|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 0.07078267947833165, |
|
"global_step": 1000, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.999998454568244e-05, |
|
"loss": 1.3539, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.9999938182748876e-05, |
|
"loss": 1.1833, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.9999870029288556e-05, |
|
"loss": 1.173, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.999976494017406e-05, |
|
"loss": 1.0772, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.999962894271507e-05, |
|
"loss": 1.0715, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.9999462037079705e-05, |
|
"loss": 1.0268, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.999926422347434e-05, |
|
"loss": 0.9807, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.999903550214352e-05, |
|
"loss": 0.9862, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.999877587337004e-05, |
|
"loss": 0.9725, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.999848533747488e-05, |
|
"loss": 0.9993, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.999816389481725e-05, |
|
"loss": 0.9596, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.999781154579456e-05, |
|
"loss": 0.979, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9997428290842444e-05, |
|
"loss": 0.9748, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.999701413043471e-05, |
|
"loss": 0.9309, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.999656906508344e-05, |
|
"loss": 0.9143, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.999609309533887e-05, |
|
"loss": 0.9439, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.999558622178947e-05, |
|
"loss": 0.9286, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.99950484450619e-05, |
|
"loss": 0.9544, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.999447976582104e-05, |
|
"loss": 0.9355, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.999388018476998e-05, |
|
"loss": 0.9154, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.999324970265001e-05, |
|
"loss": 0.9326, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.999258832024061e-05, |
|
"loss": 0.9215, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.99918960383595e-05, |
|
"loss": 0.9281, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.9991172857862555e-05, |
|
"loss": 0.935, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.99904187796439e-05, |
|
"loss": 0.941, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.9989633804635814e-05, |
|
"loss": 0.9377, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.9988817933808814e-05, |
|
"loss": 0.9014, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.9987971168171585e-05, |
|
"loss": 0.9323, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.998709350877103e-05, |
|
"loss": 0.8987, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.998618495669224e-05, |
|
"loss": 0.8933, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.9985245513058495e-05, |
|
"loss": 0.893, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.9984275179031276e-05, |
|
"loss": 0.909, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.998327395581025e-05, |
|
"loss": 0.9235, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.9982241844633265e-05, |
|
"loss": 0.8945, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.998117884677638e-05, |
|
"loss": 0.9095, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.998008496355382e-05, |
|
"loss": 0.8919, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.9978960196318006e-05, |
|
"loss": 0.9088, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.997780454645954e-05, |
|
"loss": 0.8985, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.99766180154072e-05, |
|
"loss": 0.8972, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.9975400604627957e-05, |
|
"loss": 0.8983, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.9974152315626935e-05, |
|
"loss": 0.9115, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.997287314994746e-05, |
|
"loss": 0.8957, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.997156310917103e-05, |
|
"loss": 0.8681, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.9970222194917296e-05, |
|
"loss": 0.894, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.996885040884409e-05, |
|
"loss": 0.8798, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.996744775264743e-05, |
|
"loss": 0.9034, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.996601422806147e-05, |
|
"loss": 0.9033, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.9964549836858536e-05, |
|
"loss": 0.8841, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.9963054580849134e-05, |
|
"loss": 0.8877, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.996152846188191e-05, |
|
"loss": 0.8729, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.995997148184369e-05, |
|
"loss": 0.8853, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.9958383642659414e-05, |
|
"loss": 0.8837, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.995676494629221e-05, |
|
"loss": 0.8833, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.9955115394743354e-05, |
|
"loss": 0.8843, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.995343499005225e-05, |
|
"loss": 0.892, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.995172373429646e-05, |
|
"loss": 0.8575, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.9949981629591705e-05, |
|
"loss": 0.8311, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.99482086780918e-05, |
|
"loss": 0.8669, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.994640488198874e-05, |
|
"loss": 0.8388, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.994457024351264e-05, |
|
"loss": 0.8424, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.994270476493175e-05, |
|
"loss": 0.8676, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.994080844855243e-05, |
|
"loss": 0.8598, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.993888129671921e-05, |
|
"loss": 0.824, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.993692331181469e-05, |
|
"loss": 0.8652, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.993493449625963e-05, |
|
"loss": 0.8533, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.993291485251288e-05, |
|
"loss": 0.8677, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.993086438307143e-05, |
|
"loss": 0.8459, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.9928783090470365e-05, |
|
"loss": 0.8626, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.992667097728287e-05, |
|
"loss": 0.8127, |
|
"step": 690 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.992452804612027e-05, |
|
"loss": 0.8716, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.992235429963195e-05, |
|
"loss": 0.8544, |
|
"step": 710 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.992014974050542e-05, |
|
"loss": 0.8562, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.991791437146627e-05, |
|
"loss": 0.871, |
|
"step": 730 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.9915648195278186e-05, |
|
"loss": 0.8453, |
|
"step": 740 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.9913351214742945e-05, |
|
"loss": 0.8524, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.991102343270042e-05, |
|
"loss": 0.8581, |
|
"step": 760 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.9908664852028545e-05, |
|
"loss": 0.8477, |
|
"step": 770 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.990627547564335e-05, |
|
"loss": 0.8651, |
|
"step": 780 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.990385530649891e-05, |
|
"loss": 0.8453, |
|
"step": 790 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.9901404347587404e-05, |
|
"loss": 0.8586, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.9898922601939056e-05, |
|
"loss": 0.8746, |
|
"step": 810 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.989641007262218e-05, |
|
"loss": 0.8652, |
|
"step": 820 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.98938667627431e-05, |
|
"loss": 0.8531, |
|
"step": 830 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.989129267544626e-05, |
|
"loss": 0.8686, |
|
"step": 840 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.988868781391408e-05, |
|
"loss": 0.8692, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.988605218136711e-05, |
|
"loss": 0.8274, |
|
"step": 860 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.9883385781063876e-05, |
|
"loss": 0.8502, |
|
"step": 870 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.9880688616300975e-05, |
|
"loss": 0.8445, |
|
"step": 880 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.9877960690413035e-05, |
|
"loss": 0.8475, |
|
"step": 890 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.987520200677271e-05, |
|
"loss": 0.8215, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.987241256879071e-05, |
|
"loss": 0.8389, |
|
"step": 910 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.986959237991571e-05, |
|
"loss": 0.8422, |
|
"step": 920 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.9866741443634455e-05, |
|
"loss": 0.8287, |
|
"step": 930 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.986385976347169e-05, |
|
"loss": 0.8694, |
|
"step": 940 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.986094734299016e-05, |
|
"loss": 0.847, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.985800418579063e-05, |
|
"loss": 0.8191, |
|
"step": 960 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.985503029551184e-05, |
|
"loss": 0.8419, |
|
"step": 970 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.985202567583057e-05, |
|
"loss": 0.8517, |
|
"step": 980 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.984899033046155e-05, |
|
"loss": 0.8653, |
|
"step": 990 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.9845924263157526e-05, |
|
"loss": 0.8349, |
|
"step": 1000 |
|
} |
|
], |
|
"max_steps": 28254, |
|
"num_train_epochs": 2, |
|
"total_flos": 2.4351950380204032e+17, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|