692224
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.8658513734937235e-05,
|
|
"loss": 3.6608,
|
|
"step": 692736
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.8650144166230444e-05,
|
|
"loss": 3.6583,
|
|
"step": 693248
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.8641758218719924e-05,
|
|
"loss": 3.6379,
|
|
"step": 693760
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.8633372271209404e-05,
|
|
"loss": 3.6582,
|
|
"step": 694272
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.8624986323698884e-05,
|
|
"loss": 3.6391,
|
|
"step": 694784
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.8616600376188364e-05,
|
|
"loss": 3.6497,
|
|
"step": 695296
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.8608214428677844e-05,
|
|
"loss": 3.6527,
|
|
"step": 695808
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.8599828481167324e-05,
|
|
"loss": 3.6535,
|
|
"step": 696320
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.8591442533656804e-05,
|
|
"loss": 3.6577,
|
|
"step": 696832
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.858307296495002e-05,
|
|
"loss": 3.6614,
|
|
"step": 697344
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.85746870174395e-05,
|
|
"loss": 3.6592,
|
|
"step": 697856
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.856630106992898e-05,
|
|
"loss": 3.656,
|
|
"step": 698368
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.855791512241846e-05,
|
|
"loss": 3.6473,
|
|
"step": 698880
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.854952917490794e-05,
|
|
"loss": 3.6548,
|
|
"step": 699392
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.854114322739741e-05,
|
|
"loss": 3.6563,
|
|
"step": 699904
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.853277365869063e-05,
|
|
"loss": 3.6528,
|
|
"step": 700416
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.852438771118011e-05,
|
|
"loss": 3.6491,
|
|
"step": 700928
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.851600176366959e-05,
|
|
"loss": 3.6475,
|
|
"step": 701440
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.850761581615906e-05,
|
|
"loss": 3.6507,
|
|
"step": 701952
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.849924624745228e-05,
|
|
"loss": 3.6421,
|
|
"step": 702464
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.849086029994176e-05,
|
|
"loss": 3.6606,
|
|
"step": 702976
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.8482474352431245e-05,
|
|
"loss": 3.6575,
|
|
"step": 703488
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.847408840492072e-05,
|
|
"loss": 3.6555,
|
|
"step": 704000
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.84657024574102e-05,
|
|
"loss": 3.6487,
|
|
"step": 704512
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.8457332888703414e-05,
|
|
"loss": 3.6541,
|
|
"step": 705024
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.844894694119289e-05,
|
|
"loss": 3.6542,
|
|
"step": 705536
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.844056099368237e-05,
|
|
"loss": 3.6497,
|
|
"step": 706048
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.843217504617185e-05,
|
|
"loss": 3.6399,
|
|
"step": 706560
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.842380547746506e-05,
|
|
"loss": 3.6414,
|
|
"step": 707072
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.8415419529954536e-05,
|
|
"loss": 3.6426,
|
|
"step": 707584
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.8407033582444016e-05,
|
|
"loss": 3.653,
|
|
"step": 708096
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.8398647634933496e-05,
|
|
"loss": 3.644,
|
|
"step": 708608
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.839027806622671e-05,
|
|
"loss": 3.6452,
|
|
"step": 709120
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.838189211871619e-05,
|
|
"loss": 3.6597,
|
|
"step": 709632
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.837350617120567e-05,
|
|
"loss": 3.6501,
|
|
"step": 710144
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.836512022369515e-05,
|
|
"loss": 3.6553,
|
|
"step": 710656
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.835673427618463e-05,
|
|
"loss": 3.6422,
|
|
"step": 711168
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.834836470747784e-05,
|
|
"loss": 3.6276,
|
|
"step": 711680
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.833997875996732e-05,
|
|
"loss": 3.6603,
|
|
"step": 712192
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.83315928124568e-05,
|
|
"loss": 3.6446,
|
|
"step": 712704
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.832320686494628e-05,
|
|
"loss": 3.6507,
|
|
"step": 713216
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.831482091743576e-05,
|
|
"loss": 3.6466,
|
|
"step": 713728
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.830645134872897e-05,
|
|
"loss": 3.6445,
|
|
"step": 714240
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.829806540121845e-05,
|
|
"loss": 3.6302,
|
|
"step": 714752
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.8289679453707936e-05,
|
|
"loss": 3.6419,
|
|
"step": 715264
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.8281293506197416e-05,
|
|
"loss": 3.6401,
|
|
"step": 715776
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.8272907558686896e-05,
|
|
"loss": 3.64,
|
|
"step": 716288
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.8264537989980105e-05,
|
|
"loss": 3.6582,
|
|
"step": 716800
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.8256152042469585e-05,
|
|
"loss": 3.6415,
|
|
"step": 717312
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.8247766094959065e-05,
|
|
"loss": 3.6385,
|
|
"step": 717824
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.8239380147448545e-05,
|
|
"loss": 3.6589,
|
|
"step": 718336
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.8231010578741754e-05,
|
|
"loss": 3.6303,
|
|
"step": 718848
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.8222624631231234e-05,
|
|
"loss": 3.6332,
|
|
"step": 719360
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.8214238683720714e-05,
|
|
"loss": 3.6501,
|
|
"step": 719872
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.8205852736210194e-05,
|
|
"loss": 3.6429,
|
|
"step": 720384
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.81974831675034e-05,
|
|
"loss": 3.6354,
|
|
"step": 720896
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.818909721999289e-05,
|
|
"loss": 3.6434,
|
|
"step": 721408
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.818071127248237e-05,
|
|
"loss": 3.6249,
|
|
"step": 721920
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.817232532497185e-05,
|
|
"loss": 3.6344,
|
|
"step": 722432
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.816393937746133e-05,
|
|
"loss": 3.6517,
|
|
"step": 722944
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.815556980875454e-05,
|
|
"loss": 3.6387,
|
|
"step": 723456
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.814718386124402e-05,
|
|
"loss": 3.6503,
|
|
"step": 723968
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.81387979137335e-05,
|
|
"loss": 3.6482,
|
|
"step": 724480
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.813041196622298e-05,
|
|
"loss": 3.6602,
|
|
"step": 724992
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.812204239751619e-05,
|
|
"loss": 3.6354,
|
|
"step": 725504
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.811365645000567e-05,
|
|
"loss": 3.6477,
|
|
"step": 726016
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.810527050249515e-05,
|
|
"loss": 3.647,
|
|
"step": 726528
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.809688455498463e-05,
|
|
"loss": 3.631,
|
|
"step": 727040
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.808849860747411e-05,
|
|
"loss": 3.656,
|
|
"step": 727552
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.808012903876732e-05,
|
|
"loss": 3.6423,
|
|
"step": 728064
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.80717430912568e-05,
|
|
"loss": 3.647,
|
|
"step": 728576
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.806335714374628e-05,
|
|
"loss": 3.6494,
|
|
"step": 729088
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.805497119623576e-05,
|
|
"loss": 3.6312,
|
|
"step": 729600
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.804660162752897e-05,
|
|
"loss": 3.6269,
|
|
"step": 730112
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.803821568001845e-05,
|
|
"loss": 3.6348,
|
|
"step": 730624
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.802982973250793e-05,
|
|
"loss": 3.643,
|
|
"step": 731136
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.802144378499741e-05,
|
|
"loss": 3.6412,
|
|
"step": 731648
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.801307421629062e-05,
|
|
"loss": 3.6351,
|
|
"step": 732160
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.80046882687801e-05,
|
|
"loss": 3.6364,
|
|
"step": 732672
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.799630232126958e-05,
|
|
"loss": 3.6446,
|
|
"step": 733184
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.798791637375906e-05,
|
|
"loss": 3.636,
|
|
"step": 733696
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.797954680505228e-05,
|
|
"loss": 3.6308,
|
|
"step": 734208
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.797116085754176e-05,
|
|
"loss": 3.6579,
|
|
"step": 734720
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.796277491003124e-05,
|
|
"loss": 3.653,
|
|
"step": 735232
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.795438896252072e-05,
|
|
"loss": 3.6455,
|
|
"step": 735744
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.79460030150102e-05,
|
|
"loss": 3.6297,
|
|
"step": 736256
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.793761706749967e-05,
|
|
"loss": 3.6437,
|
|
"step": 736768
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7929247498792886e-05,
|
|
"loss": 3.6361,
|
|
"step": 737280
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7920861551282366e-05,
|
|
"loss": 3.6483,
|
|
"step": 737792
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7912475603771846e-05,
|
|
"loss": 3.6391,
|
|
"step": 738304
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.790408965626132e-05,
|
|
"loss": 3.6352,
|
|
"step": 738816
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7895720087554535e-05,
|
|
"loss": 3.643,
|
|
"step": 739328
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7887334140044015e-05,
|
|
"loss": 3.6495,
|
|
"step": 739840
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7878948192533495e-05,
|
|
"loss": 3.631,
|
|
"step": 740352
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7870562245022975e-05,
|
|
"loss": 3.645,
|
|
"step": 740864
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7862176297512455e-05,
|
|
"loss": 3.6281,
|
|
"step": 741376
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.785380672880567e-05,
|
|
"loss": 3.6359,
|
|
"step": 741888
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7845420781295144e-05,
|
|
"loss": 3.6284,
|
|
"step": 742400
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7837034833784624e-05,
|
|
"loss": 3.6304,
|
|
"step": 742912
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7828648886274104e-05,
|
|
"loss": 3.6363,
|
|
"step": 743424
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.782027931756732e-05,
|
|
"loss": 3.6414,
|
|
"step": 743936
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.781189337005679e-05,
|
|
"loss": 3.639,
|
|
"step": 744448
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.780350742254627e-05,
|
|
"loss": 3.6303,
|
|
"step": 744960
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.779512147503575e-05,
|
|
"loss": 3.6459,
|
|
"step": 745472
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.778673552752523e-05,
|
|
"loss": 3.6517,
|
|
"step": 745984
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.777836595881845e-05,
|
|
"loss": 3.6416,
|
|
"step": 746496
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.776998001130793e-05,
|
|
"loss": 3.6491,
|
|
"step": 747008
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.776159406379741e-05,
|
|
"loss": 3.627,
|
|
"step": 747520
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.775320811628689e-05,
|
|
"loss": 3.6391,
|
|
"step": 748032
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.77448385475801e-05,
|
|
"loss": 3.6347,
|
|
"step": 748544
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.773645260006958e-05,
|
|
"loss": 3.6283,
|
|
"step": 749056
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.772806665255906e-05,
|
|
"loss": 3.6347,
|
|
"step": 749568
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.771968070504854e-05,
|
|
"loss": 3.635,
|
|
"step": 750080
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.771129475753802e-05,
|
|
"loss": 3.6405,
|
|
"step": 750592
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7702925188831226e-05,
|
|
"loss": 3.6357,
|
|
"step": 751104
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7694539241320706e-05,
|
|
"loss": 3.6428,
|
|
"step": 751616
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7686153293810186e-05,
|
|
"loss": 3.639,
|
|
"step": 752128
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7677767346299666e-05,
|
|
"loss": 3.6315,
|
|
"step": 752640
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.766939777759288e-05,
|
|
"loss": 3.6416,
|
|
"step": 753152
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.766101183008236e-05,
|
|
"loss": 3.6423,
|
|
"step": 753664
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.765262588257184e-05,
|
|
"loss": 3.6473,
|
|
"step": 754176
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.764423993506132e-05,
|
|
"loss": 3.6344,
|
|
"step": 754688
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.763587036635453e-05,
|
|
"loss": 3.6494,
|
|
"step": 755200
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.762748441884401e-05,
|
|
"loss": 3.637,
|
|
"step": 755712
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.761909847133349e-05,
|
|
"loss": 3.6427,
|
|
"step": 756224
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.761071252382297e-05,
|
|
"loss": 3.6394,
|
|
"step": 756736
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.760232657631245e-05,
|
|
"loss": 3.6367,
|
|
"step": 757248
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.759395700760566e-05,
|
|
"loss": 3.633,
|
|
"step": 757760
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.758557106009514e-05,
|
|
"loss": 3.6539,
|
|
"step": 758272
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.757718511258462e-05,
|
|
"loss": 3.6207,
|
|
"step": 758784
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7568799165074107e-05,
|
|
"loss": 3.6395,
|
|
"step": 759296
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7560413217563586e-05,
|
|
"loss": 3.6443,
|
|
"step": 759808
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7552043648856796e-05,
|
|
"loss": 3.6391,
|
|
"step": 760320
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7543657701346276e-05,
|
|
"loss": 3.6328,
|
|
"step": 760832
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7535271753835755e-05,
|
|
"loss": 3.6361,
|
|
"step": 761344
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7526885806325235e-05,
|
|
"loss": 3.6396,
|
|
"step": 761856
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7518516237618445e-05,
|
|
"loss": 3.642,
|
|
"step": 762368
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.7510130290107924e-05,
|
|
"loss": 3.6391,
|
|
"step": 762880
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.8440346717834473,
|
|
"eval_runtime": 305.2603,
|
|
"eval_samples_per_second": 1250.051,
|
|
"eval_steps_per_second": 39.065,
|
|
"step": 763200
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.7501744342597404e-05,
|
|
"loss": 3.6269,
|
|
"step": 763392
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.7493358395086884e-05,
|
|
"loss": 3.6289,
|
|
"step": 763904
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.7484988826380093e-05,
|
|
"loss": 3.6352,
|
|
"step": 764416
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.7476602878869573e-05,
|
|
"loss": 3.639,
|
|
"step": 764928
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.746821693135906e-05,
|
|
"loss": 3.6476,
|
|
"step": 765440
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.745983098384854e-05,
|
|
"loss": 3.6285,
|
|
"step": 765952
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.745146141514175e-05,
|
|
"loss": 3.6346,
|
|
"step": 766464
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.744307546763123e-05,
|
|
"loss": 3.6339,
|
|
"step": 766976
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.743468952012071e-05,
|
|
"loss": 3.6277,
|
|
"step": 767488
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.742630357261019e-05,
|
|
"loss": 3.6282,
|
|
"step": 768000
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.741791762509967e-05,
|
|
"loss": 3.6383,
|
|
"step": 768512
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.740954805639288e-05,
|
|
"loss": 3.641,
|
|
"step": 769024
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.740117848768609e-05,
|
|
"loss": 3.6382,
|
|
"step": 769536
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.739279254017557e-05,
|
|
"loss": 3.6136,
|
|
"step": 770048
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.738440659266505e-05,
|
|
"loss": 3.6352,
|
|
"step": 770560
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.737602064515453e-05,
|
|
"loss": 3.6157,
|
|
"step": 771072
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.7367634697644014e-05,
|
|
"loss": 3.6279,
|
|
"step": 771584
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.7359248750133494e-05,
|
|
"loss": 3.6277,
|
|
"step": 772096
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.7350862802622974e-05,
|
|
"loss": 3.6286,
|
|
"step": 772608
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.7342476855112454e-05,
|
|
"loss": 3.6339,
|
|
"step": 773120
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.733410728640566e-05,
|
|
"loss": 3.6413,
|
|
"step": 773632
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.732572133889514e-05,
|
|
"loss": 3.6336,
|
|
"step": 774144
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.731733539138462e-05,
|
|
"loss": 3.6302,
|
|
"step": 774656
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.7308949443874096e-05,
|
|
"loss": 3.6227,
|
|
"step": 775168
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.730057987516731e-05,
|
|
"loss": 3.6352,
|
|
"step": 775680
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.729219392765679e-05,
|
|
"loss": 3.6326,
|
|
"step": 776192
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.728380798014627e-05,
|
|
"loss": 3.6276,
|
|
"step": 776704
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.727542203263575e-05,
|
|
"loss": 3.6301,
|
|
"step": 777216
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.726703608512523e-05,
|
|
"loss": 3.6224,
|
|
"step": 777728
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.725866651641845e-05,
|
|
"loss": 3.6295,
|
|
"step": 778240
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.725028056890793e-05,
|
|
"loss": 3.6174,
|
|
"step": 778752
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.72418946213974e-05,
|
|
"loss": 3.6388,
|
|
"step": 779264
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.723350867388688e-05,
|
|
"loss": 3.6403,
|
|
"step": 779776
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.7225139105180096e-05,
|
|
"loss": 3.6288,
|
|
"step": 780288
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.721675315766957e-05,
|
|
"loss": 3.628,
|
|
"step": 780800
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.720836721015905e-05,
|
|
"loss": 3.6238,
|
|
"step": 781312
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.719998126264853e-05,
|
|
"loss": 3.6361,
|
|
"step": 781824
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.7191611693941745e-05,
|
|
"loss": 3.6283,
|
|
"step": 782336
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.718322574643122e-05,
|
|
"loss": 3.6128,
|
|
"step": 782848
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.7174839798920705e-05,
|
|
"loss": 3.6218,
|
|
"step": 783360
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.7166453851410185e-05,
|
|
"loss": 3.6218,
|
|
"step": 783872
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.71580842827034e-05,
|
|
"loss": 3.628,
|
|
"step": 784384
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.7149698335192874e-05,
|
|
"loss": 3.6221,
|
|
"step": 784896
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.7141312387682354e-05,
|
|
"loss": 3.6249,
|
|
"step": 785408
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.7132926440171834e-05,
|
|
"loss": 3.6339,
|
|
"step": 785920
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.7124540492661314e-05,
|
|
"loss": 3.6295,
|
|
"step": 786432
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.711617092395452e-05,
|
|
"loss": 3.6324,
|
|
"step": 786944
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.7107784976444e-05,
|
|
"loss": 3.6241,
|
|
"step": 787456
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.709939902893348e-05,
|
|
"loss": 3.6007,
|
|
"step": 787968
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.709101308142296e-05,
|
|
"loss": 3.637,
|
|
"step": 788480
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.708264351271617e-05,
|
|
"loss": 3.6236,
|
|
"step": 788992
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.707425756520566e-05,
|
|
"loss": 3.6256,
|
|
"step": 789504
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.706587161769514e-05,
|
|
"loss": 3.6292,
|
|
"step": 790016
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.705748567018462e-05,
|
|
"loss": 3.6215,
|
|
"step": 790528
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.704911610147783e-05,
|
|
"loss": 3.6066,
|
|
"step": 791040
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.704073015396731e-05,
|
|
"loss": 3.6165,
|
|
"step": 791552
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.703234420645679e-05,
|
|
"loss": 3.622,
|
|
"step": 792064
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.702395825894627e-05,
|
|
"loss": 3.6158,
|
|
"step": 792576
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.701558869023948e-05,
|
|
"loss": 3.6349,
|
|
"step": 793088
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.700720274272896e-05,
|
|
"loss": 3.6221,
|
|
"step": 793600
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.699881679521844e-05,
|
|
"loss": 3.6169,
|
|
"step": 794112
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.699043084770792e-05,
|
|
"loss": 3.6337,
|
|
"step": 794624
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.6982061279001126e-05,
|
|
"loss": 3.613,
|
|
"step": 795136
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.6973675331490606e-05,
|
|
"loss": 3.6109,
|
|
"step": 795648
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.696528938398009e-05,
|
|
"loss": 3.6284,
|
|
"step": 796160
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.695690343646957e-05,
|
|
"loss": 3.616,
|
|
"step": 796672
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.694853386776278e-05,
|
|
"loss": 3.6125,
|
|
"step": 797184
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.694014792025226e-05,
|
|
"loss": 3.6194,
|
|
"step": 797696
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.693176197274174e-05,
|
|
"loss": 3.6067,
|
|
"step": 798208
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.692337602523122e-05,
|
|
"loss": 3.6153,
|
|
"step": 798720
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.69149900777207e-05,
|
|
"loss": 3.6301,
|
|
"step": 799232
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.690662050901391e-05,
|
|
"loss": 3.6181,
|
|
"step": 799744
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.689823456150339e-05,
|
|
"loss": 3.6283,
|
|
"step": 800256
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.688984861399287e-05,
|
|
"loss": 3.622,
|
|
"step": 800768
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.688146266648235e-05,
|
|
"loss": 3.6405,
|
|
"step": 801280
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.687309309777556e-05,
|
|
"loss": 3.612,
|
|
"step": 801792
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.6864707150265046e-05,
|
|
"loss": 3.6273,
|
|
"step": 802304
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.6856321202754526e-05,
|
|
"loss": 3.6262,
|
|
"step": 802816
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.6847935255244006e-05,
|
|
"loss": 3.6078,
|
|
"step": 803328
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.6839565686537215e-05,
|
|
"loss": 3.6325,
|
|
"step": 803840
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.6831179739026695e-05,
|
|
"loss": 3.6206,
|
|
"step": 804352
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.6822793791516175e-05,
|
|
"loss": 3.624,
|
|
"step": 804864
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.6814407844005655e-05,
|
|
"loss": 3.6288,
|
|
"step": 805376
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.6806038275298864e-05,
|
|
"loss": 3.6093,
|
|
"step": 805888
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.6797652327788344e-05,
|
|
"loss": 3.6045,
|
|
"step": 806400
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.6789266380277824e-05,
|
|
"loss": 3.611,
|
|
"step": 806912
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.6780880432767304e-05,
|
|
"loss": 3.6231,
|
|
"step": 807424
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.677251086406051e-05,
|
|
"loss": 3.6152,
|
|
"step": 807936
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.676412491655e-05,
|
|
"loss": 3.6127,
|
|
"step": 808448
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.675573896903948e-05,
|
|
"loss": 3.6136,
|
|
"step": 808960
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.674735302152896e-05,
|
|
"loss": 3.6222,
|
|
"step": 809472
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.673898345282217e-05,
|
|
"loss": 3.613,
|
|
"step": 809984
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.673059750531165e-05,
|
|
"loss": 3.6132,
|
|
"step": 810496
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.672221155780113e-05,
|
|
"loss": 3.631,
|
|
"step": 811008
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.671382561029061e-05,
|
|
"loss": 3.6319,
|
|
"step": 811520
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.670545604158382e-05,
|
|
"loss": 3.6237,
|
|
"step": 812032
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.66970700940733e-05,
|
|
"loss": 3.6117,
|
|
"step": 812544
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.668868414656278e-05,
|
|
"loss": 3.6198,
|
|
"step": 813056
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.668029819905226e-05,
|
|
"loss": 3.6123,
|
|
"step": 813568
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.667192863034547e-05,
|
|
"loss": 3.6255,
|
|
"step": 814080
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6663542682834954e-05,
|
|
"loss": 3.6174,
|
|
"step": 814592
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6655156735324434e-05,
|
|
"loss": 3.6204,
|
|
"step": 815104
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6646770787813914e-05,
|
|
"loss": 3.6169,
|
|
"step": 815616
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.663840121910712e-05,
|
|
"loss": 3.6213,
|
|
"step": 816128
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.66300152715966e-05,
|
|
"loss": 3.6186,
|
|
"step": 816640
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.662162932408608e-05,
|
|
"loss": 3.6192,
|
|
"step": 817152
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.661324337657556e-05,
|
|
"loss": 3.609,
|
|
"step": 817664
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.660487380786877e-05,
|
|
"loss": 3.6157,
|
|
"step": 818176
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.659648786035825e-05,
|
|
"loss": 3.6081,
|
|
"step": 818688
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.658810191284773e-05,
|
|
"loss": 3.6076,
|
|
"step": 819200
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.657971596533721e-05,
|
|
"loss": 3.615,
|
|
"step": 819712
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.657134639663042e-05,
|
|
"loss": 3.6137,
|
|
"step": 820224
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.656296044911991e-05,
|
|
"loss": 3.619,
|
|
"step": 820736
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.655457450160939e-05,
|
|
"loss": 3.6094,
|
|
"step": 821248
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.654618855409887e-05,
|
|
"loss": 3.626,
|
|
"step": 821760
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6537818985392076e-05,
|
|
"loss": 3.6302,
|
|
"step": 822272
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6529433037881556e-05,
|
|
"loss": 3.6172,
|
|
"step": 822784
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6521047090371036e-05,
|
|
"loss": 3.6241,
|
|
"step": 823296
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6512661142860516e-05,
|
|
"loss": 3.6133,
|
|
"step": 823808
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6504275195349996e-05,
|
|
"loss": 3.6128,
|
|
"step": 824320
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6495905626643205e-05,
|
|
"loss": 3.6139,
|
|
"step": 824832
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6487519679132685e-05,
|
|
"loss": 3.6052,
|
|
"step": 825344
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6479133731622165e-05,
|
|
"loss": 3.6153,
|
|
"step": 825856
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6470747784111645e-05,
|
|
"loss": 3.6173,
|
|
"step": 826368
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6462361836601125e-05,
|
|
"loss": 3.6204,
|
|
"step": 826880
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.645399226789434e-05,
|
|
"loss": 3.6155,
|
|
"step": 827392
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.644560632038382e-05,
|
|
"loss": 3.6212,
|
|
"step": 827904
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.64372203728733e-05,
|
|
"loss": 3.6126,
|
|
"step": 828416
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6428834425362774e-05,
|
|
"loss": 3.6145,
|
|
"step": 828928
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.642046485665599e-05,
|
|
"loss": 3.6182,
|
|
"step": 829440
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.641207890914547e-05,
|
|
"loss": 3.6222,
|
|
"step": 829952
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.640369296163495e-05,
|
|
"loss": 3.6242,
|
|
"step": 830464
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.639530701412442e-05,
|
|
"loss": 3.6101,
|
|
"step": 830976
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.638693744541764e-05,
|
|
"loss": 3.6321,
|
|
"step": 831488
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.637855149790712e-05,
|
|
"loss": 3.6163,
|
|
"step": 832000
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.63701655503966e-05,
|
|
"loss": 3.6233,
|
|
"step": 832512
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.636177960288608e-05,
|
|
"loss": 3.6176,
|
|
"step": 833024
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6353410034179295e-05,
|
|
"loss": 3.6111,
|
|
"step": 833536
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6345024086668775e-05,
|
|
"loss": 3.6107,
|
|
"step": 834048
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.633663813915825e-05,
|
|
"loss": 3.6339,
|
|
"step": 834560
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.632825219164773e-05,
|
|
"loss": 3.603,
|
|
"step": 835072
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6319882622940944e-05,
|
|
"loss": 3.616,
|
|
"step": 835584
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.631149667543042e-05,
|
|
"loss": 3.6209,
|
|
"step": 836096
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.63031107279199e-05,
|
|
"loss": 3.6219,
|
|
"step": 836608
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6294724780409377e-05,
|
|
"loss": 3.6111,
|
|
"step": 837120
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.628635521170259e-05,
|
|
"loss": 3.6133,
|
|
"step": 837632
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.6277969264192066e-05,
|
|
"loss": 3.6155,
|
|
"step": 838144
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.626958331668155e-05,
|
|
"loss": 3.6253,
|
|
"step": 838656
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.626119736917103e-05,
|
|
"loss": 3.6181,
|
|
"step": 839168
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 3.8424723148345947,
|
|
"eval_runtime": 305.313,
|
|
"eval_samples_per_second": 1249.836,
|
|
"eval_steps_per_second": 39.058,
|
|
"step": 839520
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.625281142166051e-05,
|
|
"loss": 3.5922,
|
|
"step": 839680
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.624442547414999e-05,
|
|
"loss": 3.6087,
|
|
"step": 840192
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.623603952663947e-05,
|
|
"loss": 3.615,
|
|
"step": 840704
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.622765357912895e-05,
|
|
"loss": 3.6166,
|
|
"step": 841216
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.621928401042216e-05,
|
|
"loss": 3.6255,
|
|
"step": 841728
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.621089806291164e-05,
|
|
"loss": 3.6116,
|
|
"step": 842240
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.620251211540112e-05,
|
|
"loss": 3.6127,
|
|
"step": 842752
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.61941261678906e-05,
|
|
"loss": 3.6143,
|
|
"step": 843264
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.618575659918381e-05,
|
|
"loss": 3.6062,
|
|
"step": 843776
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.617737065167329e-05,
|
|
"loss": 3.6017,
|
|
"step": 844288
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.616898470416277e-05,
|
|
"loss": 3.6156,
|
|
"step": 844800
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.616059875665225e-05,
|
|
"loss": 3.6262,
|
|
"step": 845312
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.6152229187945466e-05,
|
|
"loss": 3.6144,
|
|
"step": 845824
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.6143843240434946e-05,
|
|
"loss": 3.5932,
|
|
"step": 846336
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.6135457292924426e-05,
|
|
"loss": 3.6129,
|
|
"step": 846848
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.6127071345413906e-05,
|
|
"loss": 3.5951,
|
|
"step": 847360
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.6118701776707115e-05,
|
|
"loss": 3.6079,
|
|
"step": 847872
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.6110315829196595e-05,
|
|
"loss": 3.6103,
|
|
"step": 848384
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.6101929881686075e-05,
|
|
"loss": 3.6084,
|
|
"step": 848896
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.6093543934175555e-05,
|
|
"loss": 3.6111,
|
|
"step": 849408
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.6085174365468764e-05,
|
|
"loss": 3.6198,
|
|
"step": 849920
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.6076788417958244e-05,
|
|
"loss": 3.6113,
|
|
"step": 850432
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.6068402470447724e-05,
|
|
"loss": 3.6129,
|
|
"step": 850944
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.6060016522937204e-05,
|
|
"loss": 3.6065,
|
|
"step": 851456
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.6051630575426684e-05,
|
|
"loss": 3.6134,
|
|
"step": 851968
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.60432610067199e-05,
|
|
"loss": 3.6115,
|
|
"step": 852480
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.603487505920938e-05,
|
|
"loss": 3.6047,
|
|
"step": 852992
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.602648911169886e-05,
|
|
"loss": 3.6094,
|
|
"step": 853504
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.601811954299207e-05,
|
|
"loss": 3.6016,
|
|
"step": 854016
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.600973359548155e-05,
|
|
"loss": 3.6074,
|
|
"step": 854528
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.600134764797103e-05,
|
|
"loss": 3.6001,
|
|
"step": 855040
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.599296170046051e-05,
|
|
"loss": 3.6151,
|
|
"step": 855552
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.598459213175372e-05,
|
|
"loss": 3.6118,
|
|
"step": 856064
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.59762061842432e-05,
|
|
"loss": 3.6128,
|
|
"step": 856576
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.596782023673268e-05,
|
|
"loss": 3.6072,
|
|
"step": 857088
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.595943428922216e-05,
|
|
"loss": 3.608,
|
|
"step": 857600
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.595104834171164e-05,
|
|
"loss": 3.6167,
|
|
"step": 858112
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.594267877300485e-05,
|
|
"loss": 3.6046,
|
|
"step": 858624
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.593429282549433e-05,
|
|
"loss": 3.5975,
|
|
"step": 859136
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.592590687798381e-05,
|
|
"loss": 3.6016,
|
|
"step": 859648
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.591752093047329e-05,
|
|
"loss": 3.5997,
|
|
"step": 860160
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.59091513617665e-05,
|
|
"loss": 3.6058,
|
|
"step": 860672
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.590076541425598e-05,
|
|
"loss": 3.6006,
|
|
"step": 861184
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.589237946674546e-05,
|
|
"loss": 3.6085,
|
|
"step": 861696
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.588399351923494e-05,
|
|
"loss": 3.6075,
|
|
"step": 862208
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.587562395052815e-05,
|
|
"loss": 3.6102,
|
|
"step": 862720
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.586723800301763e-05,
|
|
"loss": 3.6119,
|
|
"step": 863232
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.585885205550711e-05,
|
|
"loss": 3.6059,
|
|
"step": 863744
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.585046610799659e-05,
|
|
"loss": 3.5828,
|
|
"step": 864256
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.584209653928981e-05,
|
|
"loss": 3.6146,
|
|
"step": 864768
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.583371059177929e-05,
|
|
"loss": 3.6038,
|
|
"step": 865280
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.582532464426877e-05,
|
|
"loss": 3.6046,
|
|
"step": 865792
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.581693869675825e-05,
|
|
"loss": 3.6077,
|
|
"step": 866304
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.5808569128051456e-05,
|
|
"loss": 3.6027,
|
|
"step": 866816
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.5800183180540936e-05,
|
|
"loss": 3.5849,
|
|
"step": 867328
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.5791797233030416e-05,
|
|
"loss": 3.5958,
|
|
"step": 867840
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.5783411285519896e-05,
|
|
"loss": 3.601,
|
|
"step": 868352
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.5775041716813105e-05,
|
|
"loss": 3.5938,
|
|
"step": 868864
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.5766655769302585e-05,
|
|
"loss": 3.6161,
|
|
"step": 869376
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.5758269821792065e-05,
|
|
"loss": 3.6055,
|
|
"step": 869888
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.5749883874281545e-05,
|
|
"loss": 3.589,
|
|
"step": 870400
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.574151430557476e-05,
|
|
"loss": 3.6147,
|
|
"step": 870912
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.573312835806424e-05,
|
|
"loss": 3.5943,
|
|
"step": 871424
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.572474241055372e-05,
|
|
"loss": 3.5903,
|
|
"step": 871936
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.57163564630432e-05,
|
|
"loss": 3.6081,
|
|
"step": 872448
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.570798689433641e-05,
|
|
"loss": 3.5993,
|
|
"step": 872960
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.569960094682589e-05,
|
|
"loss": 3.5919,
|
|
"step": 873472
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.569121499931537e-05,
|
|
"loss": 3.5985,
|
|
"step": 873984
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.568282905180485e-05,
|
|
"loss": 3.586,
|
|
"step": 874496
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.567445948309806e-05,
|
|
"loss": 3.5982,
|
|
"step": 875008
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.566607353558754e-05,
|
|
"loss": 3.6048,
|
|
"step": 875520
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.565768758807702e-05,
|
|
"loss": 3.6002,
|
|
"step": 876032
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.56493016405665e-05,
|
|
"loss": 3.6099,
|
|
"step": 876544
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.5640932071859714e-05,
|
|
"loss": 3.5989,
|
|
"step": 877056
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.5632546124349194e-05,
|
|
"loss": 3.6191,
|
|
"step": 877568
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.5624160176838674e-05,
|
|
"loss": 3.5921,
|
|
"step": 878080
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.561577422932815e-05,
|
|
"loss": 3.6073,
|
|
"step": 878592
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.560740466062136e-05,
|
|
"loss": 3.6031,
|
|
"step": 879104
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.559901871311084e-05,
|
|
"loss": 3.5902,
|
|
"step": 879616
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.559063276560032e-05,
|
|
"loss": 3.6089,
|
|
"step": 880128
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.5582246818089796e-05,
|
|
"loss": 3.6062,
|
|
"step": 880640
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.557387724938301e-05,
|
|
"loss": 3.6029,
|
|
"step": 881152
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.556549130187249e-05,
|
|
"loss": 3.61,
|
|
"step": 881664
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.555710535436197e-05,
|
|
"loss": 3.5871,
|
|
"step": 882176
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.554871940685145e-05,
|
|
"loss": 3.5844,
|
|
"step": 882688
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.554034983814467e-05,
|
|
"loss": 3.5899,
|
|
"step": 883200
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.553196389063415e-05,
|
|
"loss": 3.6053,
|
|
"step": 883712
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.552357794312362e-05,
|
|
"loss": 3.5971,
|
|
"step": 884224
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.55151919956131e-05,
|
|
"loss": 3.5953,
|
|
"step": 884736
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.550682242690632e-05,
|
|
"loss": 3.5932,
|
|
"step": 885248
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.54984364793958e-05,
|
|
"loss": 3.5982,
|
|
"step": 885760
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.549005053188527e-05,
|
|
"loss": 3.5945,
|
|
"step": 886272
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.548166458437475e-05,
|
|
"loss": 3.5958,
|
|
"step": 886784
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5473295015667966e-05,
|
|
"loss": 3.6076,
|
|
"step": 887296
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5464909068157446e-05,
|
|
"loss": 3.6134,
|
|
"step": 887808
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5456523120646926e-05,
|
|
"loss": 3.6062,
|
|
"step": 888320
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5448137173136406e-05,
|
|
"loss": 3.5921,
|
|
"step": 888832
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.543976760442962e-05,
|
|
"loss": 3.5936,
|
|
"step": 889344
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5431381656919095e-05,
|
|
"loss": 3.5941,
|
|
"step": 889856
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5422995709408575e-05,
|
|
"loss": 3.6059,
|
|
"step": 890368
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5414609761898055e-05,
|
|
"loss": 3.6001,
|
|
"step": 890880
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.540624019319127e-05,
|
|
"loss": 3.598,
|
|
"step": 891392
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5397854245680744e-05,
|
|
"loss": 3.5984,
|
|
"step": 891904
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5389468298170224e-05,
|
|
"loss": 3.5989,
|
|
"step": 892416
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5381082350659704e-05,
|
|
"loss": 3.6026,
|
|
"step": 892928
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.537271278195291e-05,
|
|
"loss": 3.5956,
|
|
"step": 893440
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.53643268344424e-05,
|
|
"loss": 3.5908,
|
|
"step": 893952
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.535594088693188e-05,
|
|
"loss": 3.5918,
|
|
"step": 894464
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.534755493942136e-05,
|
|
"loss": 3.591,
|
|
"step": 894976
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.533918537071457e-05,
|
|
"loss": 3.586,
|
|
"step": 895488
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.533079942320405e-05,
|
|
"loss": 3.5933,
|
|
"step": 896000
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.532241347569353e-05,
|
|
"loss": 3.5934,
|
|
"step": 896512
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.531402752818301e-05,
|
|
"loss": 3.5978,
|
|
"step": 897024
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.530565795947622e-05,
|
|
"loss": 3.5871,
|
|
"step": 897536
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.52972720119657e-05,
|
|
"loss": 3.606,
|
|
"step": 898048
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.528888606445518e-05,
|
|
"loss": 3.6109,
|
|
"step": 898560
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.528050011694466e-05,
|
|
"loss": 3.5984,
|
|
"step": 899072
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.527211416943414e-05,
|
|
"loss": 3.6041,
|
|
"step": 899584
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.526374460072735e-05,
|
|
"loss": 3.5938,
|
|
"step": 900096
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.525535865321683e-05,
|
|
"loss": 3.5953,
|
|
"step": 900608
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.524697270570631e-05,
|
|
"loss": 3.5968,
|
|
"step": 901120
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.523858675819579e-05,
|
|
"loss": 3.5812,
|
|
"step": 901632
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5230217189489e-05,
|
|
"loss": 3.5959,
|
|
"step": 902144
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.522183124197848e-05,
|
|
"loss": 3.5985,
|
|
"step": 902656
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.521344529446796e-05,
|
|
"loss": 3.6052,
|
|
"step": 903168
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.520505934695744e-05,
|
|
"loss": 3.5922,
|
|
"step": 903680
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.519668977825065e-05,
|
|
"loss": 3.6041,
|
|
"step": 904192
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.518830383074013e-05,
|
|
"loss": 3.5944,
|
|
"step": 904704
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.517991788322961e-05,
|
|
"loss": 3.5943,
|
|
"step": 905216
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.517153193571909e-05,
|
|
"loss": 3.5977,
|
|
"step": 905728
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.51631623670123e-05,
|
|
"loss": 3.6015,
|
|
"step": 906240
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.515477641950179e-05,
|
|
"loss": 3.6051,
|
|
"step": 906752
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.514639047199127e-05,
|
|
"loss": 3.5889,
|
|
"step": 907264
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.513800452448075e-05,
|
|
"loss": 3.6119,
|
|
"step": 907776
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5129634955773956e-05,
|
|
"loss": 3.5956,
|
|
"step": 908288
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5121249008263436e-05,
|
|
"loss": 3.604,
|
|
"step": 908800
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5112863060752916e-05,
|
|
"loss": 3.6003,
|
|
"step": 909312
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5104477113242396e-05,
|
|
"loss": 3.5922,
|
|
"step": 909824
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5096107544535605e-05,
|
|
"loss": 3.5917,
|
|
"step": 910336
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5087721597025085e-05,
|
|
"loss": 3.6154,
|
|
"step": 910848
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5079335649514565e-05,
|
|
"loss": 3.585,
|
|
"step": 911360
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5070949702004045e-05,
|
|
"loss": 3.5936,
|
|
"step": 911872
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.5062580133297254e-05,
|
|
"loss": 3.6008,
|
|
"step": 912384
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.505419418578674e-05,
|
|
"loss": 3.5974,
|
|
"step": 912896
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.504580823827622e-05,
|
|
"loss": 3.5954,
|
|
"step": 913408
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.50374222907657e-05,
|
|
"loss": 3.5948,
|
|
"step": 913920
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.502905272205891e-05,
|
|
"loss": 3.5992,
|
|
"step": 914432
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.502066677454839e-05,
|
|
"loss": 3.6069,
|
|
"step": 914944
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.501228082703787e-05,
|
|
"loss": 3.5968,
|
|
"step": 915456
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.843853712081909,
|
|
"eval_runtime": 338.0063,
|
|
"eval_samples_per_second": 1128.947,
|
|
"eval_steps_per_second": 35.28,
|
|
"step": 915840
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.500389487952735e-05,
|
|
"loss": 3.5913,
|
|
"step": 915968
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.499552531082056e-05,
|
|
"loss": 3.5841,
|
|
"step": 916480
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.498713936331004e-05,
|
|
"loss": 3.5953,
|
|
"step": 916992
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.497875341579952e-05,
|
|
"loss": 3.6001,
|
|
"step": 917504
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.497038384709273e-05,
|
|
"loss": 3.5998,
|
|
"step": 918016
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.496199789958221e-05,
|
|
"loss": 3.5939,
|
|
"step": 918528
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.4953611952071694e-05,
|
|
"loss": 3.5987,
|
|
"step": 919040
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.4945226004561174e-05,
|
|
"loss": 3.5922,
|
|
"step": 919552
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.493685643585438e-05,
|
|
"loss": 3.5861,
|
|
"step": 920064
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.492847048834386e-05,
|
|
"loss": 3.5833,
|
|
"step": 920576
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.492008454083334e-05,
|
|
"loss": 3.6006,
|
|
"step": 921088
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.491169859332282e-05,
|
|
"loss": 3.6042,
|
|
"step": 921600
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.490332902461603e-05,
|
|
"loss": 3.5973,
|
|
"step": 922112
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.489494307710551e-05,
|
|
"loss": 3.5739,
|
|
"step": 922624
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.488655712959499e-05,
|
|
"loss": 3.5934,
|
|
"step": 923136
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.487817118208447e-05,
|
|
"loss": 3.5739,
|
|
"step": 923648
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.486978523457395e-05,
|
|
"loss": 3.5916,
|
|
"step": 924160
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.486139928706343e-05,
|
|
"loss": 3.5874,
|
|
"step": 924672
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.485301333955291e-05,
|
|
"loss": 3.5886,
|
|
"step": 925184
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.484462739204239e-05,
|
|
"loss": 3.5932,
|
|
"step": 925696
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.483625782333561e-05,
|
|
"loss": 3.6033,
|
|
"step": 926208
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.482787187582509e-05,
|
|
"loss": 3.5891,
|
|
"step": 926720
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.481948592831457e-05,
|
|
"loss": 3.5926,
|
|
"step": 927232
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.481109998080405e-05,
|
|
"loss": 3.5911,
|
|
"step": 927744
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.480273041209726e-05,
|
|
"loss": 3.5959,
|
|
"step": 928256
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.4794344464586737e-05,
|
|
"loss": 3.5933,
|
|
"step": 928768
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.4785958517076217e-05,
|
|
"loss": 3.5851,
|
|
"step": 929280
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.4777572569565697e-05,
|
|
"loss": 3.5898,
|
|
"step": 929792
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.4769203000858906e-05,
|
|
"loss": 3.5844,
|
|
"step": 930304
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.4760817053348386e-05,
|
|
"loss": 3.5861,
|
|
"step": 930816
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4752431105837866e-05,
|
|
"loss": 3.588,
|
|
"step": 931328
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4744045158327345e-05,
|
|
"loss": 3.5933,
|
|
"step": 931840
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.473567558962056e-05,
|
|
"loss": 3.5903,
|
|
"step": 932352
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.472728964211004e-05,
|
|
"loss": 3.5984,
|
|
"step": 932864
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.471890369459952e-05,
|
|
"loss": 3.5892,
|
|
"step": 933376
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4710517747088994e-05,
|
|
"loss": 3.5879,
|
|
"step": 933888
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.470214817838221e-05,
|
|
"loss": 3.5992,
|
|
"step": 934400
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.469376223087169e-05,
|
|
"loss": 3.5814,
|
|
"step": 934912
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.468537628336117e-05,
|
|
"loss": 3.5801,
|
|
"step": 935424
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4676990335850643e-05,
|
|
"loss": 3.5838,
|
|
"step": 935936
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.466862076714386e-05,
|
|
"loss": 3.581,
|
|
"step": 936448
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.466023481963334e-05,
|
|
"loss": 3.5837,
|
|
"step": 936960
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.465184887212282e-05,
|
|
"loss": 3.5827,
|
|
"step": 937472
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.46434629246123e-05,
|
|
"loss": 3.5871,
|
|
"step": 937984
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4635093355905515e-05,
|
|
"loss": 3.5899,
|
|
"step": 938496
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4626707408394995e-05,
|
|
"loss": 3.5929,
|
|
"step": 939008
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.461832146088447e-05,
|
|
"loss": 3.5924,
|
|
"step": 939520
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.460993551337395e-05,
|
|
"loss": 3.5875,
|
|
"step": 940032
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4601565944667164e-05,
|
|
"loss": 3.5644,
|
|
"step": 940544
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4593179997156644e-05,
|
|
"loss": 3.5949,
|
|
"step": 941056
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.458479404964612e-05,
|
|
"loss": 3.5852,
|
|
"step": 941568
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.45764081021356e-05,
|
|
"loss": 3.5882,
|
|
"step": 942080
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.456803853342881e-05,
|
|
"loss": 3.5846,
|
|
"step": 942592
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.455965258591829e-05,
|
|
"loss": 3.582,
|
|
"step": 943104
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.455126663840777e-05,
|
|
"loss": 3.5708,
|
|
"step": 943616
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.454288069089725e-05,
|
|
"loss": 3.5797,
|
|
"step": 944128
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.453451112219047e-05,
|
|
"loss": 3.5846,
|
|
"step": 944640
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.452612517467994e-05,
|
|
"loss": 3.5721,
|
|
"step": 945152
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.451773922716942e-05,
|
|
"loss": 3.5969,
|
|
"step": 945664
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.45093532796589e-05,
|
|
"loss": 3.5886,
|
|
"step": 946176
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.450098371095212e-05,
|
|
"loss": 3.5679,
|
|
"step": 946688
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.449259776344159e-05,
|
|
"loss": 3.5992,
|
|
"step": 947200
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.448421181593107e-05,
|
|
"loss": 3.5755,
|
|
"step": 947712
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.447582586842055e-05,
|
|
"loss": 3.5734,
|
|
"step": 948224
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4467456299713767e-05,
|
|
"loss": 3.5885,
|
|
"step": 948736
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4459070352203247e-05,
|
|
"loss": 3.5819,
|
|
"step": 949248
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4450684404692727e-05,
|
|
"loss": 3.5708,
|
|
"step": 949760
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4442298457182206e-05,
|
|
"loss": 3.5828,
|
|
"step": 950272
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4433928888475416e-05,
|
|
"loss": 3.5696,
|
|
"step": 950784
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4425542940964896e-05,
|
|
"loss": 3.5758,
|
|
"step": 951296
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4417156993454375e-05,
|
|
"loss": 3.5887,
|
|
"step": 951808
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4408771045943855e-05,
|
|
"loss": 3.5808,
|
|
"step": 952320
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4400401477237065e-05,
|
|
"loss": 3.5889,
|
|
"step": 952832
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4392015529726544e-05,
|
|
"loss": 3.5808,
|
|
"step": 953344
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4383629582216024e-05,
|
|
"loss": 3.6007,
|
|
"step": 953856
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4375243634705504e-05,
|
|
"loss": 3.5753,
|
|
"step": 954368
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4366874065998713e-05,
|
|
"loss": 3.5905,
|
|
"step": 954880
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.4358488118488193e-05,
|
|
"loss": 3.5822,
|
|
"step": 955392
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.435010217097768e-05,
|
|
"loss": 3.5747,
|
|
"step": 955904
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.434171622346716e-05,
|
|
"loss": 3.5856,
|
|
"step": 956416
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.433334665476037e-05,
|
|
"loss": 3.5941,
|
|
"step": 956928
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.432496070724985e-05,
|
|
"loss": 3.5818,
|
|
"step": 957440
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.431657475973933e-05,
|
|
"loss": 3.5926,
|
|
"step": 957952
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.430818881222881e-05,
|
|
"loss": 3.5696,
|
|
"step": 958464
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.429981924352202e-05,
|
|
"loss": 3.5652,
|
|
"step": 958976
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.42914332960115e-05,
|
|
"loss": 3.5731,
|
|
"step": 959488
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.428304734850098e-05,
|
|
"loss": 3.5836,
|
|
"step": 960000
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.427466140099046e-05,
|
|
"loss": 3.5805,
|
|
"step": 960512
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.426629183228367e-05,
|
|
"loss": 3.5808,
|
|
"step": 961024
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.425790588477315e-05,
|
|
"loss": 3.5731,
|
|
"step": 961536
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.4249519937262634e-05,
|
|
"loss": 3.5795,
|
|
"step": 962048
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.4241133989752114e-05,
|
|
"loss": 3.5758,
|
|
"step": 962560
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.423276442104532e-05,
|
|
"loss": 3.5763,
|
|
"step": 963072
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.42243784735348e-05,
|
|
"loss": 3.5907,
|
|
"step": 963584
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.421599252602428e-05,
|
|
"loss": 3.5936,
|
|
"step": 964096
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.420762295731749e-05,
|
|
"loss": 3.5913,
|
|
"step": 964608
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.419923700980697e-05,
|
|
"loss": 3.5703,
|
|
"step": 965120
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.419085106229645e-05,
|
|
"loss": 3.5802,
|
|
"step": 965632
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.418246511478593e-05,
|
|
"loss": 3.5765,
|
|
"step": 966144
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.417407916727541e-05,
|
|
"loss": 3.5857,
|
|
"step": 966656
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.416569321976489e-05,
|
|
"loss": 3.5851,
|
|
"step": 967168
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.415730727225437e-05,
|
|
"loss": 3.5775,
|
|
"step": 967680
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.414892132474385e-05,
|
|
"loss": 3.5818,
|
|
"step": 968192
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.414055175603707e-05,
|
|
"loss": 3.5821,
|
|
"step": 968704
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.413216580852655e-05,
|
|
"loss": 3.5847,
|
|
"step": 969216
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.412377986101603e-05,
|
|
"loss": 3.5783,
|
|
"step": 969728
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.411539391350551e-05,
|
|
"loss": 3.5755,
|
|
"step": 970240
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.4107024344798716e-05,
|
|
"loss": 3.5721,
|
|
"step": 970752
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.4098638397288196e-05,
|
|
"loss": 3.5727,
|
|
"step": 971264
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.4090252449777676e-05,
|
|
"loss": 3.5722,
|
|
"step": 971776
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.4081866502267156e-05,
|
|
"loss": 3.5721,
|
|
"step": 972288
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.4073496933560365e-05,
|
|
"loss": 3.5749,
|
|
"step": 972800
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.4065110986049845e-05,
|
|
"loss": 3.582,
|
|
"step": 973312
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.4056725038539325e-05,
|
|
"loss": 3.5673,
|
|
"step": 973824
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.4048339091028805e-05,
|
|
"loss": 3.5874,
|
|
"step": 974336
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.403996952232202e-05,
|
|
"loss": 3.594,
|
|
"step": 974848
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.40315835748115e-05,
|
|
"loss": 3.5831,
|
|
"step": 975360
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.402319762730098e-05,
|
|
"loss": 3.5859,
|
|
"step": 975872
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.401482805859419e-05,
|
|
"loss": 3.5809,
|
|
"step": 976384
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.400644211108367e-05,
|
|
"loss": 3.5725,
|
|
"step": 976896
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.399805616357315e-05,
|
|
"loss": 3.5857,
|
|
"step": 977408
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.398967021606263e-05,
|
|
"loss": 3.5608,
|
|
"step": 977920
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.398130064735584e-05,
|
|
"loss": 3.5789,
|
|
"step": 978432
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.397291469984532e-05,
|
|
"loss": 3.5783,
|
|
"step": 978944
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.39645287523348e-05,
|
|
"loss": 3.5833,
|
|
"step": 979456
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.395614280482428e-05,
|
|
"loss": 3.5776,
|
|
"step": 979968
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.394775685731376e-05,
|
|
"loss": 3.584,
|
|
"step": 980480
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.3939387288606975e-05,
|
|
"loss": 3.5734,
|
|
"step": 980992
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.3931001341096455e-05,
|
|
"loss": 3.5741,
|
|
"step": 981504
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.3922615393585935e-05,
|
|
"loss": 3.5837,
|
|
"step": 982016
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.3914245824879144e-05,
|
|
"loss": 3.5821,
|
|
"step": 982528
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.3905859877368624e-05,
|
|
"loss": 3.5888,
|
|
"step": 983040
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.3897473929858104e-05,
|
|
"loss": 3.575,
|
|
"step": 983552
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.3889087982347584e-05,
|
|
"loss": 3.5903,
|
|
"step": 984064
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.388071841364079e-05,
|
|
"loss": 3.5796,
|
|
"step": 984576
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.387233246613027e-05,
|
|
"loss": 3.5816,
|
|
"step": 985088
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.386394651861975e-05,
|
|
"loss": 3.5846,
|
|
"step": 985600
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.385556057110923e-05,
|
|
"loss": 3.5769,
|
|
"step": 986112
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.384719100240245e-05,
|
|
"loss": 3.5749,
|
|
"step": 986624
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.383880505489193e-05,
|
|
"loss": 3.5924,
|
|
"step": 987136
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.383041910738141e-05,
|
|
"loss": 3.5704,
|
|
"step": 987648
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.382203315987089e-05,
|
|
"loss": 3.5749,
|
|
"step": 988160
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.38136635911641e-05,
|
|
"loss": 3.5851,
|
|
"step": 988672
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.380527764365358e-05,
|
|
"loss": 3.5812,
|
|
"step": 989184
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.379689169614306e-05,
|
|
"loss": 3.5786,
|
|
"step": 989696
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.378850574863254e-05,
|
|
"loss": 3.5787,
|
|
"step": 990208
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.3780136179925746e-05,
|
|
"loss": 3.5788,
|
|
"step": 990720
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.3771750232415226e-05,
|
|
"loss": 3.5862,
|
|
"step": 991232
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.3763364284904706e-05,
|
|
"loss": 3.5813,
|
|
"step": 991744
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 3.841752052307129,
|
|
"eval_runtime": 335.7692,
|
|
"eval_samples_per_second": 1136.468,
|
|
"eval_steps_per_second": 35.515,
|
|
"step": 992160
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3754978337394186e-05,
|
|
"loss": 3.5706,
|
|
"step": 992256
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3746592389883666e-05,
|
|
"loss": 3.567,
|
|
"step": 992768
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3738206442373146e-05,
|
|
"loss": 3.5729,
|
|
"step": 993280
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3729820494862626e-05,
|
|
"loss": 3.5895,
|
|
"step": 993792
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.372145092615584e-05,
|
|
"loss": 3.5791,
|
|
"step": 994304
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3713064978645315e-05,
|
|
"loss": 3.5772,
|
|
"step": 994816
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3704679031134795e-05,
|
|
"loss": 3.5772,
|
|
"step": 995328
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3696293083624275e-05,
|
|
"loss": 3.5769,
|
|
"step": 995840
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.368792351491749e-05,
|
|
"loss": 3.5659,
|
|
"step": 996352
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3679537567406964e-05,
|
|
"loss": 3.5683,
|
|
"step": 996864
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3671151619896444e-05,
|
|
"loss": 3.5785,
|
|
"step": 997376
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3662765672385924e-05,
|
|
"loss": 3.5857,
|
|
"step": 997888
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.365441248248287e-05,
|
|
"loss": 3.5872,
|
|
"step": 998400
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3646026534972356e-05,
|
|
"loss": 3.5546,
|
|
"step": 998912
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3637640587461836e-05,
|
|
"loss": 3.5723,
|
|
"step": 999424
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3629254639951316e-05,
|
|
"loss": 3.5594,
|
|
"step": 999936
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.362086869244079e-05,
|
|
"loss": 3.574,
|
|
"step": 1000448
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.361248274493027e-05,
|
|
"loss": 3.5691,
|
|
"step": 1000960
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.360409679741975e-05,
|
|
"loss": 3.5717,
|
|
"step": 1001472
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.359571084990923e-05,
|
|
"loss": 3.5706,
|
|
"step": 1001984
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.358734128120244e-05,
|
|
"loss": 3.5902,
|
|
"step": 1002496
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.357895533369192e-05,
|
|
"loss": 3.5698,
|
|
"step": 1003008
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.35705693861814e-05,
|
|
"loss": 3.5748,
|
|
"step": 1003520
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.356218343867088e-05,
|
|
"loss": 3.572,
|
|
"step": 1004032
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3553813869964094e-05,
|
|
"loss": 3.5784,
|
|
"step": 1004544
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3545427922453574e-05,
|
|
"loss": 3.5741,
|
|
"step": 1005056
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3537041974943054e-05,
|
|
"loss": 3.5678,
|
|
"step": 1005568
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.3528656027432534e-05,
|
|
"loss": 3.5745,
|
|
"step": 1006080
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.352028645872574e-05,
|
|
"loss": 3.5665,
|
|
"step": 1006592
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.351190051121522e-05,
|
|
"loss": 3.5655,
|
|
"step": 1007104
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.35035145637047e-05,
|
|
"loss": 3.5714,
|
|
"step": 1007616
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.349512861619418e-05,
|
|
"loss": 3.5742,
|
|
"step": 1008128
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.348675904748739e-05,
|
|
"loss": 3.5772,
|
|
"step": 1008640
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.347837309997687e-05,
|
|
"loss": 3.5777,
|
|
"step": 1009152
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.346998715246635e-05,
|
|
"loss": 3.571,
|
|
"step": 1009664
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.346160120495583e-05,
|
|
"loss": 3.5725,
|
|
"step": 1010176
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.345323163624904e-05,
|
|
"loss": 3.5833,
|
|
"step": 1010688
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.344484568873853e-05,
|
|
"loss": 3.5595,
|
|
"step": 1011200
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.343645974122801e-05,
|
|
"loss": 3.5654,
|
|
"step": 1011712
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.342807379371749e-05,
|
|
"loss": 3.5706,
|
|
"step": 1012224
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3419704225010696e-05,
|
|
"loss": 3.5591,
|
|
"step": 1012736
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3411318277500176e-05,
|
|
"loss": 3.5703,
|
|
"step": 1013248
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3402932329989656e-05,
|
|
"loss": 3.5695,
|
|
"step": 1013760
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3394546382479136e-05,
|
|
"loss": 3.5682,
|
|
"step": 1014272
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3386176813772345e-05,
|
|
"loss": 3.5724,
|
|
"step": 1014784
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3377790866261825e-05,
|
|
"loss": 3.5734,
|
|
"step": 1015296
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3369404918751305e-05,
|
|
"loss": 3.5752,
|
|
"step": 1015808
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3361018971240785e-05,
|
|
"loss": 3.5685,
|
|
"step": 1016320
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3352649402533994e-05,
|
|
"loss": 3.5524,
|
|
"step": 1016832
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.334426345502348e-05,
|
|
"loss": 3.5737,
|
|
"step": 1017344
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.333587750751296e-05,
|
|
"loss": 3.5694,
|
|
"step": 1017856
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.332749156000244e-05,
|
|
"loss": 3.5723,
|
|
"step": 1018368
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.331912199129565e-05,
|
|
"loss": 3.5696,
|
|
"step": 1018880
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.331073604378513e-05,
|
|
"loss": 3.5602,
|
|
"step": 1019392
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.330235009627461e-05,
|
|
"loss": 3.5545,
|
|
"step": 1019904
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.329396414876409e-05,
|
|
"loss": 3.5599,
|
|
"step": 1020416
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3285610958861035e-05,
|
|
"loss": 3.5692,
|
|
"step": 1020928
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.327722501135051e-05,
|
|
"loss": 3.5549,
|
|
"step": 1021440
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.326883906383999e-05,
|
|
"loss": 3.5822,
|
|
"step": 1021952
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.326045311632947e-05,
|
|
"loss": 3.5708,
|
|
"step": 1022464
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.325206716881895e-05,
|
|
"loss": 3.5491,
|
|
"step": 1022976
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3243681221308435e-05,
|
|
"loss": 3.5778,
|
|
"step": 1023488
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3235295273797915e-05,
|
|
"loss": 3.5569,
|
|
"step": 1024000
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3226925705091124e-05,
|
|
"loss": 3.5599,
|
|
"step": 1024512
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3218539757580604e-05,
|
|
"loss": 3.5652,
|
|
"step": 1025024
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3210153810070084e-05,
|
|
"loss": 3.5712,
|
|
"step": 1025536
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3201767862559564e-05,
|
|
"loss": 3.5548,
|
|
"step": 1026048
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3193381915049043e-05,
|
|
"loss": 3.5681,
|
|
"step": 1026560
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.318501234634225e-05,
|
|
"loss": 3.5482,
|
|
"step": 1027072
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.317662639883173e-05,
|
|
"loss": 3.5583,
|
|
"step": 1027584
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.316824045132121e-05,
|
|
"loss": 3.5752,
|
|
"step": 1028096
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.315987088261442e-05,
|
|
"loss": 3.5626,
|
|
"step": 1028608
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.31514849351039e-05,
|
|
"loss": 3.5734,
|
|
"step": 1029120
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.314309898759339e-05,
|
|
"loss": 3.5641,
|
|
"step": 1029632
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.313471304008287e-05,
|
|
"loss": 3.5835,
|
|
"step": 1030144
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.312634347137608e-05,
|
|
"loss": 3.5563,
|
|
"step": 1030656
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.311795752386556e-05,
|
|
"loss": 3.5752,
|
|
"step": 1031168
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.310957157635504e-05,
|
|
"loss": 3.561,
|
|
"step": 1031680
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.310118562884452e-05,
|
|
"loss": 3.5621,
|
|
"step": 1032192
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3092816060137726e-05,
|
|
"loss": 3.5684,
|
|
"step": 1032704
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3084430112627206e-05,
|
|
"loss": 3.5755,
|
|
"step": 1033216
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3076044165116686e-05,
|
|
"loss": 3.5662,
|
|
"step": 1033728
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3067658217606166e-05,
|
|
"loss": 3.5753,
|
|
"step": 1034240
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3059288648899375e-05,
|
|
"loss": 3.5516,
|
|
"step": 1034752
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.3050902701388855e-05,
|
|
"loss": 3.5502,
|
|
"step": 1035264
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.304251675387834e-05,
|
|
"loss": 3.5563,
|
|
"step": 1035776
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.303413080636782e-05,
|
|
"loss": 3.5644,
|
|
"step": 1036288
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.302576123766103e-05,
|
|
"loss": 3.5612,
|
|
"step": 1036800
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.301737529015051e-05,
|
|
"loss": 3.5637,
|
|
"step": 1037312
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.300898934263999e-05,
|
|
"loss": 3.5623,
|
|
"step": 1037824
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.300060339512947e-05,
|
|
"loss": 3.5587,
|
|
"step": 1038336
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.299223382642268e-05,
|
|
"loss": 3.561,
|
|
"step": 1038848
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.298384787891216e-05,
|
|
"loss": 3.5574,
|
|
"step": 1039360
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.297546193140164e-05,
|
|
"loss": 3.5718,
|
|
"step": 1039872
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.296707598389112e-05,
|
|
"loss": 3.5776,
|
|
"step": 1040384
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.29586900363806e-05,
|
|
"loss": 3.5725,
|
|
"step": 1040896
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.295030408887008e-05,
|
|
"loss": 3.5537,
|
|
"step": 1041408
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.294191814135956e-05,
|
|
"loss": 3.5607,
|
|
"step": 1041920
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.293353219384904e-05,
|
|
"loss": 3.5617,
|
|
"step": 1042432
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2925162625142256e-05,
|
|
"loss": 3.5704,
|
|
"step": 1042944
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2916776677631735e-05,
|
|
"loss": 3.5667,
|
|
"step": 1043456
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2908390730121215e-05,
|
|
"loss": 3.5612,
|
|
"step": 1043968
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2900004782610695e-05,
|
|
"loss": 3.5634,
|
|
"step": 1044480
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2891635213903904e-05,
|
|
"loss": 3.5691,
|
|
"step": 1044992
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2883249266393384e-05,
|
|
"loss": 3.5653,
|
|
"step": 1045504
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2874863318882864e-05,
|
|
"loss": 3.5633,
|
|
"step": 1046016
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2866493750176073e-05,
|
|
"loss": 3.557,
|
|
"step": 1046528
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2858107802665553e-05,
|
|
"loss": 3.5549,
|
|
"step": 1047040
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2849721855155033e-05,
|
|
"loss": 3.5622,
|
|
"step": 1047552
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.284133590764451e-05,
|
|
"loss": 3.5498,
|
|
"step": 1048064
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.283296633893773e-05,
|
|
"loss": 3.5565,
|
|
"step": 1048576
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.282458039142721e-05,
|
|
"loss": 3.5569,
|
|
"step": 1049088
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.281619444391669e-05,
|
|
"loss": 3.5638,
|
|
"step": 1049600
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.280780849640617e-05,
|
|
"loss": 3.5563,
|
|
"step": 1050112
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.279943892769938e-05,
|
|
"loss": 3.566,
|
|
"step": 1050624
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.279106935899259e-05,
|
|
"loss": 3.5741,
|
|
"step": 1051136
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.278268341148207e-05,
|
|
"loss": 3.5693,
|
|
"step": 1051648
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.277429746397155e-05,
|
|
"loss": 3.5708,
|
|
"step": 1052160
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.276591151646103e-05,
|
|
"loss": 3.5585,
|
|
"step": 1052672
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.275752556895051e-05,
|
|
"loss": 3.5615,
|
|
"step": 1053184
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.274913962143999e-05,
|
|
"loss": 3.5653,
|
|
"step": 1053696
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.274075367392947e-05,
|
|
"loss": 3.5463,
|
|
"step": 1054208
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.273238410522268e-05,
|
|
"loss": 3.5636,
|
|
"step": 1054720
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.272399815771216e-05,
|
|
"loss": 3.5552,
|
|
"step": 1055232
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.271561221020164e-05,
|
|
"loss": 3.5688,
|
|
"step": 1055744
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2707226262691116e-05,
|
|
"loss": 3.5659,
|
|
"step": 1056256
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.269885669398433e-05,
|
|
"loss": 3.5652,
|
|
"step": 1056768
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.269047074647381e-05,
|
|
"loss": 3.5534,
|
|
"step": 1057280
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2682084798963285e-05,
|
|
"loss": 3.5613,
|
|
"step": 1057792
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2673698851452765e-05,
|
|
"loss": 3.5682,
|
|
"step": 1058304
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.266532928274598e-05,
|
|
"loss": 3.5657,
|
|
"step": 1058816
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.265694333523546e-05,
|
|
"loss": 3.572,
|
|
"step": 1059328
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2648557387724934e-05,
|
|
"loss": 3.5561,
|
|
"step": 1059840
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.264017144021442e-05,
|
|
"loss": 3.5718,
|
|
"step": 1060352
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2631801871507637e-05,
|
|
"loss": 3.5647,
|
|
"step": 1060864
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2623415923997117e-05,
|
|
"loss": 3.5644,
|
|
"step": 1061376
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.261502997648659e-05,
|
|
"loss": 3.5681,
|
|
"step": 1061888
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.260664402897607e-05,
|
|
"loss": 3.5608,
|
|
"step": 1062400
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2598274460269286e-05,
|
|
"loss": 3.5605,
|
|
"step": 1062912
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.258988851275876e-05,
|
|
"loss": 3.574,
|
|
"step": 1063424
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.258150256524824e-05,
|
|
"loss": 3.5585,
|
|
"step": 1063936
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.257311661773772e-05,
|
|
"loss": 3.5549,
|
|
"step": 1064448
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2564747049030934e-05,
|
|
"loss": 3.5699,
|
|
"step": 1064960
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.255636110152041e-05,
|
|
"loss": 3.567,
|
|
"step": 1065472
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.254797515400989e-05,
|
|
"loss": 3.5597,
|
|
"step": 1065984
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2539589206499374e-05,
|
|
"loss": 3.564,
|
|
"step": 1066496
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.253121963779259e-05,
|
|
"loss": 3.5566,
|
|
"step": 1067008
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.2522833690282063e-05,
|
|
"loss": 3.5745,
|
|
"step": 1067520
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.251444774277154e-05,
|
|
"loss": 3.5607,
|
|
"step": 1068032
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.842722177505493,
|
|
"eval_runtime": 306.0136,
|
|
"eval_samples_per_second": 1246.974,
|
|
"eval_steps_per_second": 38.969,
|
|
"step": 1068480
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.250606179526102e-05,
|
|
"loss": 3.5598,
|
|
"step": 1068544
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.249769222655423e-05,
|
|
"loss": 3.5513,
|
|
"step": 1069056
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.248930627904371e-05,
|
|
"loss": 3.5554,
|
|
"step": 1069568
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.248092033153319e-05,
|
|
"loss": 3.5739,
|
|
"step": 1070080
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.247253438402267e-05,
|
|
"loss": 3.5648,
|
|
"step": 1070592
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.246416481531588e-05,
|
|
"loss": 3.5612,
|
|
"step": 1071104
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.245577886780536e-05,
|
|
"loss": 3.5583,
|
|
"step": 1071616
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.244739292029484e-05,
|
|
"loss": 3.5585,
|
|
"step": 1072128
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.243900697278433e-05,
|
|
"loss": 3.5476,
|
|
"step": 1072640
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.243063740407754e-05,
|
|
"loss": 3.559,
|
|
"step": 1073152
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.242225145656702e-05,
|
|
"loss": 3.5596,
|
|
"step": 1073664
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.24138655090565e-05,
|
|
"loss": 3.5741,
|
|
"step": 1074176
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.2405495940349706e-05,
|
|
"loss": 3.5688,
|
|
"step": 1074688
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.2397109992839186e-05,
|
|
"loss": 3.5372,
|
|
"step": 1075200
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.2388724045328666e-05,
|
|
"loss": 3.5572,
|
|
"step": 1075712
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.2380338097818146e-05,
|
|
"loss": 3.5424,
|
|
"step": 1076224
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.2371952150307626e-05,
|
|
"loss": 3.5573,
|
|
"step": 1076736
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.2363566202797106e-05,
|
|
"loss": 3.5525,
|
|
"step": 1077248
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.2355180255286586e-05,
|
|
"loss": 3.5604,
|
|
"step": 1077760
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.2346794307776066e-05,
|
|
"loss": 3.552,
|
|
"step": 1078272
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.233842473906928e-05,
|
|
"loss": 3.5728,
|
|
"step": 1078784
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.233003879155876e-05,
|
|
"loss": 3.5563,
|
|
"step": 1079296
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.232165284404824e-05,
|
|
"loss": 3.5596,
|
|
"step": 1079808
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.231326689653772e-05,
|
|
"loss": 3.5541,
|
|
"step": 1080320
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.23048809490272e-05,
|
|
"loss": 3.5697,
|
|
"step": 1080832
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.229649500151668e-05,
|
|
"loss": 3.5543,
|
|
"step": 1081344
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.228810905400616e-05,
|
|
"loss": 3.5521,
|
|
"step": 1081856
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.227972310649564e-05,
|
|
"loss": 3.5581,
|
|
"step": 1082368
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.227135353778885e-05,
|
|
"loss": 3.5533,
|
|
"step": 1082880
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 3.226296759027833e-05,
|
|
"loss": 3.5488,
|
|
"step": 1083392
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.225458164276781e-05,
|
|
"loss": 3.5544,
|
|
"step": 1083904
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.224619569525729e-05,
|
|
"loss": 3.5556,
|
|
"step": 1084416
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.22378261265505e-05,
|
|
"loss": 3.561,
|
|
"step": 1084928
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.222944017903998e-05,
|
|
"loss": 3.5591,
|
|
"step": 1085440
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.2221054231529466e-05,
|
|
"loss": 3.5582,
|
|
"step": 1085952
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.2212668284018946e-05,
|
|
"loss": 3.5616,
|
|
"step": 1086464
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.2204298715312155e-05,
|
|
"loss": 3.5624,
|
|
"step": 1086976
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.2195912767801635e-05,
|
|
"loss": 3.5464,
|
|
"step": 1087488
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.2187526820291115e-05,
|
|
"loss": 3.5524,
|
|
"step": 1088000
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.2179140872780595e-05,
|
|
"loss": 3.5525,
|
|
"step": 1088512
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.217078768287753e-05,
|
|
"loss": 3.5424,
|
|
"step": 1089024
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.216240173536701e-05,
|
|
"loss": 3.5548,
|
|
"step": 1089536
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.215401578785649e-05,
|
|
"loss": 3.558,
|
|
"step": 1090048
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.214562984034597e-05,
|
|
"loss": 3.5485,
|
|
"step": 1090560
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.213724389283545e-05,
|
|
"loss": 3.5595,
|
|
"step": 1091072
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.212885794532493e-05,
|
|
"loss": 3.5577,
|
|
"step": 1091584
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.212047199781442e-05,
|
|
"loss": 3.5602,
|
|
"step": 1092096
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.211208605030389e-05,
|
|
"loss": 3.5538,
|
|
"step": 1092608
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.210371648159711e-05,
|
|
"loss": 3.537,
|
|
"step": 1093120
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.209533053408659e-05,
|
|
"loss": 3.5587,
|
|
"step": 1093632
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.208694458657607e-05,
|
|
"loss": 3.5563,
|
|
"step": 1094144
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.207855863906554e-05,
|
|
"loss": 3.5577,
|
|
"step": 1094656
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.207018907035876e-05,
|
|
"loss": 3.55,
|
|
"step": 1095168
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.206180312284824e-05,
|
|
"loss": 3.5463,
|
|
"step": 1095680
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.205343355414145e-05,
|
|
"loss": 3.5386,
|
|
"step": 1096192
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.204504760663093e-05,
|
|
"loss": 3.544,
|
|
"step": 1096704
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.203666165912041e-05,
|
|
"loss": 3.5553,
|
|
"step": 1097216
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.202827571160989e-05,
|
|
"loss": 3.5373,
|
|
"step": 1097728
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.201988976409937e-05,
|
|
"loss": 3.5656,
|
|
"step": 1098240
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.201150381658885e-05,
|
|
"loss": 3.5582,
|
|
"step": 1098752
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.2003117869078327e-05,
|
|
"loss": 3.5357,
|
|
"step": 1099264
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.199474830037154e-05,
|
|
"loss": 3.562,
|
|
"step": 1099776
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.1986362352861016e-05,
|
|
"loss": 3.542,
|
|
"step": 1100288
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.1977976405350496e-05,
|
|
"loss": 3.5481,
|
|
"step": 1100800
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.1969590457839976e-05,
|
|
"loss": 3.5446,
|
|
"step": 1101312
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.196122088913319e-05,
|
|
"loss": 3.5535,
|
|
"step": 1101824
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.1952834941622665e-05,
|
|
"loss": 3.5393,
|
|
"step": 1102336
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.194446537291588e-05,
|
|
"loss": 3.5539,
|
|
"step": 1102848
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.193607942540536e-05,
|
|
"loss": 3.5314,
|
|
"step": 1103360
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.192769347789484e-05,
|
|
"loss": 3.5451,
|
|
"step": 1103872
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.191930753038432e-05,
|
|
"loss": 3.555,
|
|
"step": 1104384
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.19109215828738e-05,
|
|
"loss": 3.5471,
|
|
"step": 1104896
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.190253563536328e-05,
|
|
"loss": 3.5558,
|
|
"step": 1105408
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.189414968785276e-05,
|
|
"loss": 3.5479,
|
|
"step": 1105920
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.188578011914597e-05,
|
|
"loss": 3.5662,
|
|
"step": 1106432
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.187739417163545e-05,
|
|
"loss": 3.545,
|
|
"step": 1106944
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.1869024602928665e-05,
|
|
"loss": 3.5614,
|
|
"step": 1107456
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.186063865541814e-05,
|
|
"loss": 3.5446,
|
|
"step": 1107968
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.185225270790762e-05,
|
|
"loss": 3.5464,
|
|
"step": 1108480
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.18438667603971e-05,
|
|
"loss": 3.5536,
|
|
"step": 1108992
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.1835497191690314e-05,
|
|
"loss": 3.5562,
|
|
"step": 1109504
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.1827111244179794e-05,
|
|
"loss": 3.5525,
|
|
"step": 1110016
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.1818725296669274e-05,
|
|
"loss": 3.5602,
|
|
"step": 1110528
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.1810339349158754e-05,
|
|
"loss": 3.5381,
|
|
"step": 1111040
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.1801953401648234e-05,
|
|
"loss": 3.5316,
|
|
"step": 1111552
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.1793567454137714e-05,
|
|
"loss": 3.5452,
|
|
"step": 1112064
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.1785181506627194e-05,
|
|
"loss": 3.5482,
|
|
"step": 1112576
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.1776795559116674e-05,
|
|
"loss": 3.5451,
|
|
"step": 1113088
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.176842599040988e-05,
|
|
"loss": 3.5443,
|
|
"step": 1113600
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 3.176004004289936e-05,
|
|
"loss": 3.5489,
|
|
"step": 1114112
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.175165409538884e-05,
|
|
"loss": 3.5418,
|
|
"step": 1114624
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.174328452668205e-05,
|
|
"loss": 3.5445,
|
|
"step": 1115136
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.173489857917153e-05,
|
|
"loss": 3.5427,
|
|
"step": 1115648
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.172651263166101e-05,
|
|
"loss": 3.5553,
|
|
"step": 1116160
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.17181266841505e-05,
|
|
"loss": 3.5609,
|
|
"step": 1116672
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.170975711544371e-05,
|
|
"loss": 3.5594,
|
|
"step": 1117184
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.170137116793319e-05,
|
|
"loss": 3.5391,
|
|
"step": 1117696
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.169298522042267e-05,
|
|
"loss": 3.5447,
|
|
"step": 1118208
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.168459927291215e-05,
|
|
"loss": 3.5451,
|
|
"step": 1118720
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.167621332540163e-05,
|
|
"loss": 3.5523,
|
|
"step": 1119232
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.166782737789111e-05,
|
|
"loss": 3.5511,
|
|
"step": 1119744
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.165944143038059e-05,
|
|
"loss": 3.5467,
|
|
"step": 1120256
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.165105548287007e-05,
|
|
"loss": 3.5469,
|
|
"step": 1120768
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1642685914163276e-05,
|
|
"loss": 3.5507,
|
|
"step": 1121280
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1634299966652756e-05,
|
|
"loss": 3.5514,
|
|
"step": 1121792
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1625914019142236e-05,
|
|
"loss": 3.5446,
|
|
"step": 1122304
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1617528071631716e-05,
|
|
"loss": 3.5472,
|
|
"step": 1122816
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.160915850292493e-05,
|
|
"loss": 3.5354,
|
|
"step": 1123328
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.160077255541441e-05,
|
|
"loss": 3.546,
|
|
"step": 1123840
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.159238660790389e-05,
|
|
"loss": 3.5351,
|
|
"step": 1124352
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.158400066039337e-05,
|
|
"loss": 3.5403,
|
|
"step": 1124864
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.157563109168658e-05,
|
|
"loss": 3.5412,
|
|
"step": 1125376
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.156724514417606e-05,
|
|
"loss": 3.5499,
|
|
"step": 1125888
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.155885919666554e-05,
|
|
"loss": 3.5407,
|
|
"step": 1126400
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.155047324915502e-05,
|
|
"loss": 3.5478,
|
|
"step": 1126912
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.154212005925196e-05,
|
|
"loss": 3.5587,
|
|
"step": 1127424
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.153373411174144e-05,
|
|
"loss": 3.5571,
|
|
"step": 1127936
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.152534816423092e-05,
|
|
"loss": 3.5535,
|
|
"step": 1128448
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1516962216720406e-05,
|
|
"loss": 3.5444,
|
|
"step": 1128960
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1508576269209886e-05,
|
|
"loss": 3.547,
|
|
"step": 1129472
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1500190321699366e-05,
|
|
"loss": 3.5454,
|
|
"step": 1129984
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1491804374188846e-05,
|
|
"loss": 3.5312,
|
|
"step": 1130496
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1483434805482055e-05,
|
|
"loss": 3.5505,
|
|
"step": 1131008
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1475048857971535e-05,
|
|
"loss": 3.541,
|
|
"step": 1131520
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1466662910461015e-05,
|
|
"loss": 3.5551,
|
|
"step": 1132032
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1458276962950495e-05,
|
|
"loss": 3.5451,
|
|
"step": 1132544
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1449907394243704e-05,
|
|
"loss": 3.5507,
|
|
"step": 1133056
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1441521446733184e-05,
|
|
"loss": 3.5413,
|
|
"step": 1133568
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.143315187802639e-05,
|
|
"loss": 3.5443,
|
|
"step": 1134080
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.142476593051587e-05,
|
|
"loss": 3.552,
|
|
"step": 1134592
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.141637998300536e-05,
|
|
"loss": 3.5491,
|
|
"step": 1135104
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.140799403549484e-05,
|
|
"loss": 3.5591,
|
|
"step": 1135616
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.139960808798432e-05,
|
|
"loss": 3.5434,
|
|
"step": 1136128
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.13912221404738e-05,
|
|
"loss": 3.5549,
|
|
"step": 1136640
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.138283619296327e-05,
|
|
"loss": 3.5498,
|
|
"step": 1137152
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.137445024545275e-05,
|
|
"loss": 3.5489,
|
|
"step": 1137664
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.136608067674597e-05,
|
|
"loss": 3.5479,
|
|
"step": 1138176
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.135769472923544e-05,
|
|
"loss": 3.5435,
|
|
"step": 1138688
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.134930878172492e-05,
|
|
"loss": 3.5514,
|
|
"step": 1139200
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.13409228342144e-05,
|
|
"loss": 3.5573,
|
|
"step": 1139712
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.133255326550762e-05,
|
|
"loss": 3.5437,
|
|
"step": 1140224
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1324183696800826e-05,
|
|
"loss": 3.5407,
|
|
"step": 1140736
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.131579774929031e-05,
|
|
"loss": 3.5507,
|
|
"step": 1141248
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.130741180177979e-05,
|
|
"loss": 3.5507,
|
|
"step": 1141760
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.129902585426927e-05,
|
|
"loss": 3.5446,
|
|
"step": 1142272
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1290639906758746e-05,
|
|
"loss": 3.5496,
|
|
"step": 1142784
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1282253959248226e-05,
|
|
"loss": 3.5433,
|
|
"step": 1143296
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1273868011737706e-05,
|
|
"loss": 3.5628,
|
|
"step": 1143808
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 3.1265482064227186e-05,
|
|
"loss": 3.5409,
|
|
"step": 1144320
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 3.8433427810668945,
|
|
"eval_runtime": 301.8981,
|
|
"eval_samples_per_second": 1263.973,
|
|
"eval_steps_per_second": 39.5,
|
|
"step": 1144800
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.1257112495520395e-05,
|
|
"loss": 3.5059,
|
|
"step": 1144832
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.124874292681361e-05,
|
|
"loss": 3.5391,
|
|
"step": 1145344
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.124035697930309e-05,
|
|
"loss": 3.5384,
|
|
"step": 1145856
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.1231971031792564e-05,
|
|
"loss": 3.5574,
|
|
"step": 1146368
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.122358508428205e-05,
|
|
"loss": 3.5501,
|
|
"step": 1146880
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.121519913677153e-05,
|
|
"loss": 3.5501,
|
|
"step": 1147392
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.120681318926101e-05,
|
|
"loss": 3.5418,
|
|
"step": 1147904
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.119842724175049e-05,
|
|
"loss": 3.5447,
|
|
"step": 1148416
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.119004129423997e-05,
|
|
"loss": 3.5357,
|
|
"step": 1148928
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.118167172553318e-05,
|
|
"loss": 3.5421,
|
|
"step": 1149440
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.117328577802266e-05,
|
|
"loss": 3.542,
|
|
"step": 1149952
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.116489983051214e-05,
|
|
"loss": 3.5591,
|
|
"step": 1150464
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.115653026180535e-05,
|
|
"loss": 3.5528,
|
|
"step": 1150976
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.114814431429483e-05,
|
|
"loss": 3.528,
|
|
"step": 1151488
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.113975836678431e-05,
|
|
"loss": 3.5409,
|
|
"step": 1152000
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.113137241927379e-05,
|
|
"loss": 3.5287,
|
|
"step": 1152512
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.112298647176327e-05,
|
|
"loss": 3.541,
|
|
"step": 1153024
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.111460052425275e-05,
|
|
"loss": 3.5381,
|
|
"step": 1153536
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.110621457674223e-05,
|
|
"loss": 3.5448,
|
|
"step": 1154048
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.1097828629231715e-05,
|
|
"loss": 3.5365,
|
|
"step": 1154560
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.1089459060524924e-05,
|
|
"loss": 3.5582,
|
|
"step": 1155072
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.1081073113014404e-05,
|
|
"loss": 3.5429,
|
|
"step": 1155584
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.1072687165503884e-05,
|
|
"loss": 3.5394,
|
|
"step": 1156096
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.1064301217993364e-05,
|
|
"loss": 3.5391,
|
|
"step": 1156608
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.1055915270482844e-05,
|
|
"loss": 3.5555,
|
|
"step": 1157120
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.1047529322972324e-05,
|
|
"loss": 3.5379,
|
|
"step": 1157632
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.1039143375461804e-05,
|
|
"loss": 3.5337,
|
|
"step": 1158144
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.103075742795128e-05,
|
|
"loss": 3.545,
|
|
"step": 1158656
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.102238785924449e-05,
|
|
"loss": 3.5356,
|
|
"step": 1159168
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 3.101400191173397e-05,
|
|
"loss": 3.5326,
|
|
"step": 1159680
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.100561596422345e-05,
|
|
"loss": 3.5405,
|
|
"step": 1160192
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.099723001671293e-05,
|
|
"loss": 3.5407,
|
|
"step": 1160704
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.098886044800615e-05,
|
|
"loss": 3.546,
|
|
"step": 1161216
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.098047450049563e-05,
|
|
"loss": 3.5424,
|
|
"step": 1161728
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.09720885529851e-05,
|
|
"loss": 3.5459,
|
|
"step": 1162240
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.096370260547458e-05,
|
|
"loss": 3.5428,
|
|
"step": 1162752
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.09553330367678e-05,
|
|
"loss": 3.5475,
|
|
"step": 1163264
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.094694708925728e-05,
|
|
"loss": 3.532,
|
|
"step": 1163776
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.093856114174675e-05,
|
|
"loss": 3.5387,
|
|
"step": 1164288
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.093017519423623e-05,
|
|
"loss": 3.5339,
|
|
"step": 1164800
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.0921822004333176e-05,
|
|
"loss": 3.5316,
|
|
"step": 1165312
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.0913436056822656e-05,
|
|
"loss": 3.5363,
|
|
"step": 1165824
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.0905050109312136e-05,
|
|
"loss": 3.5427,
|
|
"step": 1166336
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.089666416180162e-05,
|
|
"loss": 3.5338,
|
|
"step": 1166848
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.08882782142911e-05,
|
|
"loss": 3.5421,
|
|
"step": 1167360
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.0879892266780576e-05,
|
|
"loss": 3.5448,
|
|
"step": 1167872
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.0871506319270056e-05,
|
|
"loss": 3.5457,
|
|
"step": 1168384
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.0863120371759536e-05,
|
|
"loss": 3.5383,
|
|
"step": 1168896
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.085475080305275e-05,
|
|
"loss": 3.5213,
|
|
"step": 1169408
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.0846364855542225e-05,
|
|
"loss": 3.5405,
|
|
"step": 1169920
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.0837978908031705e-05,
|
|
"loss": 3.5442,
|
|
"step": 1170432
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.0829592960521185e-05,
|
|
"loss": 3.5427,
|
|
"step": 1170944
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.08212233918144e-05,
|
|
"loss": 3.5337,
|
|
"step": 1171456
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.0812837444303874e-05,
|
|
"loss": 3.5323,
|
|
"step": 1171968
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.080446787559709e-05,
|
|
"loss": 3.5246,
|
|
"step": 1172480
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.0796081928086576e-05,
|
|
"loss": 3.5254,
|
|
"step": 1172992
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.078769598057605e-05,
|
|
"loss": 3.5424,
|
|
"step": 1173504
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.077931003306553e-05,
|
|
"loss": 3.5252,
|
|
"step": 1174016
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.077092408555501e-05,
|
|
"loss": 3.5464,
|
|
"step": 1174528
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.076253813804449e-05,
|
|
"loss": 3.5382,
|
|
"step": 1175040
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.075415219053397e-05,
|
|
"loss": 3.5273,
|
|
"step": 1175552
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.074578262182718e-05,
|
|
"loss": 3.5459,
|
|
"step": 1176064
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.073739667431666e-05,
|
|
"loss": 3.526,
|
|
"step": 1176576
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.072901072680614e-05,
|
|
"loss": 3.5294,
|
|
"step": 1177088
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.072062477929562e-05,
|
|
"loss": 3.5307,
|
|
"step": 1177600
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.071225521058883e-05,
|
|
"loss": 3.543,
|
|
"step": 1178112
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.0703869263078314e-05,
|
|
"loss": 3.5227,
|
|
"step": 1178624
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.069549969437152e-05,
|
|
"loss": 3.5366,
|
|
"step": 1179136
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.0687113746861e-05,
|
|
"loss": 3.5176,
|
|
"step": 1179648
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.067872779935048e-05,
|
|
"loss": 3.5287,
|
|
"step": 1180160
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.067034185183996e-05,
|
|
"loss": 3.5437,
|
|
"step": 1180672
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.066195590432944e-05,
|
|
"loss": 3.5357,
|
|
"step": 1181184
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.065356995681892e-05,
|
|
"loss": 3.5378,
|
|
"step": 1181696
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.06451840093084e-05,
|
|
"loss": 3.5364,
|
|
"step": 1182208
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.063681444060161e-05,
|
|
"loss": 3.5499,
|
|
"step": 1182720
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.062842849309109e-05,
|
|
"loss": 3.5331,
|
|
"step": 1183232
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.062004254558057e-05,
|
|
"loss": 3.5481,
|
|
"step": 1183744
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.061165659807005e-05,
|
|
"loss": 3.5259,
|
|
"step": 1184256
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.060328702936327e-05,
|
|
"loss": 3.5324,
|
|
"step": 1184768
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.059491746065648e-05,
|
|
"loss": 3.5358,
|
|
"step": 1185280
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.058653151314596e-05,
|
|
"loss": 3.542,
|
|
"step": 1185792
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.057814556563544e-05,
|
|
"loss": 3.5361,
|
|
"step": 1186304
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.056975961812492e-05,
|
|
"loss": 3.5476,
|
|
"step": 1186816
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.05613736706144e-05,
|
|
"loss": 3.527,
|
|
"step": 1187328
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.055298772310388e-05,
|
|
"loss": 3.5192,
|
|
"step": 1187840
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.054460177559336e-05,
|
|
"loss": 3.5271,
|
|
"step": 1188352
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.053621582808284e-05,
|
|
"loss": 3.5348,
|
|
"step": 1188864
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.0527846259376046e-05,
|
|
"loss": 3.5286,
|
|
"step": 1189376
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.0519460311865526e-05,
|
|
"loss": 3.531,
|
|
"step": 1189888
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 3.051107436435501e-05,
|
|
"loss": 3.5365,
|
|
"step": 1190400
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.050268841684449e-05,
|
|
"loss": 3.5279,
|
|
"step": 1190912
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0494335226941427e-05,
|
|
"loss": 3.528,
|
|
"step": 1191424
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0485949279430907e-05,
|
|
"loss": 3.5287,
|
|
"step": 1191936
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.047756333192039e-05,
|
|
"loss": 3.5423,
|
|
"step": 1192448
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.046917738440987e-05,
|
|
"loss": 3.5464,
|
|
"step": 1192960
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.046080781570308e-05,
|
|
"loss": 3.5411,
|
|
"step": 1193472
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.045242186819256e-05,
|
|
"loss": 3.5252,
|
|
"step": 1193984
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.044403592068204e-05,
|
|
"loss": 3.5342,
|
|
"step": 1194496
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.043564997317152e-05,
|
|
"loss": 3.5292,
|
|
"step": 1195008
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0427264025661e-05,
|
|
"loss": 3.5378,
|
|
"step": 1195520
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0418878078150483e-05,
|
|
"loss": 3.5379,
|
|
"step": 1196032
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0410492130639963e-05,
|
|
"loss": 3.5302,
|
|
"step": 1196544
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0402106183129443e-05,
|
|
"loss": 3.5315,
|
|
"step": 1197056
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0393736614422652e-05,
|
|
"loss": 3.5395,
|
|
"step": 1197568
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0385350666912132e-05,
|
|
"loss": 3.5361,
|
|
"step": 1198080
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.037696471940161e-05,
|
|
"loss": 3.5332,
|
|
"step": 1198592
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.036857877189109e-05,
|
|
"loss": 3.5322,
|
|
"step": 1199104
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0360209203184304e-05,
|
|
"loss": 3.5226,
|
|
"step": 1199616
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0351823255673784e-05,
|
|
"loss": 3.5299,
|
|
"step": 1200128
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0343437308163264e-05,
|
|
"loss": 3.5212,
|
|
"step": 1200640
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0335051360652744e-05,
|
|
"loss": 3.5268,
|
|
"step": 1201152
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0326681791945953e-05,
|
|
"loss": 3.5241,
|
|
"step": 1201664
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0318295844435436e-05,
|
|
"loss": 3.537,
|
|
"step": 1202176
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0309909896924916e-05,
|
|
"loss": 3.5309,
|
|
"step": 1202688
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0301523949414396e-05,
|
|
"loss": 3.5279,
|
|
"step": 1203200
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0293170759511335e-05,
|
|
"loss": 3.544,
|
|
"step": 1203712
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0284784812000815e-05,
|
|
"loss": 3.5436,
|
|
"step": 1204224
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0276398864490298e-05,
|
|
"loss": 3.538,
|
|
"step": 1204736
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0268012916979778e-05,
|
|
"loss": 3.5294,
|
|
"step": 1205248
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0259643348272987e-05,
|
|
"loss": 3.5317,
|
|
"step": 1205760
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0251257400762467e-05,
|
|
"loss": 3.5298,
|
|
"step": 1206272
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0242871453251947e-05,
|
|
"loss": 3.5222,
|
|
"step": 1206784
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0234485505741427e-05,
|
|
"loss": 3.5309,
|
|
"step": 1207296
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0226099558230907e-05,
|
|
"loss": 3.5277,
|
|
"step": 1207808
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.021771361072039e-05,
|
|
"loss": 3.5376,
|
|
"step": 1208320
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.020932766320987e-05,
|
|
"loss": 3.5305,
|
|
"step": 1208832
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.020094171569935e-05,
|
|
"loss": 3.5382,
|
|
"step": 1209344
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.019257214699256e-05,
|
|
"loss": 3.529,
|
|
"step": 1209856
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0184202578285768e-05,
|
|
"loss": 3.5303,
|
|
"step": 1210368
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.017581663077525e-05,
|
|
"loss": 3.5324,
|
|
"step": 1210880
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.016743068326473e-05,
|
|
"loss": 3.5369,
|
|
"step": 1211392
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.015904473575421e-05,
|
|
"loss": 3.5409,
|
|
"step": 1211904
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.015065878824369e-05,
|
|
"loss": 3.5319,
|
|
"step": 1212416
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.014227284073317e-05,
|
|
"loss": 3.5407,
|
|
"step": 1212928
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.013388689322265e-05,
|
|
"loss": 3.536,
|
|
"step": 1213440
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0125500945712128e-05,
|
|
"loss": 3.5347,
|
|
"step": 1213952
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.011713137700534e-05,
|
|
"loss": 3.5331,
|
|
"step": 1214464
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0108745429494824e-05,
|
|
"loss": 3.5299,
|
|
"step": 1214976
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0100359481984304e-05,
|
|
"loss": 3.5385,
|
|
"step": 1215488
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0091973534473777e-05,
|
|
"loss": 3.5409,
|
|
"step": 1216000
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0083603965766993e-05,
|
|
"loss": 3.5336,
|
|
"step": 1216512
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0075234397060205e-05,
|
|
"loss": 3.527,
|
|
"step": 1217024
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0066848449549685e-05,
|
|
"loss": 3.5359,
|
|
"step": 1217536
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0058462502039165e-05,
|
|
"loss": 3.5378,
|
|
"step": 1218048
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0050076554528645e-05,
|
|
"loss": 3.528,
|
|
"step": 1218560
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0041690607018125e-05,
|
|
"loss": 3.5385,
|
|
"step": 1219072
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0033321038311334e-05,
|
|
"loss": 3.5271,
|
|
"step": 1219584
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0024935090800814e-05,
|
|
"loss": 3.5454,
|
|
"step": 1220096
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.0016549143290294e-05,
|
|
"loss": 3.5291,
|
|
"step": 1220608
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.0008163195779777e-05,
|
|
"loss": 3.5305,
|
|
"step": 1221120
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.8443686962127686,
|
|
"eval_runtime": 302.26,
|
|
"eval_samples_per_second": 1262.459,
|
|
"eval_steps_per_second": 39.453,
|
|
"step": 1221120
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9999793627072986e-05,
|
|
"loss": 3.524,
|
|
"step": 1221632
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9991407679562466e-05,
|
|
"loss": 3.5221,
|
|
"step": 1222144
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9983021732051946e-05,
|
|
"loss": 3.5424,
|
|
"step": 1222656
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9974635784541426e-05,
|
|
"loss": 3.537,
|
|
"step": 1223168
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9966249837030903e-05,
|
|
"loss": 3.5338,
|
|
"step": 1223680
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9957863889520383e-05,
|
|
"loss": 3.528,
|
|
"step": 1224192
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.99494943208136e-05,
|
|
"loss": 3.5287,
|
|
"step": 1224704
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9941108373303072e-05,
|
|
"loss": 3.5146,
|
|
"step": 1225216
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9932722425792552e-05,
|
|
"loss": 3.5338,
|
|
"step": 1225728
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9924336478282032e-05,
|
|
"loss": 3.5223,
|
|
"step": 1226240
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9915950530771515e-05,
|
|
"loss": 3.5468,
|
|
"step": 1226752
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9907580962064724e-05,
|
|
"loss": 3.5407,
|
|
"step": 1227264
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9899195014554204e-05,
|
|
"loss": 3.5106,
|
|
"step": 1227776
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9890809067043684e-05,
|
|
"loss": 3.5219,
|
|
"step": 1228288
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9882423119533164e-05,
|
|
"loss": 3.5203,
|
|
"step": 1228800
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9874037172022644e-05,
|
|
"loss": 3.5231,
|
|
"step": 1229312
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9865651224512124e-05,
|
|
"loss": 3.5274,
|
|
"step": 1229824
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9857265277001607e-05,
|
|
"loss": 3.5328,
|
|
"step": 1230336
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9848879329491087e-05,
|
|
"loss": 3.5198,
|
|
"step": 1230848
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9840509760784296e-05,
|
|
"loss": 3.5456,
|
|
"step": 1231360
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9832123813273776e-05,
|
|
"loss": 3.5269,
|
|
"step": 1231872
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9823737865763256e-05,
|
|
"loss": 3.5274,
|
|
"step": 1232384
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9815351918252736e-05,
|
|
"loss": 3.5258,
|
|
"step": 1232896
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.9806965970742216e-05,
|
|
"loss": 3.5405,
|
|
"step": 1233408
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.97985800232317e-05,
|
|
"loss": 3.5246,
|
|
"step": 1233920
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.979019407572118e-05,
|
|
"loss": 3.5208,
|
|
"step": 1234432
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.978180812821066e-05,
|
|
"loss": 3.5312,
|
|
"step": 1234944
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.977343855950387e-05,
|
|
"loss": 3.5227,
|
|
"step": 1235456
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.976505261199335e-05,
|
|
"loss": 3.5188,
|
|
"step": 1235968
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.975666666448283e-05,
|
|
"loss": 3.523,
|
|
"step": 1236480
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.974828071697231e-05,
|
|
"loss": 3.5278,
|
|
"step": 1236992
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.973991114826552e-05,
|
|
"loss": 3.5305,
|
|
"step": 1237504
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9731525200755e-05,
|
|
"loss": 3.5278,
|
|
"step": 1238016
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.972313925324448e-05,
|
|
"loss": 3.5303,
|
|
"step": 1238528
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.971475330573396e-05,
|
|
"loss": 3.5295,
|
|
"step": 1239040
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.970638373702717e-05,
|
|
"loss": 3.5344,
|
|
"step": 1239552
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9697997789516653e-05,
|
|
"loss": 3.5184,
|
|
"step": 1240064
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9689611842006133e-05,
|
|
"loss": 3.5232,
|
|
"step": 1240576
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9681225894495613e-05,
|
|
"loss": 3.519,
|
|
"step": 1241088
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.967287270459255e-05,
|
|
"loss": 3.5208,
|
|
"step": 1241600
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.966448675708203e-05,
|
|
"loss": 3.5194,
|
|
"step": 1242112
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9656100809571515e-05,
|
|
"loss": 3.5287,
|
|
"step": 1242624
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9647714862060995e-05,
|
|
"loss": 3.5227,
|
|
"step": 1243136
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9639328914550475e-05,
|
|
"loss": 3.5287,
|
|
"step": 1243648
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9630942967039955e-05,
|
|
"loss": 3.529,
|
|
"step": 1244160
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9622557019529435e-05,
|
|
"loss": 3.5328,
|
|
"step": 1244672
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9614171072018908e-05,
|
|
"loss": 3.5245,
|
|
"step": 1245184
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9605801503312124e-05,
|
|
"loss": 3.5098,
|
|
"step": 1245696
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9597431934605336e-05,
|
|
"loss": 3.5183,
|
|
"step": 1246208
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9589045987094816e-05,
|
|
"loss": 3.5366,
|
|
"step": 1246720
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9580660039584296e-05,
|
|
"loss": 3.5259,
|
|
"step": 1247232
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9572274092073776e-05,
|
|
"loss": 3.52,
|
|
"step": 1247744
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9563888144563256e-05,
|
|
"loss": 3.5148,
|
|
"step": 1248256
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.955551857585647e-05,
|
|
"loss": 3.5158,
|
|
"step": 1248768
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.954713262834595e-05,
|
|
"loss": 3.5073,
|
|
"step": 1249280
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9538746680835428e-05,
|
|
"loss": 3.5331,
|
|
"step": 1249792
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9530360733324908e-05,
|
|
"loss": 3.5129,
|
|
"step": 1250304
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.952197478581438e-05,
|
|
"loss": 3.5249,
|
|
"step": 1250816
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.951358883830386e-05,
|
|
"loss": 3.5262,
|
|
"step": 1251328
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9505202890793345e-05,
|
|
"loss": 3.5103,
|
|
"step": 1251840
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.949683332208656e-05,
|
|
"loss": 3.5314,
|
|
"step": 1252352
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9488447374576034e-05,
|
|
"loss": 3.5139,
|
|
"step": 1252864
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9480061427065514e-05,
|
|
"loss": 3.5211,
|
|
"step": 1253376
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9471675479554994e-05,
|
|
"loss": 3.5113,
|
|
"step": 1253888
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9463305910848206e-05,
|
|
"loss": 3.5302,
|
|
"step": 1254400
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9454919963337686e-05,
|
|
"loss": 3.5133,
|
|
"step": 1254912
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9446534015827166e-05,
|
|
"loss": 3.5157,
|
|
"step": 1255424
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9438148068316646e-05,
|
|
"loss": 3.5072,
|
|
"step": 1255936
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9429762120806126e-05,
|
|
"loss": 3.5124,
|
|
"step": 1256448
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9421376173295606e-05,
|
|
"loss": 3.5253,
|
|
"step": 1256960
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9412990225785086e-05,
|
|
"loss": 3.5274,
|
|
"step": 1257472
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9404604278274566e-05,
|
|
"loss": 3.5213,
|
|
"step": 1257984
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.939623470956778e-05,
|
|
"loss": 3.5212,
|
|
"step": 1258496
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9387848762057258e-05,
|
|
"loss": 3.533,
|
|
"step": 1259008
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9379462814546738e-05,
|
|
"loss": 3.5192,
|
|
"step": 1259520
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9371093245839947e-05,
|
|
"loss": 3.5344,
|
|
"step": 1260032
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9362707298329427e-05,
|
|
"loss": 3.515,
|
|
"step": 1260544
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9354321350818907e-05,
|
|
"loss": 3.5162,
|
|
"step": 1261056
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9345935403308387e-05,
|
|
"loss": 3.5245,
|
|
"step": 1261568
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.93375658346016e-05,
|
|
"loss": 3.5281,
|
|
"step": 1262080
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.932917988709108e-05,
|
|
"loss": 3.5214,
|
|
"step": 1262592
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.932079393958056e-05,
|
|
"loss": 3.5316,
|
|
"step": 1263104
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.931240799207004e-05,
|
|
"loss": 3.5182,
|
|
"step": 1263616
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.930402204455952e-05,
|
|
"loss": 3.4996,
|
|
"step": 1264128
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9295636097049e-05,
|
|
"loss": 3.5128,
|
|
"step": 1264640
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9287266528342212e-05,
|
|
"loss": 3.5194,
|
|
"step": 1265152
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9278880580831692e-05,
|
|
"loss": 3.5154,
|
|
"step": 1265664
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9270494633321172e-05,
|
|
"loss": 3.5171,
|
|
"step": 1266176
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.9262108685810652e-05,
|
|
"loss": 3.522,
|
|
"step": 1266688
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9253722738300132e-05,
|
|
"loss": 3.5127,
|
|
"step": 1267200
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.924535316959334e-05,
|
|
"loss": 3.5163,
|
|
"step": 1267712
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9236967222082824e-05,
|
|
"loss": 3.5158,
|
|
"step": 1268224
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9228581274572304e-05,
|
|
"loss": 3.5269,
|
|
"step": 1268736
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9220195327061784e-05,
|
|
"loss": 3.5323,
|
|
"step": 1269248
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9211825758354993e-05,
|
|
"loss": 3.5325,
|
|
"step": 1269760
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9203439810844473e-05,
|
|
"loss": 3.5049,
|
|
"step": 1270272
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9195053863333953e-05,
|
|
"loss": 3.5188,
|
|
"step": 1270784
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9186667915823433e-05,
|
|
"loss": 3.5171,
|
|
"step": 1271296
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9178281968312916e-05,
|
|
"loss": 3.5208,
|
|
"step": 1271808
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.916989602080239e-05,
|
|
"loss": 3.5258,
|
|
"step": 1272320
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.916151007329187e-05,
|
|
"loss": 3.5141,
|
|
"step": 1272832
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.915312412578135e-05,
|
|
"loss": 3.5152,
|
|
"step": 1273344
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9144754557074565e-05,
|
|
"loss": 3.527,
|
|
"step": 1273856
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9136368609564042e-05,
|
|
"loss": 3.522,
|
|
"step": 1274368
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9127982662053522e-05,
|
|
"loss": 3.5162,
|
|
"step": 1274880
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9119596714543002e-05,
|
|
"loss": 3.5206,
|
|
"step": 1275392
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9111227145836218e-05,
|
|
"loss": 3.5113,
|
|
"step": 1275904
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.910284119832569e-05,
|
|
"loss": 3.5129,
|
|
"step": 1276416
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.909445525081517e-05,
|
|
"loss": 3.5097,
|
|
"step": 1276928
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9086069303304654e-05,
|
|
"loss": 3.5135,
|
|
"step": 1277440
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9077699734597863e-05,
|
|
"loss": 3.5085,
|
|
"step": 1277952
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9069313787087343e-05,
|
|
"loss": 3.5231,
|
|
"step": 1278464
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9060927839576823e-05,
|
|
"loss": 3.5153,
|
|
"step": 1278976
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9052541892066303e-05,
|
|
"loss": 3.5175,
|
|
"step": 1279488
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.9044188702163248e-05,
|
|
"loss": 3.5305,
|
|
"step": 1280000
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.903580275465273e-05,
|
|
"loss": 3.532,
|
|
"step": 1280512
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.902741680714221e-05,
|
|
"loss": 3.5211,
|
|
"step": 1281024
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.901903085963169e-05,
|
|
"loss": 3.5212,
|
|
"step": 1281536
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.90106612909249e-05,
|
|
"loss": 3.5154,
|
|
"step": 1282048
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.900227534341438e-05,
|
|
"loss": 3.5145,
|
|
"step": 1282560
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.899388939590386e-05,
|
|
"loss": 3.5081,
|
|
"step": 1283072
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8985503448393337e-05,
|
|
"loss": 3.5138,
|
|
"step": 1283584
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8977117500882817e-05,
|
|
"loss": 3.5156,
|
|
"step": 1284096
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8968731553372297e-05,
|
|
"loss": 3.5223,
|
|
"step": 1284608
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8960345605861777e-05,
|
|
"loss": 3.5183,
|
|
"step": 1285120
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8951959658351257e-05,
|
|
"loss": 3.5239,
|
|
"step": 1285632
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.894359008964447e-05,
|
|
"loss": 3.5165,
|
|
"step": 1286144
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8935220520937685e-05,
|
|
"loss": 3.5163,
|
|
"step": 1286656
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8926834573427165e-05,
|
|
"loss": 3.5133,
|
|
"step": 1287168
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.891844862591664e-05,
|
|
"loss": 3.5284,
|
|
"step": 1287680
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.891006267840612e-05,
|
|
"loss": 3.5293,
|
|
"step": 1288192
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8901676730895598e-05,
|
|
"loss": 3.5151,
|
|
"step": 1288704
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8893290783385078e-05,
|
|
"loss": 3.5224,
|
|
"step": 1289216
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.888490483587456e-05,
|
|
"loss": 3.526,
|
|
"step": 1289728
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.887651888836404e-05,
|
|
"loss": 3.5208,
|
|
"step": 1290240
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.886814931965725e-05,
|
|
"loss": 3.5233,
|
|
"step": 1290752
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.885976337214673e-05,
|
|
"loss": 3.5172,
|
|
"step": 1291264
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.885137742463621e-05,
|
|
"loss": 3.5233,
|
|
"step": 1291776
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8843007855929423e-05,
|
|
"loss": 3.5204,
|
|
"step": 1292288
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8834621908418903e-05,
|
|
"loss": 3.5193,
|
|
"step": 1292800
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8826235960908383e-05,
|
|
"loss": 3.5162,
|
|
"step": 1293312
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8817850013397863e-05,
|
|
"loss": 3.5188,
|
|
"step": 1293824
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8809464065887343e-05,
|
|
"loss": 3.5226,
|
|
"step": 1294336
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8801078118376823e-05,
|
|
"loss": 3.5116,
|
|
"step": 1294848
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8792708549670032e-05,
|
|
"loss": 3.5278,
|
|
"step": 1295360
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8784322602159515e-05,
|
|
"loss": 3.5163,
|
|
"step": 1295872
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8775936654648995e-05,
|
|
"loss": 3.5242,
|
|
"step": 1296384
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8767550707138475e-05,
|
|
"loss": 3.5212,
|
|
"step": 1296896
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.8759164759627955e-05,
|
|
"loss": 3.5122,
|
|
"step": 1297408
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 3.8447844982147217,
|
|
"eval_runtime": 301.5751,
|
|
"eval_samples_per_second": 1265.326,
|
|
"eval_steps_per_second": 39.542,
|
|
"step": 1297440
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8750795190921164e-05,
|
|
"loss": 3.5083,
|
|
"step": 1297920
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8742409243410644e-05,
|
|
"loss": 3.5109,
|
|
"step": 1298432
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8734023295900124e-05,
|
|
"loss": 3.5306,
|
|
"step": 1298944
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8725637348389604e-05,
|
|
"loss": 3.5166,
|
|
"step": 1299456
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8717251400879087e-05,
|
|
"loss": 3.5234,
|
|
"step": 1299968
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8708881832172297e-05,
|
|
"loss": 3.514,
|
|
"step": 1300480
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8700495884661776e-05,
|
|
"loss": 3.5123,
|
|
"step": 1300992
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8692109937151256e-05,
|
|
"loss": 3.5055,
|
|
"step": 1301504
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8683723989640736e-05,
|
|
"loss": 3.5183,
|
|
"step": 1302016
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8675338042130216e-05,
|
|
"loss": 3.5091,
|
|
"step": 1302528
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8666952094619696e-05,
|
|
"loss": 3.5332,
|
|
"step": 1303040
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8658566147109173e-05,
|
|
"loss": 3.5213,
|
|
"step": 1303552
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.865019657840239e-05,
|
|
"loss": 3.5059,
|
|
"step": 1304064
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.864181063089187e-05,
|
|
"loss": 3.5066,
|
|
"step": 1304576
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.863342468338135e-05,
|
|
"loss": 3.512,
|
|
"step": 1305088
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8625038735870825e-05,
|
|
"loss": 3.5024,
|
|
"step": 1305600
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8616652788360305e-05,
|
|
"loss": 3.5184,
|
|
"step": 1306112
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8608266840849785e-05,
|
|
"loss": 3.5164,
|
|
"step": 1306624
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8599880893339265e-05,
|
|
"loss": 3.5054,
|
|
"step": 1307136
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8591511324632474e-05,
|
|
"loss": 3.5324,
|
|
"step": 1307648
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8583125377121954e-05,
|
|
"loss": 3.5127,
|
|
"step": 1308160
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8574739429611434e-05,
|
|
"loss": 3.5154,
|
|
"step": 1308672
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8566353482100917e-05,
|
|
"loss": 3.5102,
|
|
"step": 1309184
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8557967534590397e-05,
|
|
"loss": 3.5269,
|
|
"step": 1309696
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8549581587079877e-05,
|
|
"loss": 3.5066,
|
|
"step": 1310208
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8541195639569357e-05,
|
|
"loss": 3.5134,
|
|
"step": 1310720
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8532809692058837e-05,
|
|
"loss": 3.5122,
|
|
"step": 1311232
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8524440123352046e-05,
|
|
"loss": 3.5119,
|
|
"step": 1311744
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.8516054175841526e-05,
|
|
"loss": 3.5049,
|
|
"step": 1312256
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.850766822833101e-05,
|
|
"loss": 3.5117,
|
|
"step": 1312768
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.849928228082049e-05,
|
|
"loss": 3.5156,
|
|
"step": 1313280
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.84909127121137e-05,
|
|
"loss": 3.5152,
|
|
"step": 1313792
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.848252676460318e-05,
|
|
"loss": 3.5142,
|
|
"step": 1314304
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.847414081709266e-05,
|
|
"loss": 3.5198,
|
|
"step": 1314816
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.846575486958214e-05,
|
|
"loss": 3.5134,
|
|
"step": 1315328
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.845738530087535e-05,
|
|
"loss": 3.5188,
|
|
"step": 1315840
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.844899935336483e-05,
|
|
"loss": 3.5075,
|
|
"step": 1316352
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.844061340585431e-05,
|
|
"loss": 3.5106,
|
|
"step": 1316864
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.843222745834379e-05,
|
|
"loss": 3.5013,
|
|
"step": 1317376
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8423874268440733e-05,
|
|
"loss": 3.5091,
|
|
"step": 1317888
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8415488320930213e-05,
|
|
"loss": 3.5057,
|
|
"step": 1318400
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8407102373419692e-05,
|
|
"loss": 3.5186,
|
|
"step": 1318912
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8398716425909172e-05,
|
|
"loss": 3.5027,
|
|
"step": 1319424
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8390330478398652e-05,
|
|
"loss": 3.517,
|
|
"step": 1319936
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8381944530888132e-05,
|
|
"loss": 3.5156,
|
|
"step": 1320448
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8373558583377612e-05,
|
|
"loss": 3.5161,
|
|
"step": 1320960
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8365172635867092e-05,
|
|
"loss": 3.5105,
|
|
"step": 1321472
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8356803067160305e-05,
|
|
"loss": 3.4982,
|
|
"step": 1321984
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8348433498453514e-05,
|
|
"loss": 3.5091,
|
|
"step": 1322496
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8340047550942994e-05,
|
|
"loss": 3.5194,
|
|
"step": 1323008
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8331661603432474e-05,
|
|
"loss": 3.5129,
|
|
"step": 1323520
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8323275655921954e-05,
|
|
"loss": 3.5073,
|
|
"step": 1324032
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8314889708411434e-05,
|
|
"loss": 3.5011,
|
|
"step": 1324544
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8306520139704646e-05,
|
|
"loss": 3.5067,
|
|
"step": 1325056
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8298134192194126e-05,
|
|
"loss": 3.4906,
|
|
"step": 1325568
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8289748244683606e-05,
|
|
"loss": 3.522,
|
|
"step": 1326080
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8281362297173086e-05,
|
|
"loss": 3.4976,
|
|
"step": 1326592
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8272976349662566e-05,
|
|
"loss": 3.5134,
|
|
"step": 1327104
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8264590402152046e-05,
|
|
"loss": 3.5106,
|
|
"step": 1327616
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8256204454641526e-05,
|
|
"loss": 3.5001,
|
|
"step": 1328128
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.824781850713101e-05,
|
|
"loss": 3.514,
|
|
"step": 1328640
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8239448938424218e-05,
|
|
"loss": 3.5053,
|
|
"step": 1329152
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8231062990913698e-05,
|
|
"loss": 3.5056,
|
|
"step": 1329664
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8222693422206907e-05,
|
|
"loss": 3.4978,
|
|
"step": 1330176
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8214307474696387e-05,
|
|
"loss": 3.519,
|
|
"step": 1330688
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.820592152718587e-05,
|
|
"loss": 3.4993,
|
|
"step": 1331200
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.819753557967535e-05,
|
|
"loss": 3.5039,
|
|
"step": 1331712
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.818914963216483e-05,
|
|
"loss": 3.49,
|
|
"step": 1332224
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.818078006345804e-05,
|
|
"loss": 3.5022,
|
|
"step": 1332736
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.817239411594752e-05,
|
|
"loss": 3.5101,
|
|
"step": 1333248
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8164008168437e-05,
|
|
"loss": 3.5106,
|
|
"step": 1333760
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.815562222092648e-05,
|
|
"loss": 3.5099,
|
|
"step": 1334272
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8147236273415956e-05,
|
|
"loss": 3.511,
|
|
"step": 1334784
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8138850325905436e-05,
|
|
"loss": 3.5178,
|
|
"step": 1335296
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8130464378394916e-05,
|
|
"loss": 3.5077,
|
|
"step": 1335808
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8122078430884396e-05,
|
|
"loss": 3.5151,
|
|
"step": 1336320
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.811370886217761e-05,
|
|
"loss": 3.5044,
|
|
"step": 1336832
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.810532291466709e-05,
|
|
"loss": 3.5064,
|
|
"step": 1337344
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.809693696715657e-05,
|
|
"loss": 3.5076,
|
|
"step": 1337856
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.808855101964605e-05,
|
|
"loss": 3.5117,
|
|
"step": 1338368
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8080165072135528e-05,
|
|
"loss": 3.5097,
|
|
"step": 1338880
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8071779124625008e-05,
|
|
"loss": 3.5188,
|
|
"step": 1339392
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8063393177114488e-05,
|
|
"loss": 3.5083,
|
|
"step": 1339904
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.8055007229603968e-05,
|
|
"loss": 3.4869,
|
|
"step": 1340416
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.804663766089718e-05,
|
|
"loss": 3.4983,
|
|
"step": 1340928
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.803826809219039e-05,
|
|
"loss": 3.5051,
|
|
"step": 1341440
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.802988214467987e-05,
|
|
"loss": 3.5031,
|
|
"step": 1341952
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.802149619716935e-05,
|
|
"loss": 3.5027,
|
|
"step": 1342464
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.801311024965883e-05,
|
|
"loss": 3.5127,
|
|
"step": 1342976
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.800472430214831e-05,
|
|
"loss": 3.4988,
|
|
"step": 1343488
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7996354733441522e-05,
|
|
"loss": 3.4993,
|
|
"step": 1344000
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7987968785931002e-05,
|
|
"loss": 3.5031,
|
|
"step": 1344512
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7979582838420482e-05,
|
|
"loss": 3.5148,
|
|
"step": 1345024
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7971196890909962e-05,
|
|
"loss": 3.5145,
|
|
"step": 1345536
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.796282732220317e-05,
|
|
"loss": 3.521,
|
|
"step": 1346048
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.795444137469265e-05,
|
|
"loss": 3.4939,
|
|
"step": 1346560
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7946055427182134e-05,
|
|
"loss": 3.5072,
|
|
"step": 1347072
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7937669479671614e-05,
|
|
"loss": 3.5081,
|
|
"step": 1347584
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7929283532161094e-05,
|
|
"loss": 3.5032,
|
|
"step": 1348096
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7920897584650574e-05,
|
|
"loss": 3.5143,
|
|
"step": 1348608
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7912511637140054e-05,
|
|
"loss": 3.5004,
|
|
"step": 1349120
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7904125689629534e-05,
|
|
"loss": 3.5031,
|
|
"step": 1349632
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7895756120922743e-05,
|
|
"loss": 3.5141,
|
|
"step": 1350144
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7887386552215956e-05,
|
|
"loss": 3.5097,
|
|
"step": 1350656
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7879000604705436e-05,
|
|
"loss": 3.497,
|
|
"step": 1351168
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7870614657194916e-05,
|
|
"loss": 3.5087,
|
|
"step": 1351680
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7862228709684396e-05,
|
|
"loss": 3.501,
|
|
"step": 1352192
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7853842762173875e-05,
|
|
"loss": 3.4984,
|
|
"step": 1352704
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7845456814663355e-05,
|
|
"loss": 3.4955,
|
|
"step": 1353216
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7837070867152835e-05,
|
|
"loss": 3.5012,
|
|
"step": 1353728
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.782868491964232e-05,
|
|
"loss": 3.4945,
|
|
"step": 1354240
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7820315350935528e-05,
|
|
"loss": 3.5095,
|
|
"step": 1354752
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7811929403425008e-05,
|
|
"loss": 3.5007,
|
|
"step": 1355264
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7803543455914488e-05,
|
|
"loss": 3.5041,
|
|
"step": 1355776
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7795173887207697e-05,
|
|
"loss": 3.5194,
|
|
"step": 1356288
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.778678793969718e-05,
|
|
"loss": 3.5144,
|
|
"step": 1356800
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.777840199218666e-05,
|
|
"loss": 3.5075,
|
|
"step": 1357312
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.777001604467614e-05,
|
|
"loss": 3.5091,
|
|
"step": 1357824
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.776164647596935e-05,
|
|
"loss": 3.5034,
|
|
"step": 1358336
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.775326052845883e-05,
|
|
"loss": 3.5019,
|
|
"step": 1358848
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.774487458094831e-05,
|
|
"loss": 3.4988,
|
|
"step": 1359360
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.773648863343779e-05,
|
|
"loss": 3.4966,
|
|
"step": 1359872
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7728102685927266e-05,
|
|
"loss": 3.5043,
|
|
"step": 1360384
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7719716738416746e-05,
|
|
"loss": 3.5115,
|
|
"step": 1360896
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7711330790906226e-05,
|
|
"loss": 3.5068,
|
|
"step": 1361408
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7702944843395705e-05,
|
|
"loss": 3.508,
|
|
"step": 1361920
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7694575274688918e-05,
|
|
"loss": 3.5048,
|
|
"step": 1362432
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7686205705982134e-05,
|
|
"loss": 3.5002,
|
|
"step": 1362944
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7677819758471614e-05,
|
|
"loss": 3.4999,
|
|
"step": 1363456
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7669433810961087e-05,
|
|
"loss": 3.5167,
|
|
"step": 1363968
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7661047863450567e-05,
|
|
"loss": 3.5142,
|
|
"step": 1364480
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7652661915940047e-05,
|
|
"loss": 3.5089,
|
|
"step": 1364992
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7644275968429527e-05,
|
|
"loss": 3.5041,
|
|
"step": 1365504
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.763589002091901e-05,
|
|
"loss": 3.511,
|
|
"step": 1366016
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.762750407340849e-05,
|
|
"loss": 3.5059,
|
|
"step": 1366528
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.76191345047017e-05,
|
|
"loss": 3.5124,
|
|
"step": 1367040
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.761074855719118e-05,
|
|
"loss": 3.5054,
|
|
"step": 1367552
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.760236260968066e-05,
|
|
"loss": 3.5081,
|
|
"step": 1368064
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.759397666217014e-05,
|
|
"loss": 3.5085,
|
|
"step": 1368576
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.758560709346335e-05,
|
|
"loss": 3.5082,
|
|
"step": 1369088
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.757722114595283e-05,
|
|
"loss": 3.5006,
|
|
"step": 1369600
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.756883519844231e-05,
|
|
"loss": 3.5065,
|
|
"step": 1370112
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.756044925093179e-05,
|
|
"loss": 3.5115,
|
|
"step": 1370624
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.755206330342127e-05,
|
|
"loss": 3.4971,
|
|
"step": 1371136
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.754369373471448e-05,
|
|
"loss": 3.5153,
|
|
"step": 1371648
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7535307787203964e-05,
|
|
"loss": 3.5068,
|
|
"step": 1372160
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7526921839693444e-05,
|
|
"loss": 3.5079,
|
|
"step": 1372672
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7518535892182924e-05,
|
|
"loss": 3.5067,
|
|
"step": 1373184
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7510149944672404e-05,
|
|
"loss": 3.4994,
|
|
"step": 1373696
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.845662832260132,
|
|
"eval_runtime": 303.1687,
|
|
"eval_samples_per_second": 1258.675,
|
|
"eval_steps_per_second": 39.335,
|
|
"step": 1373760
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7501780375965613e-05,
|
|
"loss": 3.5006,
|
|
"step": 1374208
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7493394428455093e-05,
|
|
"loss": 3.4969,
|
|
"step": 1374720
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7485008480944573e-05,
|
|
"loss": 3.5164,
|
|
"step": 1375232
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7476622533434053e-05,
|
|
"loss": 3.5032,
|
|
"step": 1375744
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7468236585923536e-05,
|
|
"loss": 3.5158,
|
|
"step": 1376256
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7459867017216745e-05,
|
|
"loss": 3.499,
|
|
"step": 1376768
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7451481069706225e-05,
|
|
"loss": 3.5026,
|
|
"step": 1377280
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7443095122195705e-05,
|
|
"loss": 3.4919,
|
|
"step": 1377792
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7434709174685185e-05,
|
|
"loss": 3.5042,
|
|
"step": 1378304
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7426323227174665e-05,
|
|
"loss": 3.4986,
|
|
"step": 1378816
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7417937279664145e-05,
|
|
"loss": 3.5172,
|
|
"step": 1379328
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.740955133215362e-05,
|
|
"loss": 3.5094,
|
|
"step": 1379840
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7401181763446837e-05,
|
|
"loss": 3.4941,
|
|
"step": 1380352
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7392795815936317e-05,
|
|
"loss": 3.4927,
|
|
"step": 1380864
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7384409868425797e-05,
|
|
"loss": 3.5002,
|
|
"step": 1381376
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7376023920915274e-05,
|
|
"loss": 3.4873,
|
|
"step": 1381888
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7367637973404754e-05,
|
|
"loss": 3.5006,
|
|
"step": 1382400
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7359252025894234e-05,
|
|
"loss": 3.5033,
|
|
"step": 1382912
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7350866078383714e-05,
|
|
"loss": 3.4929,
|
|
"step": 1383424
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7342496509676923e-05,
|
|
"loss": 3.5206,
|
|
"step": 1383936
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7334110562166403e-05,
|
|
"loss": 3.4989,
|
|
"step": 1384448
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7325724614655883e-05,
|
|
"loss": 3.5044,
|
|
"step": 1384960
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7317338667145366e-05,
|
|
"loss": 3.4996,
|
|
"step": 1385472
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7308952719634846e-05,
|
|
"loss": 3.5143,
|
|
"step": 1385984
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7300566772124326e-05,
|
|
"loss": 3.4895,
|
|
"step": 1386496
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7292180824613806e-05,
|
|
"loss": 3.5015,
|
|
"step": 1387008
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7283794877103286e-05,
|
|
"loss": 3.4953,
|
|
"step": 1387520
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7275441687200227e-05,
|
|
"loss": 3.5014,
|
|
"step": 1388032
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.7267055739689707e-05,
|
|
"loss": 3.4912,
|
|
"step": 1388544
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7258669792179187e-05,
|
|
"loss": 3.498,
|
|
"step": 1389056
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7250283844668667e-05,
|
|
"loss": 3.5029,
|
|
"step": 1389568
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7241897897158147e-05,
|
|
"loss": 3.5036,
|
|
"step": 1390080
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7233528328451356e-05,
|
|
"loss": 3.4978,
|
|
"step": 1390592
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7225142380940836e-05,
|
|
"loss": 3.5102,
|
|
"step": 1391104
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.721675643343032e-05,
|
|
"loss": 3.4962,
|
|
"step": 1391616
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.72083704859198e-05,
|
|
"loss": 3.5069,
|
|
"step": 1392128
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.719998453840928e-05,
|
|
"loss": 3.4941,
|
|
"step": 1392640
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.719159859089876e-05,
|
|
"loss": 3.4997,
|
|
"step": 1393152
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.718321264338824e-05,
|
|
"loss": 3.4868,
|
|
"step": 1393664
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.717484307468145e-05,
|
|
"loss": 3.4982,
|
|
"step": 1394176
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.716645712717093e-05,
|
|
"loss": 3.4898,
|
|
"step": 1394688
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7158071179660412e-05,
|
|
"loss": 3.5055,
|
|
"step": 1395200
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7149685232149892e-05,
|
|
"loss": 3.4918,
|
|
"step": 1395712
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7141299284639372e-05,
|
|
"loss": 3.501,
|
|
"step": 1396224
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7132913337128852e-05,
|
|
"loss": 3.5062,
|
|
"step": 1396736
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7124527389618332e-05,
|
|
"loss": 3.5002,
|
|
"step": 1397248
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7116141442107812e-05,
|
|
"loss": 3.5016,
|
|
"step": 1397760
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.710777187340102e-05,
|
|
"loss": 3.4848,
|
|
"step": 1398272
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7099402304694233e-05,
|
|
"loss": 3.4929,
|
|
"step": 1398784
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7091016357183713e-05,
|
|
"loss": 3.5076,
|
|
"step": 1399296
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7082630409673193e-05,
|
|
"loss": 3.4968,
|
|
"step": 1399808
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7074244462162673e-05,
|
|
"loss": 3.4962,
|
|
"step": 1400320
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7065858514652153e-05,
|
|
"loss": 3.4859,
|
|
"step": 1400832
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7057488945945366e-05,
|
|
"loss": 3.4945,
|
|
"step": 1401344
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7049102998434846e-05,
|
|
"loss": 3.4747,
|
|
"step": 1401856
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7040717050924325e-05,
|
|
"loss": 3.5101,
|
|
"step": 1402368
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7032331103413805e-05,
|
|
"loss": 3.4849,
|
|
"step": 1402880
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7023961534707015e-05,
|
|
"loss": 3.5017,
|
|
"step": 1403392
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7015575587196495e-05,
|
|
"loss": 3.4985,
|
|
"step": 1403904
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.7007189639685974e-05,
|
|
"loss": 3.4869,
|
|
"step": 1404416
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6998803692175458e-05,
|
|
"loss": 3.5029,
|
|
"step": 1404928
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.699041774466493e-05,
|
|
"loss": 3.4903,
|
|
"step": 1405440
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.698203179715441e-05,
|
|
"loss": 3.4913,
|
|
"step": 1405952
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6973662228447627e-05,
|
|
"loss": 3.4824,
|
|
"step": 1406464
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6965276280937107e-05,
|
|
"loss": 3.5058,
|
|
"step": 1406976
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6956890333426583e-05,
|
|
"loss": 3.4871,
|
|
"step": 1407488
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6948504385916063e-05,
|
|
"loss": 3.4924,
|
|
"step": 1408000
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6940118438405543e-05,
|
|
"loss": 3.4794,
|
|
"step": 1408512
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6931732490895023e-05,
|
|
"loss": 3.4888,
|
|
"step": 1409024
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6923346543384503e-05,
|
|
"loss": 3.4989,
|
|
"step": 1409536
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6914976974677712e-05,
|
|
"loss": 3.4999,
|
|
"step": 1410048
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6906591027167196e-05,
|
|
"loss": 3.493,
|
|
"step": 1410560
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6898205079656676e-05,
|
|
"loss": 3.5006,
|
|
"step": 1411072
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6889819132146156e-05,
|
|
"loss": 3.5023,
|
|
"step": 1411584
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6881449563439365e-05,
|
|
"loss": 3.4975,
|
|
"step": 1412096
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6873063615928845e-05,
|
|
"loss": 3.4973,
|
|
"step": 1412608
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6864677668418325e-05,
|
|
"loss": 3.4961,
|
|
"step": 1413120
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6856291720907804e-05,
|
|
"loss": 3.492,
|
|
"step": 1413632
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6847922152201017e-05,
|
|
"loss": 3.4939,
|
|
"step": 1414144
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6839536204690497e-05,
|
|
"loss": 3.4978,
|
|
"step": 1414656
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6831150257179977e-05,
|
|
"loss": 3.4994,
|
|
"step": 1415168
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6822764309669457e-05,
|
|
"loss": 3.5073,
|
|
"step": 1415680
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6814378362158937e-05,
|
|
"loss": 3.4929,
|
|
"step": 1416192
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6805992414648417e-05,
|
|
"loss": 3.4709,
|
|
"step": 1416704
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6797606467137897e-05,
|
|
"loss": 3.4881,
|
|
"step": 1417216
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.6789220519627377e-05,
|
|
"loss": 3.4915,
|
|
"step": 1417728
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.678085095092059e-05,
|
|
"loss": 3.4916,
|
|
"step": 1418240
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.677246500341007e-05,
|
|
"loss": 3.4887,
|
|
"step": 1418752
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.676407905589955e-05,
|
|
"loss": 3.4957,
|
|
"step": 1419264
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.675569310838903e-05,
|
|
"loss": 3.4903,
|
|
"step": 1419776
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6747323539682238e-05,
|
|
"loss": 3.479,
|
|
"step": 1420288
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.673893759217172e-05,
|
|
"loss": 3.4913,
|
|
"step": 1420800
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.67305516446612e-05,
|
|
"loss": 3.5018,
|
|
"step": 1421312
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.672216569715068e-05,
|
|
"loss": 3.5,
|
|
"step": 1421824
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.671379612844389e-05,
|
|
"loss": 3.505,
|
|
"step": 1422336
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.670541018093337e-05,
|
|
"loss": 3.4872,
|
|
"step": 1422848
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.669702423342285e-05,
|
|
"loss": 3.4922,
|
|
"step": 1423360
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.668863828591233e-05,
|
|
"loss": 3.4949,
|
|
"step": 1423872
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6680252338401814e-05,
|
|
"loss": 3.4934,
|
|
"step": 1424384
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6671866390891294e-05,
|
|
"loss": 3.4986,
|
|
"step": 1424896
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6663480443380767e-05,
|
|
"loss": 3.4894,
|
|
"step": 1425408
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6655094495870247e-05,
|
|
"loss": 3.4901,
|
|
"step": 1425920
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6646724927163463e-05,
|
|
"loss": 3.4987,
|
|
"step": 1426432
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6638338979652943e-05,
|
|
"loss": 3.4995,
|
|
"step": 1426944
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.662995303214242e-05,
|
|
"loss": 3.4841,
|
|
"step": 1427456
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.66215670846319e-05,
|
|
"loss": 3.4925,
|
|
"step": 1427968
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6613197515925115e-05,
|
|
"loss": 3.4922,
|
|
"step": 1428480
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6604811568414588e-05,
|
|
"loss": 3.4864,
|
|
"step": 1428992
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6596425620904068e-05,
|
|
"loss": 3.4815,
|
|
"step": 1429504
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6588056052197284e-05,
|
|
"loss": 3.4872,
|
|
"step": 1430016
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6579670104686767e-05,
|
|
"loss": 3.4889,
|
|
"step": 1430528
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.657128415717624e-05,
|
|
"loss": 3.4935,
|
|
"step": 1431040
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.656289820966572e-05,
|
|
"loss": 3.4852,
|
|
"step": 1431552
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.65545122621552e-05,
|
|
"loss": 3.4914,
|
|
"step": 1432064
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6546142693448416e-05,
|
|
"loss": 3.5031,
|
|
"step": 1432576
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6537756745937893e-05,
|
|
"loss": 3.5073,
|
|
"step": 1433088
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6529370798427373e-05,
|
|
"loss": 3.4935,
|
|
"step": 1433600
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6520984850916853e-05,
|
|
"loss": 3.4975,
|
|
"step": 1434112
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6512615282210062e-05,
|
|
"loss": 3.492,
|
|
"step": 1434624
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6504229334699542e-05,
|
|
"loss": 3.4892,
|
|
"step": 1435136
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6495843387189022e-05,
|
|
"loss": 3.4828,
|
|
"step": 1435648
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6487457439678505e-05,
|
|
"loss": 3.4859,
|
|
"step": 1436160
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6479071492167985e-05,
|
|
"loss": 3.4908,
|
|
"step": 1436672
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6470685544657465e-05,
|
|
"loss": 3.4994,
|
|
"step": 1437184
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6462299597146945e-05,
|
|
"loss": 3.4936,
|
|
"step": 1437696
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6453930028440154e-05,
|
|
"loss": 3.4946,
|
|
"step": 1438208
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6445544080929634e-05,
|
|
"loss": 3.4933,
|
|
"step": 1438720
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6437158133419114e-05,
|
|
"loss": 3.4881,
|
|
"step": 1439232
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6428772185908597e-05,
|
|
"loss": 3.4872,
|
|
"step": 1439744
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6420402617201806e-05,
|
|
"loss": 3.5041,
|
|
"step": 1440256
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6412016669691286e-05,
|
|
"loss": 3.5029,
|
|
"step": 1440768
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6403630722180766e-05,
|
|
"loss": 3.4944,
|
|
"step": 1441280
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6395244774670246e-05,
|
|
"loss": 3.4932,
|
|
"step": 1441792
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6386858827159726e-05,
|
|
"loss": 3.4955,
|
|
"step": 1442304
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6378472879649206e-05,
|
|
"loss": 3.4949,
|
|
"step": 1442816
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.637008693213869e-05,
|
|
"loss": 3.5057,
|
|
"step": 1443328
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.636170098462817e-05,
|
|
"loss": 3.4917,
|
|
"step": 1443840
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.635333141592138e-05,
|
|
"loss": 3.4928,
|
|
"step": 1444352
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.634494546841086e-05,
|
|
"loss": 3.4939,
|
|
"step": 1444864
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.633655952090034e-05,
|
|
"loss": 3.4951,
|
|
"step": 1445376
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.632818995219355e-05,
|
|
"loss": 3.4904,
|
|
"step": 1445888
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.631980400468303e-05,
|
|
"loss": 3.4926,
|
|
"step": 1446400
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.631141805717251e-05,
|
|
"loss": 3.4993,
|
|
"step": 1446912
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.630303210966199e-05,
|
|
"loss": 3.4862,
|
|
"step": 1447424
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.629464616215147e-05,
|
|
"loss": 3.4999,
|
|
"step": 1447936
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.628627659344468e-05,
|
|
"loss": 3.4975,
|
|
"step": 1448448
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.627789064593416e-05,
|
|
"loss": 3.4927,
|
|
"step": 1448960
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.626950469842364e-05,
|
|
"loss": 3.4932,
|
|
"step": 1449472
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.6261118750913123e-05,
|
|
"loss": 3.4845,
|
|
"step": 1449984
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 3.8475229740142822,
|
|
"eval_runtime": 302.856,
|
|
"eval_samples_per_second": 1259.975,
|
|
"eval_steps_per_second": 39.375,
|
|
"step": 1450080
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6252749182206332e-05,
|
|
"loss": 3.4853,
|
|
"step": 1450496
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6244363234695812e-05,
|
|
"loss": 3.4831,
|
|
"step": 1451008
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6235977287185292e-05,
|
|
"loss": 3.5021,
|
|
"step": 1451520
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6227591339674772e-05,
|
|
"loss": 3.4884,
|
|
"step": 1452032
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.621920539216425e-05,
|
|
"loss": 3.5051,
|
|
"step": 1452544
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6210835823457465e-05,
|
|
"loss": 3.4842,
|
|
"step": 1453056
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6202449875946945e-05,
|
|
"loss": 3.4942,
|
|
"step": 1453568
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6194063928436424e-05,
|
|
"loss": 3.4777,
|
|
"step": 1454080
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6185677980925898e-05,
|
|
"loss": 3.4932,
|
|
"step": 1454592
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6177292033415378e-05,
|
|
"loss": 3.4838,
|
|
"step": 1455104
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.616890608590486e-05,
|
|
"loss": 3.5035,
|
|
"step": 1455616
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.616052013839434e-05,
|
|
"loss": 3.4934,
|
|
"step": 1456128
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.615215056968755e-05,
|
|
"loss": 3.4793,
|
|
"step": 1456640
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.614376462217703e-05,
|
|
"loss": 3.4798,
|
|
"step": 1457152
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.613537867466651e-05,
|
|
"loss": 3.4882,
|
|
"step": 1457664
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.612699272715599e-05,
|
|
"loss": 3.4768,
|
|
"step": 1458176
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.611860677964547e-05,
|
|
"loss": 3.4859,
|
|
"step": 1458688
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6110220832134953e-05,
|
|
"loss": 3.4888,
|
|
"step": 1459200
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6101834884624433e-05,
|
|
"loss": 3.4816,
|
|
"step": 1459712
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6093465315917642e-05,
|
|
"loss": 3.5092,
|
|
"step": 1460224
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6085079368407122e-05,
|
|
"loss": 3.4869,
|
|
"step": 1460736
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6076693420896602e-05,
|
|
"loss": 3.4884,
|
|
"step": 1461248
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6068307473386082e-05,
|
|
"loss": 3.489,
|
|
"step": 1461760
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6059921525875562e-05,
|
|
"loss": 3.4946,
|
|
"step": 1462272
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6051535578365045e-05,
|
|
"loss": 3.4822,
|
|
"step": 1462784
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6043149630854525e-05,
|
|
"loss": 3.4892,
|
|
"step": 1463296
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6034780062147734e-05,
|
|
"loss": 3.4845,
|
|
"step": 1463808
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6026394114637214e-05,
|
|
"loss": 3.4892,
|
|
"step": 1464320
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.6018008167126694e-05,
|
|
"loss": 3.4791,
|
|
"step": 1464832
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.6009622219616174e-05,
|
|
"loss": 3.4843,
|
|
"step": 1465344
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.6001252650909387e-05,
|
|
"loss": 3.4872,
|
|
"step": 1465856
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5992866703398867e-05,
|
|
"loss": 3.4924,
|
|
"step": 1466368
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5984480755888347e-05,
|
|
"loss": 3.4909,
|
|
"step": 1466880
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5976094808377827e-05,
|
|
"loss": 3.4952,
|
|
"step": 1467392
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5967708860867307e-05,
|
|
"loss": 3.4816,
|
|
"step": 1467904
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5959339292160516e-05,
|
|
"loss": 3.4944,
|
|
"step": 1468416
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.595095334465e-05,
|
|
"loss": 3.4784,
|
|
"step": 1468928
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.594256739713948e-05,
|
|
"loss": 3.4931,
|
|
"step": 1469440
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.593418144962896e-05,
|
|
"loss": 3.475,
|
|
"step": 1469952
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5925811880922168e-05,
|
|
"loss": 3.4828,
|
|
"step": 1470464
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5917425933411648e-05,
|
|
"loss": 3.4788,
|
|
"step": 1470976
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5909039985901128e-05,
|
|
"loss": 3.4955,
|
|
"step": 1471488
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5900654038390608e-05,
|
|
"loss": 3.4797,
|
|
"step": 1472000
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5892268090880085e-05,
|
|
"loss": 3.4846,
|
|
"step": 1472512
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5883882143369564e-05,
|
|
"loss": 3.4973,
|
|
"step": 1473024
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5875496195859044e-05,
|
|
"loss": 3.4912,
|
|
"step": 1473536
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5867110248348524e-05,
|
|
"loss": 3.4908,
|
|
"step": 1474048
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5858740679641737e-05,
|
|
"loss": 3.4744,
|
|
"step": 1474560
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5850371110934953e-05,
|
|
"loss": 3.4775,
|
|
"step": 1475072
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5841985163424433e-05,
|
|
"loss": 3.4907,
|
|
"step": 1475584
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5833599215913913e-05,
|
|
"loss": 3.4899,
|
|
"step": 1476096
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5825213268403386e-05,
|
|
"loss": 3.4828,
|
|
"step": 1476608
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5816827320892866e-05,
|
|
"loss": 3.4753,
|
|
"step": 1477120
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.580845775218608e-05,
|
|
"loss": 3.4823,
|
|
"step": 1477632
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5800071804675558e-05,
|
|
"loss": 3.4662,
|
|
"step": 1478144
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5791685857165038e-05,
|
|
"loss": 3.4893,
|
|
"step": 1478656
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5783299909654518e-05,
|
|
"loss": 3.4745,
|
|
"step": 1479168
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5774930340947734e-05,
|
|
"loss": 3.4841,
|
|
"step": 1479680
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5766544393437207e-05,
|
|
"loss": 3.4904,
|
|
"step": 1480192
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.575815844592669e-05,
|
|
"loss": 3.4759,
|
|
"step": 1480704
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.574977249841617e-05,
|
|
"loss": 3.4864,
|
|
"step": 1481216
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.574138655090565e-05,
|
|
"loss": 3.4804,
|
|
"step": 1481728
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.573300060339513e-05,
|
|
"loss": 3.4777,
|
|
"step": 1482240
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.572463103468834e-05,
|
|
"loss": 3.476,
|
|
"step": 1482752
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.571624508717782e-05,
|
|
"loss": 3.4895,
|
|
"step": 1483264
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.57078591396673e-05,
|
|
"loss": 3.4759,
|
|
"step": 1483776
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5699473192156783e-05,
|
|
"loss": 3.4793,
|
|
"step": 1484288
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5691103623449992e-05,
|
|
"loss": 3.4691,
|
|
"step": 1484800
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5682717675939472e-05,
|
|
"loss": 3.4738,
|
|
"step": 1485312
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5674331728428952e-05,
|
|
"loss": 3.4831,
|
|
"step": 1485824
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5665945780918432e-05,
|
|
"loss": 3.4894,
|
|
"step": 1486336
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5657576212211644e-05,
|
|
"loss": 3.4802,
|
|
"step": 1486848
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5649190264701124e-05,
|
|
"loss": 3.4891,
|
|
"step": 1487360
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5640804317190604e-05,
|
|
"loss": 3.49,
|
|
"step": 1487872
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5632418369680084e-05,
|
|
"loss": 3.4851,
|
|
"step": 1488384
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5624032422169564e-05,
|
|
"loss": 3.4865,
|
|
"step": 1488896
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5615662853462773e-05,
|
|
"loss": 3.4834,
|
|
"step": 1489408
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5607276905952253e-05,
|
|
"loss": 3.4816,
|
|
"step": 1489920
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5598890958441733e-05,
|
|
"loss": 3.4805,
|
|
"step": 1490432
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5590505010931216e-05,
|
|
"loss": 3.4856,
|
|
"step": 1490944
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5582119063420696e-05,
|
|
"loss": 3.4889,
|
|
"step": 1491456
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5573749494713905e-05,
|
|
"loss": 3.4873,
|
|
"step": 1491968
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5565363547203385e-05,
|
|
"loss": 3.4833,
|
|
"step": 1492480
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5556977599692865e-05,
|
|
"loss": 3.4653,
|
|
"step": 1492992
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5548591652182345e-05,
|
|
"loss": 3.4721,
|
|
"step": 1493504
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5540222083475558e-05,
|
|
"loss": 3.4792,
|
|
"step": 1494016
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5531836135965038e-05,
|
|
"loss": 3.482,
|
|
"step": 1494528
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5523450188454518e-05,
|
|
"loss": 3.4777,
|
|
"step": 1495040
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5515064240943998e-05,
|
|
"loss": 3.4831,
|
|
"step": 1495552
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5506678293433478e-05,
|
|
"loss": 3.4774,
|
|
"step": 1496064
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5498308724726687e-05,
|
|
"loss": 3.4715,
|
|
"step": 1496576
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.548992277721617e-05,
|
|
"loss": 3.4779,
|
|
"step": 1497088
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.548153682970565e-05,
|
|
"loss": 3.4871,
|
|
"step": 1497600
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.547315088219513e-05,
|
|
"loss": 3.4875,
|
|
"step": 1498112
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.546478131348834e-05,
|
|
"loss": 3.4945,
|
|
"step": 1498624
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.545639536597782e-05,
|
|
"loss": 3.478,
|
|
"step": 1499136
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.54480094184673e-05,
|
|
"loss": 3.4756,
|
|
"step": 1499648
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.543962347095678e-05,
|
|
"loss": 3.482,
|
|
"step": 1500160
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5431237523446262e-05,
|
|
"loss": 3.4812,
|
|
"step": 1500672
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5422851575935742e-05,
|
|
"loss": 3.489,
|
|
"step": 1501184
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5414465628425215e-05,
|
|
"loss": 3.4778,
|
|
"step": 1501696
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5406079680914695e-05,
|
|
"loss": 3.4781,
|
|
"step": 1502208
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.539771011220791e-05,
|
|
"loss": 3.4841,
|
|
"step": 1502720
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5389340543501124e-05,
|
|
"loss": 3.4901,
|
|
"step": 1503232
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5380954595990604e-05,
|
|
"loss": 3.4747,
|
|
"step": 1503744
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5372568648480084e-05,
|
|
"loss": 3.4825,
|
|
"step": 1504256
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5364182700969564e-05,
|
|
"loss": 3.4798,
|
|
"step": 1504768
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5355796753459044e-05,
|
|
"loss": 3.4746,
|
|
"step": 1505280
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5347410805948517e-05,
|
|
"loss": 3.4662,
|
|
"step": 1505792
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5339041237241733e-05,
|
|
"loss": 3.478,
|
|
"step": 1506304
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5330655289731216e-05,
|
|
"loss": 3.4778,
|
|
"step": 1506816
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.532226934222069e-05,
|
|
"loss": 3.4785,
|
|
"step": 1507328
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.531388339471017e-05,
|
|
"loss": 3.478,
|
|
"step": 1507840
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.530549744719965e-05,
|
|
"loss": 3.4749,
|
|
"step": 1508352
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5297127878492865e-05,
|
|
"loss": 3.4932,
|
|
"step": 1508864
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.528874193098234e-05,
|
|
"loss": 3.4932,
|
|
"step": 1509376
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.528035598347182e-05,
|
|
"loss": 3.4838,
|
|
"step": 1509888
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.52719700359613e-05,
|
|
"loss": 3.4839,
|
|
"step": 1510400
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5263600467254517e-05,
|
|
"loss": 3.4774,
|
|
"step": 1510912
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.525521451974399e-05,
|
|
"loss": 3.4757,
|
|
"step": 1511424
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.524682857223347e-05,
|
|
"loss": 3.4721,
|
|
"step": 1511936
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5238442624722954e-05,
|
|
"loss": 3.4726,
|
|
"step": 1512448
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5230056677212434e-05,
|
|
"loss": 3.4773,
|
|
"step": 1512960
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5221670729701914e-05,
|
|
"loss": 3.4865,
|
|
"step": 1513472
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5213284782191394e-05,
|
|
"loss": 3.4789,
|
|
"step": 1513984
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5204915213484603e-05,
|
|
"loss": 3.4848,
|
|
"step": 1514496
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5196529265974083e-05,
|
|
"loss": 3.4798,
|
|
"step": 1515008
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5188143318463563e-05,
|
|
"loss": 3.4764,
|
|
"step": 1515520
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5179757370953046e-05,
|
|
"loss": 3.474,
|
|
"step": 1516032
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5171387802246255e-05,
|
|
"loss": 3.4909,
|
|
"step": 1516544
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5163001854735735e-05,
|
|
"loss": 3.4938,
|
|
"step": 1517056
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5154615907225215e-05,
|
|
"loss": 3.4817,
|
|
"step": 1517568
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5146229959714695e-05,
|
|
"loss": 3.4804,
|
|
"step": 1518080
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5137844012204175e-05,
|
|
"loss": 3.4852,
|
|
"step": 1518592
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5129458064693655e-05,
|
|
"loss": 3.4776,
|
|
"step": 1519104
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5121072117183138e-05,
|
|
"loss": 3.4924,
|
|
"step": 1519616
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5112686169672618e-05,
|
|
"loss": 3.4749,
|
|
"step": 1520128
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5104316600965827e-05,
|
|
"loss": 3.4848,
|
|
"step": 1520640
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5095947032259036e-05,
|
|
"loss": 3.4786,
|
|
"step": 1521152
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5087561084748516e-05,
|
|
"loss": 3.489,
|
|
"step": 1521664
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5079175137238e-05,
|
|
"loss": 3.475,
|
|
"step": 1522176
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.507078918972748e-05,
|
|
"loss": 3.4798,
|
|
"step": 1522688
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.506240324221696e-05,
|
|
"loss": 3.4861,
|
|
"step": 1523200
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.505403367351017e-05,
|
|
"loss": 3.4772,
|
|
"step": 1523712
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.504564772599965e-05,
|
|
"loss": 3.4848,
|
|
"step": 1524224
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.503726177848913e-05,
|
|
"loss": 3.4886,
|
|
"step": 1524736
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.502887583097861e-05,
|
|
"loss": 3.4799,
|
|
"step": 1525248
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.502048988346809e-05,
|
|
"loss": 3.4803,
|
|
"step": 1525760
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.5012103935957572e-05,
|
|
"loss": 3.4776,
|
|
"step": 1526272
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.8492953777313232,
|
|
"eval_runtime": 306.1225,
|
|
"eval_samples_per_second": 1246.53,
|
|
"eval_steps_per_second": 38.955,
|
|
"step": 1526400
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.500373436725078e-05,
|
|
"loss": 3.4794,
|
|
"step": 1526784
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4995348419740257e-05,
|
|
"loss": 3.469,
|
|
"step": 1527296
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4986962472229737e-05,
|
|
"loss": 3.4871,
|
|
"step": 1527808
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4978576524719217e-05,
|
|
"loss": 3.4807,
|
|
"step": 1528320
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.49701905772087e-05,
|
|
"loss": 3.4933,
|
|
"step": 1528832
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.496182100850191e-05,
|
|
"loss": 3.4718,
|
|
"step": 1529344
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.495343506099139e-05,
|
|
"loss": 3.4823,
|
|
"step": 1529856
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.494504911348087e-05,
|
|
"loss": 3.4642,
|
|
"step": 1530368
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.493666316597035e-05,
|
|
"loss": 3.4832,
|
|
"step": 1530880
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.492827721845983e-05,
|
|
"loss": 3.4703,
|
|
"step": 1531392
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.491989127094931e-05,
|
|
"loss": 3.486,
|
|
"step": 1531904
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4911521702242522e-05,
|
|
"loss": 3.4866,
|
|
"step": 1532416
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4903135754732002e-05,
|
|
"loss": 3.4634,
|
|
"step": 1532928
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4894749807221482e-05,
|
|
"loss": 3.4729,
|
|
"step": 1533440
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4886363859710962e-05,
|
|
"loss": 3.4756,
|
|
"step": 1533952
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.487797791220044e-05,
|
|
"loss": 3.4666,
|
|
"step": 1534464
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.486959196468992e-05,
|
|
"loss": 3.4732,
|
|
"step": 1534976
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4861206017179402e-05,
|
|
"loss": 3.4775,
|
|
"step": 1535488
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4852836448472614e-05,
|
|
"loss": 3.4754,
|
|
"step": 1536000
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4844466879765823e-05,
|
|
"loss": 3.4941,
|
|
"step": 1536512
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4836080932255303e-05,
|
|
"loss": 3.4741,
|
|
"step": 1537024
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4827694984744783e-05,
|
|
"loss": 3.4752,
|
|
"step": 1537536
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4819309037234263e-05,
|
|
"loss": 3.4781,
|
|
"step": 1538048
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4810923089723743e-05,
|
|
"loss": 3.4843,
|
|
"step": 1538560
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4802553521016956e-05,
|
|
"loss": 3.4716,
|
|
"step": 1539072
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4794167573506436e-05,
|
|
"loss": 3.4819,
|
|
"step": 1539584
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4785781625995912e-05,
|
|
"loss": 3.4698,
|
|
"step": 1540096
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4777395678485392e-05,
|
|
"loss": 3.4744,
|
|
"step": 1540608
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4769009730974872e-05,
|
|
"loss": 3.4671,
|
|
"step": 1541120
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.4760623783464355e-05,
|
|
"loss": 3.472,
|
|
"step": 1541632
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4752254214757565e-05,
|
|
"loss": 3.475,
|
|
"step": 1542144
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4743868267247045e-05,
|
|
"loss": 3.4804,
|
|
"step": 1542656
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4735482319736524e-05,
|
|
"loss": 3.4808,
|
|
"step": 1543168
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4727096372226004e-05,
|
|
"loss": 3.4845,
|
|
"step": 1543680
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4718710424715484e-05,
|
|
"loss": 3.4702,
|
|
"step": 1544192
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4710340856008697e-05,
|
|
"loss": 3.4837,
|
|
"step": 1544704
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4701954908498177e-05,
|
|
"loss": 3.4736,
|
|
"step": 1545216
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4693568960987657e-05,
|
|
"loss": 3.4735,
|
|
"step": 1545728
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4685183013477137e-05,
|
|
"loss": 3.4636,
|
|
"step": 1546240
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4676813444770346e-05,
|
|
"loss": 3.4694,
|
|
"step": 1546752
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4668427497259826e-05,
|
|
"loss": 3.4685,
|
|
"step": 1547264
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.466004154974931e-05,
|
|
"loss": 3.4831,
|
|
"step": 1547776
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.465165560223879e-05,
|
|
"loss": 3.4668,
|
|
"step": 1548288
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.464326965472827e-05,
|
|
"loss": 3.4697,
|
|
"step": 1548800
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4634883707217746e-05,
|
|
"loss": 3.4876,
|
|
"step": 1549312
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4626497759707226e-05,
|
|
"loss": 3.4775,
|
|
"step": 1549824
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4618111812196706e-05,
|
|
"loss": 3.4795,
|
|
"step": 1550336
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4609742243489918e-05,
|
|
"loss": 3.4651,
|
|
"step": 1550848
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.460137267478313e-05,
|
|
"loss": 3.459,
|
|
"step": 1551360
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.459298672727261e-05,
|
|
"loss": 3.4809,
|
|
"step": 1551872
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.458460077976209e-05,
|
|
"loss": 3.4767,
|
|
"step": 1552384
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4576214832251567e-05,
|
|
"loss": 3.469,
|
|
"step": 1552896
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4567828884741047e-05,
|
|
"loss": 3.4664,
|
|
"step": 1553408
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4559459316034263e-05,
|
|
"loss": 3.4727,
|
|
"step": 1553920
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4551073368523743e-05,
|
|
"loss": 3.4528,
|
|
"step": 1554432
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.454268742101322e-05,
|
|
"loss": 3.4782,
|
|
"step": 1554944
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.45343014735027e-05,
|
|
"loss": 3.4594,
|
|
"step": 1555456
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4525931904795912e-05,
|
|
"loss": 3.4755,
|
|
"step": 1555968
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4517545957285392e-05,
|
|
"loss": 3.4762,
|
|
"step": 1556480
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.450916000977487e-05,
|
|
"loss": 3.4645,
|
|
"step": 1556992
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.450077406226435e-05,
|
|
"loss": 3.4714,
|
|
"step": 1557504
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.449238811475383e-05,
|
|
"loss": 3.4698,
|
|
"step": 1558016
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.448401854604704e-05,
|
|
"loss": 3.464,
|
|
"step": 1558528
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.447563259853652e-05,
|
|
"loss": 3.4689,
|
|
"step": 1559040
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4467246651026e-05,
|
|
"loss": 3.4722,
|
|
"step": 1559552
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.445886070351548e-05,
|
|
"loss": 3.4677,
|
|
"step": 1560064
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4450491134808693e-05,
|
|
"loss": 3.4656,
|
|
"step": 1560576
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4442105187298173e-05,
|
|
"loss": 3.4576,
|
|
"step": 1561088
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4433719239787653e-05,
|
|
"loss": 3.4594,
|
|
"step": 1561600
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4425333292277133e-05,
|
|
"loss": 3.4701,
|
|
"step": 1562112
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4416963723570342e-05,
|
|
"loss": 3.4807,
|
|
"step": 1562624
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4408577776059825e-05,
|
|
"loss": 3.4649,
|
|
"step": 1563136
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4400191828549305e-05,
|
|
"loss": 3.4795,
|
|
"step": 1563648
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4391805881038785e-05,
|
|
"loss": 3.4785,
|
|
"step": 1564160
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4383436312331994e-05,
|
|
"loss": 3.469,
|
|
"step": 1564672
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4375050364821474e-05,
|
|
"loss": 3.4705,
|
|
"step": 1565184
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4366664417310954e-05,
|
|
"loss": 3.4768,
|
|
"step": 1565696
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4358278469800434e-05,
|
|
"loss": 3.4711,
|
|
"step": 1566208
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4349908901093647e-05,
|
|
"loss": 3.4697,
|
|
"step": 1566720
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4341522953583127e-05,
|
|
"loss": 3.4753,
|
|
"step": 1567232
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4333137006072607e-05,
|
|
"loss": 3.4715,
|
|
"step": 1567744
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4324751058562087e-05,
|
|
"loss": 3.4754,
|
|
"step": 1568256
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4316365111051567e-05,
|
|
"loss": 3.4707,
|
|
"step": 1568768
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.430799554234478e-05,
|
|
"loss": 3.4562,
|
|
"step": 1569280
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.429960959483426e-05,
|
|
"loss": 3.4571,
|
|
"step": 1569792
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.429122364732374e-05,
|
|
"loss": 3.4669,
|
|
"step": 1570304
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.428283769981322e-05,
|
|
"loss": 3.4727,
|
|
"step": 1570816
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4274451752302695e-05,
|
|
"loss": 3.4676,
|
|
"step": 1571328
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.4266082183595908e-05,
|
|
"loss": 3.4704,
|
|
"step": 1571840
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.4257696236085388e-05,
|
|
"loss": 3.4642,
|
|
"step": 1572352
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.424931028857487e-05,
|
|
"loss": 3.4612,
|
|
"step": 1572864
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.4240924341064348e-05,
|
|
"loss": 3.4669,
|
|
"step": 1573376
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.423255477235756e-05,
|
|
"loss": 3.4695,
|
|
"step": 1573888
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.422416882484704e-05,
|
|
"loss": 3.4789,
|
|
"step": 1574400
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.4215782877336517e-05,
|
|
"loss": 3.4774,
|
|
"step": 1574912
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.4207396929825997e-05,
|
|
"loss": 3.4715,
|
|
"step": 1575424
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.419901098231548e-05,
|
|
"loss": 3.461,
|
|
"step": 1575936
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.419062503480496e-05,
|
|
"loss": 3.4771,
|
|
"step": 1576448
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.418225546609817e-05,
|
|
"loss": 3.4657,
|
|
"step": 1576960
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.417386951858765e-05,
|
|
"loss": 3.4759,
|
|
"step": 1577472
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.416548357107713e-05,
|
|
"loss": 3.4676,
|
|
"step": 1577984
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.415709762356661e-05,
|
|
"loss": 3.4622,
|
|
"step": 1578496
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.414871167605609e-05,
|
|
"loss": 3.4789,
|
|
"step": 1579008
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.41403421073493e-05,
|
|
"loss": 3.4736,
|
|
"step": 1579520
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.413195615983878e-05,
|
|
"loss": 3.4671,
|
|
"step": 1580032
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.412357021232826e-05,
|
|
"loss": 3.4689,
|
|
"step": 1580544
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.411518426481774e-05,
|
|
"loss": 3.4685,
|
|
"step": 1581056
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.410679831730722e-05,
|
|
"loss": 3.455,
|
|
"step": 1581568
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.40984123697967e-05,
|
|
"loss": 3.4563,
|
|
"step": 1582080
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.4090042801089914e-05,
|
|
"loss": 3.4699,
|
|
"step": 1582592
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.4081656853579394e-05,
|
|
"loss": 3.4639,
|
|
"step": 1583104
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.4073270906068874e-05,
|
|
"loss": 3.4664,
|
|
"step": 1583616
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.406488495855835e-05,
|
|
"loss": 3.4655,
|
|
"step": 1584128
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.405649901104783e-05,
|
|
"loss": 3.4627,
|
|
"step": 1584640
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.4048129442341043e-05,
|
|
"loss": 3.4808,
|
|
"step": 1585152
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.4039743494830526e-05,
|
|
"loss": 3.4826,
|
|
"step": 1585664
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.4031357547320003e-05,
|
|
"loss": 3.4724,
|
|
"step": 1586176
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.4022971599809482e-05,
|
|
"loss": 3.4704,
|
|
"step": 1586688
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.4014602031102695e-05,
|
|
"loss": 3.4649,
|
|
"step": 1587200
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.400621608359217e-05,
|
|
"loss": 3.4623,
|
|
"step": 1587712
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3997830136081655e-05,
|
|
"loss": 3.4621,
|
|
"step": 1588224
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3989444188571135e-05,
|
|
"loss": 3.466,
|
|
"step": 1588736
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3981058241060615e-05,
|
|
"loss": 3.4659,
|
|
"step": 1589248
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3972672293550095e-05,
|
|
"loss": 3.4731,
|
|
"step": 1589760
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3964286346039575e-05,
|
|
"loss": 3.4671,
|
|
"step": 1590272
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3955916777332784e-05,
|
|
"loss": 3.4698,
|
|
"step": 1590784
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3947530829822264e-05,
|
|
"loss": 3.4679,
|
|
"step": 1591296
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3939144882311744e-05,
|
|
"loss": 3.4646,
|
|
"step": 1591808
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3930758934801227e-05,
|
|
"loss": 3.4665,
|
|
"step": 1592320
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3922389366094436e-05,
|
|
"loss": 3.4761,
|
|
"step": 1592832
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3914003418583916e-05,
|
|
"loss": 3.4807,
|
|
"step": 1593344
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3905617471073396e-05,
|
|
"loss": 3.4726,
|
|
"step": 1593856
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3897231523562876e-05,
|
|
"loss": 3.4668,
|
|
"step": 1594368
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3888845576052356e-05,
|
|
"loss": 3.4756,
|
|
"step": 1594880
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3880459628541836e-05,
|
|
"loss": 3.4665,
|
|
"step": 1595392
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3872073681031316e-05,
|
|
"loss": 3.4785,
|
|
"step": 1595904
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3863687733520796e-05,
|
|
"loss": 3.4654,
|
|
"step": 1596416
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3855318164814005e-05,
|
|
"loss": 3.4712,
|
|
"step": 1596928
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3846932217303485e-05,
|
|
"loss": 3.4691,
|
|
"step": 1597440
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3838562648596697e-05,
|
|
"loss": 3.4799,
|
|
"step": 1597952
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3830176701086177e-05,
|
|
"loss": 3.4617,
|
|
"step": 1598464
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3821790753575657e-05,
|
|
"loss": 3.4667,
|
|
"step": 1598976
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3813404806065137e-05,
|
|
"loss": 3.4741,
|
|
"step": 1599488
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.380503523735835e-05,
|
|
"loss": 3.4656,
|
|
"step": 1600000
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3796649289847826e-05,
|
|
"loss": 3.473,
|
|
"step": 1600512
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.378826334233731e-05,
|
|
"loss": 3.4758,
|
|
"step": 1601024
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.377987739482679e-05,
|
|
"loss": 3.4699,
|
|
"step": 1601536
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.3771507826120002e-05,
|
|
"loss": 3.4654,
|
|
"step": 1602048
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.376312187860948e-05,
|
|
"loss": 3.4657,
|
|
"step": 1602560
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 3.8506855964660645,
|
|
"eval_runtime": 305.8897,
|
|
"eval_samples_per_second": 1247.479,
|
|
"eval_steps_per_second": 38.985,
|
|
"step": 1602720
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.375473593109896e-05,
|
|
"loss": 3.4618,
|
|
"step": 1603072
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.374636636239217e-05,
|
|
"loss": 3.4558,
|
|
"step": 1603584
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.373798041488165e-05,
|
|
"loss": 3.4761,
|
|
"step": 1604096
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.372959446737113e-05,
|
|
"loss": 3.4693,
|
|
"step": 1604608
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.372120851986061e-05,
|
|
"loss": 3.4782,
|
|
"step": 1605120
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3712838951153823e-05,
|
|
"loss": 3.4667,
|
|
"step": 1605632
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.37044530036433e-05,
|
|
"loss": 3.4663,
|
|
"step": 1606144
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.369606705613278e-05,
|
|
"loss": 3.4581,
|
|
"step": 1606656
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3687681108622263e-05,
|
|
"loss": 3.4648,
|
|
"step": 1607168
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3679295161111743e-05,
|
|
"loss": 3.4635,
|
|
"step": 1607680
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3670909213601223e-05,
|
|
"loss": 3.473,
|
|
"step": 1608192
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3662539644894432e-05,
|
|
"loss": 3.4747,
|
|
"step": 1608704
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3654153697383912e-05,
|
|
"loss": 3.45,
|
|
"step": 1609216
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3645767749873392e-05,
|
|
"loss": 3.4627,
|
|
"step": 1609728
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3637381802362872e-05,
|
|
"loss": 3.4651,
|
|
"step": 1610240
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3629012233656085e-05,
|
|
"loss": 3.4535,
|
|
"step": 1610752
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3620626286145565e-05,
|
|
"loss": 3.4606,
|
|
"step": 1611264
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3612240338635045e-05,
|
|
"loss": 3.464,
|
|
"step": 1611776
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3603854391124525e-05,
|
|
"loss": 3.4621,
|
|
"step": 1612288
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3595484822417734e-05,
|
|
"loss": 3.4789,
|
|
"step": 1612800
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3587098874907214e-05,
|
|
"loss": 3.4661,
|
|
"step": 1613312
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3578712927396697e-05,
|
|
"loss": 3.469,
|
|
"step": 1613824
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3570326979886177e-05,
|
|
"loss": 3.4669,
|
|
"step": 1614336
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3561941032375657e-05,
|
|
"loss": 3.4703,
|
|
"step": 1614848
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3553571463668866e-05,
|
|
"loss": 3.46,
|
|
"step": 1615360
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3545185516158346e-05,
|
|
"loss": 3.4674,
|
|
"step": 1615872
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3536799568647826e-05,
|
|
"loss": 3.4636,
|
|
"step": 1616384
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3528413621137306e-05,
|
|
"loss": 3.4596,
|
|
"step": 1616896
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3520027673626786e-05,
|
|
"loss": 3.4552,
|
|
"step": 1617408
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.3511658104919998e-05,
|
|
"loss": 3.4619,
|
|
"step": 1617920
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3503272157409478e-05,
|
|
"loss": 3.463,
|
|
"step": 1618432
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3494886209898955e-05,
|
|
"loss": 3.4678,
|
|
"step": 1618944
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3486500262388435e-05,
|
|
"loss": 3.4707,
|
|
"step": 1619456
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.347813069368165e-05,
|
|
"loss": 3.473,
|
|
"step": 1619968
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.346974474617113e-05,
|
|
"loss": 3.458,
|
|
"step": 1620480
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3461358798660607e-05,
|
|
"loss": 3.4679,
|
|
"step": 1620992
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3452972851150087e-05,
|
|
"loss": 3.4622,
|
|
"step": 1621504
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3444586903639567e-05,
|
|
"loss": 3.4644,
|
|
"step": 1622016
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.343621733493278e-05,
|
|
"loss": 3.4523,
|
|
"step": 1622528
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.342783138742226e-05,
|
|
"loss": 3.4554,
|
|
"step": 1623040
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.341944543991174e-05,
|
|
"loss": 3.4595,
|
|
"step": 1623552
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.341105949240122e-05,
|
|
"loss": 3.4667,
|
|
"step": 1624064
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.34026735448907e-05,
|
|
"loss": 3.458,
|
|
"step": 1624576
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.339430397618391e-05,
|
|
"loss": 3.4581,
|
|
"step": 1625088
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.338591802867339e-05,
|
|
"loss": 3.4764,
|
|
"step": 1625600
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3377532081162872e-05,
|
|
"loss": 3.4652,
|
|
"step": 1626112
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3369146133652352e-05,
|
|
"loss": 3.4679,
|
|
"step": 1626624
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.336076018614183e-05,
|
|
"loss": 3.4565,
|
|
"step": 1627136
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.335239061743504e-05,
|
|
"loss": 3.4453,
|
|
"step": 1627648
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.334400466992452e-05,
|
|
"loss": 3.4697,
|
|
"step": 1628160
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3335618722414e-05,
|
|
"loss": 3.4641,
|
|
"step": 1628672
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.332723277490348e-05,
|
|
"loss": 3.4582,
|
|
"step": 1629184
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.331884682739296e-05,
|
|
"loss": 3.4568,
|
|
"step": 1629696
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3310477258686173e-05,
|
|
"loss": 3.4648,
|
|
"step": 1630208
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3302091311175653e-05,
|
|
"loss": 3.4359,
|
|
"step": 1630720
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3293705363665133e-05,
|
|
"loss": 3.4667,
|
|
"step": 1631232
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.328531941615461e-05,
|
|
"loss": 3.4536,
|
|
"step": 1631744
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.327693346864409e-05,
|
|
"loss": 3.4573,
|
|
"step": 1632256
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3268563899937305e-05,
|
|
"loss": 3.467,
|
|
"step": 1632768
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3260177952426785e-05,
|
|
"loss": 3.4567,
|
|
"step": 1633280
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3251792004916262e-05,
|
|
"loss": 3.4567,
|
|
"step": 1633792
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3243406057405742e-05,
|
|
"loss": 3.4632,
|
|
"step": 1634304
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3235036488698954e-05,
|
|
"loss": 3.4522,
|
|
"step": 1634816
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3226650541188434e-05,
|
|
"loss": 3.4529,
|
|
"step": 1635328
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3218264593677914e-05,
|
|
"loss": 3.4627,
|
|
"step": 1635840
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3209878646167394e-05,
|
|
"loss": 3.4588,
|
|
"step": 1636352
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3201509077460607e-05,
|
|
"loss": 3.4525,
|
|
"step": 1636864
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3193123129950083e-05,
|
|
"loss": 3.4494,
|
|
"step": 1637376
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3184737182439563e-05,
|
|
"loss": 3.4468,
|
|
"step": 1637888
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3176351234929043e-05,
|
|
"loss": 3.4568,
|
|
"step": 1638400
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3167965287418527e-05,
|
|
"loss": 3.4668,
|
|
"step": 1638912
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3159595718711736e-05,
|
|
"loss": 3.4543,
|
|
"step": 1639424
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3151209771201216e-05,
|
|
"loss": 3.4685,
|
|
"step": 1639936
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3142823823690696e-05,
|
|
"loss": 3.4668,
|
|
"step": 1640448
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3134454254983905e-05,
|
|
"loss": 3.4613,
|
|
"step": 1640960
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3126068307473388e-05,
|
|
"loss": 3.4528,
|
|
"step": 1641472
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3117682359962868e-05,
|
|
"loss": 3.4671,
|
|
"step": 1641984
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3109296412452348e-05,
|
|
"loss": 3.4583,
|
|
"step": 1642496
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3100926843745557e-05,
|
|
"loss": 3.4542,
|
|
"step": 1643008
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3092540896235037e-05,
|
|
"loss": 3.4663,
|
|
"step": 1643520
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3084154948724517e-05,
|
|
"loss": 3.4596,
|
|
"step": 1644032
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3075769001213997e-05,
|
|
"loss": 3.4668,
|
|
"step": 1644544
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.306738305370348e-05,
|
|
"loss": 3.4597,
|
|
"step": 1645056
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.305901348499669e-05,
|
|
"loss": 3.4479,
|
|
"step": 1645568
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.305062753748617e-05,
|
|
"loss": 3.4425,
|
|
"step": 1646080
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.304224158997565e-05,
|
|
"loss": 3.4566,
|
|
"step": 1646592
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.303385564246513e-05,
|
|
"loss": 3.4608,
|
|
"step": 1647104
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.302546969495461e-05,
|
|
"loss": 3.4552,
|
|
"step": 1647616
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.301710012624782e-05,
|
|
"loss": 3.4567,
|
|
"step": 1648128
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.30087141787373e-05,
|
|
"loss": 3.4565,
|
|
"step": 1648640
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.300032823122678e-05,
|
|
"loss": 3.4458,
|
|
"step": 1649152
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.299194228371626e-05,
|
|
"loss": 3.4577,
|
|
"step": 1649664
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.298357271500947e-05,
|
|
"loss": 3.455,
|
|
"step": 1650176
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.297518676749895e-05,
|
|
"loss": 3.467,
|
|
"step": 1650688
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.296680081998843e-05,
|
|
"loss": 3.4691,
|
|
"step": 1651200
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.295841487247791e-05,
|
|
"loss": 3.4601,
|
|
"step": 1651712
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.295002892496739e-05,
|
|
"loss": 3.4502,
|
|
"step": 1652224
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.294164297745687e-05,
|
|
"loss": 3.4587,
|
|
"step": 1652736
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2933273408750083e-05,
|
|
"loss": 3.4558,
|
|
"step": 1653248
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.292488746123956e-05,
|
|
"loss": 3.4672,
|
|
"step": 1653760
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2916501513729043e-05,
|
|
"loss": 3.4537,
|
|
"step": 1654272
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2908115566218523e-05,
|
|
"loss": 3.4516,
|
|
"step": 1654784
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2899729618708003e-05,
|
|
"loss": 3.4639,
|
|
"step": 1655296
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2891360050001212e-05,
|
|
"loss": 3.4607,
|
|
"step": 1655808
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.288297410249069e-05,
|
|
"loss": 3.456,
|
|
"step": 1656320
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.287458815498017e-05,
|
|
"loss": 3.4605,
|
|
"step": 1656832
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.286620220746965e-05,
|
|
"loss": 3.4528,
|
|
"step": 1657344
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2857816259959135e-05,
|
|
"loss": 3.4516,
|
|
"step": 1657856
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2849430312448615e-05,
|
|
"loss": 3.4393,
|
|
"step": 1658368
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2841060743741824e-05,
|
|
"loss": 3.4596,
|
|
"step": 1658880
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2832674796231304e-05,
|
|
"loss": 3.4585,
|
|
"step": 1659392
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2824288848720784e-05,
|
|
"loss": 3.4545,
|
|
"step": 1659904
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2815902901210264e-05,
|
|
"loss": 3.4543,
|
|
"step": 1660416
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2807516953699744e-05,
|
|
"loss": 3.4476,
|
|
"step": 1660928
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2799147384992956e-05,
|
|
"loss": 3.4696,
|
|
"step": 1661440
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2790761437482436e-05,
|
|
"loss": 3.4711,
|
|
"step": 1661952
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2782375489971916e-05,
|
|
"loss": 3.4631,
|
|
"step": 1662464
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2773989542461393e-05,
|
|
"loss": 3.46,
|
|
"step": 1662976
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2765603594950873e-05,
|
|
"loss": 3.4531,
|
|
"step": 1663488
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.275723402624409e-05,
|
|
"loss": 3.4517,
|
|
"step": 1664000
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2748848078733565e-05,
|
|
"loss": 3.449,
|
|
"step": 1664512
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2740462131223045e-05,
|
|
"loss": 3.4514,
|
|
"step": 1665024
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2732076183712525e-05,
|
|
"loss": 3.4575,
|
|
"step": 1665536
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2723690236202005e-05,
|
|
"loss": 3.4585,
|
|
"step": 1666048
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2715304288691485e-05,
|
|
"loss": 3.4605,
|
|
"step": 1666560
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2706934719984697e-05,
|
|
"loss": 3.4556,
|
|
"step": 1667072
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2698548772474177e-05,
|
|
"loss": 3.4585,
|
|
"step": 1667584
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2690162824963657e-05,
|
|
"loss": 3.4534,
|
|
"step": 1668096
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2681776877453137e-05,
|
|
"loss": 3.4568,
|
|
"step": 1668608
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2673407308746346e-05,
|
|
"loss": 3.4565,
|
|
"step": 1669120
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2665021361235826e-05,
|
|
"loss": 3.4682,
|
|
"step": 1669632
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2656635413725306e-05,
|
|
"loss": 3.4662,
|
|
"step": 1670144
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.264824946621479e-05,
|
|
"loss": 3.4549,
|
|
"step": 1670656
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.263986351870427e-05,
|
|
"loss": 3.4625,
|
|
"step": 1671168
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.263149394999748e-05,
|
|
"loss": 3.4565,
|
|
"step": 1671680
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.262310800248696e-05,
|
|
"loss": 3.4639,
|
|
"step": 1672192
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.261472205497644e-05,
|
|
"loss": 3.4546,
|
|
"step": 1672704
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.260633610746592e-05,
|
|
"loss": 3.4603,
|
|
"step": 1673216
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.25979501599554e-05,
|
|
"loss": 3.4567,
|
|
"step": 1673728
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.258958059124861e-05,
|
|
"loss": 3.4645,
|
|
"step": 1674240
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.258119464373809e-05,
|
|
"loss": 3.4504,
|
|
"step": 1674752
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2572808696227568e-05,
|
|
"loss": 3.4572,
|
|
"step": 1675264
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2564422748717048e-05,
|
|
"loss": 3.4597,
|
|
"step": 1675776
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.255605318001026e-05,
|
|
"loss": 3.4613,
|
|
"step": 1676288
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2547667232499743e-05,
|
|
"loss": 3.4621,
|
|
"step": 1676800
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.253928128498922e-05,
|
|
"loss": 3.4582,
|
|
"step": 1677312
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.25308953374787e-05,
|
|
"loss": 3.4663,
|
|
"step": 1677824
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.252250938996818e-05,
|
|
"loss": 3.4473,
|
|
"step": 1678336
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.251412344245766e-05,
|
|
"loss": 3.4584,
|
|
"step": 1678848
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.8512587547302246,
|
|
"eval_runtime": 302.9603,
|
|
"eval_samples_per_second": 1259.541,
|
|
"eval_steps_per_second": 39.362,
|
|
"step": 1679040
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.250573749494714e-05,
|
|
"loss": 3.4567,
|
|
"step": 1679360
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2497367926240352e-05,
|
|
"loss": 3.4434,
|
|
"step": 1679872
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2488981978729832e-05,
|
|
"loss": 3.466,
|
|
"step": 1680384
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2480596031219312e-05,
|
|
"loss": 3.456,
|
|
"step": 1680896
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2472210083708792e-05,
|
|
"loss": 3.4683,
|
|
"step": 1681408
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2463840515002e-05,
|
|
"loss": 3.4527,
|
|
"step": 1681920
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.245545456749148e-05,
|
|
"loss": 3.4512,
|
|
"step": 1682432
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.244706861998096e-05,
|
|
"loss": 3.451,
|
|
"step": 1682944
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2438682672470444e-05,
|
|
"loss": 3.4515,
|
|
"step": 1683456
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2430296724959924e-05,
|
|
"loss": 3.451,
|
|
"step": 1683968
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.24219107774494e-05,
|
|
"loss": 3.4618,
|
|
"step": 1684480
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2413541208742613e-05,
|
|
"loss": 3.4653,
|
|
"step": 1684992
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2405155261232093e-05,
|
|
"loss": 3.4413,
|
|
"step": 1685504
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2396769313721573e-05,
|
|
"loss": 3.4477,
|
|
"step": 1686016
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2388383366211053e-05,
|
|
"loss": 3.4557,
|
|
"step": 1686528
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2379997418700533e-05,
|
|
"loss": 3.4441,
|
|
"step": 1687040
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2371611471190013e-05,
|
|
"loss": 3.4493,
|
|
"step": 1687552
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2363241902483222e-05,
|
|
"loss": 3.4512,
|
|
"step": 1688064
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2354855954972702e-05,
|
|
"loss": 3.4515,
|
|
"step": 1688576
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2346470007462182e-05,
|
|
"loss": 3.4655,
|
|
"step": 1689088
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2338084059951662e-05,
|
|
"loss": 3.4566,
|
|
"step": 1689600
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2329698112441146e-05,
|
|
"loss": 3.4533,
|
|
"step": 1690112
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2321328543734355e-05,
|
|
"loss": 3.4563,
|
|
"step": 1690624
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2312942596223835e-05,
|
|
"loss": 3.4579,
|
|
"step": 1691136
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2304556648713315e-05,
|
|
"loss": 3.4512,
|
|
"step": 1691648
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2296170701202795e-05,
|
|
"loss": 3.4552,
|
|
"step": 1692160
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2287801132496007e-05,
|
|
"loss": 3.4494,
|
|
"step": 1692672
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2279415184985487e-05,
|
|
"loss": 3.4476,
|
|
"step": 1693184
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2271029237474967e-05,
|
|
"loss": 3.4475,
|
|
"step": 1693696
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2262643289964447e-05,
|
|
"loss": 3.4518,
|
|
"step": 1694208
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2254273721257656e-05,
|
|
"loss": 3.4529,
|
|
"step": 1694720
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2245887773747136e-05,
|
|
"loss": 3.4536,
|
|
"step": 1695232
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2237501826236616e-05,
|
|
"loss": 3.4602,
|
|
"step": 1695744
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.22291158787261e-05,
|
|
"loss": 3.4605,
|
|
"step": 1696256
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.222072993121558e-05,
|
|
"loss": 3.4472,
|
|
"step": 1696768
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2212360362508788e-05,
|
|
"loss": 3.4562,
|
|
"step": 1697280
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2203974414998268e-05,
|
|
"loss": 3.4533,
|
|
"step": 1697792
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2195588467487748e-05,
|
|
"loss": 3.4504,
|
|
"step": 1698304
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2187202519977228e-05,
|
|
"loss": 3.4401,
|
|
"step": 1698816
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.217883295127044e-05,
|
|
"loss": 3.4443,
|
|
"step": 1699328
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.217044700375992e-05,
|
|
"loss": 3.4481,
|
|
"step": 1699840
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.21620610562494e-05,
|
|
"loss": 3.4552,
|
|
"step": 1700352
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2153675108738877e-05,
|
|
"loss": 3.4465,
|
|
"step": 1700864
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2145289161228357e-05,
|
|
"loss": 3.4457,
|
|
"step": 1701376
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2136903213717837e-05,
|
|
"loss": 3.4682,
|
|
"step": 1701888
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.212851726620732e-05,
|
|
"loss": 3.4524,
|
|
"step": 1702400
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.21201313186968e-05,
|
|
"loss": 3.4576,
|
|
"step": 1702912
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2111778128793742e-05,
|
|
"loss": 3.4424,
|
|
"step": 1703424
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2103392181283222e-05,
|
|
"loss": 3.4343,
|
|
"step": 1703936
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2095006233772702e-05,
|
|
"loss": 3.4625,
|
|
"step": 1704448
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2086620286262182e-05,
|
|
"loss": 3.4498,
|
|
"step": 1704960
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2078234338751662e-05,
|
|
"loss": 3.4457,
|
|
"step": 1705472
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2069864770044874e-05,
|
|
"loss": 3.4469,
|
|
"step": 1705984
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.206147882253435e-05,
|
|
"loss": 3.4506,
|
|
"step": 1706496
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.205309287502383e-05,
|
|
"loss": 3.4289,
|
|
"step": 1707008
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.204470692751331e-05,
|
|
"loss": 3.4515,
|
|
"step": 1707520
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.203632098000279e-05,
|
|
"loss": 3.4464,
|
|
"step": 1708032
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.202793503249227e-05,
|
|
"loss": 3.4416,
|
|
"step": 1708544
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2019565463785483e-05,
|
|
"loss": 3.4602,
|
|
"step": 1709056
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2011179516274963e-05,
|
|
"loss": 3.446,
|
|
"step": 1709568
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2002793568764443e-05,
|
|
"loss": 3.445,
|
|
"step": 1710080
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1994407621253923e-05,
|
|
"loss": 3.4529,
|
|
"step": 1710592
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1986038052547135e-05,
|
|
"loss": 3.4402,
|
|
"step": 1711104
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1977652105036615e-05,
|
|
"loss": 3.4431,
|
|
"step": 1711616
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1969266157526095e-05,
|
|
"loss": 3.4476,
|
|
"step": 1712128
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1960880210015575e-05,
|
|
"loss": 3.4449,
|
|
"step": 1712640
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1952510641308784e-05,
|
|
"loss": 3.4468,
|
|
"step": 1713152
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1944124693798264e-05,
|
|
"loss": 3.4393,
|
|
"step": 1713664
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1935738746287744e-05,
|
|
"loss": 3.4315,
|
|
"step": 1714176
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1927352798777224e-05,
|
|
"loss": 3.4446,
|
|
"step": 1714688
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1918966851266708e-05,
|
|
"loss": 3.4563,
|
|
"step": 1715200
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1910597282559917e-05,
|
|
"loss": 3.4405,
|
|
"step": 1715712
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1902211335049397e-05,
|
|
"loss": 3.461,
|
|
"step": 1716224
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1893825387538877e-05,
|
|
"loss": 3.455,
|
|
"step": 1716736
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1885439440028353e-05,
|
|
"loss": 3.4542,
|
|
"step": 1717248
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.187706987132157e-05,
|
|
"loss": 3.4402,
|
|
"step": 1717760
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.186868392381105e-05,
|
|
"loss": 3.4584,
|
|
"step": 1718272
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.186029797630053e-05,
|
|
"loss": 3.448,
|
|
"step": 1718784
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1851912028790006e-05,
|
|
"loss": 3.4434,
|
|
"step": 1719296
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1843542460083218e-05,
|
|
"loss": 3.4578,
|
|
"step": 1719808
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1835156512572698e-05,
|
|
"loss": 3.4486,
|
|
"step": 1720320
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1826770565062178e-05,
|
|
"loss": 3.4553,
|
|
"step": 1720832
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1818384617551658e-05,
|
|
"loss": 3.4485,
|
|
"step": 1721344
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.181001504884487e-05,
|
|
"loss": 3.4321,
|
|
"step": 1721856
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.180162910133435e-05,
|
|
"loss": 3.4323,
|
|
"step": 1722368
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1793243153823827e-05,
|
|
"loss": 3.4412,
|
|
"step": 1722880
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1784857206313307e-05,
|
|
"loss": 3.4508,
|
|
"step": 1723392
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.177647125880279e-05,
|
|
"loss": 3.4461,
|
|
"step": 1723904
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.1768101690096003e-05,
|
|
"loss": 3.4438,
|
|
"step": 1724416
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.175971574258548e-05,
|
|
"loss": 3.4469,
|
|
"step": 1724928
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.175132979507496e-05,
|
|
"loss": 3.4373,
|
|
"step": 1725440
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.174294384756444e-05,
|
|
"loss": 3.4419,
|
|
"step": 1725952
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.173457427885765e-05,
|
|
"loss": 3.4419,
|
|
"step": 1726464
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.172618833134713e-05,
|
|
"loss": 3.4609,
|
|
"step": 1726976
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.171780238383661e-05,
|
|
"loss": 3.4574,
|
|
"step": 1727488
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.170941643632609e-05,
|
|
"loss": 3.4477,
|
|
"step": 1728000
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.170103048881557e-05,
|
|
"loss": 3.4405,
|
|
"step": 1728512
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.169264454130505e-05,
|
|
"loss": 3.4506,
|
|
"step": 1729024
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.168427497259826e-05,
|
|
"loss": 3.4387,
|
|
"step": 1729536
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1675889025087744e-05,
|
|
"loss": 3.4593,
|
|
"step": 1730048
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1667503077577224e-05,
|
|
"loss": 3.4422,
|
|
"step": 1730560
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1659117130066704e-05,
|
|
"loss": 3.4447,
|
|
"step": 1731072
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1650731182556184e-05,
|
|
"loss": 3.4497,
|
|
"step": 1731584
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1642361613849393e-05,
|
|
"loss": 3.4539,
|
|
"step": 1732096
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1633975666338873e-05,
|
|
"loss": 3.4395,
|
|
"step": 1732608
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1625589718828353e-05,
|
|
"loss": 3.4524,
|
|
"step": 1733120
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1617203771317833e-05,
|
|
"loss": 3.4388,
|
|
"step": 1733632
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1608817823807313e-05,
|
|
"loss": 3.4379,
|
|
"step": 1734144
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1600431876296793e-05,
|
|
"loss": 3.4302,
|
|
"step": 1734656
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1592062307590005e-05,
|
|
"loss": 3.4462,
|
|
"step": 1735168
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.158367636007948e-05,
|
|
"loss": 3.4459,
|
|
"step": 1735680
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.157529041256896e-05,
|
|
"loss": 3.4379,
|
|
"step": 1736192
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1566904465058445e-05,
|
|
"loss": 3.4455,
|
|
"step": 1736704
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1558518517547925e-05,
|
|
"loss": 3.4406,
|
|
"step": 1737216
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1550148948841134e-05,
|
|
"loss": 3.4543,
|
|
"step": 1737728
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1541763001330614e-05,
|
|
"loss": 3.4611,
|
|
"step": 1738240
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1533377053820094e-05,
|
|
"loss": 3.4514,
|
|
"step": 1738752
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1524991106309574e-05,
|
|
"loss": 3.448,
|
|
"step": 1739264
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1516605158799054e-05,
|
|
"loss": 3.4406,
|
|
"step": 1739776
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1508235590092266e-05,
|
|
"loss": 3.4433,
|
|
"step": 1740288
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1499849642581746e-05,
|
|
"loss": 3.435,
|
|
"step": 1740800
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1491463695071226e-05,
|
|
"loss": 3.4406,
|
|
"step": 1741312
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1483077747560706e-05,
|
|
"loss": 3.4453,
|
|
"step": 1741824
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1474691800050186e-05,
|
|
"loss": 3.4466,
|
|
"step": 1742336
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1466305852539663e-05,
|
|
"loss": 3.4488,
|
|
"step": 1742848
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.145793628383288e-05,
|
|
"loss": 3.4452,
|
|
"step": 1743360
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.144955033632236e-05,
|
|
"loss": 3.4518,
|
|
"step": 1743872
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.144116438881184e-05,
|
|
"loss": 3.439,
|
|
"step": 1744384
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1432778441301315e-05,
|
|
"loss": 3.4482,
|
|
"step": 1744896
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1424408872594528e-05,
|
|
"loss": 3.4468,
|
|
"step": 1745408
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1416022925084008e-05,
|
|
"loss": 3.4518,
|
|
"step": 1745920
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.140763697757349e-05,
|
|
"loss": 3.4582,
|
|
"step": 1746432
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1399251030062967e-05,
|
|
"loss": 3.4434,
|
|
"step": 1746944
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1390865082552447e-05,
|
|
"loss": 3.4541,
|
|
"step": 1747456
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1382479135041927e-05,
|
|
"loss": 3.4451,
|
|
"step": 1747968
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1374093187531407e-05,
|
|
"loss": 3.4533,
|
|
"step": 1748480
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1365723618824616e-05,
|
|
"loss": 3.4421,
|
|
"step": 1748992
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.13573376713141e-05,
|
|
"loss": 3.452,
|
|
"step": 1749504
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.134895172380358e-05,
|
|
"loss": 3.4429,
|
|
"step": 1750016
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.134056577629306e-05,
|
|
"loss": 3.4567,
|
|
"step": 1750528
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.133217982878254e-05,
|
|
"loss": 3.4367,
|
|
"step": 1751040
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.132381026007575e-05,
|
|
"loss": 3.4466,
|
|
"step": 1751552
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.131542431256523e-05,
|
|
"loss": 3.4488,
|
|
"step": 1752064
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.130703836505471e-05,
|
|
"loss": 3.4496,
|
|
"step": 1752576
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1298652417544192e-05,
|
|
"loss": 3.4449,
|
|
"step": 1753088
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.12902828488374e-05,
|
|
"loss": 3.4491,
|
|
"step": 1753600
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.128189690132688e-05,
|
|
"loss": 3.4546,
|
|
"step": 1754112
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.127351095381636e-05,
|
|
"loss": 3.4377,
|
|
"step": 1754624
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.126512500630584e-05,
|
|
"loss": 3.4537,
|
|
"step": 1755136
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 3.852203607559204,
|
|
"eval_runtime": 301.7568,
|
|
"eval_samples_per_second": 1264.565,
|
|
"eval_steps_per_second": 39.519,
|
|
"step": 1755360
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1256755437599053e-05,
|
|
"loss": 3.4412,
|
|
"step": 1755648
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1248385868892263e-05,
|
|
"loss": 3.4351,
|
|
"step": 1756160
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1239999921381742e-05,
|
|
"loss": 3.4496,
|
|
"step": 1756672
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1231613973871222e-05,
|
|
"loss": 3.4483,
|
|
"step": 1757184
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1223228026360702e-05,
|
|
"loss": 3.4549,
|
|
"step": 1757696
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1214858457653915e-05,
|
|
"loss": 3.442,
|
|
"step": 1758208
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1206472510143395e-05,
|
|
"loss": 3.4422,
|
|
"step": 1758720
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1198086562632875e-05,
|
|
"loss": 3.4406,
|
|
"step": 1759232
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1189700615122355e-05,
|
|
"loss": 3.4419,
|
|
"step": 1759744
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1181314667611835e-05,
|
|
"loss": 3.4373,
|
|
"step": 1760256
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1172928720101315e-05,
|
|
"loss": 3.4518,
|
|
"step": 1760768
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1164559151394524e-05,
|
|
"loss": 3.4524,
|
|
"step": 1761280
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1156173203884007e-05,
|
|
"loss": 3.4383,
|
|
"step": 1761792
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1147787256373487e-05,
|
|
"loss": 3.4327,
|
|
"step": 1762304
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1139401308862967e-05,
|
|
"loss": 3.4444,
|
|
"step": 1762816
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1131015361352444e-05,
|
|
"loss": 3.4276,
|
|
"step": 1763328
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1122629413841924e-05,
|
|
"loss": 3.4405,
|
|
"step": 1763840
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1114259845135136e-05,
|
|
"loss": 3.4396,
|
|
"step": 1764352
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1105873897624616e-05,
|
|
"loss": 3.4426,
|
|
"step": 1764864
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1097487950114096e-05,
|
|
"loss": 3.4516,
|
|
"step": 1765376
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1089102002603576e-05,
|
|
"loss": 3.4497,
|
|
"step": 1765888
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1080716055093056e-05,
|
|
"loss": 3.4425,
|
|
"step": 1766400
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1072346486386265e-05,
|
|
"loss": 3.4435,
|
|
"step": 1766912
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1063960538875745e-05,
|
|
"loss": 3.4448,
|
|
"step": 1767424
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1055574591365225e-05,
|
|
"loss": 3.4384,
|
|
"step": 1767936
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1047188643854708e-05,
|
|
"loss": 3.448,
|
|
"step": 1768448
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1038819075147917e-05,
|
|
"loss": 3.4394,
|
|
"step": 1768960
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1030433127637397e-05,
|
|
"loss": 3.437,
|
|
"step": 1769472
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1022047180126877e-05,
|
|
"loss": 3.4349,
|
|
"step": 1769984
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.1013661232616357e-05,
|
|
"loss": 3.4436,
|
|
"step": 1770496
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.100529166390957e-05,
|
|
"loss": 3.4372,
|
|
"step": 1771008
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.099690571639905e-05,
|
|
"loss": 3.4453,
|
|
"step": 1771520
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.098851976888853e-05,
|
|
"loss": 3.453,
|
|
"step": 1772032
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.098013382137801e-05,
|
|
"loss": 3.4477,
|
|
"step": 1772544
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.097174787386749e-05,
|
|
"loss": 3.4358,
|
|
"step": 1773056
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.09633783051607e-05,
|
|
"loss": 3.4459,
|
|
"step": 1773568
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.095499235765018e-05,
|
|
"loss": 3.4434,
|
|
"step": 1774080
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0946606410139662e-05,
|
|
"loss": 3.4369,
|
|
"step": 1774592
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0938220462629142e-05,
|
|
"loss": 3.4313,
|
|
"step": 1775104
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0929834515118622e-05,
|
|
"loss": 3.4294,
|
|
"step": 1775616
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.092146494641183e-05,
|
|
"loss": 3.4393,
|
|
"step": 1776128
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.091307899890131e-05,
|
|
"loss": 3.4446,
|
|
"step": 1776640
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.090469305139079e-05,
|
|
"loss": 3.4344,
|
|
"step": 1777152
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.089630710388027e-05,
|
|
"loss": 3.4369,
|
|
"step": 1777664
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.088792115636975e-05,
|
|
"loss": 3.4534,
|
|
"step": 1778176
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0879551587662963e-05,
|
|
"loss": 3.4403,
|
|
"step": 1778688
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0871165640152443e-05,
|
|
"loss": 3.4488,
|
|
"step": 1779200
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.086277969264192e-05,
|
|
"loss": 3.4343,
|
|
"step": 1779712
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.08543937451314e-05,
|
|
"loss": 3.4205,
|
|
"step": 1780224
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0846024176424616e-05,
|
|
"loss": 3.4531,
|
|
"step": 1780736
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0837638228914095e-05,
|
|
"loss": 3.4393,
|
|
"step": 1781248
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0829252281403572e-05,
|
|
"loss": 3.4368,
|
|
"step": 1781760
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0820866333893052e-05,
|
|
"loss": 3.4416,
|
|
"step": 1782272
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0812480386382532e-05,
|
|
"loss": 3.4336,
|
|
"step": 1782784
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.080411081767574e-05,
|
|
"loss": 3.4238,
|
|
"step": 1783296
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0795724870165224e-05,
|
|
"loss": 3.4329,
|
|
"step": 1783808
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0787338922654704e-05,
|
|
"loss": 3.4332,
|
|
"step": 1784320
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0778952975144184e-05,
|
|
"loss": 3.4345,
|
|
"step": 1784832
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0770583406437393e-05,
|
|
"loss": 3.4438,
|
|
"step": 1785344
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0762197458926873e-05,
|
|
"loss": 3.4375,
|
|
"step": 1785856
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0753811511416353e-05,
|
|
"loss": 3.4331,
|
|
"step": 1786368
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0745425563905833e-05,
|
|
"loss": 3.4436,
|
|
"step": 1786880
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0737055995199046e-05,
|
|
"loss": 3.4296,
|
|
"step": 1787392
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0728670047688526e-05,
|
|
"loss": 3.4337,
|
|
"step": 1787904
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0720284100178006e-05,
|
|
"loss": 3.4366,
|
|
"step": 1788416
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0711898152667486e-05,
|
|
"loss": 3.4383,
|
|
"step": 1788928
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0703512205156966e-05,
|
|
"loss": 3.4312,
|
|
"step": 1789440
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0695142636450178e-05,
|
|
"loss": 3.4326,
|
|
"step": 1789952
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0686756688939658e-05,
|
|
"loss": 3.4177,
|
|
"step": 1790464
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0678370741429138e-05,
|
|
"loss": 3.434,
|
|
"step": 1790976
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0669984793918618e-05,
|
|
"loss": 3.4449,
|
|
"step": 1791488
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0661615225211827e-05,
|
|
"loss": 3.4307,
|
|
"step": 1792000
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0653229277701307e-05,
|
|
"loss": 3.447,
|
|
"step": 1792512
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0644843330190787e-05,
|
|
"loss": 3.4421,
|
|
"step": 1793024
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.063645738268027e-05,
|
|
"loss": 3.4485,
|
|
"step": 1793536
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.062808781397348e-05,
|
|
"loss": 3.4309,
|
|
"step": 1794048
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.061970186646296e-05,
|
|
"loss": 3.4442,
|
|
"step": 1794560
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.061131591895244e-05,
|
|
"loss": 3.4371,
|
|
"step": 1795072
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.060294635024565e-05,
|
|
"loss": 3.4322,
|
|
"step": 1795584
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0594560402735132e-05,
|
|
"loss": 3.4504,
|
|
"step": 1796096
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.058617445522461e-05,
|
|
"loss": 3.4358,
|
|
"step": 1796608
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.057778850771409e-05,
|
|
"loss": 3.4438,
|
|
"step": 1797120
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.056940256020357e-05,
|
|
"loss": 3.4397,
|
|
"step": 1797632
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0561016612693048e-05,
|
|
"loss": 3.4176,
|
|
"step": 1798144
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.055264704398626e-05,
|
|
"loss": 3.4269,
|
|
"step": 1798656
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.054426109647574e-05,
|
|
"loss": 3.4273,
|
|
"step": 1799168
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.053587514896522e-05,
|
|
"loss": 3.4399,
|
|
"step": 1799680
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.05274892014547e-05,
|
|
"loss": 3.4358,
|
|
"step": 1800192
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0519119632747913e-05,
|
|
"loss": 3.4337,
|
|
"step": 1800704
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0510733685237393e-05,
|
|
"loss": 3.434,
|
|
"step": 1801216
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.050234773772687e-05,
|
|
"loss": 3.4278,
|
|
"step": 1801728
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.049396179021635e-05,
|
|
"loss": 3.4315,
|
|
"step": 1802240
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0485592221509565e-05,
|
|
"loss": 3.4313,
|
|
"step": 1802752
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0477206273999045e-05,
|
|
"loss": 3.4486,
|
|
"step": 1803264
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0468820326488522e-05,
|
|
"loss": 3.4532,
|
|
"step": 1803776
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0460434378978002e-05,
|
|
"loss": 3.4325,
|
|
"step": 1804288
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0452048431467482e-05,
|
|
"loss": 3.4308,
|
|
"step": 1804800
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0443662483956962e-05,
|
|
"loss": 3.4399,
|
|
"step": 1805312
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0435292915250174e-05,
|
|
"loss": 3.4299,
|
|
"step": 1805824
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0426906967739654e-05,
|
|
"loss": 3.4463,
|
|
"step": 1806336
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0418521020229134e-05,
|
|
"loss": 3.4338,
|
|
"step": 1806848
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0410135072718614e-05,
|
|
"loss": 3.4348,
|
|
"step": 1807360
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0401749125208094e-05,
|
|
"loss": 3.44,
|
|
"step": 1807872
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0393379556501303e-05,
|
|
"loss": 3.4407,
|
|
"step": 1808384
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0384993608990786e-05,
|
|
"loss": 3.4301,
|
|
"step": 1808896
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0376607661480266e-05,
|
|
"loss": 3.4378,
|
|
"step": 1809408
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0368221713969746e-05,
|
|
"loss": 3.431,
|
|
"step": 1809920
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0359835766459226e-05,
|
|
"loss": 3.4274,
|
|
"step": 1810432
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0351466197752435e-05,
|
|
"loss": 3.4236,
|
|
"step": 1810944
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0343080250241915e-05,
|
|
"loss": 3.4342,
|
|
"step": 1811456
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0334694302731395e-05,
|
|
"loss": 3.4361,
|
|
"step": 1811968
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0326308355220875e-05,
|
|
"loss": 3.4267,
|
|
"step": 1812480
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0317922407710355e-05,
|
|
"loss": 3.4358,
|
|
"step": 1812992
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0309536460199835e-05,
|
|
"loss": 3.4263,
|
|
"step": 1813504
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0301166891493048e-05,
|
|
"loss": 3.4452,
|
|
"step": 1814016
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0292780943982524e-05,
|
|
"loss": 3.4479,
|
|
"step": 1814528
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0284394996472008e-05,
|
|
"loss": 3.442,
|
|
"step": 1815040
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0276009048961488e-05,
|
|
"loss": 3.4357,
|
|
"step": 1815552
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0267623101450968e-05,
|
|
"loss": 3.4265,
|
|
"step": 1816064
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0259253532744177e-05,
|
|
"loss": 3.4347,
|
|
"step": 1816576
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0250867585233657e-05,
|
|
"loss": 3.4241,
|
|
"step": 1817088
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0242481637723137e-05,
|
|
"loss": 3.4334,
|
|
"step": 1817600
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0234095690212616e-05,
|
|
"loss": 3.4314,
|
|
"step": 1818112
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0225709742702096e-05,
|
|
"loss": 3.4351,
|
|
"step": 1818624
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.021732379519158e-05,
|
|
"loss": 3.4359,
|
|
"step": 1819136
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.020895422648479e-05,
|
|
"loss": 3.4348,
|
|
"step": 1819648
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.020056827897427e-05,
|
|
"loss": 3.4421,
|
|
"step": 1820160
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.019218233146375e-05,
|
|
"loss": 3.4302,
|
|
"step": 1820672
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.018379638395323e-05,
|
|
"loss": 3.4364,
|
|
"step": 1821184
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.017542681524644e-05,
|
|
"loss": 3.4376,
|
|
"step": 1821696
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.016704086773592e-05,
|
|
"loss": 3.4373,
|
|
"step": 1822208
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.01586549202254e-05,
|
|
"loss": 3.4472,
|
|
"step": 1822720
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.015026897271488e-05,
|
|
"loss": 3.4314,
|
|
"step": 1823232
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0141883025204358e-05,
|
|
"loss": 3.4443,
|
|
"step": 1823744
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0133497077693838e-05,
|
|
"loss": 3.4348,
|
|
"step": 1824256
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0125111130183318e-05,
|
|
"loss": 3.4404,
|
|
"step": 1824768
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.011674156147653e-05,
|
|
"loss": 3.4331,
|
|
"step": 1825280
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.010835561396601e-05,
|
|
"loss": 3.4401,
|
|
"step": 1825792
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.009996966645549e-05,
|
|
"loss": 3.4321,
|
|
"step": 1826304
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.009158371894497e-05,
|
|
"loss": 3.4471,
|
|
"step": 1826816
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.008319777143445e-05,
|
|
"loss": 3.4242,
|
|
"step": 1827328
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0074828202727662e-05,
|
|
"loss": 3.434,
|
|
"step": 1827840
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0066442255217142e-05,
|
|
"loss": 3.4411,
|
|
"step": 1828352
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0058056307706622e-05,
|
|
"loss": 3.4395,
|
|
"step": 1828864
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.0049670360196102e-05,
|
|
"loss": 3.4329,
|
|
"step": 1829376
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.004130079148931e-05,
|
|
"loss": 3.4338,
|
|
"step": 1829888
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.003291484397879e-05,
|
|
"loss": 3.4464,
|
|
"step": 1830400
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.002452889646827e-05,
|
|
"loss": 3.4262,
|
|
"step": 1830912
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.001614294895775e-05,
|
|
"loss": 3.4416,
|
|
"step": 1831424
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.8545172214508057,
|
|
"eval_runtime": 304.5516,
|
|
"eval_samples_per_second": 1252.96,
|
|
"eval_steps_per_second": 39.156,
|
|
"step": 1831680
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 2.0007789759054696e-05,
|
|
"loss": 3.4241,
|
|
"step": 1831936
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9999403811544176e-05,
|
|
"loss": 3.4306,
|
|
"step": 1832448
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9991017864033653e-05,
|
|
"loss": 3.4357,
|
|
"step": 1832960
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9982631916523133e-05,
|
|
"loss": 3.4367,
|
|
"step": 1833472
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9974245969012616e-05,
|
|
"loss": 3.4462,
|
|
"step": 1833984
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9965876400305825e-05,
|
|
"loss": 3.4295,
|
|
"step": 1834496
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9957490452795305e-05,
|
|
"loss": 3.4357,
|
|
"step": 1835008
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9949104505284785e-05,
|
|
"loss": 3.4335,
|
|
"step": 1835520
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9940718557774265e-05,
|
|
"loss": 3.4278,
|
|
"step": 1836032
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9932332610263745e-05,
|
|
"loss": 3.4284,
|
|
"step": 1836544
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9923963041556957e-05,
|
|
"loss": 3.4392,
|
|
"step": 1837056
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9915577094046437e-05,
|
|
"loss": 3.4413,
|
|
"step": 1837568
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9907191146535917e-05,
|
|
"loss": 3.4246,
|
|
"step": 1838080
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9898805199025397e-05,
|
|
"loss": 3.4244,
|
|
"step": 1838592
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9890435630318606e-05,
|
|
"loss": 3.4318,
|
|
"step": 1839104
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9882049682808086e-05,
|
|
"loss": 3.4163,
|
|
"step": 1839616
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9873663735297566e-05,
|
|
"loss": 3.4301,
|
|
"step": 1840128
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.986527778778705e-05,
|
|
"loss": 3.4282,
|
|
"step": 1840640
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.985690821908026e-05,
|
|
"loss": 3.4346,
|
|
"step": 1841152
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.984852227156974e-05,
|
|
"loss": 3.4355,
|
|
"step": 1841664
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.984013632405922e-05,
|
|
"loss": 3.4407,
|
|
"step": 1842176
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.98317503765487e-05,
|
|
"loss": 3.433,
|
|
"step": 1842688
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.982336442903818e-05,
|
|
"loss": 3.4311,
|
|
"step": 1843200
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.981497848152766e-05,
|
|
"loss": 3.4299,
|
|
"step": 1843712
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.980660891282087e-05,
|
|
"loss": 3.4311,
|
|
"step": 1844224
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.979822296531035e-05,
|
|
"loss": 3.4385,
|
|
"step": 1844736
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.978983701779983e-05,
|
|
"loss": 3.4261,
|
|
"step": 1845248
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9781451070289307e-05,
|
|
"loss": 3.425,
|
|
"step": 1845760
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9773065122778787e-05,
|
|
"loss": 3.4301,
|
|
"step": 1846272
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9764695554072003e-05,
|
|
"loss": 3.4265,
|
|
"step": 1846784
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.975630960656148e-05,
|
|
"loss": 3.4251,
|
|
"step": 1847296
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.974792365905096e-05,
|
|
"loss": 3.4397,
|
|
"step": 1847808
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.973953771154044e-05,
|
|
"loss": 3.4369,
|
|
"step": 1848320
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9731168142833652e-05,
|
|
"loss": 3.4342,
|
|
"step": 1848832
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9722782195323132e-05,
|
|
"loss": 3.4278,
|
|
"step": 1849344
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9714396247812612e-05,
|
|
"loss": 3.4395,
|
|
"step": 1849856
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9706010300302092e-05,
|
|
"loss": 3.4324,
|
|
"step": 1850368
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9697640731595305e-05,
|
|
"loss": 3.4307,
|
|
"step": 1850880
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.968925478408478e-05,
|
|
"loss": 3.4179,
|
|
"step": 1851392
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.968086883657426e-05,
|
|
"loss": 3.4199,
|
|
"step": 1851904
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.967248288906374e-05,
|
|
"loss": 3.4257,
|
|
"step": 1852416
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9664113320356954e-05,
|
|
"loss": 3.4349,
|
|
"step": 1852928
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9655727372846434e-05,
|
|
"loss": 3.4248,
|
|
"step": 1853440
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9647341425335914e-05,
|
|
"loss": 3.4246,
|
|
"step": 1853952
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9638955477825393e-05,
|
|
"loss": 3.4407,
|
|
"step": 1854464
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9630569530314873e-05,
|
|
"loss": 3.4303,
|
|
"step": 1854976
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9622183582804353e-05,
|
|
"loss": 3.4395,
|
|
"step": 1855488
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9613797635293833e-05,
|
|
"loss": 3.4235,
|
|
"step": 1856000
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9605428066587046e-05,
|
|
"loss": 3.4067,
|
|
"step": 1856512
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9597042119076526e-05,
|
|
"loss": 3.4442,
|
|
"step": 1857024
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9588656171566006e-05,
|
|
"loss": 3.4308,
|
|
"step": 1857536
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9580270224055486e-05,
|
|
"loss": 3.4273,
|
|
"step": 1858048
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9571884276544962e-05,
|
|
"loss": 3.4274,
|
|
"step": 1858560
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9563514707838175e-05,
|
|
"loss": 3.4244,
|
|
"step": 1859072
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9555128760327658e-05,
|
|
"loss": 3.4119,
|
|
"step": 1859584
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9546742812817135e-05,
|
|
"loss": 3.4205,
|
|
"step": 1860096
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9538356865306615e-05,
|
|
"loss": 3.4244,
|
|
"step": 1860608
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9529970917796095e-05,
|
|
"loss": 3.4239,
|
|
"step": 1861120
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9521601349089307e-05,
|
|
"loss": 3.436,
|
|
"step": 1861632
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9513215401578787e-05,
|
|
"loss": 3.4259,
|
|
"step": 1862144
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9504829454068267e-05,
|
|
"loss": 3.4194,
|
|
"step": 1862656
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9496443506557747e-05,
|
|
"loss": 3.438,
|
|
"step": 1863168
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9488073937850956e-05,
|
|
"loss": 3.4133,
|
|
"step": 1863680
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9479687990340436e-05,
|
|
"loss": 3.4206,
|
|
"step": 1864192
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9471302042829916e-05,
|
|
"loss": 3.4256,
|
|
"step": 1864704
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9462916095319396e-05,
|
|
"loss": 3.4261,
|
|
"step": 1865216
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.945454652661261e-05,
|
|
"loss": 3.4204,
|
|
"step": 1865728
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.944616057910209e-05,
|
|
"loss": 3.4227,
|
|
"step": 1866240
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9437774631591568e-05,
|
|
"loss": 3.4106,
|
|
"step": 1866752
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9429388684081048e-05,
|
|
"loss": 3.4211,
|
|
"step": 1867264
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9421002736570528e-05,
|
|
"loss": 3.4329,
|
|
"step": 1867776
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.941263316786374e-05,
|
|
"loss": 3.4215,
|
|
"step": 1868288
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.940424722035322e-05,
|
|
"loss": 3.4362,
|
|
"step": 1868800
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.93958612728427e-05,
|
|
"loss": 3.4311,
|
|
"step": 1869312
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.938747532533218e-05,
|
|
"loss": 3.441,
|
|
"step": 1869824
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.937910575662539e-05,
|
|
"loss": 3.4185,
|
|
"step": 1870336
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.937071980911487e-05,
|
|
"loss": 3.4306,
|
|
"step": 1870848
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.936233386160435e-05,
|
|
"loss": 3.4294,
|
|
"step": 1871360
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9353947914093833e-05,
|
|
"loss": 3.4173,
|
|
"step": 1871872
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9345578345387042e-05,
|
|
"loss": 3.4391,
|
|
"step": 1872384
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9337192397876522e-05,
|
|
"loss": 3.4259,
|
|
"step": 1872896
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9328806450366002e-05,
|
|
"loss": 3.4294,
|
|
"step": 1873408
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9320420502855482e-05,
|
|
"loss": 3.4312,
|
|
"step": 1873920
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9312034555344962e-05,
|
|
"loss": 3.4105,
|
|
"step": 1874432
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9303664986638174e-05,
|
|
"loss": 3.4152,
|
|
"step": 1874944
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9295279039127654e-05,
|
|
"loss": 3.4188,
|
|
"step": 1875456
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9286893091617134e-05,
|
|
"loss": 3.4283,
|
|
"step": 1875968
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.927850714410661e-05,
|
|
"loss": 3.424,
|
|
"step": 1876480
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9270137575399823e-05,
|
|
"loss": 3.4215,
|
|
"step": 1876992
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9261751627889303e-05,
|
|
"loss": 3.4213,
|
|
"step": 1877504
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9253365680378783e-05,
|
|
"loss": 3.4225,
|
|
"step": 1878016
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9244979732868263e-05,
|
|
"loss": 3.4177,
|
|
"step": 1878528
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9236610164161476e-05,
|
|
"loss": 3.4191,
|
|
"step": 1879040
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9228224216650956e-05,
|
|
"loss": 3.4409,
|
|
"step": 1879552
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9219838269140436e-05,
|
|
"loss": 3.4369,
|
|
"step": 1880064
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9211452321629912e-05,
|
|
"loss": 3.4274,
|
|
"step": 1880576
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9203066374119395e-05,
|
|
"loss": 3.415,
|
|
"step": 1881088
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9194680426608875e-05,
|
|
"loss": 3.4284,
|
|
"step": 1881600
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9186310857902084e-05,
|
|
"loss": 3.4207,
|
|
"step": 1882112
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9177924910391564e-05,
|
|
"loss": 3.4352,
|
|
"step": 1882624
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9169538962881044e-05,
|
|
"loss": 3.4232,
|
|
"step": 1883136
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9161153015370524e-05,
|
|
"loss": 3.4215,
|
|
"step": 1883648
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9152767067860004e-05,
|
|
"loss": 3.4277,
|
|
"step": 1884160
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9144397499153217e-05,
|
|
"loss": 3.4339,
|
|
"step": 1884672
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9136011551642697e-05,
|
|
"loss": 3.4182,
|
|
"step": 1885184
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9127625604132177e-05,
|
|
"loss": 3.4288,
|
|
"step": 1885696
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9119239656621657e-05,
|
|
"loss": 3.4197,
|
|
"step": 1886208
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9110853709111137e-05,
|
|
"loss": 3.4202,
|
|
"step": 1886720
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9102467761600617e-05,
|
|
"loss": 3.4123,
|
|
"step": 1887232
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9094081814090097e-05,
|
|
"loss": 3.4217,
|
|
"step": 1887744
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9085695866579576e-05,
|
|
"loss": 3.4204,
|
|
"step": 1888256
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9077342676676518e-05,
|
|
"loss": 3.4234,
|
|
"step": 1888768
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9068956729165998e-05,
|
|
"loss": 3.4255,
|
|
"step": 1889280
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9060570781655478e-05,
|
|
"loss": 3.416,
|
|
"step": 1889792
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9052184834144958e-05,
|
|
"loss": 3.4296,
|
|
"step": 1890304
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.904379888663444e-05,
|
|
"loss": 3.4399,
|
|
"step": 1890816
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9035412939123918e-05,
|
|
"loss": 3.4277,
|
|
"step": 1891328
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.902704337041713e-05,
|
|
"loss": 3.4324,
|
|
"step": 1891840
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.901865742290661e-05,
|
|
"loss": 3.411,
|
|
"step": 1892352
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.901027147539609e-05,
|
|
"loss": 3.4278,
|
|
"step": 1892864
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.9001885527885567e-05,
|
|
"loss": 3.4197,
|
|
"step": 1893376
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.899349958037505e-05,
|
|
"loss": 3.418,
|
|
"step": 1893888
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.8985130011668263e-05,
|
|
"loss": 3.4205,
|
|
"step": 1894400
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.897674406415774e-05,
|
|
"loss": 3.421,
|
|
"step": 1894912
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.896835811664722e-05,
|
|
"loss": 3.4255,
|
|
"step": 1895424
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.89599721691367e-05,
|
|
"loss": 3.4229,
|
|
"step": 1895936
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.895158622162618e-05,
|
|
"loss": 3.4313,
|
|
"step": 1896448
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.894321665291939e-05,
|
|
"loss": 3.4183,
|
|
"step": 1896960
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.893483070540887e-05,
|
|
"loss": 3.4219,
|
|
"step": 1897472
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.892644475789835e-05,
|
|
"loss": 3.4279,
|
|
"step": 1897984
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.891805881038783e-05,
|
|
"loss": 3.4292,
|
|
"step": 1898496
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.890967286287731e-05,
|
|
"loss": 3.4369,
|
|
"step": 1899008
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.890130329417052e-05,
|
|
"loss": 3.4202,
|
|
"step": 1899520
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.8892917346660004e-05,
|
|
"loss": 3.4369,
|
|
"step": 1900032
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.8884531399149484e-05,
|
|
"loss": 3.4237,
|
|
"step": 1900544
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.8876145451638964e-05,
|
|
"loss": 3.429,
|
|
"step": 1901056
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.8867775882932173e-05,
|
|
"loss": 3.4232,
|
|
"step": 1901568
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.8859389935421653e-05,
|
|
"loss": 3.4277,
|
|
"step": 1902080
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.8851003987911133e-05,
|
|
"loss": 3.4206,
|
|
"step": 1902592
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.8842618040400613e-05,
|
|
"loss": 3.4402,
|
|
"step": 1903104
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.8834232092890096e-05,
|
|
"loss": 3.4154,
|
|
"step": 1903616
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.8825862524183305e-05,
|
|
"loss": 3.4254,
|
|
"step": 1904128
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.8817476576672785e-05,
|
|
"loss": 3.4292,
|
|
"step": 1904640
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.8809090629162265e-05,
|
|
"loss": 3.4319,
|
|
"step": 1905152
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.880070468165174e-05,
|
|
"loss": 3.42,
|
|
"step": 1905664
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.8792335112944958e-05,
|
|
"loss": 3.4261,
|
|
"step": 1906176
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.8783949165434437e-05,
|
|
"loss": 3.431,
|
|
"step": 1906688
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.8775563217923917e-05,
|
|
"loss": 3.4245,
|
|
"step": 1907200
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.8767177270413394e-05,
|
|
"loss": 3.4273,
|
|
"step": 1907712
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.8565003871917725,
|
|
"eval_runtime": 303.4104,
|
|
"eval_samples_per_second": 1257.673,
|
|
"eval_steps_per_second": 39.303,
|
|
"step": 1908000
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.8758807701706606e-05,
|
|
"loss": 3.4153,
|
|
"step": 1908224
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.8750421754196086e-05,
|
|
"loss": 3.4197,
|
|
"step": 1908736
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.8742035806685566e-05,
|
|
"loss": 3.4243,
|
|
"step": 1909248
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.8733649859175046e-05,
|
|
"loss": 3.4273,
|
|
"step": 1909760
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.872528029046826e-05,
|
|
"loss": 3.4347,
|
|
"step": 1910272
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.871689434295774e-05,
|
|
"loss": 3.4181,
|
|
"step": 1910784
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.8708508395447215e-05,
|
|
"loss": 3.425,
|
|
"step": 1911296
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.8700122447936695e-05,
|
|
"loss": 3.423,
|
|
"step": 1911808
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.869175287922991e-05,
|
|
"loss": 3.4144,
|
|
"step": 1912320
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.868336693171939e-05,
|
|
"loss": 3.4192,
|
|
"step": 1912832
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.8674980984208868e-05,
|
|
"loss": 3.4254,
|
|
"step": 1913344
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.8666595036698348e-05,
|
|
"loss": 3.427,
|
|
"step": 1913856
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.865822546799156e-05,
|
|
"loss": 3.426,
|
|
"step": 1914368
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.864983952048104e-05,
|
|
"loss": 3.4047,
|
|
"step": 1914880
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.864145357297052e-05,
|
|
"loss": 3.4257,
|
|
"step": 1915392
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.863306762546e-05,
|
|
"loss": 3.4062,
|
|
"step": 1915904
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.862468167794948e-05,
|
|
"loss": 3.4166,
|
|
"step": 1916416
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.861629573043896e-05,
|
|
"loss": 3.4212,
|
|
"step": 1916928
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.860790978292844e-05,
|
|
"loss": 3.4205,
|
|
"step": 1917440
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.859954021422165e-05,
|
|
"loss": 3.4235,
|
|
"step": 1917952
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.859115426671113e-05,
|
|
"loss": 3.4298,
|
|
"step": 1918464
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.8582768319200612e-05,
|
|
"loss": 3.424,
|
|
"step": 1918976
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.8574382371690092e-05,
|
|
"loss": 3.4232,
|
|
"step": 1919488
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.8565996424179572e-05,
|
|
"loss": 3.414,
|
|
"step": 1920000
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.855762685547278e-05,
|
|
"loss": 3.422,
|
|
"step": 1920512
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.854924090796226e-05,
|
|
"loss": 3.425,
|
|
"step": 1921024
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.854085496045174e-05,
|
|
"loss": 3.4205,
|
|
"step": 1921536
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.853246901294122e-05,
|
|
"loss": 3.4187,
|
|
"step": 1922048
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.85240830654307e-05,
|
|
"loss": 3.4131,
|
|
"step": 1922560
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.8515713496723914e-05,
|
|
"loss": 3.4183,
|
|
"step": 1923072
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8507327549213394e-05,
|
|
"loss": 3.4122,
|
|
"step": 1923584
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.849894160170287e-05,
|
|
"loss": 3.4287,
|
|
"step": 1924096
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.849055565419235e-05,
|
|
"loss": 3.4257,
|
|
"step": 1924608
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8482186085485566e-05,
|
|
"loss": 3.4263,
|
|
"step": 1925120
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8473800137975046e-05,
|
|
"loss": 3.4167,
|
|
"step": 1925632
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8465414190464522e-05,
|
|
"loss": 3.4236,
|
|
"step": 1926144
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8457028242954002e-05,
|
|
"loss": 3.4245,
|
|
"step": 1926656
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8448658674247215e-05,
|
|
"loss": 3.4191,
|
|
"step": 1927168
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8440272726736695e-05,
|
|
"loss": 3.4091,
|
|
"step": 1927680
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8431886779226175e-05,
|
|
"loss": 3.4089,
|
|
"step": 1928192
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8423500831715655e-05,
|
|
"loss": 3.4146,
|
|
"step": 1928704
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8415131263008867e-05,
|
|
"loss": 3.4235,
|
|
"step": 1929216
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8406745315498344e-05,
|
|
"loss": 3.4144,
|
|
"step": 1929728
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8398359367987824e-05,
|
|
"loss": 3.4153,
|
|
"step": 1930240
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8389973420477304e-05,
|
|
"loss": 3.4289,
|
|
"step": 1930752
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.838160385177052e-05,
|
|
"loss": 3.4197,
|
|
"step": 1931264
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8373217904259996e-05,
|
|
"loss": 3.4265,
|
|
"step": 1931776
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8364831956749476e-05,
|
|
"loss": 3.4149,
|
|
"step": 1932288
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8356446009238956e-05,
|
|
"loss": 3.3969,
|
|
"step": 1932800
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8348060061728436e-05,
|
|
"loss": 3.4314,
|
|
"step": 1933312
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8339690493021645e-05,
|
|
"loss": 3.4181,
|
|
"step": 1933824
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.833130454551113e-05,
|
|
"loss": 3.4189,
|
|
"step": 1934336
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.832291859800061e-05,
|
|
"loss": 3.4177,
|
|
"step": 1934848
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.831453265049009e-05,
|
|
"loss": 3.4163,
|
|
"step": 1935360
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8306163081783297e-05,
|
|
"loss": 3.3999,
|
|
"step": 1935872
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8297777134272777e-05,
|
|
"loss": 3.4112,
|
|
"step": 1936384
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8289391186762257e-05,
|
|
"loss": 3.4135,
|
|
"step": 1936896
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8281005239251737e-05,
|
|
"loss": 3.4124,
|
|
"step": 1937408
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.827263567054495e-05,
|
|
"loss": 3.4296,
|
|
"step": 1937920
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.826424972303443e-05,
|
|
"loss": 3.4136,
|
|
"step": 1938432
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.825586377552391e-05,
|
|
"loss": 3.4098,
|
|
"step": 1938944
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.824747782801339e-05,
|
|
"loss": 3.4296,
|
|
"step": 1939456
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.82391082593066e-05,
|
|
"loss": 3.4018,
|
|
"step": 1939968
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8230722311796082e-05,
|
|
"loss": 3.4055,
|
|
"step": 1940480
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8222336364285562e-05,
|
|
"loss": 3.4205,
|
|
"step": 1940992
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8213950416775042e-05,
|
|
"loss": 3.4138,
|
|
"step": 1941504
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.820558084806825e-05,
|
|
"loss": 3.4093,
|
|
"step": 1942016
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.819719490055773e-05,
|
|
"loss": 3.414,
|
|
"step": 1942528
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.818880895304721e-05,
|
|
"loss": 3.3984,
|
|
"step": 1943040
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.818042300553669e-05,
|
|
"loss": 3.4084,
|
|
"step": 1943552
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8172037058026174e-05,
|
|
"loss": 3.4243,
|
|
"step": 1944064
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8163667489319383e-05,
|
|
"loss": 3.4121,
|
|
"step": 1944576
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8155281541808863e-05,
|
|
"loss": 3.4204,
|
|
"step": 1945088
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8146895594298343e-05,
|
|
"loss": 3.421,
|
|
"step": 1945600
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.813850964678782e-05,
|
|
"loss": 3.4357,
|
|
"step": 1946112
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8130140078081036e-05,
|
|
"loss": 3.4079,
|
|
"step": 1946624
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8121754130570516e-05,
|
|
"loss": 3.4214,
|
|
"step": 1947136
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8113368183059996e-05,
|
|
"loss": 3.4203,
|
|
"step": 1947648
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8104982235549472e-05,
|
|
"loss": 3.4038,
|
|
"step": 1948160
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8096612666842685e-05,
|
|
"loss": 3.4308,
|
|
"step": 1948672
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8088226719332165e-05,
|
|
"loss": 3.4153,
|
|
"step": 1949184
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8079840771821645e-05,
|
|
"loss": 3.42,
|
|
"step": 1949696
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8071454824311125e-05,
|
|
"loss": 3.4209,
|
|
"step": 1950208
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8063068876800605e-05,
|
|
"loss": 3.4037,
|
|
"step": 1950720
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8054699308093817e-05,
|
|
"loss": 3.4002,
|
|
"step": 1951232
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8046313360583294e-05,
|
|
"loss": 3.4081,
|
|
"step": 1951744
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8037927413072774e-05,
|
|
"loss": 3.4166,
|
|
"step": 1952256
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.8029541465562254e-05,
|
|
"loss": 3.413,
|
|
"step": 1952768
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.802117189685547e-05,
|
|
"loss": 3.4096,
|
|
"step": 1953280
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.8012785949344946e-05,
|
|
"loss": 3.4081,
|
|
"step": 1953792
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.8004400001834426e-05,
|
|
"loss": 3.415,
|
|
"step": 1954304
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7996014054323906e-05,
|
|
"loss": 3.4091,
|
|
"step": 1954816
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.798764448561712e-05,
|
|
"loss": 3.4073,
|
|
"step": 1955328
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.79792585381066e-05,
|
|
"loss": 3.4324,
|
|
"step": 1955840
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.797087259059608e-05,
|
|
"loss": 3.4262,
|
|
"step": 1956352
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7962486643085558e-05,
|
|
"loss": 3.4201,
|
|
"step": 1956864
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7954100695575038e-05,
|
|
"loss": 3.4041,
|
|
"step": 1957376
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7945714748064518e-05,
|
|
"loss": 3.4186,
|
|
"step": 1957888
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7937345179357727e-05,
|
|
"loss": 3.4092,
|
|
"step": 1958400
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7928959231847207e-05,
|
|
"loss": 3.4215,
|
|
"step": 1958912
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.792057328433669e-05,
|
|
"loss": 3.413,
|
|
"step": 1959424
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.791218733682617e-05,
|
|
"loss": 3.4087,
|
|
"step": 1959936
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.790380138931565e-05,
|
|
"loss": 3.4152,
|
|
"step": 1960448
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.789543182060886e-05,
|
|
"loss": 3.4248,
|
|
"step": 1960960
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.788704587309834e-05,
|
|
"loss": 3.4067,
|
|
"step": 1961472
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.787865992558782e-05,
|
|
"loss": 3.4186,
|
|
"step": 1961984
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.78702739780773e-05,
|
|
"loss": 3.4036,
|
|
"step": 1962496
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.786188803056678e-05,
|
|
"loss": 3.4122,
|
|
"step": 1963008
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7853518461859992e-05,
|
|
"loss": 3.4047,
|
|
"step": 1963520
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7845132514349472e-05,
|
|
"loss": 3.4068,
|
|
"step": 1964032
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.783674656683895e-05,
|
|
"loss": 3.4107,
|
|
"step": 1964544
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.782836061932843e-05,
|
|
"loss": 3.4147,
|
|
"step": 1965056
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7819974671817912e-05,
|
|
"loss": 3.4149,
|
|
"step": 1965568
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.781158872430739e-05,
|
|
"loss": 3.407,
|
|
"step": 1966080
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.78032191556006e-05,
|
|
"loss": 3.4217,
|
|
"step": 1966592
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.779483320809008e-05,
|
|
"loss": 3.4279,
|
|
"step": 1967104
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.778644726057956e-05,
|
|
"loss": 3.4172,
|
|
"step": 1967616
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.777806131306904e-05,
|
|
"loss": 3.4264,
|
|
"step": 1968128
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.776967536555852e-05,
|
|
"loss": 3.4036,
|
|
"step": 1968640
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7761305796851733e-05,
|
|
"loss": 3.4132,
|
|
"step": 1969152
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7752919849341213e-05,
|
|
"loss": 3.4099,
|
|
"step": 1969664
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7744533901830693e-05,
|
|
"loss": 3.4064,
|
|
"step": 1970176
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7736147954320173e-05,
|
|
"loss": 3.4123,
|
|
"step": 1970688
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7727762006809653e-05,
|
|
"loss": 3.4117,
|
|
"step": 1971200
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.771937605929913e-05,
|
|
"loss": 3.4173,
|
|
"step": 1971712
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7711006490592345e-05,
|
|
"loss": 3.412,
|
|
"step": 1972224
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7702620543081825e-05,
|
|
"loss": 3.4208,
|
|
"step": 1972736
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7694234595571305e-05,
|
|
"loss": 3.4131,
|
|
"step": 1973248
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7685848648060782e-05,
|
|
"loss": 3.4097,
|
|
"step": 1973760
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7677479079353994e-05,
|
|
"loss": 3.417,
|
|
"step": 1974272
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7669093131843474e-05,
|
|
"loss": 3.4195,
|
|
"step": 1974784
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7660707184332954e-05,
|
|
"loss": 3.4242,
|
|
"step": 1975296
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7652321236822434e-05,
|
|
"loss": 3.4112,
|
|
"step": 1975808
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7643935289311914e-05,
|
|
"loss": 3.4279,
|
|
"step": 1976320
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7635549341801394e-05,
|
|
"loss": 3.4143,
|
|
"step": 1976832
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7627179773094603e-05,
|
|
"loss": 3.4174,
|
|
"step": 1977344
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7618793825584083e-05,
|
|
"loss": 3.4148,
|
|
"step": 1977856
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7610407878073566e-05,
|
|
"loss": 3.4132,
|
|
"step": 1978368
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7602021930563046e-05,
|
|
"loss": 3.4106,
|
|
"step": 1978880
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7593652361856256e-05,
|
|
"loss": 3.4327,
|
|
"step": 1979392
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7585266414345735e-05,
|
|
"loss": 3.3989,
|
|
"step": 1979904
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7576880466835215e-05,
|
|
"loss": 3.4202,
|
|
"step": 1980416
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7568494519324695e-05,
|
|
"loss": 3.419,
|
|
"step": 1980928
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7560124950617908e-05,
|
|
"loss": 3.4159,
|
|
"step": 1981440
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7551739003107388e-05,
|
|
"loss": 3.4113,
|
|
"step": 1981952
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7543353055596868e-05,
|
|
"loss": 3.4154,
|
|
"step": 1982464
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7534967108086348e-05,
|
|
"loss": 3.418,
|
|
"step": 1982976
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7526581160575828e-05,
|
|
"loss": 3.4171,
|
|
"step": 1983488
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.7518211591869037e-05,
|
|
"loss": 3.4147,
|
|
"step": 1984000
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 3.8563754558563232,
|
|
"eval_runtime": 303.3839,
|
|
"eval_samples_per_second": 1257.782,
|
|
"eval_steps_per_second": 39.307,
|
|
"step": 1984320
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.750982564435852e-05,
|
|
"loss": 3.3953,
|
|
"step": 1984512
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.750145607565173e-05,
|
|
"loss": 3.4115,
|
|
"step": 1985024
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.749307012814121e-05,
|
|
"loss": 3.4118,
|
|
"step": 1985536
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.748468418063069e-05,
|
|
"loss": 3.4163,
|
|
"step": 1986048
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.747629823312017e-05,
|
|
"loss": 3.4256,
|
|
"step": 1986560
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.746792866441338e-05,
|
|
"loss": 3.4067,
|
|
"step": 1987072
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.745954271690286e-05,
|
|
"loss": 3.4148,
|
|
"step": 1987584
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.745115676939234e-05,
|
|
"loss": 3.4127,
|
|
"step": 1988096
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.744277082188182e-05,
|
|
"loss": 3.4062,
|
|
"step": 1988608
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.74343848743713e-05,
|
|
"loss": 3.4061,
|
|
"step": 1989120
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.742599892686078e-05,
|
|
"loss": 3.4167,
|
|
"step": 1989632
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.741762935815399e-05,
|
|
"loss": 3.42,
|
|
"step": 1990144
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.740924341064347e-05,
|
|
"loss": 3.4182,
|
|
"step": 1990656
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7400857463132954e-05,
|
|
"loss": 3.3922,
|
|
"step": 1991168
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7392471515622434e-05,
|
|
"loss": 3.4154,
|
|
"step": 1991680
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7384101946915643e-05,
|
|
"loss": 3.3932,
|
|
"step": 1992192
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7375715999405123e-05,
|
|
"loss": 3.4094,
|
|
"step": 1992704
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7367330051894603e-05,
|
|
"loss": 3.4093,
|
|
"step": 1993216
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7358944104384083e-05,
|
|
"loss": 3.4091,
|
|
"step": 1993728
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7350574535677295e-05,
|
|
"loss": 3.4139,
|
|
"step": 1994240
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7342188588166775e-05,
|
|
"loss": 3.4216,
|
|
"step": 1994752
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7333802640656255e-05,
|
|
"loss": 3.416,
|
|
"step": 1995264
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.732541669314573e-05,
|
|
"loss": 3.4096,
|
|
"step": 1995776
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7317047124438944e-05,
|
|
"loss": 3.4022,
|
|
"step": 1996288
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7308661176928424e-05,
|
|
"loss": 3.4156,
|
|
"step": 1996800
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7300275229417907e-05,
|
|
"loss": 3.4164,
|
|
"step": 1997312
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7291889281907384e-05,
|
|
"loss": 3.4061,
|
|
"step": 1997824
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7283503334396864e-05,
|
|
"loss": 3.4105,
|
|
"step": 1998336
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7275133765690076e-05,
|
|
"loss": 3.4027,
|
|
"step": 1998848
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7266747818179553e-05,
|
|
"loss": 3.4099,
|
|
"step": 1999360
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7258361870669036e-05,
|
|
"loss": 3.3973,
|
|
"step": 1999872
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7249975923158516e-05,
|
|
"loss": 3.4207,
|
|
"step": 2000384
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7241589975647996e-05,
|
|
"loss": 3.4226,
|
|
"step": 2000896
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7233220406941205e-05,
|
|
"loss": 3.4116,
|
|
"step": 2001408
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7224834459430685e-05,
|
|
"loss": 3.4084,
|
|
"step": 2001920
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7216448511920165e-05,
|
|
"loss": 3.406,
|
|
"step": 2002432
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7208062564409645e-05,
|
|
"loss": 3.4168,
|
|
"step": 2002944
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7199692995702858e-05,
|
|
"loss": 3.4108,
|
|
"step": 2003456
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7191307048192338e-05,
|
|
"loss": 3.3961,
|
|
"step": 2003968
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7182921100681818e-05,
|
|
"loss": 3.4023,
|
|
"step": 2004480
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7174535153171298e-05,
|
|
"loss": 3.4032,
|
|
"step": 2004992
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7166165584464507e-05,
|
|
"loss": 3.4115,
|
|
"step": 2005504
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.715777963695399e-05,
|
|
"loss": 3.4053,
|
|
"step": 2006016
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.714939368944347e-05,
|
|
"loss": 3.4068,
|
|
"step": 2006528
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.714100774193295e-05,
|
|
"loss": 3.4167,
|
|
"step": 2007040
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.713262179442243e-05,
|
|
"loss": 3.41,
|
|
"step": 2007552
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.712423584691191e-05,
|
|
"loss": 3.4154,
|
|
"step": 2008064
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7115849899401386e-05,
|
|
"loss": 3.4081,
|
|
"step": 2008576
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.71074803306946e-05,
|
|
"loss": 3.3845,
|
|
"step": 2009088
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.709909438318408e-05,
|
|
"loss": 3.4178,
|
|
"step": 2009600
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7090708435673562e-05,
|
|
"loss": 3.4092,
|
|
"step": 2010112
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.708232248816304e-05,
|
|
"loss": 3.4083,
|
|
"step": 2010624
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.707395291945625e-05,
|
|
"loss": 3.411,
|
|
"step": 2011136
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.706556697194573e-05,
|
|
"loss": 3.4039,
|
|
"step": 2011648
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7057181024435208e-05,
|
|
"loss": 3.3898,
|
|
"step": 2012160
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.704879507692469e-05,
|
|
"loss": 3.3973,
|
|
"step": 2012672
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7040425508217904e-05,
|
|
"loss": 3.407,
|
|
"step": 2013184
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7032039560707384e-05,
|
|
"loss": 3.4007,
|
|
"step": 2013696
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.702365361319686e-05,
|
|
"loss": 3.4168,
|
|
"step": 2014208
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.701526766568634e-05,
|
|
"loss": 3.4043,
|
|
"step": 2014720
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7006898096979553e-05,
|
|
"loss": 3.4005,
|
|
"step": 2015232
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6998512149469033e-05,
|
|
"loss": 3.4176,
|
|
"step": 2015744
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6990126201958512e-05,
|
|
"loss": 3.3966,
|
|
"step": 2016256
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6981740254447992e-05,
|
|
"loss": 3.3939,
|
|
"step": 2016768
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6973354306937472e-05,
|
|
"loss": 3.4114,
|
|
"step": 2017280
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.696498473823068e-05,
|
|
"loss": 3.4022,
|
|
"step": 2017792
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.695659879072016e-05,
|
|
"loss": 3.3974,
|
|
"step": 2018304
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6948212843209645e-05,
|
|
"loss": 3.4019,
|
|
"step": 2018816
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6939826895699125e-05,
|
|
"loss": 3.392,
|
|
"step": 2019328
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6931457326992334e-05,
|
|
"loss": 3.3997,
|
|
"step": 2019840
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6923071379481814e-05,
|
|
"loss": 3.4141,
|
|
"step": 2020352
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6914685431971294e-05,
|
|
"loss": 3.404,
|
|
"step": 2020864
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6906299484460774e-05,
|
|
"loss": 3.4093,
|
|
"step": 2021376
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6897929915753986e-05,
|
|
"loss": 3.4062,
|
|
"step": 2021888
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6889543968243466e-05,
|
|
"loss": 3.4254,
|
|
"step": 2022400
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6881158020732946e-05,
|
|
"loss": 3.3975,
|
|
"step": 2022912
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6872772073222426e-05,
|
|
"loss": 3.4111,
|
|
"step": 2023424
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6864402504515635e-05,
|
|
"loss": 3.4099,
|
|
"step": 2023936
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6856016557005115e-05,
|
|
"loss": 3.3948,
|
|
"step": 2024448
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.68476306094946e-05,
|
|
"loss": 3.4182,
|
|
"step": 2024960
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.683924466198408e-05,
|
|
"loss": 3.4053,
|
|
"step": 2025472
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.683085871447356e-05,
|
|
"loss": 3.4081,
|
|
"step": 2025984
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6822489145766767e-05,
|
|
"loss": 3.4116,
|
|
"step": 2026496
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6814103198256247e-05,
|
|
"loss": 3.3949,
|
|
"step": 2027008
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6805717250745727e-05,
|
|
"loss": 3.3908,
|
|
"step": 2027520
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.6797331303235207e-05,
|
|
"loss": 3.3966,
|
|
"step": 2028032
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.678896173452842e-05,
|
|
"loss": 3.4082,
|
|
"step": 2028544
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.67805757870179e-05,
|
|
"loss": 3.4011,
|
|
"step": 2029056
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.677218983950738e-05,
|
|
"loss": 3.401,
|
|
"step": 2029568
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.676380389199686e-05,
|
|
"loss": 3.3982,
|
|
"step": 2030080
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.675543432329007e-05,
|
|
"loss": 3.4067,
|
|
"step": 2030592
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6747048375779552e-05,
|
|
"loss": 3.3967,
|
|
"step": 2031104
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6738662428269032e-05,
|
|
"loss": 3.3992,
|
|
"step": 2031616
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6730276480758512e-05,
|
|
"loss": 3.4181,
|
|
"step": 2032128
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.672189053324799e-05,
|
|
"loss": 3.4159,
|
|
"step": 2032640
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.67135209645412e-05,
|
|
"loss": 3.4105,
|
|
"step": 2033152
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.670513501703068e-05,
|
|
"loss": 3.396,
|
|
"step": 2033664
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.669674906952016e-05,
|
|
"loss": 3.4067,
|
|
"step": 2034176
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.668836312200964e-05,
|
|
"loss": 3.3973,
|
|
"step": 2034688
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6679993553302853e-05,
|
|
"loss": 3.4129,
|
|
"step": 2035200
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6671607605792333e-05,
|
|
"loss": 3.4015,
|
|
"step": 2035712
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.666322165828181e-05,
|
|
"loss": 3.4062,
|
|
"step": 2036224
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.665483571077129e-05,
|
|
"loss": 3.4013,
|
|
"step": 2036736
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6646466142064502e-05,
|
|
"loss": 3.4089,
|
|
"step": 2037248
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6638080194553986e-05,
|
|
"loss": 3.4056,
|
|
"step": 2037760
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6629694247043462e-05,
|
|
"loss": 3.4058,
|
|
"step": 2038272
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6621308299532942e-05,
|
|
"loss": 3.3942,
|
|
"step": 2038784
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6612922352022422e-05,
|
|
"loss": 3.4014,
|
|
"step": 2039296
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.660455278331563e-05,
|
|
"loss": 3.3932,
|
|
"step": 2039808
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6596166835805115e-05,
|
|
"loss": 3.3955,
|
|
"step": 2040320
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6587780888294595e-05,
|
|
"loss": 3.4,
|
|
"step": 2040832
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6579394940784075e-05,
|
|
"loss": 3.4002,
|
|
"step": 2041344
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6571008993273555e-05,
|
|
"loss": 3.4056,
|
|
"step": 2041856
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6562639424566764e-05,
|
|
"loss": 3.3958,
|
|
"step": 2042368
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6554253477056244e-05,
|
|
"loss": 3.4105,
|
|
"step": 2042880
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6545867529545724e-05,
|
|
"loss": 3.4196,
|
|
"step": 2043392
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6537481582035207e-05,
|
|
"loss": 3.4065,
|
|
"step": 2043904
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6529112013328416e-05,
|
|
"loss": 3.412,
|
|
"step": 2044416
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6520726065817896e-05,
|
|
"loss": 3.3988,
|
|
"step": 2044928
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6512340118307376e-05,
|
|
"loss": 3.4014,
|
|
"step": 2045440
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6503954170796856e-05,
|
|
"loss": 3.4002,
|
|
"step": 2045952
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.649558460209007e-05,
|
|
"loss": 3.3908,
|
|
"step": 2046464
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6487198654579548e-05,
|
|
"loss": 3.4012,
|
|
"step": 2046976
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6478812707069028e-05,
|
|
"loss": 3.4044,
|
|
"step": 2047488
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6470426759558508e-05,
|
|
"loss": 3.4074,
|
|
"step": 2048000
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6462057190851717e-05,
|
|
"loss": 3.4023,
|
|
"step": 2048512
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6453671243341197e-05,
|
|
"loss": 3.4095,
|
|
"step": 2049024
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6445285295830677e-05,
|
|
"loss": 3.3979,
|
|
"step": 2049536
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.643689934832016e-05,
|
|
"loss": 3.404,
|
|
"step": 2050048
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.642852977961337e-05,
|
|
"loss": 3.4052,
|
|
"step": 2050560
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.642014383210285e-05,
|
|
"loss": 3.4103,
|
|
"step": 2051072
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.641175788459233e-05,
|
|
"loss": 3.4119,
|
|
"step": 2051584
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.640337193708181e-05,
|
|
"loss": 3.4003,
|
|
"step": 2052096
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6394985989571286e-05,
|
|
"loss": 3.4185,
|
|
"step": 2052608
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.638660004206077e-05,
|
|
"loss": 3.4021,
|
|
"step": 2053120
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6378230473353982e-05,
|
|
"loss": 3.4096,
|
|
"step": 2053632
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6369844525843462e-05,
|
|
"loss": 3.4046,
|
|
"step": 2054144
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.636145857833294e-05,
|
|
"loss": 3.3987,
|
|
"step": 2054656
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.635307263082242e-05,
|
|
"loss": 3.3978,
|
|
"step": 2055168
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.634470306211563e-05,
|
|
"loss": 3.4228,
|
|
"step": 2055680
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.633631711460511e-05,
|
|
"loss": 3.3933,
|
|
"step": 2056192
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.632793116709459e-05,
|
|
"loss": 3.4043,
|
|
"step": 2056704
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.631954521958407e-05,
|
|
"loss": 3.4088,
|
|
"step": 2057216
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6311175650877283e-05,
|
|
"loss": 3.4089,
|
|
"step": 2057728
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.630278970336676e-05,
|
|
"loss": 3.4012,
|
|
"step": 2058240
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.629440375585624e-05,
|
|
"loss": 3.404,
|
|
"step": 2058752
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6286017808345723e-05,
|
|
"loss": 3.4054,
|
|
"step": 2059264
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6277631860835203e-05,
|
|
"loss": 3.4123,
|
|
"step": 2059776
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.6269262292128412e-05,
|
|
"loss": 3.4051,
|
|
"step": 2060288
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.857787609100342,
|
|
"eval_runtime": 306.714,
|
|
"eval_samples_per_second": 1244.126,
|
|
"eval_steps_per_second": 38.88,
|
|
"step": 2060640
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6260876344617892e-05,
|
|
"loss": 3.3943,
|
|
"step": 2060800
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6252490397107372e-05,
|
|
"loss": 3.3999,
|
|
"step": 2061312
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6244104449596852e-05,
|
|
"loss": 3.402,
|
|
"step": 2061824
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6235718502086332e-05,
|
|
"loss": 3.4063,
|
|
"step": 2062336
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6227348933379544e-05,
|
|
"loss": 3.4147,
|
|
"step": 2062848
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6218962985869024e-05,
|
|
"loss": 3.3997,
|
|
"step": 2063360
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6210577038358504e-05,
|
|
"loss": 3.4036,
|
|
"step": 2063872
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6202191090847984e-05,
|
|
"loss": 3.4044,
|
|
"step": 2064384
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6193805143337464e-05,
|
|
"loss": 3.3973,
|
|
"step": 2064896
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6185435574630677e-05,
|
|
"loss": 3.3916,
|
|
"step": 2065408
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6177049627120157e-05,
|
|
"loss": 3.4058,
|
|
"step": 2065920
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6168663679609637e-05,
|
|
"loss": 3.4154,
|
|
"step": 2066432
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6160277732099117e-05,
|
|
"loss": 3.4042,
|
|
"step": 2066944
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6151908163392326e-05,
|
|
"loss": 3.3831,
|
|
"step": 2067456
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6143522215881806e-05,
|
|
"loss": 3.4044,
|
|
"step": 2067968
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6135136268371286e-05,
|
|
"loss": 3.3859,
|
|
"step": 2068480
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6126750320860766e-05,
|
|
"loss": 3.3995,
|
|
"step": 2068992
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6118364373350246e-05,
|
|
"loss": 3.4009,
|
|
"step": 2069504
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6109994804643458e-05,
|
|
"loss": 3.3991,
|
|
"step": 2070016
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6101608857132938e-05,
|
|
"loss": 3.4031,
|
|
"step": 2070528
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6093222909622415e-05,
|
|
"loss": 3.4094,
|
|
"step": 2071040
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6084836962111895e-05,
|
|
"loss": 3.4029,
|
|
"step": 2071552
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6076451014601378e-05,
|
|
"loss": 3.4016,
|
|
"step": 2072064
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6068065067090858e-05,
|
|
"loss": 3.3928,
|
|
"step": 2072576
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6059695498384067e-05,
|
|
"loss": 3.4034,
|
|
"step": 2073088
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6051309550873547e-05,
|
|
"loss": 3.4055,
|
|
"step": 2073600
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6042923603363027e-05,
|
|
"loss": 3.3932,
|
|
"step": 2074112
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.6034537655852507e-05,
|
|
"loss": 3.4027,
|
|
"step": 2074624
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.602616808714572e-05,
|
|
"loss": 3.395,
|
|
"step": 2075136
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.60177821396352e-05,
|
|
"loss": 3.3969,
|
|
"step": 2075648
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.600939619212468e-05,
|
|
"loss": 3.3933,
|
|
"step": 2076160
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.600101024461416e-05,
|
|
"loss": 3.4058,
|
|
"step": 2076672
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.599262429710364e-05,
|
|
"loss": 3.4043,
|
|
"step": 2077184
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5984254728396848e-05,
|
|
"loss": 3.4055,
|
|
"step": 2077696
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.597586878088633e-05,
|
|
"loss": 3.3975,
|
|
"step": 2078208
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.596748283337581e-05,
|
|
"loss": 3.3988,
|
|
"step": 2078720
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.595909688586529e-05,
|
|
"loss": 3.4056,
|
|
"step": 2079232
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.59507273171585e-05,
|
|
"loss": 3.3964,
|
|
"step": 2079744
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.594234136964798e-05,
|
|
"loss": 3.3907,
|
|
"step": 2080256
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.593395542213746e-05,
|
|
"loss": 3.3932,
|
|
"step": 2080768
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.592556947462694e-05,
|
|
"loss": 3.3921,
|
|
"step": 2081280
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.591718352711642e-05,
|
|
"loss": 3.3979,
|
|
"step": 2081792
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5908813958409633e-05,
|
|
"loss": 3.3932,
|
|
"step": 2082304
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5900428010899113e-05,
|
|
"loss": 3.3992,
|
|
"step": 2082816
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5892042063388593e-05,
|
|
"loss": 3.399,
|
|
"step": 2083328
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.588365611587807e-05,
|
|
"loss": 3.4042,
|
|
"step": 2083840
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.587527016836755e-05,
|
|
"loss": 3.4055,
|
|
"step": 2084352
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5866884220857033e-05,
|
|
"loss": 3.3985,
|
|
"step": 2084864
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5858498273346513e-05,
|
|
"loss": 3.3757,
|
|
"step": 2085376
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.585012870463972e-05,
|
|
"loss": 3.4066,
|
|
"step": 2085888
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.58417427571292e-05,
|
|
"loss": 3.3998,
|
|
"step": 2086400
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.583335680961868e-05,
|
|
"loss": 3.3962,
|
|
"step": 2086912
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.582497086210816e-05,
|
|
"loss": 3.3988,
|
|
"step": 2087424
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5816601293401374e-05,
|
|
"loss": 3.3955,
|
|
"step": 2087936
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5808215345890854e-05,
|
|
"loss": 3.378,
|
|
"step": 2088448
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5799829398380334e-05,
|
|
"loss": 3.3877,
|
|
"step": 2088960
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5791443450869814e-05,
|
|
"loss": 3.3963,
|
|
"step": 2089472
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5783057503359294e-05,
|
|
"loss": 3.3883,
|
|
"step": 2089984
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5774687934652503e-05,
|
|
"loss": 3.4075,
|
|
"step": 2090496
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5766301987141986e-05,
|
|
"loss": 3.3974,
|
|
"step": 2091008
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5757916039631466e-05,
|
|
"loss": 3.3834,
|
|
"step": 2091520
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5749530092120946e-05,
|
|
"loss": 3.4085,
|
|
"step": 2092032
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5741160523414155e-05,
|
|
"loss": 3.3878,
|
|
"step": 2092544
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5732774575903635e-05,
|
|
"loss": 3.3828,
|
|
"step": 2093056
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5724388628393115e-05,
|
|
"loss": 3.4019,
|
|
"step": 2093568
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5716002680882595e-05,
|
|
"loss": 3.3941,
|
|
"step": 2094080
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5707633112175808e-05,
|
|
"loss": 3.3867,
|
|
"step": 2094592
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5699247164665288e-05,
|
|
"loss": 3.3904,
|
|
"step": 2095104
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5690861217154768e-05,
|
|
"loss": 3.383,
|
|
"step": 2095616
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5682475269644247e-05,
|
|
"loss": 3.3902,
|
|
"step": 2096128
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5674089322133724e-05,
|
|
"loss": 3.4,
|
|
"step": 2096640
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5665703374623204e-05,
|
|
"loss": 3.397,
|
|
"step": 2097152
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.565733380591642e-05,
|
|
"loss": 3.4028,
|
|
"step": 2097664
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5648947858405896e-05,
|
|
"loss": 3.3936,
|
|
"step": 2098176
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5640561910895376e-05,
|
|
"loss": 3.4139,
|
|
"step": 2098688
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5632175963384856e-05,
|
|
"loss": 3.3878,
|
|
"step": 2099200
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.562380639467807e-05,
|
|
"loss": 3.4028,
|
|
"step": 2099712
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.561542044716755e-05,
|
|
"loss": 3.3958,
|
|
"step": 2100224
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.560703449965703e-05,
|
|
"loss": 3.3869,
|
|
"step": 2100736
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.559864855214651e-05,
|
|
"loss": 3.4043,
|
|
"step": 2101248
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.559027898343972e-05,
|
|
"loss": 3.3997,
|
|
"step": 2101760
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5581893035929198e-05,
|
|
"loss": 3.397,
|
|
"step": 2102272
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5573507088418678e-05,
|
|
"loss": 3.4051,
|
|
"step": 2102784
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5565121140908158e-05,
|
|
"loss": 3.3824,
|
|
"step": 2103296
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.555675157220137e-05,
|
|
"loss": 3.381,
|
|
"step": 2103808
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.554836562469085e-05,
|
|
"loss": 3.387,
|
|
"step": 2104320
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.553997967718033e-05,
|
|
"loss": 3.3994,
|
|
"step": 2104832
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.553159372966981e-05,
|
|
"loss": 3.3925,
|
|
"step": 2105344
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.552322416096302e-05,
|
|
"loss": 3.3925,
|
|
"step": 2105856
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5514838213452502e-05,
|
|
"loss": 3.3879,
|
|
"step": 2106368
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5506452265941982e-05,
|
|
"loss": 3.3933,
|
|
"step": 2106880
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5498066318431462e-05,
|
|
"loss": 3.3875,
|
|
"step": 2107392
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.548969674972467e-05,
|
|
"loss": 3.3931,
|
|
"step": 2107904
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.548131080221415e-05,
|
|
"loss": 3.4022,
|
|
"step": 2108416
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.547292485470363e-05,
|
|
"loss": 3.4054,
|
|
"step": 2108928
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.546453890719311e-05,
|
|
"loss": 3.4014,
|
|
"step": 2109440
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5456152959682595e-05,
|
|
"loss": 3.3862,
|
|
"step": 2109952
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5447767012172075e-05,
|
|
"loss": 3.3924,
|
|
"step": 2110464
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5439397443465284e-05,
|
|
"loss": 3.39,
|
|
"step": 2110976
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5431011495954764e-05,
|
|
"loss": 3.401,
|
|
"step": 2111488
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5422625548444244e-05,
|
|
"loss": 3.3941,
|
|
"step": 2112000
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5414239600933724e-05,
|
|
"loss": 3.3966,
|
|
"step": 2112512
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5405853653423204e-05,
|
|
"loss": 3.3937,
|
|
"step": 2113024
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5397484084716416e-05,
|
|
"loss": 3.3934,
|
|
"step": 2113536
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5389098137205896e-05,
|
|
"loss": 3.3986,
|
|
"step": 2114048
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5380712189695376e-05,
|
|
"loss": 3.3918,
|
|
"step": 2114560
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5372326242184853e-05,
|
|
"loss": 3.3837,
|
|
"step": 2115072
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5363940294674332e-05,
|
|
"loss": 3.3893,
|
|
"step": 2115584
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.535557072596755e-05,
|
|
"loss": 3.3876,
|
|
"step": 2116096
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5347184778457025e-05,
|
|
"loss": 3.3849,
|
|
"step": 2116608
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5338798830946505e-05,
|
|
"loss": 3.3899,
|
|
"step": 2117120
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5330412883435985e-05,
|
|
"loss": 3.3896,
|
|
"step": 2117632
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5322026935925465e-05,
|
|
"loss": 3.396,
|
|
"step": 2118144
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5313657367218677e-05,
|
|
"loss": 3.3834,
|
|
"step": 2118656
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5305271419708157e-05,
|
|
"loss": 3.4009,
|
|
"step": 2119168
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5296885472197637e-05,
|
|
"loss": 3.4094,
|
|
"step": 2119680
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5288499524687117e-05,
|
|
"loss": 3.3966,
|
|
"step": 2120192
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5280129955980326e-05,
|
|
"loss": 3.4019,
|
|
"step": 2120704
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5271744008469806e-05,
|
|
"loss": 3.3898,
|
|
"step": 2121216
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5263358060959286e-05,
|
|
"loss": 3.3916,
|
|
"step": 2121728
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5254972113448768e-05,
|
|
"loss": 3.3921,
|
|
"step": 2122240
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5246586165938248e-05,
|
|
"loss": 3.3775,
|
|
"step": 2122752
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5238216597231459e-05,
|
|
"loss": 3.3938,
|
|
"step": 2123264
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5229830649720939e-05,
|
|
"loss": 3.3946,
|
|
"step": 2123776
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5221444702210418e-05,
|
|
"loss": 3.4,
|
|
"step": 2124288
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.52130587546999e-05,
|
|
"loss": 3.3879,
|
|
"step": 2124800
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.520467280718938e-05,
|
|
"loss": 3.4015,
|
|
"step": 2125312
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.519630323848259e-05,
|
|
"loss": 3.3896,
|
|
"step": 2125824
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.518791729097207e-05,
|
|
"loss": 3.3914,
|
|
"step": 2126336
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.517953134346155e-05,
|
|
"loss": 3.3958,
|
|
"step": 2126848
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.517114539595103e-05,
|
|
"loss": 3.3963,
|
|
"step": 2127360
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5162759448440509e-05,
|
|
"loss": 3.4041,
|
|
"step": 2127872
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5154389879733721e-05,
|
|
"loss": 3.3889,
|
|
"step": 2128384
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5146003932223201e-05,
|
|
"loss": 3.4107,
|
|
"step": 2128896
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.513761798471268e-05,
|
|
"loss": 3.3899,
|
|
"step": 2129408
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.512923203720216e-05,
|
|
"loss": 3.4022,
|
|
"step": 2129920
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.512084608969164e-05,
|
|
"loss": 3.3954,
|
|
"step": 2130432
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5112476520984852e-05,
|
|
"loss": 3.3872,
|
|
"step": 2130944
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.510409057347433e-05,
|
|
"loss": 3.3905,
|
|
"step": 2131456
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.509570462596381e-05,
|
|
"loss": 3.4132,
|
|
"step": 2131968
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.508731867845329e-05,
|
|
"loss": 3.383,
|
|
"step": 2132480
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5078949109746501e-05,
|
|
"loss": 3.3935,
|
|
"step": 2132992
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5070563162235981e-05,
|
|
"loss": 3.3979,
|
|
"step": 2133504
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5062177214725463e-05,
|
|
"loss": 3.3958,
|
|
"step": 2134016
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5053791267214943e-05,
|
|
"loss": 3.3939,
|
|
"step": 2134528
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5045421698508152e-05,
|
|
"loss": 3.3927,
|
|
"step": 2135040
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5037035750997633e-05,
|
|
"loss": 3.3941,
|
|
"step": 2135552
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5028649803487113e-05,
|
|
"loss": 3.404,
|
|
"step": 2136064
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5020263855976593e-05,
|
|
"loss": 3.3912,
|
|
"step": 2136576
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 3.8584606647491455,
|
|
"eval_runtime": 305.9781,
|
|
"eval_samples_per_second": 1247.119,
|
|
"eval_steps_per_second": 38.973,
|
|
"step": 2136960
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.5011894287269804e-05,
|
|
"loss": 3.3777,
|
|
"step": 2137088
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.5003524718563017e-05,
|
|
"loss": 3.3865,
|
|
"step": 2137600
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4995138771052497e-05,
|
|
"loss": 3.395,
|
|
"step": 2138112
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4986752823541975e-05,
|
|
"loss": 3.3991,
|
|
"step": 2138624
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4978366876031455e-05,
|
|
"loss": 3.3974,
|
|
"step": 2139136
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4969997307324667e-05,
|
|
"loss": 3.3914,
|
|
"step": 2139648
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4961611359814149e-05,
|
|
"loss": 3.3984,
|
|
"step": 2140160
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4953225412303625e-05,
|
|
"loss": 3.3911,
|
|
"step": 2140672
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4944839464793105e-05,
|
|
"loss": 3.3858,
|
|
"step": 2141184
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4936453517282587e-05,
|
|
"loss": 3.3809,
|
|
"step": 2141696
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.49280839485758e-05,
|
|
"loss": 3.3983,
|
|
"step": 2142208
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4919698001065278e-05,
|
|
"loss": 3.4027,
|
|
"step": 2142720
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4911312053554758e-05,
|
|
"loss": 3.3961,
|
|
"step": 2143232
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4902926106044238e-05,
|
|
"loss": 3.375,
|
|
"step": 2143744
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4894556537337448e-05,
|
|
"loss": 3.392,
|
|
"step": 2144256
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4886170589826928e-05,
|
|
"loss": 3.3736,
|
|
"step": 2144768
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4877784642316408e-05,
|
|
"loss": 3.3912,
|
|
"step": 2145280
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4869398694805888e-05,
|
|
"loss": 3.3881,
|
|
"step": 2145792
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.486101274729537e-05,
|
|
"loss": 3.3883,
|
|
"step": 2146304
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4852643178588579e-05,
|
|
"loss": 3.3926,
|
|
"step": 2146816
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4844257231078059e-05,
|
|
"loss": 3.4017,
|
|
"step": 2147328
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.483587128356754e-05,
|
|
"loss": 3.3907,
|
|
"step": 2147840
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.482748533605702e-05,
|
|
"loss": 3.3916,
|
|
"step": 2148352
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.48190993885465e-05,
|
|
"loss": 3.3883,
|
|
"step": 2148864
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.481071344103598e-05,
|
|
"loss": 3.3954,
|
|
"step": 2149376
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4802343872329191e-05,
|
|
"loss": 3.3952,
|
|
"step": 2149888
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4793957924818671e-05,
|
|
"loss": 3.385,
|
|
"step": 2150400
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4785571977308151e-05,
|
|
"loss": 3.3914,
|
|
"step": 2150912
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.477718602979763e-05,
|
|
"loss": 3.3847,
|
|
"step": 2151424
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.4768816461090842e-05,
|
|
"loss": 3.3836,
|
|
"step": 2151936
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4760430513580322e-05,
|
|
"loss": 3.3872,
|
|
"step": 2152448
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4752044566069804e-05,
|
|
"loss": 3.3917,
|
|
"step": 2152960
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.474365861855928e-05,
|
|
"loss": 3.3899,
|
|
"step": 2153472
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.473527267104876e-05,
|
|
"loss": 3.3985,
|
|
"step": 2153984
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4726886723538242e-05,
|
|
"loss": 3.3887,
|
|
"step": 2154496
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4718500776027722e-05,
|
|
"loss": 3.3879,
|
|
"step": 2155008
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4710131207320933e-05,
|
|
"loss": 3.3979,
|
|
"step": 2155520
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4701745259810412e-05,
|
|
"loss": 3.3838,
|
|
"step": 2156032
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4693359312299892e-05,
|
|
"loss": 3.3828,
|
|
"step": 2156544
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4684973364789372e-05,
|
|
"loss": 3.384,
|
|
"step": 2157056
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4676603796082583e-05,
|
|
"loss": 3.3827,
|
|
"step": 2157568
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4668217848572063e-05,
|
|
"loss": 3.3845,
|
|
"step": 2158080
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4659831901061543e-05,
|
|
"loss": 3.3849,
|
|
"step": 2158592
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4651445953551025e-05,
|
|
"loss": 3.3894,
|
|
"step": 2159104
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4643060006040505e-05,
|
|
"loss": 3.3894,
|
|
"step": 2159616
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4634690437333714e-05,
|
|
"loss": 3.394,
|
|
"step": 2160128
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4626304489823195e-05,
|
|
"loss": 3.3919,
|
|
"step": 2160640
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4617918542312675e-05,
|
|
"loss": 3.3896,
|
|
"step": 2161152
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4609532594802155e-05,
|
|
"loss": 3.3674,
|
|
"step": 2161664
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4601163026095366e-05,
|
|
"loss": 3.396,
|
|
"step": 2162176
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4592777078584846e-05,
|
|
"loss": 3.3872,
|
|
"step": 2162688
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4584391131074326e-05,
|
|
"loss": 3.3905,
|
|
"step": 2163200
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4576005183563806e-05,
|
|
"loss": 3.3857,
|
|
"step": 2163712
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4567635614857017e-05,
|
|
"loss": 3.3842,
|
|
"step": 2164224
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4559249667346497e-05,
|
|
"loss": 3.3722,
|
|
"step": 2164736
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4550863719835978e-05,
|
|
"loss": 3.3788,
|
|
"step": 2165248
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4542477772325458e-05,
|
|
"loss": 3.3875,
|
|
"step": 2165760
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4534091824814935e-05,
|
|
"loss": 3.3769,
|
|
"step": 2166272
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4525722256108149e-05,
|
|
"loss": 3.3968,
|
|
"step": 2166784
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4517336308597629e-05,
|
|
"loss": 3.3883,
|
|
"step": 2167296
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4508950361087109e-05,
|
|
"loss": 3.3714,
|
|
"step": 2167808
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4500564413576587e-05,
|
|
"loss": 3.4017,
|
|
"step": 2168320
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.44921948448698e-05,
|
|
"loss": 3.3772,
|
|
"step": 2168832
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.448380889735928e-05,
|
|
"loss": 3.3733,
|
|
"step": 2169344
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4475422949848758e-05,
|
|
"loss": 3.3897,
|
|
"step": 2169856
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4467037002338238e-05,
|
|
"loss": 3.3831,
|
|
"step": 2170368
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.445866743363145e-05,
|
|
"loss": 3.3739,
|
|
"step": 2170880
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4450281486120932e-05,
|
|
"loss": 3.3852,
|
|
"step": 2171392
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4441895538610409e-05,
|
|
"loss": 3.3722,
|
|
"step": 2171904
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4433509591099889e-05,
|
|
"loss": 3.3765,
|
|
"step": 2172416
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4425123643589369e-05,
|
|
"loss": 3.3916,
|
|
"step": 2172928
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.441673769607885e-05,
|
|
"loss": 3.3859,
|
|
"step": 2173440
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.440836812737206e-05,
|
|
"loss": 3.3894,
|
|
"step": 2173952
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4399982179861541e-05,
|
|
"loss": 3.3853,
|
|
"step": 2174464
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4391596232351021e-05,
|
|
"loss": 3.4036,
|
|
"step": 2174976
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4383210284840501e-05,
|
|
"loss": 3.3776,
|
|
"step": 2175488
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4374840716133712e-05,
|
|
"loss": 3.3942,
|
|
"step": 2176000
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4366454768623192e-05,
|
|
"loss": 3.3825,
|
|
"step": 2176512
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4358068821112672e-05,
|
|
"loss": 3.3808,
|
|
"step": 2177024
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4349682873602152e-05,
|
|
"loss": 3.3911,
|
|
"step": 2177536
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4341313304895362e-05,
|
|
"loss": 3.3953,
|
|
"step": 2178048
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4332927357384842e-05,
|
|
"loss": 3.3851,
|
|
"step": 2178560
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4324541409874322e-05,
|
|
"loss": 3.3958,
|
|
"step": 2179072
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4316155462363804e-05,
|
|
"loss": 3.3732,
|
|
"step": 2179584
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4307785893657013e-05,
|
|
"loss": 3.3681,
|
|
"step": 2180096
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4299399946146495e-05,
|
|
"loss": 3.3776,
|
|
"step": 2180608
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4291013998635975e-05,
|
|
"loss": 3.3874,
|
|
"step": 2181120
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4282628051125455e-05,
|
|
"loss": 3.3839,
|
|
"step": 2181632
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4274258482418665e-05,
|
|
"loss": 3.3853,
|
|
"step": 2182144
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.4265872534908145e-05,
|
|
"loss": 3.377,
|
|
"step": 2182656
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4257486587397625e-05,
|
|
"loss": 3.381,
|
|
"step": 2183168
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4249100639887105e-05,
|
|
"loss": 3.378,
|
|
"step": 2183680
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4240731071180316e-05,
|
|
"loss": 3.3812,
|
|
"step": 2184192
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4232345123669796e-05,
|
|
"loss": 3.3944,
|
|
"step": 2184704
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4223959176159276e-05,
|
|
"loss": 3.3935,
|
|
"step": 2185216
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4215573228648758e-05,
|
|
"loss": 3.395,
|
|
"step": 2185728
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4207203659941967e-05,
|
|
"loss": 3.3727,
|
|
"step": 2186240
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4198817712431448e-05,
|
|
"loss": 3.3845,
|
|
"step": 2186752
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4190431764920928e-05,
|
|
"loss": 3.3806,
|
|
"step": 2187264
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4182045817410408e-05,
|
|
"loss": 3.3901,
|
|
"step": 2187776
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4173676248703619e-05,
|
|
"loss": 3.3856,
|
|
"step": 2188288
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4165290301193099e-05,
|
|
"loss": 3.3831,
|
|
"step": 2188800
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4156904353682579e-05,
|
|
"loss": 3.384,
|
|
"step": 2189312
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4148518406172059e-05,
|
|
"loss": 3.3863,
|
|
"step": 2189824
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4140132458661537e-05,
|
|
"loss": 3.3882,
|
|
"step": 2190336
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.413176288995475e-05,
|
|
"loss": 3.3838,
|
|
"step": 2190848
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.412337694244423e-05,
|
|
"loss": 3.3784,
|
|
"step": 2191360
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4114990994933708e-05,
|
|
"loss": 3.3762,
|
|
"step": 2191872
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4106605047423188e-05,
|
|
"loss": 3.3791,
|
|
"step": 2192384
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4098235478716402e-05,
|
|
"loss": 3.3786,
|
|
"step": 2192896
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4089849531205882e-05,
|
|
"loss": 3.377,
|
|
"step": 2193408
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4081463583695358e-05,
|
|
"loss": 3.3794,
|
|
"step": 2193920
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.407307763618484e-05,
|
|
"loss": 3.387,
|
|
"step": 2194432
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.406469168867432e-05,
|
|
"loss": 3.3727,
|
|
"step": 2194944
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4056322119967533e-05,
|
|
"loss": 3.3902,
|
|
"step": 2195456
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.404793617245701e-05,
|
|
"loss": 3.4011,
|
|
"step": 2195968
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.403955022494649e-05,
|
|
"loss": 3.389,
|
|
"step": 2196480
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.403116427743597e-05,
|
|
"loss": 3.3879,
|
|
"step": 2196992
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4022794708729182e-05,
|
|
"loss": 3.386,
|
|
"step": 2197504
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4014408761218661e-05,
|
|
"loss": 3.3749,
|
|
"step": 2198016
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.4006022813708141e-05,
|
|
"loss": 3.3891,
|
|
"step": 2198528
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3997636866197621e-05,
|
|
"loss": 3.3658,
|
|
"step": 2199040
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3989250918687103e-05,
|
|
"loss": 3.3846,
|
|
"step": 2199552
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3980864971176583e-05,
|
|
"loss": 3.3816,
|
|
"step": 2200064
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3972495402469792e-05,
|
|
"loss": 3.389,
|
|
"step": 2200576
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3964109454959274e-05,
|
|
"loss": 3.3817,
|
|
"step": 2201088
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3955723507448754e-05,
|
|
"loss": 3.3901,
|
|
"step": 2201600
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3947337559938234e-05,
|
|
"loss": 3.3773,
|
|
"step": 2202112
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3938951612427714e-05,
|
|
"loss": 3.3791,
|
|
"step": 2202624
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3930582043720924e-05,
|
|
"loss": 3.3902,
|
|
"step": 2203136
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3922196096210404e-05,
|
|
"loss": 3.3863,
|
|
"step": 2203648
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3913810148699884e-05,
|
|
"loss": 3.396,
|
|
"step": 2204160
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3905424201189363e-05,
|
|
"loss": 3.38,
|
|
"step": 2204672
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3897038253678843e-05,
|
|
"loss": 3.396,
|
|
"step": 2205184
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3888652306168322e-05,
|
|
"loss": 3.3845,
|
|
"step": 2205696
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3880266358657804e-05,
|
|
"loss": 3.387,
|
|
"step": 2206208
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3871880411147284e-05,
|
|
"loss": 3.3875,
|
|
"step": 2206720
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3863510842440495e-05,
|
|
"loss": 3.3804,
|
|
"step": 2207232
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3855124894929975e-05,
|
|
"loss": 3.3811,
|
|
"step": 2207744
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3846738947419455e-05,
|
|
"loss": 3.3989,
|
|
"step": 2208256
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3838352999908935e-05,
|
|
"loss": 3.3753,
|
|
"step": 2208768
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3829983431202146e-05,
|
|
"loss": 3.3827,
|
|
"step": 2209280
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3821597483691626e-05,
|
|
"loss": 3.3884,
|
|
"step": 2209792
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3813211536181105e-05,
|
|
"loss": 3.389,
|
|
"step": 2210304
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3804825588670587e-05,
|
|
"loss": 3.3841,
|
|
"step": 2210816
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3796456019963796e-05,
|
|
"loss": 3.3844,
|
|
"step": 2211328
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3788070072453276e-05,
|
|
"loss": 3.3837,
|
|
"step": 2211840
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3779684124942758e-05,
|
|
"loss": 3.3917,
|
|
"step": 2212352
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.3771298177432238e-05,
|
|
"loss": 3.3848,
|
|
"step": 2212864
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.8589565753936768,
|
|
"eval_runtime": 303.4359,
|
|
"eval_samples_per_second": 1257.567,
|
|
"eval_steps_per_second": 39.3,
|
|
"step": 2213280
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3762928608725449e-05,
|
|
"loss": 3.3788,
|
|
"step": 2213376
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3754542661214929e-05,
|
|
"loss": 3.375,
|
|
"step": 2213888
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3746156713704408e-05,
|
|
"loss": 3.3792,
|
|
"step": 2214400
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3737770766193888e-05,
|
|
"loss": 3.3959,
|
|
"step": 2214912
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.37294011974871e-05,
|
|
"loss": 3.3844,
|
|
"step": 2215424
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.372101524997658e-05,
|
|
"loss": 3.3827,
|
|
"step": 2215936
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3712629302466059e-05,
|
|
"loss": 3.3864,
|
|
"step": 2216448
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3704243354955539e-05,
|
|
"loss": 3.3845,
|
|
"step": 2216960
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.369587378624875e-05,
|
|
"loss": 3.373,
|
|
"step": 2217472
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.368748783873823e-05,
|
|
"loss": 3.3778,
|
|
"step": 2217984
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3679101891227711e-05,
|
|
"loss": 3.3842,
|
|
"step": 2218496
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3670715943717191e-05,
|
|
"loss": 3.3919,
|
|
"step": 2219008
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3662329996206668e-05,
|
|
"loss": 3.3912,
|
|
"step": 2219520
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3653960427499882e-05,
|
|
"loss": 3.3618,
|
|
"step": 2220032
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3645574479989362e-05,
|
|
"loss": 3.3785,
|
|
"step": 2220544
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3637188532478839e-05,
|
|
"loss": 3.3675,
|
|
"step": 2221056
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.362880258496832e-05,
|
|
"loss": 3.3833,
|
|
"step": 2221568
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3620433016261533e-05,
|
|
"loss": 3.3774,
|
|
"step": 2222080
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3612047068751013e-05,
|
|
"loss": 3.3805,
|
|
"step": 2222592
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3603661121240491e-05,
|
|
"loss": 3.377,
|
|
"step": 2223104
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3595275173729971e-05,
|
|
"loss": 3.3957,
|
|
"step": 2223616
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3586905605023183e-05,
|
|
"loss": 3.3795,
|
|
"step": 2224128
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3578519657512665e-05,
|
|
"loss": 3.3825,
|
|
"step": 2224640
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3570133710002142e-05,
|
|
"loss": 3.3769,
|
|
"step": 2225152
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3561747762491622e-05,
|
|
"loss": 3.3868,
|
|
"step": 2225664
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3553361814981103e-05,
|
|
"loss": 3.386,
|
|
"step": 2226176
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3544992246274312e-05,
|
|
"loss": 3.3738,
|
|
"step": 2226688
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3536606298763792e-05,
|
|
"loss": 3.3819,
|
|
"step": 2227200
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3528220351253274e-05,
|
|
"loss": 3.3751,
|
|
"step": 2227712
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.3519834403742754e-05,
|
|
"loss": 3.3716,
|
|
"step": 2228224
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3511464835035965e-05,
|
|
"loss": 3.381,
|
|
"step": 2228736
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3503078887525445e-05,
|
|
"loss": 3.3799,
|
|
"step": 2229248
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3494692940014925e-05,
|
|
"loss": 3.384,
|
|
"step": 2229760
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3486306992504405e-05,
|
|
"loss": 3.386,
|
|
"step": 2230272
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3477937423797615e-05,
|
|
"loss": 3.3799,
|
|
"step": 2230784
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3469551476287095e-05,
|
|
"loss": 3.3782,
|
|
"step": 2231296
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3461165528776575e-05,
|
|
"loss": 3.3899,
|
|
"step": 2231808
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3452779581266057e-05,
|
|
"loss": 3.3682,
|
|
"step": 2232320
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3444410012559266e-05,
|
|
"loss": 3.376,
|
|
"step": 2232832
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3436024065048746e-05,
|
|
"loss": 3.3785,
|
|
"step": 2233344
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3427638117538228e-05,
|
|
"loss": 3.3678,
|
|
"step": 2233856
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3419252170027708e-05,
|
|
"loss": 3.3774,
|
|
"step": 2234368
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3410882601320918e-05,
|
|
"loss": 3.3785,
|
|
"step": 2234880
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3402496653810398e-05,
|
|
"loss": 3.3784,
|
|
"step": 2235392
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3394110706299878e-05,
|
|
"loss": 3.38,
|
|
"step": 2235904
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3385724758789358e-05,
|
|
"loss": 3.3825,
|
|
"step": 2236416
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3377355190082569e-05,
|
|
"loss": 3.3866,
|
|
"step": 2236928
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3368969242572049e-05,
|
|
"loss": 3.3795,
|
|
"step": 2237440
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3360583295061529e-05,
|
|
"loss": 3.3608,
|
|
"step": 2237952
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.335219734755101e-05,
|
|
"loss": 3.3828,
|
|
"step": 2238464
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.334382777884422e-05,
|
|
"loss": 3.3786,
|
|
"step": 2238976
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.33354418313337e-05,
|
|
"loss": 3.3817,
|
|
"step": 2239488
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3327055883823181e-05,
|
|
"loss": 3.3776,
|
|
"step": 2240000
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3318669936312661e-05,
|
|
"loss": 3.3704,
|
|
"step": 2240512
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3310300367605872e-05,
|
|
"loss": 3.3646,
|
|
"step": 2241024
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3301914420095352e-05,
|
|
"loss": 3.3679,
|
|
"step": 2241536
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3293528472584832e-05,
|
|
"loss": 3.3803,
|
|
"step": 2242048
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3285142525074312e-05,
|
|
"loss": 3.3655,
|
|
"step": 2242560
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3276772956367523e-05,
|
|
"loss": 3.3891,
|
|
"step": 2243072
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3268387008857003e-05,
|
|
"loss": 3.3816,
|
|
"step": 2243584
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3260001061346483e-05,
|
|
"loss": 3.3597,
|
|
"step": 2244096
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3251615113835963e-05,
|
|
"loss": 3.3873,
|
|
"step": 2244608
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3243229166325441e-05,
|
|
"loss": 3.3678,
|
|
"step": 2245120
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3234859597618653e-05,
|
|
"loss": 3.3687,
|
|
"step": 2245632
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3226473650108135e-05,
|
|
"loss": 3.3728,
|
|
"step": 2246144
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3218087702597615e-05,
|
|
"loss": 3.3797,
|
|
"step": 2246656
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3209701755087092e-05,
|
|
"loss": 3.3656,
|
|
"step": 2247168
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3201332186380306e-05,
|
|
"loss": 3.3774,
|
|
"step": 2247680
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3192946238869786e-05,
|
|
"loss": 3.3603,
|
|
"step": 2248192
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3184560291359262e-05,
|
|
"loss": 3.3679,
|
|
"step": 2248704
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3176174343848744e-05,
|
|
"loss": 3.3847,
|
|
"step": 2249216
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3167788396338224e-05,
|
|
"loss": 3.3713,
|
|
"step": 2249728
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3159418827631436e-05,
|
|
"loss": 3.3844,
|
|
"step": 2250240
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3151032880120915e-05,
|
|
"loss": 3.3749,
|
|
"step": 2250752
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3142646932610395e-05,
|
|
"loss": 3.3933,
|
|
"step": 2251264
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3134260985099875e-05,
|
|
"loss": 3.3669,
|
|
"step": 2251776
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3125891416393089e-05,
|
|
"loss": 3.3862,
|
|
"step": 2252288
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3117505468882565e-05,
|
|
"loss": 3.3718,
|
|
"step": 2252800
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3109119521372045e-05,
|
|
"loss": 3.3733,
|
|
"step": 2253312
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3100733573861527e-05,
|
|
"loss": 3.3805,
|
|
"step": 2253824
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3092364005154736e-05,
|
|
"loss": 3.3842,
|
|
"step": 2254336
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3083978057644216e-05,
|
|
"loss": 3.3775,
|
|
"step": 2254848
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3075592110133698e-05,
|
|
"loss": 3.3841,
|
|
"step": 2255360
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3067206162623178e-05,
|
|
"loss": 3.3628,
|
|
"step": 2255872
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3058836593916388e-05,
|
|
"loss": 3.3607,
|
|
"step": 2256384
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3050450646405868e-05,
|
|
"loss": 3.3701,
|
|
"step": 2256896
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3042064698895348e-05,
|
|
"loss": 3.3746,
|
|
"step": 2257408
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3033678751384828e-05,
|
|
"loss": 3.373,
|
|
"step": 2257920
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3025309182678039e-05,
|
|
"loss": 3.3725,
|
|
"step": 2258432
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.3016923235167519e-05,
|
|
"loss": 3.3725,
|
|
"step": 2258944
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.3008537287656999e-05,
|
|
"loss": 3.3692,
|
|
"step": 2259456
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.300015134014648e-05,
|
|
"loss": 3.3719,
|
|
"step": 2259968
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.299176539263596e-05,
|
|
"loss": 3.3703,
|
|
"step": 2260480
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.298339582392917e-05,
|
|
"loss": 3.3813,
|
|
"step": 2260992
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2975009876418651e-05,
|
|
"loss": 3.3866,
|
|
"step": 2261504
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2966623928908131e-05,
|
|
"loss": 3.3813,
|
|
"step": 2262016
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2958237981397611e-05,
|
|
"loss": 3.3668,
|
|
"step": 2262528
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2949852033887091e-05,
|
|
"loss": 3.3724,
|
|
"step": 2263040
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2941482465180302e-05,
|
|
"loss": 3.3727,
|
|
"step": 2263552
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2933096517669782e-05,
|
|
"loss": 3.3821,
|
|
"step": 2264064
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2924710570159262e-05,
|
|
"loss": 3.3728,
|
|
"step": 2264576
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2916324622648743e-05,
|
|
"loss": 3.3751,
|
|
"step": 2265088
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.290793867513822e-05,
|
|
"loss": 3.3739,
|
|
"step": 2265600
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.28995527276277e-05,
|
|
"loss": 3.3802,
|
|
"step": 2266112
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2891183158920914e-05,
|
|
"loss": 3.376,
|
|
"step": 2266624
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.288279721141039e-05,
|
|
"loss": 3.3748,
|
|
"step": 2267136
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.287441126389987e-05,
|
|
"loss": 3.3675,
|
|
"step": 2267648
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2866025316389352e-05,
|
|
"loss": 3.3644,
|
|
"step": 2268160
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2857655747682565e-05,
|
|
"loss": 3.3747,
|
|
"step": 2268672
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2849269800172043e-05,
|
|
"loss": 3.3633,
|
|
"step": 2269184
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2840883852661523e-05,
|
|
"loss": 3.3692,
|
|
"step": 2269696
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2832497905151003e-05,
|
|
"loss": 3.3686,
|
|
"step": 2270208
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2824111957640483e-05,
|
|
"loss": 3.3771,
|
|
"step": 2270720
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2815742388933694e-05,
|
|
"loss": 3.3682,
|
|
"step": 2271232
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2807356441423174e-05,
|
|
"loss": 3.3762,
|
|
"step": 2271744
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2798970493912654e-05,
|
|
"loss": 3.3869,
|
|
"step": 2272256
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2790584546402135e-05,
|
|
"loss": 3.3813,
|
|
"step": 2272768
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2782214977695344e-05,
|
|
"loss": 3.3796,
|
|
"step": 2273280
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2773829030184824e-05,
|
|
"loss": 3.3723,
|
|
"step": 2273792
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2765443082674306e-05,
|
|
"loss": 3.372,
|
|
"step": 2274304
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2757057135163786e-05,
|
|
"loss": 3.3754,
|
|
"step": 2274816
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2748671187653266e-05,
|
|
"loss": 3.3582,
|
|
"step": 2275328
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2740301618946477e-05,
|
|
"loss": 3.3762,
|
|
"step": 2275840
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2731915671435957e-05,
|
|
"loss": 3.3683,
|
|
"step": 2276352
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2723529723925437e-05,
|
|
"loss": 3.3794,
|
|
"step": 2276864
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2715143776414917e-05,
|
|
"loss": 3.3778,
|
|
"step": 2277376
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2706757828904398e-05,
|
|
"loss": 3.3787,
|
|
"step": 2277888
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2698388260197607e-05,
|
|
"loss": 3.3677,
|
|
"step": 2278400
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2690002312687089e-05,
|
|
"loss": 3.371,
|
|
"step": 2278912
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2681616365176569e-05,
|
|
"loss": 3.3806,
|
|
"step": 2279424
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2673230417666045e-05,
|
|
"loss": 3.3754,
|
|
"step": 2279936
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2664844470155527e-05,
|
|
"loss": 3.3846,
|
|
"step": 2280448
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.265647490144874e-05,
|
|
"loss": 3.3708,
|
|
"step": 2280960
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.264808895393822e-05,
|
|
"loss": 3.3831,
|
|
"step": 2281472
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2639703006427698e-05,
|
|
"loss": 3.3762,
|
|
"step": 2281984
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2631317058917178e-05,
|
|
"loss": 3.3758,
|
|
"step": 2282496
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.262294749021039e-05,
|
|
"loss": 3.3783,
|
|
"step": 2283008
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.261456154269987e-05,
|
|
"loss": 3.3725,
|
|
"step": 2283520
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2606175595189348e-05,
|
|
"loss": 3.3748,
|
|
"step": 2284032
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2597789647678828e-05,
|
|
"loss": 3.3863,
|
|
"step": 2284544
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2589403700168308e-05,
|
|
"loss": 3.3699,
|
|
"step": 2285056
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.258103413146152e-05,
|
|
"loss": 3.3687,
|
|
"step": 2285568
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2572648183950999e-05,
|
|
"loss": 3.3815,
|
|
"step": 2286080
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.256426223644048e-05,
|
|
"loss": 3.3799,
|
|
"step": 2286592
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.255587628892996e-05,
|
|
"loss": 3.3713,
|
|
"step": 2287104
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.254749034141944e-05,
|
|
"loss": 3.3786,
|
|
"step": 2287616
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2539120772712651e-05,
|
|
"loss": 3.3691,
|
|
"step": 2288128
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2530734825202131e-05,
|
|
"loss": 3.3858,
|
|
"step": 2288640
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.2522348877691611e-05,
|
|
"loss": 3.3723,
|
|
"step": 2289152
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 3.8598251342773438,
|
|
"eval_runtime": 302.9631,
|
|
"eval_samples_per_second": 1259.53,
|
|
"eval_steps_per_second": 39.361,
|
|
"step": 2289600
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2513962930181091e-05,
|
|
"loss": 3.3633,
|
|
"step": 2289664
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2505593361474302e-05,
|
|
"loss": 3.3661,
|
|
"step": 2290176
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2497207413963782e-05,
|
|
"loss": 3.3698,
|
|
"step": 2290688
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2488821466453262e-05,
|
|
"loss": 3.3857,
|
|
"step": 2291200
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2480435518942742e-05,
|
|
"loss": 3.3774,
|
|
"step": 2291712
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2472049571432222e-05,
|
|
"loss": 3.3753,
|
|
"step": 2292224
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2463680002725433e-05,
|
|
"loss": 3.3731,
|
|
"step": 2292736
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2455294055214914e-05,
|
|
"loss": 3.3734,
|
|
"step": 2293248
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2446908107704393e-05,
|
|
"loss": 3.3612,
|
|
"step": 2293760
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2438522160193873e-05,
|
|
"loss": 3.3734,
|
|
"step": 2294272
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2430136212683353e-05,
|
|
"loss": 3.3708,
|
|
"step": 2294784
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2421766643976565e-05,
|
|
"loss": 3.3874,
|
|
"step": 2295296
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2413380696466043e-05,
|
|
"loss": 3.3807,
|
|
"step": 2295808
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2404994748955525e-05,
|
|
"loss": 3.3524,
|
|
"step": 2296320
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2396625180248736e-05,
|
|
"loss": 3.3693,
|
|
"step": 2296832
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2388239232738216e-05,
|
|
"loss": 3.3566,
|
|
"step": 2297344
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2379853285227696e-05,
|
|
"loss": 3.3716,
|
|
"step": 2297856
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2371467337717176e-05,
|
|
"loss": 3.3669,
|
|
"step": 2298368
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2363081390206656e-05,
|
|
"loss": 3.3734,
|
|
"step": 2298880
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2354711821499866e-05,
|
|
"loss": 3.3665,
|
|
"step": 2299392
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2346325873989346e-05,
|
|
"loss": 3.3868,
|
|
"step": 2299904
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2337939926478826e-05,
|
|
"loss": 3.3697,
|
|
"step": 2300416
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2329553978968306e-05,
|
|
"loss": 3.3734,
|
|
"step": 2300928
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2321168031457786e-05,
|
|
"loss": 3.365,
|
|
"step": 2301440
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2312798462750997e-05,
|
|
"loss": 3.3835,
|
|
"step": 2301952
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2304412515240479e-05,
|
|
"loss": 3.3698,
|
|
"step": 2302464
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2296026567729957e-05,
|
|
"loss": 3.3647,
|
|
"step": 2302976
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2287640620219437e-05,
|
|
"loss": 3.3728,
|
|
"step": 2303488
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.227927105151265e-05,
|
|
"loss": 3.3698,
|
|
"step": 2304000
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.227088510400213e-05,
|
|
"loss": 3.3605,
|
|
"step": 2304512
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2262499156491608e-05,
|
|
"loss": 3.3698,
|
|
"step": 2305024
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.225411320898109e-05,
|
|
"loss": 3.3677,
|
|
"step": 2305536
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.224572726147057e-05,
|
|
"loss": 3.3749,
|
|
"step": 2306048
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2237357692763778e-05,
|
|
"loss": 3.3748,
|
|
"step": 2306560
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.222897174525326e-05,
|
|
"loss": 3.3716,
|
|
"step": 2307072
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.222058579774274e-05,
|
|
"loss": 3.375,
|
|
"step": 2307584
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.221219985023222e-05,
|
|
"loss": 3.3769,
|
|
"step": 2308096
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2203813902721698e-05,
|
|
"loss": 3.3607,
|
|
"step": 2308608
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.219544433401491e-05,
|
|
"loss": 3.3671,
|
|
"step": 2309120
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.218705838650439e-05,
|
|
"loss": 3.3655,
|
|
"step": 2309632
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.217867243899387e-05,
|
|
"loss": 3.357,
|
|
"step": 2310144
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.217028649148335e-05,
|
|
"loss": 3.3686,
|
|
"step": 2310656
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2161916922776561e-05,
|
|
"loss": 3.3737,
|
|
"step": 2311168
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2153530975266041e-05,
|
|
"loss": 3.3642,
|
|
"step": 2311680
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2145145027755521e-05,
|
|
"loss": 3.3716,
|
|
"step": 2312192
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2136759080245001e-05,
|
|
"loss": 3.372,
|
|
"step": 2312704
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2128373132734481e-05,
|
|
"loss": 3.3752,
|
|
"step": 2313216
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2119987185223961e-05,
|
|
"loss": 3.3689,
|
|
"step": 2313728
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2111617616517172e-05,
|
|
"loss": 3.353,
|
|
"step": 2314240
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2103231669006652e-05,
|
|
"loss": 3.3724,
|
|
"step": 2314752
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2094845721496133e-05,
|
|
"loss": 3.3725,
|
|
"step": 2315264
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2086459773985612e-05,
|
|
"loss": 3.3738,
|
|
"step": 2315776
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2078090205278824e-05,
|
|
"loss": 3.366,
|
|
"step": 2316288
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2069704257768304e-05,
|
|
"loss": 3.3616,
|
|
"step": 2316800
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2061318310257784e-05,
|
|
"loss": 3.3533,
|
|
"step": 2317312
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2052932362747262e-05,
|
|
"loss": 3.3565,
|
|
"step": 2317824
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2044546415236744e-05,
|
|
"loss": 3.3711,
|
|
"step": 2318336
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2036176846529955e-05,
|
|
"loss": 3.3529,
|
|
"step": 2318848
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2027790899019433e-05,
|
|
"loss": 3.3793,
|
|
"step": 2319360
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2019404951508915e-05,
|
|
"loss": 3.3707,
|
|
"step": 2319872
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2011035382802125e-05,
|
|
"loss": 3.351,
|
|
"step": 2320384
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.2002649435291605e-05,
|
|
"loss": 3.3784,
|
|
"step": 2320896
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1994263487781085e-05,
|
|
"loss": 3.3578,
|
|
"step": 2321408
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1985877540270565e-05,
|
|
"loss": 3.3612,
|
|
"step": 2321920
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1977491592760045e-05,
|
|
"loss": 3.3586,
|
|
"step": 2322432
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1969122024053256e-05,
|
|
"loss": 3.3694,
|
|
"step": 2322944
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1960736076542736e-05,
|
|
"loss": 3.3554,
|
|
"step": 2323456
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1952350129032216e-05,
|
|
"loss": 3.3696,
|
|
"step": 2323968
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1943964181521698e-05,
|
|
"loss": 3.3473,
|
|
"step": 2324480
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1935578234011176e-05,
|
|
"loss": 3.3615,
|
|
"step": 2324992
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1927208665304387e-05,
|
|
"loss": 3.3724,
|
|
"step": 2325504
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1918822717793868e-05,
|
|
"loss": 3.365,
|
|
"step": 2326016
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1910436770283347e-05,
|
|
"loss": 3.3713,
|
|
"step": 2326528
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1902050822772827e-05,
|
|
"loss": 3.3642,
|
|
"step": 2327040
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1893681254066039e-05,
|
|
"loss": 3.3844,
|
|
"step": 2327552
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1885295306555519e-05,
|
|
"loss": 3.3614,
|
|
"step": 2328064
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1876909359044997e-05,
|
|
"loss": 3.3791,
|
|
"step": 2328576
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1868523411534479e-05,
|
|
"loss": 3.3583,
|
|
"step": 2329088
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.186015384282769e-05,
|
|
"loss": 3.364,
|
|
"step": 2329600
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.185176789531717e-05,
|
|
"loss": 3.3705,
|
|
"step": 2330112
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.184338194780665e-05,
|
|
"loss": 3.372,
|
|
"step": 2330624
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.183499600029613e-05,
|
|
"loss": 3.3676,
|
|
"step": 2331136
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.182662643158934e-05,
|
|
"loss": 3.3734,
|
|
"step": 2331648
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.181824048407882e-05,
|
|
"loss": 3.3543,
|
|
"step": 2332160
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.18098545365683e-05,
|
|
"loss": 3.3489,
|
|
"step": 2332672
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.180146858905778e-05,
|
|
"loss": 3.3613,
|
|
"step": 2333184
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.179308264154726e-05,
|
|
"loss": 3.3659,
|
|
"step": 2333696
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1784713072840471e-05,
|
|
"loss": 3.3629,
|
|
"step": 2334208
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1776327125329951e-05,
|
|
"loss": 3.3619,
|
|
"step": 2334720
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.1767941177819433e-05,
|
|
"loss": 3.3651,
|
|
"step": 2335232
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.175955523030891e-05,
|
|
"loss": 3.3593,
|
|
"step": 2335744
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.175116928279839e-05,
|
|
"loss": 3.3617,
|
|
"step": 2336256
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1742799714091603e-05,
|
|
"loss": 3.3611,
|
|
"step": 2336768
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1734413766581083e-05,
|
|
"loss": 3.3717,
|
|
"step": 2337280
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1726027819070562e-05,
|
|
"loss": 3.3779,
|
|
"step": 2337792
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1717641871560041e-05,
|
|
"loss": 3.375,
|
|
"step": 2338304
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1709272302853254e-05,
|
|
"loss": 3.3547,
|
|
"step": 2338816
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1700886355342734e-05,
|
|
"loss": 3.3639,
|
|
"step": 2339328
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1692500407832214e-05,
|
|
"loss": 3.3642,
|
|
"step": 2339840
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1684114460321694e-05,
|
|
"loss": 3.3705,
|
|
"step": 2340352
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1675728512811174e-05,
|
|
"loss": 3.3673,
|
|
"step": 2340864
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1667358944104385e-05,
|
|
"loss": 3.3618,
|
|
"step": 2341376
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1658972996593865e-05,
|
|
"loss": 3.3647,
|
|
"step": 2341888
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1650587049083344e-05,
|
|
"loss": 3.3664,
|
|
"step": 2342400
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1642201101572824e-05,
|
|
"loss": 3.3699,
|
|
"step": 2342912
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1633815154062304e-05,
|
|
"loss": 3.3625,
|
|
"step": 2343424
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1625445585355515e-05,
|
|
"loss": 3.3633,
|
|
"step": 2343936
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1617059637844995e-05,
|
|
"loss": 3.3541,
|
|
"step": 2344448
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1608673690334475e-05,
|
|
"loss": 3.3634,
|
|
"step": 2344960
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1600287742823955e-05,
|
|
"loss": 3.3539,
|
|
"step": 2345472
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1591918174117168e-05,
|
|
"loss": 3.3596,
|
|
"step": 2345984
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1583532226606647e-05,
|
|
"loss": 3.3572,
|
|
"step": 2346496
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1575146279096126e-05,
|
|
"loss": 3.3692,
|
|
"step": 2347008
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1566760331585606e-05,
|
|
"loss": 3.3599,
|
|
"step": 2347520
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1558390762878818e-05,
|
|
"loss": 3.3661,
|
|
"step": 2348032
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1550004815368298e-05,
|
|
"loss": 3.3759,
|
|
"step": 2348544
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1541618867857776e-05,
|
|
"loss": 3.376,
|
|
"step": 2349056
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1533232920347258e-05,
|
|
"loss": 3.3709,
|
|
"step": 2349568
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1524846972836738e-05,
|
|
"loss": 3.3621,
|
|
"step": 2350080
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1516477404129949e-05,
|
|
"loss": 3.3651,
|
|
"step": 2350592
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1508091456619429e-05,
|
|
"loss": 3.3631,
|
|
"step": 2351104
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1499705509108909e-05,
|
|
"loss": 3.3478,
|
|
"step": 2351616
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1491319561598389e-05,
|
|
"loss": 3.3682,
|
|
"step": 2352128
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.14829499928916e-05,
|
|
"loss": 3.359,
|
|
"step": 2352640
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.147456404538108e-05,
|
|
"loss": 3.3722,
|
|
"step": 2353152
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.146617809787056e-05,
|
|
"loss": 3.3636,
|
|
"step": 2353664
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.145779215036004e-05,
|
|
"loss": 3.3681,
|
|
"step": 2354176
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.144940620284952e-05,
|
|
"loss": 3.3596,
|
|
"step": 2354688
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.144103663414273e-05,
|
|
"loss": 3.3604,
|
|
"step": 2355200
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1432650686632212e-05,
|
|
"loss": 3.3699,
|
|
"step": 2355712
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.142426473912169e-05,
|
|
"loss": 3.3659,
|
|
"step": 2356224
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.141587879161117e-05,
|
|
"loss": 3.3781,
|
|
"step": 2356736
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1407509222904382e-05,
|
|
"loss": 3.3616,
|
|
"step": 2357248
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1399123275393862e-05,
|
|
"loss": 3.3731,
|
|
"step": 2357760
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.139073732788334e-05,
|
|
"loss": 3.3686,
|
|
"step": 2358272
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1382351380372822e-05,
|
|
"loss": 3.3667,
|
|
"step": 2358784
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1373965432862302e-05,
|
|
"loss": 3.363,
|
|
"step": 2359296
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1365595864155511e-05,
|
|
"loss": 3.3608,
|
|
"step": 2359808
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1357209916644993e-05,
|
|
"loss": 3.3701,
|
|
"step": 2360320
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1348823969134473e-05,
|
|
"loss": 3.3731,
|
|
"step": 2360832
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1340438021623951e-05,
|
|
"loss": 3.3636,
|
|
"step": 2361344
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1332068452917164e-05,
|
|
"loss": 3.3598,
|
|
"step": 2361856
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1323682505406644e-05,
|
|
"loss": 3.3676,
|
|
"step": 2362368
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1315296557896124e-05,
|
|
"loss": 3.3678,
|
|
"step": 2362880
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1306910610385604e-05,
|
|
"loss": 3.3621,
|
|
"step": 2363392
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1298541041678814e-05,
|
|
"loss": 3.368,
|
|
"step": 2363904
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1290155094168294e-05,
|
|
"loss": 3.359,
|
|
"step": 2364416
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1281769146657776e-05,
|
|
"loss": 3.3792,
|
|
"step": 2364928
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.1273383199147254e-05,
|
|
"loss": 3.3586,
|
|
"step": 2365440
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.8602099418640137,
|
|
"eval_runtime": 303.6541,
|
|
"eval_samples_per_second": 1256.664,
|
|
"eval_steps_per_second": 39.272,
|
|
"step": 2365920
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1265013630440465e-05,
|
|
"loss": 3.3624,
|
|
"step": 2365952
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1256627682929947e-05,
|
|
"loss": 3.3608,
|
|
"step": 2366464
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1248241735419425e-05,
|
|
"loss": 3.3582,
|
|
"step": 2366976
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1239855787908905e-05,
|
|
"loss": 3.3772,
|
|
"step": 2367488
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1231469840398385e-05,
|
|
"loss": 3.3678,
|
|
"step": 2368000
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1223100271691597e-05,
|
|
"loss": 3.3662,
|
|
"step": 2368512
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1214714324181076e-05,
|
|
"loss": 3.3605,
|
|
"step": 2369024
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1206328376670557e-05,
|
|
"loss": 3.3648,
|
|
"step": 2369536
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1197942429160037e-05,
|
|
"loss": 3.354,
|
|
"step": 2370048
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1189556481649515e-05,
|
|
"loss": 3.362,
|
|
"step": 2370560
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1181170534138995e-05,
|
|
"loss": 3.3622,
|
|
"step": 2371072
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1172800965432208e-05,
|
|
"loss": 3.3767,
|
|
"step": 2371584
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1164415017921688e-05,
|
|
"loss": 3.3703,
|
|
"step": 2372096
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1156029070411168e-05,
|
|
"loss": 3.3466,
|
|
"step": 2372608
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1147643122900648e-05,
|
|
"loss": 3.3598,
|
|
"step": 2373120
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1139273554193859e-05,
|
|
"loss": 3.3493,
|
|
"step": 2373632
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1130887606683338e-05,
|
|
"loss": 3.3606,
|
|
"step": 2374144
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1122501659172818e-05,
|
|
"loss": 3.3594,
|
|
"step": 2374656
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1114115711662298e-05,
|
|
"loss": 3.3636,
|
|
"step": 2375168
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1105746142955511e-05,
|
|
"loss": 3.3569,
|
|
"step": 2375680
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1097360195444989e-05,
|
|
"loss": 3.376,
|
|
"step": 2376192
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1088974247934469e-05,
|
|
"loss": 3.3644,
|
|
"step": 2376704
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1080588300423949e-05,
|
|
"loss": 3.357,
|
|
"step": 2377216
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.107220235291343e-05,
|
|
"loss": 3.3556,
|
|
"step": 2377728
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1063816405402909e-05,
|
|
"loss": 3.3738,
|
|
"step": 2378240
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.105544683669612e-05,
|
|
"loss": 3.361,
|
|
"step": 2378752
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1047060889185601e-05,
|
|
"loss": 3.3552,
|
|
"step": 2379264
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.103867494167508e-05,
|
|
"loss": 3.366,
|
|
"step": 2379776
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.103028899416456e-05,
|
|
"loss": 3.3579,
|
|
"step": 2380288
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.1021903046654041e-05,
|
|
"loss": 3.3535,
|
|
"step": 2380800
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.1013533477947252e-05,
|
|
"loss": 3.3599,
|
|
"step": 2381312
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.100514753043673e-05,
|
|
"loss": 3.3587,
|
|
"step": 2381824
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0996761582926212e-05,
|
|
"loss": 3.3654,
|
|
"step": 2382336
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0988375635415692e-05,
|
|
"loss": 3.3639,
|
|
"step": 2382848
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.097998968790517e-05,
|
|
"loss": 3.3668,
|
|
"step": 2383360
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0971620119198383e-05,
|
|
"loss": 3.3592,
|
|
"step": 2383872
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0963234171687863e-05,
|
|
"loss": 3.3681,
|
|
"step": 2384384
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0954848224177343e-05,
|
|
"loss": 3.3536,
|
|
"step": 2384896
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0946462276666823e-05,
|
|
"loss": 3.3601,
|
|
"step": 2385408
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0938092707960033e-05,
|
|
"loss": 3.3546,
|
|
"step": 2385920
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0929706760449513e-05,
|
|
"loss": 3.3514,
|
|
"step": 2386432
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0921320812938993e-05,
|
|
"loss": 3.3574,
|
|
"step": 2386944
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0912934865428473e-05,
|
|
"loss": 3.3645,
|
|
"step": 2387456
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0904548917917953e-05,
|
|
"loss": 3.3552,
|
|
"step": 2387968
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0896162970407433e-05,
|
|
"loss": 3.3614,
|
|
"step": 2388480
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0887777022896913e-05,
|
|
"loss": 3.3641,
|
|
"step": 2388992
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0879407454190124e-05,
|
|
"loss": 3.3664,
|
|
"step": 2389504
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0871021506679604e-05,
|
|
"loss": 3.3596,
|
|
"step": 2390016
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0862635559169085e-05,
|
|
"loss": 3.34,
|
|
"step": 2390528
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0854249611658564e-05,
|
|
"loss": 3.3608,
|
|
"step": 2391040
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0845880042951776e-05,
|
|
"loss": 3.366,
|
|
"step": 2391552
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0837494095441256e-05,
|
|
"loss": 3.3628,
|
|
"step": 2392064
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0829108147930734e-05,
|
|
"loss": 3.3527,
|
|
"step": 2392576
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0820722200420214e-05,
|
|
"loss": 3.3529,
|
|
"step": 2393088
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0812336252909696e-05,
|
|
"loss": 3.3447,
|
|
"step": 2393600
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0803966684202907e-05,
|
|
"loss": 3.3453,
|
|
"step": 2394112
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0795580736692385e-05,
|
|
"loss": 3.3642,
|
|
"step": 2394624
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0787194789181867e-05,
|
|
"loss": 3.3479,
|
|
"step": 2395136
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0778825220475078e-05,
|
|
"loss": 3.3676,
|
|
"step": 2395648
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0770439272964557e-05,
|
|
"loss": 3.3592,
|
|
"step": 2396160
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0762053325454037e-05,
|
|
"loss": 3.3498,
|
|
"step": 2396672
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0753667377943517e-05,
|
|
"loss": 3.3659,
|
|
"step": 2397184
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0745281430432997e-05,
|
|
"loss": 3.3457,
|
|
"step": 2397696
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0736911861726208e-05,
|
|
"loss": 3.3513,
|
|
"step": 2398208
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0728525914215688e-05,
|
|
"loss": 3.3504,
|
|
"step": 2398720
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0720139966705168e-05,
|
|
"loss": 3.364,
|
|
"step": 2399232
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0711754019194648e-05,
|
|
"loss": 3.3443,
|
|
"step": 2399744
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0703384450487859e-05,
|
|
"loss": 3.3572,
|
|
"step": 2400256
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0694998502977339e-05,
|
|
"loss": 3.3375,
|
|
"step": 2400768
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.068661255546682e-05,
|
|
"loss": 3.3502,
|
|
"step": 2401280
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0678226607956299e-05,
|
|
"loss": 3.3616,
|
|
"step": 2401792
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0669840660445779e-05,
|
|
"loss": 3.3589,
|
|
"step": 2402304
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0661471091738991e-05,
|
|
"loss": 3.3573,
|
|
"step": 2402816
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0653085144228471e-05,
|
|
"loss": 3.358,
|
|
"step": 2403328
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.064469919671795e-05,
|
|
"loss": 3.3713,
|
|
"step": 2403840
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0636313249207431e-05,
|
|
"loss": 3.3544,
|
|
"step": 2404352
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0627943680500642e-05,
|
|
"loss": 3.3685,
|
|
"step": 2404864
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.061955773299012e-05,
|
|
"loss": 3.3463,
|
|
"step": 2405376
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0611171785479602e-05,
|
|
"loss": 3.3548,
|
|
"step": 2405888
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0602785837969082e-05,
|
|
"loss": 3.3593,
|
|
"step": 2406400
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0594416269262292e-05,
|
|
"loss": 3.3634,
|
|
"step": 2406912
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0586030321751772e-05,
|
|
"loss": 3.3572,
|
|
"step": 2407424
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0577644374241252e-05,
|
|
"loss": 3.3678,
|
|
"step": 2407936
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0569258426730732e-05,
|
|
"loss": 3.3498,
|
|
"step": 2408448
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0560888858023945e-05,
|
|
"loss": 3.3395,
|
|
"step": 2408960
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0552502910513423e-05,
|
|
"loss": 3.3491,
|
|
"step": 2409472
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0544116963002903e-05,
|
|
"loss": 3.3546,
|
|
"step": 2409984
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0535731015492385e-05,
|
|
"loss": 3.3508,
|
|
"step": 2410496
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0527361446785594e-05,
|
|
"loss": 3.3545,
|
|
"step": 2411008
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.0518975499275074e-05,
|
|
"loss": 3.358,
|
|
"step": 2411520
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0510589551764555e-05,
|
|
"loss": 3.3481,
|
|
"step": 2412032
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0502203604254035e-05,
|
|
"loss": 3.348,
|
|
"step": 2412544
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0493834035547246e-05,
|
|
"loss": 3.3531,
|
|
"step": 2413056
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0485448088036726e-05,
|
|
"loss": 3.3638,
|
|
"step": 2413568
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0477062140526206e-05,
|
|
"loss": 3.3688,
|
|
"step": 2414080
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0468676193015684e-05,
|
|
"loss": 3.3628,
|
|
"step": 2414592
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0460290245505166e-05,
|
|
"loss": 3.3455,
|
|
"step": 2415104
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0451920676798377e-05,
|
|
"loss": 3.3573,
|
|
"step": 2415616
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0443534729287857e-05,
|
|
"loss": 3.3511,
|
|
"step": 2416128
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0435148781777337e-05,
|
|
"loss": 3.361,
|
|
"step": 2416640
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0426762834266817e-05,
|
|
"loss": 3.3555,
|
|
"step": 2417152
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0418393265560027e-05,
|
|
"loss": 3.3523,
|
|
"step": 2417664
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0410007318049509e-05,
|
|
"loss": 3.3532,
|
|
"step": 2418176
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0401621370538987e-05,
|
|
"loss": 3.3615,
|
|
"step": 2418688
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0393235423028467e-05,
|
|
"loss": 3.36,
|
|
"step": 2419200
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0384849475517947e-05,
|
|
"loss": 3.3535,
|
|
"step": 2419712
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0376479906811158e-05,
|
|
"loss": 3.3542,
|
|
"step": 2420224
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0368093959300638e-05,
|
|
"loss": 3.3457,
|
|
"step": 2420736
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.035970801179012e-05,
|
|
"loss": 3.3519,
|
|
"step": 2421248
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.03513220642796e-05,
|
|
"loss": 3.3451,
|
|
"step": 2421760
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0342952495572809e-05,
|
|
"loss": 3.3503,
|
|
"step": 2422272
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.033456654806229e-05,
|
|
"loss": 3.3462,
|
|
"step": 2422784
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.032618060055177e-05,
|
|
"loss": 3.3601,
|
|
"step": 2423296
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0317794653041248e-05,
|
|
"loss": 3.3532,
|
|
"step": 2423808
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0309425084334461e-05,
|
|
"loss": 3.3493,
|
|
"step": 2424320
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0301039136823941e-05,
|
|
"loss": 3.3672,
|
|
"step": 2424832
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0292653189313421e-05,
|
|
"loss": 3.3693,
|
|
"step": 2425344
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0284267241802901e-05,
|
|
"loss": 3.3609,
|
|
"step": 2425856
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.027588129429238e-05,
|
|
"loss": 3.3515,
|
|
"step": 2426368
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0267511725585592e-05,
|
|
"loss": 3.3552,
|
|
"step": 2426880
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0259125778075073e-05,
|
|
"loss": 3.3507,
|
|
"step": 2427392
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0250739830564552e-05,
|
|
"loss": 3.3433,
|
|
"step": 2427904
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0242353883054031e-05,
|
|
"loss": 3.3553,
|
|
"step": 2428416
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0233984314347244e-05,
|
|
"loss": 3.3506,
|
|
"step": 2428928
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0225598366836722e-05,
|
|
"loss": 3.3592,
|
|
"step": 2429440
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0217212419326202e-05,
|
|
"loss": 3.3582,
|
|
"step": 2429952
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0208826471815682e-05,
|
|
"loss": 3.3612,
|
|
"step": 2430464
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0200440524305164e-05,
|
|
"loss": 3.354,
|
|
"step": 2430976
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0192070955598373e-05,
|
|
"loss": 3.3513,
|
|
"step": 2431488
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0183685008087855e-05,
|
|
"loss": 3.3572,
|
|
"step": 2432000
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0175299060577334e-05,
|
|
"loss": 3.3594,
|
|
"step": 2432512
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0166913113066813e-05,
|
|
"loss": 3.3669,
|
|
"step": 2433024
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0158543544360025e-05,
|
|
"loss": 3.3552,
|
|
"step": 2433536
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0150157596849505e-05,
|
|
"loss": 3.3625,
|
|
"step": 2434048
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0141771649338985e-05,
|
|
"loss": 3.3584,
|
|
"step": 2434560
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0133385701828463e-05,
|
|
"loss": 3.3588,
|
|
"step": 2435072
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0125016133121676e-05,
|
|
"loss": 3.3552,
|
|
"step": 2435584
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0116630185611156e-05,
|
|
"loss": 3.3522,
|
|
"step": 2436096
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0108244238100636e-05,
|
|
"loss": 3.3616,
|
|
"step": 2436608
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0099858290590116e-05,
|
|
"loss": 3.3633,
|
|
"step": 2437120
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0091472343079596e-05,
|
|
"loss": 3.356,
|
|
"step": 2437632
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0083102774372808e-05,
|
|
"loss": 3.3522,
|
|
"step": 2438144
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0074716826862286e-05,
|
|
"loss": 3.3571,
|
|
"step": 2438656
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0066330879351766e-05,
|
|
"loss": 3.36,
|
|
"step": 2439168
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0057944931841246e-05,
|
|
"loss": 3.3519,
|
|
"step": 2439680
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0049575363134459e-05,
|
|
"loss": 3.3632,
|
|
"step": 2440192
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0041189415623937e-05,
|
|
"loss": 3.3481,
|
|
"step": 2440704
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0032803468113417e-05,
|
|
"loss": 3.3674,
|
|
"step": 2441216
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.0024417520602899e-05,
|
|
"loss": 3.3527,
|
|
"step": 2441728
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 1.0016047951896108e-05,
|
|
"loss": 3.3545,
|
|
"step": 2442240
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 3.8610332012176514,
|
|
"eval_runtime": 303.2873,
|
|
"eval_samples_per_second": 1258.183,
|
|
"eval_steps_per_second": 39.319,
|
|
"step": 2442240
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.000767838318932e-05,
|
|
"loss": 3.3475,
|
|
"step": 2442752
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.9992924356788e-06,
|
|
"loss": 3.3468,
|
|
"step": 2443264
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.99090648816828e-06,
|
|
"loss": 3.3666,
|
|
"step": 2443776
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.98252054065776e-06,
|
|
"loss": 3.3594,
|
|
"step": 2444288
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.97413459314724e-06,
|
|
"loss": 3.3574,
|
|
"step": 2444800
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.965765024440451e-06,
|
|
"loss": 3.3517,
|
|
"step": 2445312
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.957379076929933e-06,
|
|
"loss": 3.3536,
|
|
"step": 2445824
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.94899312941941e-06,
|
|
"loss": 3.3385,
|
|
"step": 2446336
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.94060718190889e-06,
|
|
"loss": 3.36,
|
|
"step": 2446848
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.932237613202103e-06,
|
|
"loss": 3.3478,
|
|
"step": 2447360
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.923851665691582e-06,
|
|
"loss": 3.3709,
|
|
"step": 2447872
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.915465718181061e-06,
|
|
"loss": 3.3621,
|
|
"step": 2448384
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.907079770670543e-06,
|
|
"loss": 3.336,
|
|
"step": 2448896
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.898693823160023e-06,
|
|
"loss": 3.3469,
|
|
"step": 2449408
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.890324254453232e-06,
|
|
"loss": 3.3456,
|
|
"step": 2449920
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.881938306942714e-06,
|
|
"loss": 3.3463,
|
|
"step": 2450432
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.873552359432194e-06,
|
|
"loss": 3.3522,
|
|
"step": 2450944
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.865166411921672e-06,
|
|
"loss": 3.3548,
|
|
"step": 2451456
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.856796843214885e-06,
|
|
"loss": 3.345,
|
|
"step": 2451968
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.848410895704364e-06,
|
|
"loss": 3.3683,
|
|
"step": 2452480
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.840024948193844e-06,
|
|
"loss": 3.3506,
|
|
"step": 2452992
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.831639000683324e-06,
|
|
"loss": 3.3512,
|
|
"step": 2453504
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.823253053172804e-06,
|
|
"loss": 3.3474,
|
|
"step": 2454016
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.814883484466015e-06,
|
|
"loss": 3.3647,
|
|
"step": 2454528
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.806497536955497e-06,
|
|
"loss": 3.3501,
|
|
"step": 2455040
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.798111589444975e-06,
|
|
"loss": 3.3453,
|
|
"step": 2455552
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.789725641934455e-06,
|
|
"loss": 3.3544,
|
|
"step": 2456064
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.781356073227667e-06,
|
|
"loss": 3.3481,
|
|
"step": 2456576
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 9.772970125717146e-06,
|
|
"loss": 3.3441,
|
|
"step": 2457088
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.764584178206626e-06,
|
|
"loss": 3.3474,
|
|
"step": 2457600
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.756198230696106e-06,
|
|
"loss": 3.3507,
|
|
"step": 2458112
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.747812283185587e-06,
|
|
"loss": 3.3551,
|
|
"step": 2458624
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.739442714478796e-06,
|
|
"loss": 3.3541,
|
|
"step": 2459136
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.731056766968278e-06,
|
|
"loss": 3.3568,
|
|
"step": 2459648
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.722670819457758e-06,
|
|
"loss": 3.3516,
|
|
"step": 2460160
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.714284871947236e-06,
|
|
"loss": 3.3585,
|
|
"step": 2460672
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.705898924436716e-06,
|
|
"loss": 3.3439,
|
|
"step": 2461184
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.697529355729929e-06,
|
|
"loss": 3.3501,
|
|
"step": 2461696
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.689143408219409e-06,
|
|
"loss": 3.3435,
|
|
"step": 2462208
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.680757460708887e-06,
|
|
"loss": 3.3465,
|
|
"step": 2462720
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.672371513198369e-06,
|
|
"loss": 3.3437,
|
|
"step": 2463232
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.66400194449158e-06,
|
|
"loss": 3.3555,
|
|
"step": 2463744
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.65561599698106e-06,
|
|
"loss": 3.3475,
|
|
"step": 2464256
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.64723004947054e-06,
|
|
"loss": 3.353,
|
|
"step": 2464768
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.63884410196002e-06,
|
|
"loss": 3.3546,
|
|
"step": 2465280
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.6304581544495e-06,
|
|
"loss": 3.3591,
|
|
"step": 2465792
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.622072206938979e-06,
|
|
"loss": 3.3494,
|
|
"step": 2466304
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.61370263823219e-06,
|
|
"loss": 3.3344,
|
|
"step": 2466816
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.60531669072167e-06,
|
|
"loss": 3.3438,
|
|
"step": 2467328
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.596930743211152e-06,
|
|
"loss": 3.3616,
|
|
"step": 2467840
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.58854479570063e-06,
|
|
"loss": 3.3526,
|
|
"step": 2468352
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.58017522699384e-06,
|
|
"loss": 3.3425,
|
|
"step": 2468864
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.571789279483322e-06,
|
|
"loss": 3.3397,
|
|
"step": 2469376
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.5634033319728e-06,
|
|
"loss": 3.3411,
|
|
"step": 2469888
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.55501738446228e-06,
|
|
"loss": 3.3305,
|
|
"step": 2470400
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.54663143695176e-06,
|
|
"loss": 3.3591,
|
|
"step": 2470912
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.538245489441242e-06,
|
|
"loss": 3.3391,
|
|
"step": 2471424
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.529875920734451e-06,
|
|
"loss": 3.3513,
|
|
"step": 2471936
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.521489973223933e-06,
|
|
"loss": 3.3509,
|
|
"step": 2472448
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.513104025713413e-06,
|
|
"loss": 3.3384,
|
|
"step": 2472960
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.504718078202891e-06,
|
|
"loss": 3.3555,
|
|
"step": 2473472
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.496348509496104e-06,
|
|
"loss": 3.3398,
|
|
"step": 2473984
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.487962561985583e-06,
|
|
"loss": 3.3478,
|
|
"step": 2474496
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.479576614475063e-06,
|
|
"loss": 3.3365,
|
|
"step": 2475008
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.471190666964543e-06,
|
|
"loss": 3.3549,
|
|
"step": 2475520
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.462804719454023e-06,
|
|
"loss": 3.34,
|
|
"step": 2476032
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.454435150747234e-06,
|
|
"loss": 3.3413,
|
|
"step": 2476544
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.446049203236714e-06,
|
|
"loss": 3.3341,
|
|
"step": 2477056
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.437663255726194e-06,
|
|
"loss": 3.3386,
|
|
"step": 2477568
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.429277308215674e-06,
|
|
"loss": 3.3505,
|
|
"step": 2478080
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.420907739508886e-06,
|
|
"loss": 3.3539,
|
|
"step": 2478592
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.412521791998365e-06,
|
|
"loss": 3.3477,
|
|
"step": 2479104
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.404135844487845e-06,
|
|
"loss": 3.346,
|
|
"step": 2479616
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.395749896977325e-06,
|
|
"loss": 3.361,
|
|
"step": 2480128
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.387380328270537e-06,
|
|
"loss": 3.3452,
|
|
"step": 2480640
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.378994380760015e-06,
|
|
"loss": 3.3629,
|
|
"step": 2481152
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.370608433249495e-06,
|
|
"loss": 3.3397,
|
|
"step": 2481664
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.362222485738977e-06,
|
|
"loss": 3.3451,
|
|
"step": 2482176
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.353852917032186e-06,
|
|
"loss": 3.3498,
|
|
"step": 2482688
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.345466969521668e-06,
|
|
"loss": 3.3527,
|
|
"step": 2483200
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.337081022011148e-06,
|
|
"loss": 3.3469,
|
|
"step": 2483712
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.328695074500628e-06,
|
|
"loss": 3.3566,
|
|
"step": 2484224
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.320309126990106e-06,
|
|
"loss": 3.3452,
|
|
"step": 2484736
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.311939558283318e-06,
|
|
"loss": 3.3259,
|
|
"step": 2485248
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.303553610772798e-06,
|
|
"loss": 3.3403,
|
|
"step": 2485760
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.295167663262278e-06,
|
|
"loss": 3.3455,
|
|
"step": 2486272
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.286781715751758e-06,
|
|
"loss": 3.3413,
|
|
"step": 2486784
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.278412147044969e-06,
|
|
"loss": 3.3449,
|
|
"step": 2487296
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.270026199534449e-06,
|
|
"loss": 3.3493,
|
|
"step": 2487808
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.261640252023929e-06,
|
|
"loss": 3.3391,
|
|
"step": 2488320
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.253254304513409e-06,
|
|
"loss": 3.3416,
|
|
"step": 2488832
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.244884735806621e-06,
|
|
"loss": 3.3426,
|
|
"step": 2489344
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.236498788296101e-06,
|
|
"loss": 3.3525,
|
|
"step": 2489856
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.22811284078558e-06,
|
|
"loss": 3.3579,
|
|
"step": 2490368
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.21972689327506e-06,
|
|
"loss": 3.3585,
|
|
"step": 2490880
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.211340945764541e-06,
|
|
"loss": 3.3332,
|
|
"step": 2491392
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.20295499825402e-06,
|
|
"loss": 3.3476,
|
|
"step": 2491904
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.1945690507435e-06,
|
|
"loss": 3.3461,
|
|
"step": 2492416
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.186199482036712e-06,
|
|
"loss": 3.3501,
|
|
"step": 2492928
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.177813534526192e-06,
|
|
"loss": 3.3505,
|
|
"step": 2493440
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.16942758701567e-06,
|
|
"loss": 3.3421,
|
|
"step": 2493952
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.161041639505152e-06,
|
|
"loss": 3.341,
|
|
"step": 2494464
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.152655691994632e-06,
|
|
"loss": 3.3529,
|
|
"step": 2494976
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.144286123287841e-06,
|
|
"loss": 3.3512,
|
|
"step": 2495488
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.135900175777323e-06,
|
|
"loss": 3.3414,
|
|
"step": 2496000
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.127514228266802e-06,
|
|
"loss": 3.3472,
|
|
"step": 2496512
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.119128280756282e-06,
|
|
"loss": 3.3381,
|
|
"step": 2497024
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.110758712049493e-06,
|
|
"loss": 3.3401,
|
|
"step": 2497536
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.102372764538973e-06,
|
|
"loss": 3.3374,
|
|
"step": 2498048
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.093986817028453e-06,
|
|
"loss": 3.3426,
|
|
"step": 2498560
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.085600869517933e-06,
|
|
"loss": 3.3349,
|
|
"step": 2499072
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.077231300811144e-06,
|
|
"loss": 3.3495,
|
|
"step": 2499584
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.068845353300624e-06,
|
|
"loss": 3.3434,
|
|
"step": 2500096
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.060459405790104e-06,
|
|
"loss": 3.3427,
|
|
"step": 2500608
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.052073458279584e-06,
|
|
"loss": 3.359,
|
|
"step": 2501120
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.043687510769064e-06,
|
|
"loss": 3.3605,
|
|
"step": 2501632
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.035317942062276e-06,
|
|
"loss": 3.3474,
|
|
"step": 2502144
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.026931994551756e-06,
|
|
"loss": 3.3491,
|
|
"step": 2502656
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.018546047041234e-06,
|
|
"loss": 3.3435,
|
|
"step": 2503168
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.010160099530714e-06,
|
|
"loss": 3.341,
|
|
"step": 2503680
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 9.001774152020196e-06,
|
|
"loss": 3.336,
|
|
"step": 2504192
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.993404583313405e-06,
|
|
"loss": 3.3417,
|
|
"step": 2504704
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.985018635802887e-06,
|
|
"loss": 3.3432,
|
|
"step": 2505216
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.976632688292367e-06,
|
|
"loss": 3.3507,
|
|
"step": 2505728
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.968246740781847e-06,
|
|
"loss": 3.3455,
|
|
"step": 2506240
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.959860793271325e-06,
|
|
"loss": 3.3534,
|
|
"step": 2506752
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.951474845760807e-06,
|
|
"loss": 3.3443,
|
|
"step": 2507264
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.943105277054017e-06,
|
|
"loss": 3.3449,
|
|
"step": 2507776
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.934719329543496e-06,
|
|
"loss": 3.3408,
|
|
"step": 2508288
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.926333382032977e-06,
|
|
"loss": 3.3548,
|
|
"step": 2508800
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.917947434522457e-06,
|
|
"loss": 3.3563,
|
|
"step": 2509312
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.909577865815668e-06,
|
|
"loss": 3.3433,
|
|
"step": 2509824
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.901191918305148e-06,
|
|
"loss": 3.3487,
|
|
"step": 2510336
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.892805970794628e-06,
|
|
"loss": 3.3512,
|
|
"step": 2510848
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.884420023284108e-06,
|
|
"loss": 3.3493,
|
|
"step": 2511360
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.876034075773588e-06,
|
|
"loss": 3.3488,
|
|
"step": 2511872
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.867664507066799e-06,
|
|
"loss": 3.3424,
|
|
"step": 2512384
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.859278559556279e-06,
|
|
"loss": 3.3528,
|
|
"step": 2512896
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.85089261204576e-06,
|
|
"loss": 3.3488,
|
|
"step": 2513408
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.842506664535239e-06,
|
|
"loss": 3.3484,
|
|
"step": 2513920
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.83413709582845e-06,
|
|
"loss": 3.3451,
|
|
"step": 2514432
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.825751148317931e-06,
|
|
"loss": 3.3455,
|
|
"step": 2514944
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.81736520080741e-06,
|
|
"loss": 3.3524,
|
|
"step": 2515456
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.80897925329689e-06,
|
|
"loss": 3.3409,
|
|
"step": 2515968
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.800609684590102e-06,
|
|
"loss": 3.3547,
|
|
"step": 2516480
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.792223737079582e-06,
|
|
"loss": 3.3424,
|
|
"step": 2516992
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.78383778956906e-06,
|
|
"loss": 3.3539,
|
|
"step": 2517504
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.775451842058542e-06,
|
|
"loss": 3.3479,
|
|
"step": 2518016
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 8.767082273351752e-06,
|
|
"loss": 3.3408,
|
|
"step": 2518528
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.861509323120117,
|
|
"eval_runtime": 304.7971,
|
|
"eval_samples_per_second": 1251.951,
|
|
"eval_steps_per_second": 39.124,
|
|
"step": 2518560
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.758696325841232e-06,
|
|
"loss": 3.3419,
|
|
"step": 2519040
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.750310378330712e-06,
|
|
"loss": 3.3388,
|
|
"step": 2519552
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.741924430820192e-06,
|
|
"loss": 3.3589,
|
|
"step": 2520064
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.733538483309672e-06,
|
|
"loss": 3.3418,
|
|
"step": 2520576
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.72515253579915e-06,
|
|
"loss": 3.353,
|
|
"step": 2521088
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.716782967092363e-06,
|
|
"loss": 3.3447,
|
|
"step": 2521600
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.708397019581843e-06,
|
|
"loss": 3.3408,
|
|
"step": 2522112
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.700011072071323e-06,
|
|
"loss": 3.3335,
|
|
"step": 2522624
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.691625124560803e-06,
|
|
"loss": 3.3472,
|
|
"step": 2523136
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.683239177050283e-06,
|
|
"loss": 3.3368,
|
|
"step": 2523648
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.674869608343495e-06,
|
|
"loss": 3.3618,
|
|
"step": 2524160
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.666483660832973e-06,
|
|
"loss": 3.3506,
|
|
"step": 2524672
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.658097713322453e-06,
|
|
"loss": 3.3323,
|
|
"step": 2525184
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.649711765811933e-06,
|
|
"loss": 3.334,
|
|
"step": 2525696
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.641342197105146e-06,
|
|
"loss": 3.3391,
|
|
"step": 2526208
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.632956249594624e-06,
|
|
"loss": 3.3302,
|
|
"step": 2526720
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.624570302084104e-06,
|
|
"loss": 3.3447,
|
|
"step": 2527232
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.616184354573586e-06,
|
|
"loss": 3.3452,
|
|
"step": 2527744
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.607814785866796e-06,
|
|
"loss": 3.3347,
|
|
"step": 2528256
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.599428838356276e-06,
|
|
"loss": 3.3618,
|
|
"step": 2528768
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.591042890845756e-06,
|
|
"loss": 3.3433,
|
|
"step": 2529280
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.582656943335236e-06,
|
|
"loss": 3.3433,
|
|
"step": 2529792
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.574287374628447e-06,
|
|
"loss": 3.338,
|
|
"step": 2530304
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.565901427117927e-06,
|
|
"loss": 3.3556,
|
|
"step": 2530816
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.557515479607407e-06,
|
|
"loss": 3.3355,
|
|
"step": 2531328
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.549129532096887e-06,
|
|
"loss": 3.3421,
|
|
"step": 2531840
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.540743584586367e-06,
|
|
"loss": 3.3407,
|
|
"step": 2532352
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.532374015879578e-06,
|
|
"loss": 3.3416,
|
|
"step": 2532864
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 8.523988068369058e-06,
|
|
"loss": 3.3335,
|
|
"step": 2533376
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.515602120858538e-06,
|
|
"loss": 3.3412,
|
|
"step": 2533888
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.507216173348018e-06,
|
|
"loss": 3.3442,
|
|
"step": 2534400
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.498830225837498e-06,
|
|
"loss": 3.344,
|
|
"step": 2534912
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.49046065713071e-06,
|
|
"loss": 3.3452,
|
|
"step": 2535424
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.482074709620188e-06,
|
|
"loss": 3.3488,
|
|
"step": 2535936
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.473688762109668e-06,
|
|
"loss": 3.3399,
|
|
"step": 2536448
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.46530281459915e-06,
|
|
"loss": 3.3477,
|
|
"step": 2536960
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.456916867088628e-06,
|
|
"loss": 3.3356,
|
|
"step": 2537472
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.448547298381839e-06,
|
|
"loss": 3.3421,
|
|
"step": 2537984
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.44016135087132e-06,
|
|
"loss": 3.33,
|
|
"step": 2538496
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.4317754033608e-06,
|
|
"loss": 3.3368,
|
|
"step": 2539008
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.423389455850279e-06,
|
|
"loss": 3.3347,
|
|
"step": 2539520
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.415019887143491e-06,
|
|
"loss": 3.3496,
|
|
"step": 2540032
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.406633939632971e-06,
|
|
"loss": 3.3309,
|
|
"step": 2540544
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.398247992122451e-06,
|
|
"loss": 3.3472,
|
|
"step": 2541056
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.389862044611931e-06,
|
|
"loss": 3.3448,
|
|
"step": 2541568
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.381476097101411e-06,
|
|
"loss": 3.3482,
|
|
"step": 2542080
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.373090149590891e-06,
|
|
"loss": 3.341,
|
|
"step": 2542592
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.36470420208037e-06,
|
|
"loss": 3.3285,
|
|
"step": 2543104
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.356334633373582e-06,
|
|
"loss": 3.3399,
|
|
"step": 2543616
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.347948685863062e-06,
|
|
"loss": 3.3476,
|
|
"step": 2544128
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.339562738352542e-06,
|
|
"loss": 3.3418,
|
|
"step": 2544640
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.331176790842022e-06,
|
|
"loss": 3.3354,
|
|
"step": 2545152
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.322790843331502e-06,
|
|
"loss": 3.3309,
|
|
"step": 2545664
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.314421274624712e-06,
|
|
"loss": 3.3349,
|
|
"step": 2546176
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.306035327114192e-06,
|
|
"loss": 3.3181,
|
|
"step": 2546688
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.297649379603672e-06,
|
|
"loss": 3.3506,
|
|
"step": 2547200
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.289263432093152e-06,
|
|
"loss": 3.328,
|
|
"step": 2547712
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.280893863386365e-06,
|
|
"loss": 3.3457,
|
|
"step": 2548224
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.272507915875843e-06,
|
|
"loss": 3.3399,
|
|
"step": 2548736
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.264121968365323e-06,
|
|
"loss": 3.3304,
|
|
"step": 2549248
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.255736020854805e-06,
|
|
"loss": 3.3441,
|
|
"step": 2549760
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.247366452148015e-06,
|
|
"loss": 3.3343,
|
|
"step": 2550272
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.238980504637495e-06,
|
|
"loss": 3.3367,
|
|
"step": 2550784
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.230594557126975e-06,
|
|
"loss": 3.3273,
|
|
"step": 2551296
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.222208609616455e-06,
|
|
"loss": 3.3497,
|
|
"step": 2551808
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.213822662105934e-06,
|
|
"loss": 3.3297,
|
|
"step": 2552320
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.205453093399146e-06,
|
|
"loss": 3.3337,
|
|
"step": 2552832
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.197067145888626e-06,
|
|
"loss": 3.3207,
|
|
"step": 2553344
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.188681198378104e-06,
|
|
"loss": 3.3305,
|
|
"step": 2553856
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.180295250867586e-06,
|
|
"loss": 3.3395,
|
|
"step": 2554368
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.171909303357066e-06,
|
|
"loss": 3.343,
|
|
"step": 2554880
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.163539734650277e-06,
|
|
"loss": 3.3388,
|
|
"step": 2555392
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.155153787139757e-06,
|
|
"loss": 3.3401,
|
|
"step": 2555904
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.146767839629237e-06,
|
|
"loss": 3.3484,
|
|
"step": 2556416
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.138381892118717e-06,
|
|
"loss": 3.3391,
|
|
"step": 2556928
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.130012323411929e-06,
|
|
"loss": 3.3478,
|
|
"step": 2557440
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.121626375901407e-06,
|
|
"loss": 3.3352,
|
|
"step": 2557952
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.113240428390887e-06,
|
|
"loss": 3.3368,
|
|
"step": 2558464
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.104854480880367e-06,
|
|
"loss": 3.3378,
|
|
"step": 2558976
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.096484912173578e-06,
|
|
"loss": 3.343,
|
|
"step": 2559488
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.088098964663058e-06,
|
|
"loss": 3.3407,
|
|
"step": 2560000
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.07971301715254e-06,
|
|
"loss": 3.3467,
|
|
"step": 2560512
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.07132706964202e-06,
|
|
"loss": 3.3396,
|
|
"step": 2561024
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.06295750093523e-06,
|
|
"loss": 3.3186,
|
|
"step": 2561536
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.05457155342471e-06,
|
|
"loss": 3.3298,
|
|
"step": 2562048
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.04618560591419e-06,
|
|
"loss": 3.3363,
|
|
"step": 2562560
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.037799658403669e-06,
|
|
"loss": 3.3338,
|
|
"step": 2563072
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.029430089696881e-06,
|
|
"loss": 3.3345,
|
|
"step": 2563584
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 8.021044142186361e-06,
|
|
"loss": 3.3425,
|
|
"step": 2564096
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 8.012658194675841e-06,
|
|
"loss": 3.3296,
|
|
"step": 2564608
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 8.004272247165321e-06,
|
|
"loss": 3.3287,
|
|
"step": 2565120
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.995902678458532e-06,
|
|
"loss": 3.3347,
|
|
"step": 2565632
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.987516730948012e-06,
|
|
"loss": 3.3435,
|
|
"step": 2566144
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.979130783437493e-06,
|
|
"loss": 3.3469,
|
|
"step": 2566656
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.970744835926972e-06,
|
|
"loss": 3.35,
|
|
"step": 2567168
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.962358888416452e-06,
|
|
"loss": 3.3248,
|
|
"step": 2567680
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.953972940905931e-06,
|
|
"loss": 3.3376,
|
|
"step": 2568192
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.945586993395411e-06,
|
|
"loss": 3.3404,
|
|
"step": 2568704
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.937217424688622e-06,
|
|
"loss": 3.3361,
|
|
"step": 2569216
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.928831477178104e-06,
|
|
"loss": 3.3427,
|
|
"step": 2569728
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.920445529667584e-06,
|
|
"loss": 3.3321,
|
|
"step": 2570240
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.912059582157062e-06,
|
|
"loss": 3.3334,
|
|
"step": 2570752
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.903690013450275e-06,
|
|
"loss": 3.3453,
|
|
"step": 2571264
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.895304065939755e-06,
|
|
"loss": 3.3416,
|
|
"step": 2571776
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.886918118429233e-06,
|
|
"loss": 3.3279,
|
|
"step": 2572288
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.878532170918713e-06,
|
|
"loss": 3.3403,
|
|
"step": 2572800
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.870162602211925e-06,
|
|
"loss": 3.3322,
|
|
"step": 2573312
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.861776654701405e-06,
|
|
"loss": 3.331,
|
|
"step": 2573824
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.853390707190885e-06,
|
|
"loss": 3.327,
|
|
"step": 2574336
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.845004759680365e-06,
|
|
"loss": 3.3323,
|
|
"step": 2574848
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.836618812169845e-06,
|
|
"loss": 3.3271,
|
|
"step": 2575360
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.828249243463056e-06,
|
|
"loss": 3.3419,
|
|
"step": 2575872
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.819863295952536e-06,
|
|
"loss": 3.3324,
|
|
"step": 2576384
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.811477348442016e-06,
|
|
"loss": 3.3341,
|
|
"step": 2576896
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.803091400931496e-06,
|
|
"loss": 3.3503,
|
|
"step": 2577408
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.794705453420976e-06,
|
|
"loss": 3.3479,
|
|
"step": 2577920
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.786319505910456e-06,
|
|
"loss": 3.3378,
|
|
"step": 2578432
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.777949937203666e-06,
|
|
"loss": 3.3412,
|
|
"step": 2578944
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.769563989693148e-06,
|
|
"loss": 3.3345,
|
|
"step": 2579456
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.761178042182626e-06,
|
|
"loss": 3.3298,
|
|
"step": 2579968
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.752792094672106e-06,
|
|
"loss": 3.3304,
|
|
"step": 2580480
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.744422525965319e-06,
|
|
"loss": 3.3286,
|
|
"step": 2580992
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.736036578454797e-06,
|
|
"loss": 3.3349,
|
|
"step": 2581504
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.727650630944277e-06,
|
|
"loss": 3.3415,
|
|
"step": 2582016
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.719264683433759e-06,
|
|
"loss": 3.339,
|
|
"step": 2582528
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.710878735923239e-06,
|
|
"loss": 3.3391,
|
|
"step": 2583040
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.702492788412717e-06,
|
|
"loss": 3.3352,
|
|
"step": 2583552
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.69412321970593e-06,
|
|
"loss": 3.3311,
|
|
"step": 2584064
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.68573727219541e-06,
|
|
"loss": 3.3332,
|
|
"step": 2584576
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.677351324684888e-06,
|
|
"loss": 3.3463,
|
|
"step": 2585088
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.668965377174368e-06,
|
|
"loss": 3.3465,
|
|
"step": 2585600
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.66059580846758e-06,
|
|
"loss": 3.3396,
|
|
"step": 2586112
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.65220986095706e-06,
|
|
"loss": 3.3357,
|
|
"step": 2586624
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.64382391344654e-06,
|
|
"loss": 3.3418,
|
|
"step": 2587136
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.63543796593602e-06,
|
|
"loss": 3.3386,
|
|
"step": 2587648
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.6270683972292315e-06,
|
|
"loss": 3.3436,
|
|
"step": 2588160
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.6186824497187114e-06,
|
|
"loss": 3.3373,
|
|
"step": 2588672
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.6102965022081906e-06,
|
|
"loss": 3.3411,
|
|
"step": 2589184
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.6019105546976705e-06,
|
|
"loss": 3.3388,
|
|
"step": 2589696
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.593524607187151e-06,
|
|
"loss": 3.3414,
|
|
"step": 2590208
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.585155038480361e-06,
|
|
"loss": 3.3331,
|
|
"step": 2590720
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.576769090969841e-06,
|
|
"loss": 3.3382,
|
|
"step": 2591232
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.568383143459322e-06,
|
|
"loss": 3.3455,
|
|
"step": 2591744
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.559997195948801e-06,
|
|
"loss": 3.3284,
|
|
"step": 2592256
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.551627627242013e-06,
|
|
"loss": 3.3475,
|
|
"step": 2592768
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.543241679731493e-06,
|
|
"loss": 3.3363,
|
|
"step": 2593280
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.5348557322209735e-06,
|
|
"loss": 3.3396,
|
|
"step": 2593792
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.526469784710452e-06,
|
|
"loss": 3.3378,
|
|
"step": 2594304
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 7.518083837199933e-06,
|
|
"loss": 3.3335,
|
|
"step": 2594816
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 3.862386465072632,
|
|
"eval_runtime": 304.6216,
|
|
"eval_samples_per_second": 1252.672,
|
|
"eval_steps_per_second": 39.147,
|
|
"step": 2594880
|
|
}
|
|
],
|
|
"logging_steps": 512,
|
|
"max_steps": 3052726,
|
|
"num_train_epochs": 9223372036854775807,
|
|
"save_steps": 10,
|
|
"total_flos": 1.8059709543885435e+18,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|