2732 lines
55 KiB
JSON
2732 lines
55 KiB
JSON
{
|
|
"best_metric": 3.9520537853240967,
|
|
"best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/binding-c-command/transformer/0/checkpoints/checkpoint-228960",
|
|
"epoch": 0.025000606015738065,
|
|
"eval_steps": 10,
|
|
"global_step": 228960,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.999998362119627e-05,
|
|
"loss": 10.9624,
|
|
"step": 1
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.999161405248948e-05,
|
|
"loss": 6.822,
|
|
"step": 512
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.998322810497896e-05,
|
|
"loss": 6.1893,
|
|
"step": 1024
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.997484215746844e-05,
|
|
"loss": 5.9774,
|
|
"step": 1536
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.996645620995792e-05,
|
|
"loss": 5.8082,
|
|
"step": 2048
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.99580702624474e-05,
|
|
"loss": 5.6938,
|
|
"step": 2560
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.994968431493688e-05,
|
|
"loss": 5.5921,
|
|
"step": 3072
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.994129836742636e-05,
|
|
"loss": 5.5277,
|
|
"step": 3584
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.993291241991584e-05,
|
|
"loss": 5.4338,
|
|
"step": 4096
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.992452647240532e-05,
|
|
"loss": 5.4008,
|
|
"step": 4608
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.99161405248948e-05,
|
|
"loss": 5.3401,
|
|
"step": 5120
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.990775457738428e-05,
|
|
"loss": 5.3007,
|
|
"step": 5632
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.989938500867749e-05,
|
|
"loss": 5.2593,
|
|
"step": 6144
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.989099906116697e-05,
|
|
"loss": 5.1918,
|
|
"step": 6656
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.988261311365645e-05,
|
|
"loss": 5.1743,
|
|
"step": 7168
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.987422716614593e-05,
|
|
"loss": 5.1266,
|
|
"step": 7680
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.986584121863541e-05,
|
|
"loss": 5.1042,
|
|
"step": 8192
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.985745527112489e-05,
|
|
"loss": 5.0762,
|
|
"step": 8704
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.984906932361437e-05,
|
|
"loss": 5.0383,
|
|
"step": 9216
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.984068337610385e-05,
|
|
"loss": 5.0149,
|
|
"step": 9728
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.983229742859333e-05,
|
|
"loss": 5.0018,
|
|
"step": 10240
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9823927859886547e-05,
|
|
"loss": 4.972,
|
|
"step": 10752
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9815541912376026e-05,
|
|
"loss": 4.9446,
|
|
"step": 11264
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9807155964865506e-05,
|
|
"loss": 4.9231,
|
|
"step": 11776
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9798770017354986e-05,
|
|
"loss": 4.9142,
|
|
"step": 12288
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9790400448648195e-05,
|
|
"loss": 4.8885,
|
|
"step": 12800
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9782014501137675e-05,
|
|
"loss": 4.8615,
|
|
"step": 13312
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9773628553627155e-05,
|
|
"loss": 4.8521,
|
|
"step": 13824
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9765242606116635e-05,
|
|
"loss": 4.8317,
|
|
"step": 14336
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.9756873037409844e-05,
|
|
"loss": 4.8177,
|
|
"step": 14848
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9748487089899324e-05,
|
|
"loss": 4.8048,
|
|
"step": 15360
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9740101142388804e-05,
|
|
"loss": 4.7885,
|
|
"step": 15872
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9731715194878284e-05,
|
|
"loss": 4.7788,
|
|
"step": 16384
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.97233456261715e-05,
|
|
"loss": 4.7606,
|
|
"step": 16896
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.971495967866098e-05,
|
|
"loss": 4.7554,
|
|
"step": 17408
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.970657373115046e-05,
|
|
"loss": 4.756,
|
|
"step": 17920
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.969818778363994e-05,
|
|
"loss": 4.7316,
|
|
"step": 18432
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.968981821493315e-05,
|
|
"loss": 4.6953,
|
|
"step": 18944
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.968143226742263e-05,
|
|
"loss": 4.7007,
|
|
"step": 19456
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.967304631991211e-05,
|
|
"loss": 4.6801,
|
|
"step": 19968
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.966466037240159e-05,
|
|
"loss": 4.6846,
|
|
"step": 20480
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.96562908036948e-05,
|
|
"loss": 4.6714,
|
|
"step": 20992
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.964790485618428e-05,
|
|
"loss": 4.6626,
|
|
"step": 21504
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.963951890867376e-05,
|
|
"loss": 4.6429,
|
|
"step": 22016
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.963113296116324e-05,
|
|
"loss": 4.6519,
|
|
"step": 22528
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.962274701365272e-05,
|
|
"loss": 4.6346,
|
|
"step": 23040
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9614377444945934e-05,
|
|
"loss": 4.6331,
|
|
"step": 23552
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9605991497435414e-05,
|
|
"loss": 4.623,
|
|
"step": 24064
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9597605549924894e-05,
|
|
"loss": 4.6016,
|
|
"step": 24576
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.95892359812181e-05,
|
|
"loss": 4.6024,
|
|
"step": 25088
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.958085003370758e-05,
|
|
"loss": 4.6049,
|
|
"step": 25600
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.957246408619706e-05,
|
|
"loss": 4.5845,
|
|
"step": 26112
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.956407813868654e-05,
|
|
"loss": 4.5752,
|
|
"step": 26624
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9555692191176016e-05,
|
|
"loss": 4.5654,
|
|
"step": 27136
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9547306243665496e-05,
|
|
"loss": 4.5651,
|
|
"step": 27648
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9538920296154976e-05,
|
|
"loss": 4.5488,
|
|
"step": 28160
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9530534348644456e-05,
|
|
"loss": 4.5677,
|
|
"step": 28672
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.952216477993767e-05,
|
|
"loss": 4.5365,
|
|
"step": 29184
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.951377883242715e-05,
|
|
"loss": 4.5446,
|
|
"step": 29696
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.950539288491663e-05,
|
|
"loss": 4.5359,
|
|
"step": 30208
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.949700693740611e-05,
|
|
"loss": 4.5111,
|
|
"step": 30720
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.948862098989559e-05,
|
|
"loss": 4.5192,
|
|
"step": 31232
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.94802514211888e-05,
|
|
"loss": 4.5095,
|
|
"step": 31744
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.947186547367828e-05,
|
|
"loss": 4.4953,
|
|
"step": 32256
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.946347952616776e-05,
|
|
"loss": 4.4871,
|
|
"step": 32768
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.945509357865724e-05,
|
|
"loss": 4.5026,
|
|
"step": 33280
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.944672400995045e-05,
|
|
"loss": 4.482,
|
|
"step": 33792
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.943833806243993e-05,
|
|
"loss": 4.4749,
|
|
"step": 34304
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.942995211492941e-05,
|
|
"loss": 4.464,
|
|
"step": 34816
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9421582546222625e-05,
|
|
"loss": 4.4767,
|
|
"step": 35328
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9413196598712105e-05,
|
|
"loss": 4.481,
|
|
"step": 35840
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9404810651201585e-05,
|
|
"loss": 4.466,
|
|
"step": 36352
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9396424703691065e-05,
|
|
"loss": 4.4608,
|
|
"step": 36864
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9388038756180545e-05,
|
|
"loss": 4.4643,
|
|
"step": 37376
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9379652808670025e-05,
|
|
"loss": 4.4623,
|
|
"step": 37888
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9371266861159505e-05,
|
|
"loss": 4.4513,
|
|
"step": 38400
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9362897292452714e-05,
|
|
"loss": 4.4343,
|
|
"step": 38912
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9354511344942194e-05,
|
|
"loss": 4.4415,
|
|
"step": 39424
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9346125397431674e-05,
|
|
"loss": 4.4303,
|
|
"step": 39936
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9337739449921154e-05,
|
|
"loss": 4.4284,
|
|
"step": 40448
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9329353502410634e-05,
|
|
"loss": 4.417,
|
|
"step": 40960
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9320967554900114e-05,
|
|
"loss": 4.4235,
|
|
"step": 41472
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9312581607389594e-05,
|
|
"loss": 4.428,
|
|
"step": 41984
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.9304195659879074e-05,
|
|
"loss": 4.4195,
|
|
"step": 42496
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.929582609117229e-05,
|
|
"loss": 4.3942,
|
|
"step": 43008
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.92874565224655e-05,
|
|
"loss": 4.3995,
|
|
"step": 43520
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.927907057495498e-05,
|
|
"loss": 4.4045,
|
|
"step": 44032
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.927068462744446e-05,
|
|
"loss": 4.3984,
|
|
"step": 44544
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.926229867993394e-05,
|
|
"loss": 4.3933,
|
|
"step": 45056
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.925391273242342e-05,
|
|
"loss": 4.3903,
|
|
"step": 45568
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.92455267849129e-05,
|
|
"loss": 4.3846,
|
|
"step": 46080
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.923714083740238e-05,
|
|
"loss": 4.3713,
|
|
"step": 46592
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.922877126869559e-05,
|
|
"loss": 4.3764,
|
|
"step": 47104
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.922038532118507e-05,
|
|
"loss": 4.3763,
|
|
"step": 47616
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.921199937367455e-05,
|
|
"loss": 4.3682,
|
|
"step": 48128
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.920361342616403e-05,
|
|
"loss": 4.3771,
|
|
"step": 48640
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.919522747865351e-05,
|
|
"loss": 4.3572,
|
|
"step": 49152
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.918684153114299e-05,
|
|
"loss": 4.358,
|
|
"step": 49664
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.917845558363247e-05,
|
|
"loss": 4.3554,
|
|
"step": 50176
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.917006963612195e-05,
|
|
"loss": 4.3545,
|
|
"step": 50688
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.916171644621889e-05,
|
|
"loss": 4.3583,
|
|
"step": 51200
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.915333049870837e-05,
|
|
"loss": 4.3416,
|
|
"step": 51712
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.914494455119785e-05,
|
|
"loss": 4.3392,
|
|
"step": 52224
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.9136558603687325e-05,
|
|
"loss": 4.3352,
|
|
"step": 52736
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.9128172656176805e-05,
|
|
"loss": 4.34,
|
|
"step": 53248
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.9119786708666285e-05,
|
|
"loss": 4.3212,
|
|
"step": 53760
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.91114171399595e-05,
|
|
"loss": 4.3129,
|
|
"step": 54272
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.910303119244898e-05,
|
|
"loss": 4.324,
|
|
"step": 54784
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.909464524493846e-05,
|
|
"loss": 4.3285,
|
|
"step": 55296
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.908625929742794e-05,
|
|
"loss": 4.3214,
|
|
"step": 55808
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.907787334991742e-05,
|
|
"loss": 4.32,
|
|
"step": 56320
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.90694874024069e-05,
|
|
"loss": 4.3071,
|
|
"step": 56832
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.906110145489638e-05,
|
|
"loss": 4.3175,
|
|
"step": 57344
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.905273188618959e-05,
|
|
"loss": 4.3005,
|
|
"step": 57856
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.904434593867907e-05,
|
|
"loss": 4.3082,
|
|
"step": 58368
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.903595999116855e-05,
|
|
"loss": 4.3083,
|
|
"step": 58880
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.902757404365803e-05,
|
|
"loss": 4.3094,
|
|
"step": 59392
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.901920447495124e-05,
|
|
"loss": 4.2979,
|
|
"step": 59904
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.901081852744072e-05,
|
|
"loss": 4.2957,
|
|
"step": 60416
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.90024325799302e-05,
|
|
"loss": 4.2959,
|
|
"step": 60928
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8994046632419686e-05,
|
|
"loss": 4.2984,
|
|
"step": 61440
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8985660684909166e-05,
|
|
"loss": 4.2861,
|
|
"step": 61952
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8977274737398646e-05,
|
|
"loss": 4.2889,
|
|
"step": 62464
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8968888789888125e-05,
|
|
"loss": 4.285,
|
|
"step": 62976
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8960519221181335e-05,
|
|
"loss": 4.2868,
|
|
"step": 63488
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8952133273670815e-05,
|
|
"loss": 4.2715,
|
|
"step": 64000
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8943747326160294e-05,
|
|
"loss": 4.2681,
|
|
"step": 64512
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8935361378649774e-05,
|
|
"loss": 4.274,
|
|
"step": 65024
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8926991809942984e-05,
|
|
"loss": 4.2763,
|
|
"step": 65536
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8918605862432463e-05,
|
|
"loss": 4.2667,
|
|
"step": 66048
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8910219914921943e-05,
|
|
"loss": 4.2809,
|
|
"step": 66560
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8901833967411423e-05,
|
|
"loss": 4.2703,
|
|
"step": 67072
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.889346439870464e-05,
|
|
"loss": 4.2557,
|
|
"step": 67584
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.888507845119412e-05,
|
|
"loss": 4.2638,
|
|
"step": 68096
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.88766925036836e-05,
|
|
"loss": 4.2722,
|
|
"step": 68608
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.886830655617308e-05,
|
|
"loss": 4.2618,
|
|
"step": 69120
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.885993698746629e-05,
|
|
"loss": 4.2588,
|
|
"step": 69632
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.885155103995577e-05,
|
|
"loss": 4.2492,
|
|
"step": 70144
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.884316509244525e-05,
|
|
"loss": 4.257,
|
|
"step": 70656
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.883479552373846e-05,
|
|
"loss": 4.2614,
|
|
"step": 71168
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.882640957622794e-05,
|
|
"loss": 4.2466,
|
|
"step": 71680
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.881802362871742e-05,
|
|
"loss": 4.2398,
|
|
"step": 72192
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.88096376812069e-05,
|
|
"loss": 4.2453,
|
|
"step": 72704
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.880125173369638e-05,
|
|
"loss": 4.2382,
|
|
"step": 73216
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8792882164989586e-05,
|
|
"loss": 4.2353,
|
|
"step": 73728
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.878449621747907e-05,
|
|
"loss": 4.2438,
|
|
"step": 74240
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.877611026996855e-05,
|
|
"loss": 4.2317,
|
|
"step": 74752
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.876772432245803e-05,
|
|
"loss": 4.2344,
|
|
"step": 75264
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.875933837494751e-05,
|
|
"loss": 4.2301,
|
|
"step": 75776
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.8750952427436986e-05,
|
|
"loss": 4.2256,
|
|
"step": 76288
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 4.193907260894775,
|
|
"eval_runtime": 305.9942,
|
|
"eval_samples_per_second": 1247.053,
|
|
"eval_steps_per_second": 38.971,
|
|
"step": 76320
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8742566479926466e-05,
|
|
"loss": 4.2149,
|
|
"step": 76800
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8734180532415946e-05,
|
|
"loss": 4.2221,
|
|
"step": 77312
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.872581096370916e-05,
|
|
"loss": 4.2301,
|
|
"step": 77824
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8717425016198635e-05,
|
|
"loss": 4.219,
|
|
"step": 78336
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8709039068688115e-05,
|
|
"loss": 4.2201,
|
|
"step": 78848
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8700653121177595e-05,
|
|
"loss": 4.2136,
|
|
"step": 79360
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.869228355247081e-05,
|
|
"loss": 4.2104,
|
|
"step": 79872
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.868389760496029e-05,
|
|
"loss": 4.1966,
|
|
"step": 80384
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8675528036253507e-05,
|
|
"loss": 4.2225,
|
|
"step": 80896
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8667142088742986e-05,
|
|
"loss": 4.2065,
|
|
"step": 81408
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.865875614123246e-05,
|
|
"loss": 4.2163,
|
|
"step": 81920
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.865037019372194e-05,
|
|
"loss": 4.2125,
|
|
"step": 82432
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8642000625015155e-05,
|
|
"loss": 4.1956,
|
|
"step": 82944
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8633614677504635e-05,
|
|
"loss": 4.1966,
|
|
"step": 83456
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.862522872999411e-05,
|
|
"loss": 4.1915,
|
|
"step": 83968
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.861684278248359e-05,
|
|
"loss": 4.1838,
|
|
"step": 84480
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.860845683497307e-05,
|
|
"loss": 4.1903,
|
|
"step": 84992
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.860007088746255e-05,
|
|
"loss": 4.1828,
|
|
"step": 85504
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.859168493995203e-05,
|
|
"loss": 4.1796,
|
|
"step": 86016
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.858329899244151e-05,
|
|
"loss": 4.2006,
|
|
"step": 86528
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8574913044930995e-05,
|
|
"loss": 4.1851,
|
|
"step": 87040
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8566527097420475e-05,
|
|
"loss": 4.1773,
|
|
"step": 87552
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8558141149909955e-05,
|
|
"loss": 4.175,
|
|
"step": 88064
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.8549771581203164e-05,
|
|
"loss": 4.1887,
|
|
"step": 88576
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.854140201249637e-05,
|
|
"loss": 4.167,
|
|
"step": 89088
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.853301606498585e-05,
|
|
"loss": 4.163,
|
|
"step": 89600
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.852463011747533e-05,
|
|
"loss": 4.1674,
|
|
"step": 90112
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.851624416996481e-05,
|
|
"loss": 4.168,
|
|
"step": 90624
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 4.850785822245429e-05,
|
|
"loss": 4.1622,
|
|
"step": 91136
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.849947227494377e-05,
|
|
"loss": 4.1625,
|
|
"step": 91648
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.849110270623698e-05,
|
|
"loss": 4.1634,
|
|
"step": 92160
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.848271675872646e-05,
|
|
"loss": 4.1675,
|
|
"step": 92672
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.847433081121595e-05,
|
|
"loss": 4.1523,
|
|
"step": 93184
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.846594486370543e-05,
|
|
"loss": 4.1616,
|
|
"step": 93696
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.845755891619491e-05,
|
|
"loss": 4.1674,
|
|
"step": 94208
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.844918934748812e-05,
|
|
"loss": 4.1636,
|
|
"step": 94720
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.84408033999776e-05,
|
|
"loss": 4.1358,
|
|
"step": 95232
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.843241745246708e-05,
|
|
"loss": 4.1439,
|
|
"step": 95744
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.842403150495656e-05,
|
|
"loss": 4.1424,
|
|
"step": 96256
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.841564555744604e-05,
|
|
"loss": 4.1458,
|
|
"step": 96768
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.840725960993552e-05,
|
|
"loss": 4.1399,
|
|
"step": 97280
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8398873662425e-05,
|
|
"loss": 4.146,
|
|
"step": 97792
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.839050409371821e-05,
|
|
"loss": 4.1351,
|
|
"step": 98304
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8382118146207687e-05,
|
|
"loss": 4.1434,
|
|
"step": 98816
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8373732198697167e-05,
|
|
"loss": 4.1413,
|
|
"step": 99328
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8365346251186646e-05,
|
|
"loss": 4.1428,
|
|
"step": 99840
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8356960303676126e-05,
|
|
"loss": 4.1404,
|
|
"step": 100352
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8348574356165606e-05,
|
|
"loss": 4.1203,
|
|
"step": 100864
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.834020478745882e-05,
|
|
"loss": 4.1345,
|
|
"step": 101376
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8331818839948295e-05,
|
|
"loss": 4.1379,
|
|
"step": 101888
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8323432892437775e-05,
|
|
"loss": 4.1326,
|
|
"step": 102400
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8315046944927255e-05,
|
|
"loss": 4.1246,
|
|
"step": 102912
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8306660997416735e-05,
|
|
"loss": 4.1176,
|
|
"step": 103424
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8298291428709944e-05,
|
|
"loss": 4.1278,
|
|
"step": 103936
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8289905481199424e-05,
|
|
"loss": 4.1091,
|
|
"step": 104448
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8281519533688904e-05,
|
|
"loss": 4.1373,
|
|
"step": 104960
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8273133586178384e-05,
|
|
"loss": 4.1155,
|
|
"step": 105472
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.826474763866787e-05,
|
|
"loss": 4.1284,
|
|
"step": 105984
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.825636169115735e-05,
|
|
"loss": 4.1272,
|
|
"step": 106496
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.824797574364683e-05,
|
|
"loss": 4.1017,
|
|
"step": 107008
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.823960617494004e-05,
|
|
"loss": 4.1165,
|
|
"step": 107520
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.823122022742952e-05,
|
|
"loss": 4.1112,
|
|
"step": 108032
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8222834279919e-05,
|
|
"loss": 4.0996,
|
|
"step": 108544
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.821444833240848e-05,
|
|
"loss": 4.0981,
|
|
"step": 109056
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.820607876370169e-05,
|
|
"loss": 4.1153,
|
|
"step": 109568
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.819769281619117e-05,
|
|
"loss": 4.1039,
|
|
"step": 110080
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.818930686868065e-05,
|
|
"loss": 4.0985,
|
|
"step": 110592
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.818092092117013e-05,
|
|
"loss": 4.0888,
|
|
"step": 111104
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.817255135246334e-05,
|
|
"loss": 4.1075,
|
|
"step": 111616
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.816416540495282e-05,
|
|
"loss": 4.1111,
|
|
"step": 112128
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8155779457442305e-05,
|
|
"loss": 4.1065,
|
|
"step": 112640
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8147393509931785e-05,
|
|
"loss": 4.1007,
|
|
"step": 113152
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8139007562421265e-05,
|
|
"loss": 4.1118,
|
|
"step": 113664
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8130637993714474e-05,
|
|
"loss": 4.1066,
|
|
"step": 114176
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8122252046203954e-05,
|
|
"loss": 4.1054,
|
|
"step": 114688
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8113866098693434e-05,
|
|
"loss": 4.0887,
|
|
"step": 115200
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8105480151182914e-05,
|
|
"loss": 4.1013,
|
|
"step": 115712
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.809711058247612e-05,
|
|
"loss": 4.0927,
|
|
"step": 116224
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.80887246349656e-05,
|
|
"loss": 4.091,
|
|
"step": 116736
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.808033868745508e-05,
|
|
"loss": 4.0877,
|
|
"step": 117248
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.807195273994456e-05,
|
|
"loss": 4.0923,
|
|
"step": 117760
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.806356679243404e-05,
|
|
"loss": 4.1039,
|
|
"step": 118272
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.805519722372726e-05,
|
|
"loss": 4.0968,
|
|
"step": 118784
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.804681127621674e-05,
|
|
"loss": 4.0725,
|
|
"step": 119296
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.803842532870622e-05,
|
|
"loss": 4.0783,
|
|
"step": 119808
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.80300393811957e-05,
|
|
"loss": 4.0856,
|
|
"step": 120320
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.802165343368518e-05,
|
|
"loss": 4.0833,
|
|
"step": 120832
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.801328386497839e-05,
|
|
"loss": 4.0852,
|
|
"step": 121344
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 4.8004914296271596e-05,
|
|
"loss": 4.0805,
|
|
"step": 121856
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7996528348761076e-05,
|
|
"loss": 4.0797,
|
|
"step": 122368
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7988142401250556e-05,
|
|
"loss": 4.0686,
|
|
"step": 122880
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7979756453740036e-05,
|
|
"loss": 4.0794,
|
|
"step": 123392
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7971370506229516e-05,
|
|
"loss": 4.0773,
|
|
"step": 123904
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7962984558718996e-05,
|
|
"loss": 4.0762,
|
|
"step": 124416
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.795461499001221e-05,
|
|
"loss": 4.0835,
|
|
"step": 124928
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.794622904250169e-05,
|
|
"loss": 4.0746,
|
|
"step": 125440
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.793784309499117e-05,
|
|
"loss": 4.0674,
|
|
"step": 125952
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.792945714748065e-05,
|
|
"loss": 4.0728,
|
|
"step": 126464
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.792107119997013e-05,
|
|
"loss": 4.0732,
|
|
"step": 126976
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.791270163126334e-05,
|
|
"loss": 4.0752,
|
|
"step": 127488
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.790431568375282e-05,
|
|
"loss": 4.0651,
|
|
"step": 128000
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.78959297362423e-05,
|
|
"loss": 4.0608,
|
|
"step": 128512
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7887543788731774e-05,
|
|
"loss": 4.0612,
|
|
"step": 129024
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7879157841221254e-05,
|
|
"loss": 4.0713,
|
|
"step": 129536
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7870771893710734e-05,
|
|
"loss": 4.0542,
|
|
"step": 130048
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.786240232500395e-05,
|
|
"loss": 4.0424,
|
|
"step": 130560
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.785401637749343e-05,
|
|
"loss": 4.0614,
|
|
"step": 131072
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.784563042998291e-05,
|
|
"loss": 4.0617,
|
|
"step": 131584
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.783724448247239e-05,
|
|
"loss": 4.0555,
|
|
"step": 132096
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.782885853496187e-05,
|
|
"loss": 4.0567,
|
|
"step": 132608
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.782047258745135e-05,
|
|
"loss": 4.0518,
|
|
"step": 133120
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.781208663994083e-05,
|
|
"loss": 4.06,
|
|
"step": 133632
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.780370069243031e-05,
|
|
"loss": 4.0467,
|
|
"step": 134144
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.779531474491979e-05,
|
|
"loss": 4.0517,
|
|
"step": 134656
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7786945176213e-05,
|
|
"loss": 4.061,
|
|
"step": 135168
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.777855922870248e-05,
|
|
"loss": 4.0615,
|
|
"step": 135680
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.777018965999569e-05,
|
|
"loss": 4.048,
|
|
"step": 136192
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.776180371248517e-05,
|
|
"loss": 4.0478,
|
|
"step": 136704
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.775341776497465e-05,
|
|
"loss": 4.0537,
|
|
"step": 137216
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7745031817464134e-05,
|
|
"loss": 4.0553,
|
|
"step": 137728
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7736645869953614e-05,
|
|
"loss": 4.0399,
|
|
"step": 138240
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7728259922443094e-05,
|
|
"loss": 4.0511,
|
|
"step": 138752
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7719873974932574e-05,
|
|
"loss": 4.0416,
|
|
"step": 139264
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7711488027422054e-05,
|
|
"loss": 4.0541,
|
|
"step": 139776
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.770311845871526e-05,
|
|
"loss": 4.0341,
|
|
"step": 140288
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.769473251120474e-05,
|
|
"loss": 4.0366,
|
|
"step": 140800
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.768634656369422e-05,
|
|
"loss": 4.0399,
|
|
"step": 141312
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.76779606161837e-05,
|
|
"loss": 4.0428,
|
|
"step": 141824
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.766959104747691e-05,
|
|
"loss": 4.0351,
|
|
"step": 142336
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.766120509996639e-05,
|
|
"loss": 4.0529,
|
|
"step": 142848
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.765281915245587e-05,
|
|
"loss": 4.0416,
|
|
"step": 143360
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.764443320494535e-05,
|
|
"loss": 4.0332,
|
|
"step": 143872
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.763606363623857e-05,
|
|
"loss": 4.0343,
|
|
"step": 144384
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.762767768872805e-05,
|
|
"loss": 4.0495,
|
|
"step": 144896
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.761929174121753e-05,
|
|
"loss": 4.0427,
|
|
"step": 145408
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.761090579370701e-05,
|
|
"loss": 4.0376,
|
|
"step": 145920
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.760253622500022e-05,
|
|
"loss": 4.0297,
|
|
"step": 146432
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.75941502774897e-05,
|
|
"loss": 4.0398,
|
|
"step": 146944
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.758576432997918e-05,
|
|
"loss": 4.0432,
|
|
"step": 147456
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.757737838246866e-05,
|
|
"loss": 4.0326,
|
|
"step": 147968
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7569008813761866e-05,
|
|
"loss": 4.0278,
|
|
"step": 148480
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7560622866251346e-05,
|
|
"loss": 4.03,
|
|
"step": 148992
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7552236918740826e-05,
|
|
"loss": 4.0257,
|
|
"step": 149504
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7543867350034035e-05,
|
|
"loss": 4.0224,
|
|
"step": 150016
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.753548140252352e-05,
|
|
"loss": 4.0376,
|
|
"step": 150528
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7527095455013e-05,
|
|
"loss": 4.0228,
|
|
"step": 151040
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.751870950750248e-05,
|
|
"loss": 4.0267,
|
|
"step": 151552
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.751032355999196e-05,
|
|
"loss": 4.0247,
|
|
"step": 152064
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 4.7501937612481435e-05,
|
|
"loss": 4.0197,
|
|
"step": 152576
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"eval_loss": 4.026294708251953,
|
|
"eval_runtime": 306.1917,
|
|
"eval_samples_per_second": 1246.249,
|
|
"eval_steps_per_second": 38.946,
|
|
"step": 152640
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7493551664970914e-05,
|
|
"loss": 4.0101,
|
|
"step": 153088
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7485165717460394e-05,
|
|
"loss": 4.0192,
|
|
"step": 153600
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7476779769949874e-05,
|
|
"loss": 4.0327,
|
|
"step": 154112
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7468393822439354e-05,
|
|
"loss": 4.0198,
|
|
"step": 154624
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7460007874928834e-05,
|
|
"loss": 4.0238,
|
|
"step": 155136
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7451621927418314e-05,
|
|
"loss": 4.015,
|
|
"step": 155648
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7443235979907794e-05,
|
|
"loss": 4.0183,
|
|
"step": 156160
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7434850032397274e-05,
|
|
"loss": 4.0012,
|
|
"step": 156672
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.742648046369049e-05,
|
|
"loss": 4.0258,
|
|
"step": 157184
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.741809451617997e-05,
|
|
"loss": 4.0216,
|
|
"step": 157696
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.740970856866945e-05,
|
|
"loss": 4.0176,
|
|
"step": 158208
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.740132262115893e-05,
|
|
"loss": 4.0246,
|
|
"step": 158720
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.739295305245214e-05,
|
|
"loss": 4.0054,
|
|
"step": 159232
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.738456710494162e-05,
|
|
"loss": 4.0074,
|
|
"step": 159744
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.73761811574311e-05,
|
|
"loss": 4.0111,
|
|
"step": 160256
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.736779520992058e-05,
|
|
"loss": 3.9957,
|
|
"step": 160768
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.735940926241006e-05,
|
|
"loss": 4.0045,
|
|
"step": 161280
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.735102331489954e-05,
|
|
"loss": 3.9984,
|
|
"step": 161792
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.734263736738902e-05,
|
|
"loss": 3.9972,
|
|
"step": 162304
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.73342514198785e-05,
|
|
"loss": 4.0236,
|
|
"step": 162816
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.732586547236797e-05,
|
|
"loss": 4.0041,
|
|
"step": 163328
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.731747952485746e-05,
|
|
"loss": 4.0046,
|
|
"step": 163840
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.730909357734694e-05,
|
|
"loss": 3.9952,
|
|
"step": 164352
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7300724008640154e-05,
|
|
"loss": 4.0105,
|
|
"step": 164864
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7292354439933364e-05,
|
|
"loss": 3.9854,
|
|
"step": 165376
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7283968492422844e-05,
|
|
"loss": 3.9907,
|
|
"step": 165888
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7275582544912323e-05,
|
|
"loss": 3.9903,
|
|
"step": 166400
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.72671965974018e-05,
|
|
"loss": 3.9992,
|
|
"step": 166912
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.7258810649891277e-05,
|
|
"loss": 3.9893,
|
|
"step": 167424
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7250424702380757e-05,
|
|
"loss": 3.9887,
|
|
"step": 167936
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7242038754870237e-05,
|
|
"loss": 3.9958,
|
|
"step": 168448
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7233652807359716e-05,
|
|
"loss": 3.9995,
|
|
"step": 168960
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7225283238652926e-05,
|
|
"loss": 3.9806,
|
|
"step": 169472
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.721689729114241e-05,
|
|
"loss": 3.9949,
|
|
"step": 169984
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.720851134363189e-05,
|
|
"loss": 3.9978,
|
|
"step": 170496
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.720012539612137e-05,
|
|
"loss": 3.9939,
|
|
"step": 171008
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.719175582741458e-05,
|
|
"loss": 3.9727,
|
|
"step": 171520
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.718336987990406e-05,
|
|
"loss": 3.9834,
|
|
"step": 172032
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.717498393239354e-05,
|
|
"loss": 3.973,
|
|
"step": 172544
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.716659798488302e-05,
|
|
"loss": 3.9852,
|
|
"step": 173056
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.715822841617623e-05,
|
|
"loss": 3.9788,
|
|
"step": 173568
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.714984246866571e-05,
|
|
"loss": 3.9865,
|
|
"step": 174080
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.714145652115519e-05,
|
|
"loss": 3.9769,
|
|
"step": 174592
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.713307057364467e-05,
|
|
"loss": 3.9808,
|
|
"step": 175104
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.712470100493788e-05,
|
|
"loss": 3.9854,
|
|
"step": 175616
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7116315057427366e-05,
|
|
"loss": 3.9878,
|
|
"step": 176128
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7107929109916846e-05,
|
|
"loss": 3.983,
|
|
"step": 176640
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7099543162406326e-05,
|
|
"loss": 3.963,
|
|
"step": 177152
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7091173593699535e-05,
|
|
"loss": 3.9769,
|
|
"step": 177664
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7082787646189015e-05,
|
|
"loss": 3.9864,
|
|
"step": 178176
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7074401698678495e-05,
|
|
"loss": 3.9742,
|
|
"step": 178688
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7066032129971704e-05,
|
|
"loss": 3.9706,
|
|
"step": 179200
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7057646182461184e-05,
|
|
"loss": 3.968,
|
|
"step": 179712
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7049260234950664e-05,
|
|
"loss": 3.9735,
|
|
"step": 180224
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7040874287440144e-05,
|
|
"loss": 3.959,
|
|
"step": 180736
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7032488339929624e-05,
|
|
"loss": 3.9818,
|
|
"step": 181248
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7024102392419104e-05,
|
|
"loss": 3.9704,
|
|
"step": 181760
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7015716444908584e-05,
|
|
"loss": 3.9748,
|
|
"step": 182272
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.7007330497398064e-05,
|
|
"loss": 3.984,
|
|
"step": 182784
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.699896092869128e-05,
|
|
"loss": 3.9512,
|
|
"step": 183296
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.699057498118076e-05,
|
|
"loss": 3.969,
|
|
"step": 183808
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.698218903367024e-05,
|
|
"loss": 3.963,
|
|
"step": 184320
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.697380308615972e-05,
|
|
"loss": 3.957,
|
|
"step": 184832
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.696543351745293e-05,
|
|
"loss": 3.9524,
|
|
"step": 185344
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.695704756994241e-05,
|
|
"loss": 3.9674,
|
|
"step": 185856
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.694866162243189e-05,
|
|
"loss": 3.9615,
|
|
"step": 186368
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.694027567492137e-05,
|
|
"loss": 3.9541,
|
|
"step": 186880
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.693190610621458e-05,
|
|
"loss": 3.9458,
|
|
"step": 187392
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.692352015870406e-05,
|
|
"loss": 3.9663,
|
|
"step": 187904
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.691513421119354e-05,
|
|
"loss": 3.9716,
|
|
"step": 188416
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.690674826368302e-05,
|
|
"loss": 3.9639,
|
|
"step": 188928
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.689837869497623e-05,
|
|
"loss": 3.9618,
|
|
"step": 189440
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.688999274746571e-05,
|
|
"loss": 3.9716,
|
|
"step": 189952
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.688160679995519e-05,
|
|
"loss": 3.9654,
|
|
"step": 190464
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.687322085244467e-05,
|
|
"loss": 3.9699,
|
|
"step": 190976
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.686485128373788e-05,
|
|
"loss": 3.9469,
|
|
"step": 191488
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.685646533622736e-05,
|
|
"loss": 3.9644,
|
|
"step": 192000
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.684807938871684e-05,
|
|
"loss": 3.9589,
|
|
"step": 192512
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.683969344120632e-05,
|
|
"loss": 3.952,
|
|
"step": 193024
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.683132387249953e-05,
|
|
"loss": 3.9567,
|
|
"step": 193536
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.682293792498901e-05,
|
|
"loss": 3.9586,
|
|
"step": 194048
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.681455197747849e-05,
|
|
"loss": 3.969,
|
|
"step": 194560
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.680616602996797e-05,
|
|
"loss": 3.9566,
|
|
"step": 195072
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.679779646126119e-05,
|
|
"loss": 3.9384,
|
|
"step": 195584
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.678941051375067e-05,
|
|
"loss": 3.9439,
|
|
"step": 196096
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.678102456624015e-05,
|
|
"loss": 3.9511,
|
|
"step": 196608
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.677263861872963e-05,
|
|
"loss": 3.9541,
|
|
"step": 197120
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.6764285428826565e-05,
|
|
"loss": 3.9536,
|
|
"step": 197632
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 4.6755899481316045e-05,
|
|
"loss": 3.9478,
|
|
"step": 198144
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6747513533805525e-05,
|
|
"loss": 3.9491,
|
|
"step": 198656
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6739127586295005e-05,
|
|
"loss": 3.9381,
|
|
"step": 199168
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6730741638784485e-05,
|
|
"loss": 3.9527,
|
|
"step": 199680
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6722355691273965e-05,
|
|
"loss": 3.9436,
|
|
"step": 200192
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6713969743763445e-05,
|
|
"loss": 3.9478,
|
|
"step": 200704
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.670560017505666e-05,
|
|
"loss": 3.9575,
|
|
"step": 201216
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.669721422754614e-05,
|
|
"loss": 3.9477,
|
|
"step": 201728
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.668882828003562e-05,
|
|
"loss": 3.9403,
|
|
"step": 202240
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.66804423325251e-05,
|
|
"loss": 3.9468,
|
|
"step": 202752
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.667205638501458e-05,
|
|
"loss": 3.9453,
|
|
"step": 203264
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.666368681630779e-05,
|
|
"loss": 3.948,
|
|
"step": 203776
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.665530086879727e-05,
|
|
"loss": 3.9415,
|
|
"step": 204288
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.664691492128675e-05,
|
|
"loss": 3.9387,
|
|
"step": 204800
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.663852897377623e-05,
|
|
"loss": 3.9327,
|
|
"step": 205312
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.66301430262657e-05,
|
|
"loss": 3.9468,
|
|
"step": 205824
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.662175707875518e-05,
|
|
"loss": 3.9344,
|
|
"step": 206336
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.66133875100484e-05,
|
|
"loss": 3.9226,
|
|
"step": 206848
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.660500156253788e-05,
|
|
"loss": 3.9345,
|
|
"step": 207360
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.659661561502736e-05,
|
|
"loss": 3.9402,
|
|
"step": 207872
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.658822966751684e-05,
|
|
"loss": 3.9311,
|
|
"step": 208384
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.657984372000632e-05,
|
|
"loss": 3.9402,
|
|
"step": 208896
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.65714577724958e-05,
|
|
"loss": 3.9285,
|
|
"step": 209408
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.656307182498528e-05,
|
|
"loss": 3.941,
|
|
"step": 209920
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.655468587747476e-05,
|
|
"loss": 3.9271,
|
|
"step": 210432
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.654631630876797e-05,
|
|
"loss": 3.9285,
|
|
"step": 210944
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6537946740061176e-05,
|
|
"loss": 3.9455,
|
|
"step": 211456
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6529560792550656e-05,
|
|
"loss": 3.9411,
|
|
"step": 211968
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6521174845040136e-05,
|
|
"loss": 3.9305,
|
|
"step": 212480
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6512788897529616e-05,
|
|
"loss": 3.9321,
|
|
"step": 212992
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6504402950019096e-05,
|
|
"loss": 3.9302,
|
|
"step": 213504
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.649603338131231e-05,
|
|
"loss": 3.9368,
|
|
"step": 214016
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.648764743380179e-05,
|
|
"loss": 3.9238,
|
|
"step": 214528
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.647926148629127e-05,
|
|
"loss": 3.9362,
|
|
"step": 215040
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.647087553878075e-05,
|
|
"loss": 3.9261,
|
|
"step": 215552
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.646248959127023e-05,
|
|
"loss": 3.9368,
|
|
"step": 216064
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.645410364375971e-05,
|
|
"loss": 3.9161,
|
|
"step": 216576
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.644571769624919e-05,
|
|
"loss": 3.9263,
|
|
"step": 217088
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.64373481275424e-05,
|
|
"loss": 3.9212,
|
|
"step": 217600
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.642896218003188e-05,
|
|
"loss": 3.9272,
|
|
"step": 218112
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.642057623252136e-05,
|
|
"loss": 3.9233,
|
|
"step": 218624
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.641219028501084e-05,
|
|
"loss": 3.9417,
|
|
"step": 219136
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.640382071630405e-05,
|
|
"loss": 3.9288,
|
|
"step": 219648
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6395434768793536e-05,
|
|
"loss": 3.9228,
|
|
"step": 220160
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6387048821283016e-05,
|
|
"loss": 3.9169,
|
|
"step": 220672
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6378662873772496e-05,
|
|
"loss": 3.9355,
|
|
"step": 221184
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6370276926261976e-05,
|
|
"loss": 3.9314,
|
|
"step": 221696
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6361907357555185e-05,
|
|
"loss": 3.9288,
|
|
"step": 222208
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6353521410044665e-05,
|
|
"loss": 3.9203,
|
|
"step": 222720
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6345135462534145e-05,
|
|
"loss": 3.9267,
|
|
"step": 223232
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6336749515023625e-05,
|
|
"loss": 3.9319,
|
|
"step": 223744
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6328363567513105e-05,
|
|
"loss": 3.9228,
|
|
"step": 224256
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6319977620002585e-05,
|
|
"loss": 3.9174,
|
|
"step": 224768
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.631159167249206e-05,
|
|
"loss": 3.9234,
|
|
"step": 225280
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.630320572498154e-05,
|
|
"loss": 3.9186,
|
|
"step": 225792
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.6294836156274754e-05,
|
|
"loss": 3.9116,
|
|
"step": 226304
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.628646658756797e-05,
|
|
"loss": 3.9268,
|
|
"step": 226816
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.627808064005745e-05,
|
|
"loss": 3.9225,
|
|
"step": 227328
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.626969469254693e-05,
|
|
"loss": 3.9167,
|
|
"step": 227840
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.626130874503641e-05,
|
|
"loss": 3.9151,
|
|
"step": 228352
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 4.625293917632962e-05,
|
|
"loss": 3.9124,
|
|
"step": 228864
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"eval_loss": 3.9520537853240967,
|
|
"eval_runtime": 306.6915,
|
|
"eval_samples_per_second": 1244.218,
|
|
"eval_steps_per_second": 38.883,
|
|
"step": 228960
|
|
}
|
|
],
|
|
"logging_steps": 512,
|
|
"max_steps": 3052726,
|
|
"num_train_epochs": 9223372036854775807,
|
|
"save_steps": 10,
|
|
"total_flos": 1.6001973826759066e+17,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|