| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 2036, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004911591355599214, | |
| "grad_norm": 4.0625, | |
| "learning_rate": 4.977897838899804e-05, | |
| "loss": 2.9953, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.009823182711198428, | |
| "grad_norm": 1.6328125, | |
| "learning_rate": 4.953339882121808e-05, | |
| "loss": 0.0935, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.014734774066797643, | |
| "grad_norm": 1.578125, | |
| "learning_rate": 4.928781925343812e-05, | |
| "loss": 0.073, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.019646365422396856, | |
| "grad_norm": 1.2890625, | |
| "learning_rate": 4.904223968565815e-05, | |
| "loss": 0.0679, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.02455795677799607, | |
| "grad_norm": 1.5078125, | |
| "learning_rate": 4.87966601178782e-05, | |
| "loss": 0.055, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.029469548133595286, | |
| "grad_norm": 1.234375, | |
| "learning_rate": 4.855108055009823e-05, | |
| "loss": 0.0542, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0343811394891945, | |
| "grad_norm": 0.984375, | |
| "learning_rate": 4.830550098231827e-05, | |
| "loss": 0.0458, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.03929273084479371, | |
| "grad_norm": 0.95703125, | |
| "learning_rate": 4.805992141453831e-05, | |
| "loss": 0.0485, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.04420432220039293, | |
| "grad_norm": 1.21875, | |
| "learning_rate": 4.781434184675835e-05, | |
| "loss": 0.0563, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.04911591355599214, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 4.756876227897839e-05, | |
| "loss": 0.0525, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.054027504911591355, | |
| "grad_norm": 1.09375, | |
| "learning_rate": 4.732318271119843e-05, | |
| "loss": 0.0501, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.05893909626719057, | |
| "grad_norm": 0.94921875, | |
| "learning_rate": 4.7077603143418466e-05, | |
| "loss": 0.0466, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.06385068762278978, | |
| "grad_norm": 0.90625, | |
| "learning_rate": 4.683202357563851e-05, | |
| "loss": 0.0472, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.068762278978389, | |
| "grad_norm": 1.1015625, | |
| "learning_rate": 4.658644400785855e-05, | |
| "loss": 0.0432, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.07367387033398821, | |
| "grad_norm": 1.1015625, | |
| "learning_rate": 4.634086444007859e-05, | |
| "loss": 0.0497, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.07858546168958742, | |
| "grad_norm": 1.3125, | |
| "learning_rate": 4.609528487229863e-05, | |
| "loss": 0.0406, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.08349705304518663, | |
| "grad_norm": 1.1015625, | |
| "learning_rate": 4.584970530451866e-05, | |
| "loss": 0.048, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.08840864440078586, | |
| "grad_norm": 0.875, | |
| "learning_rate": 4.560412573673871e-05, | |
| "loss": 0.0437, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.09332023575638507, | |
| "grad_norm": 0.8515625, | |
| "learning_rate": 4.535854616895874e-05, | |
| "loss": 0.0478, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.09823182711198428, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 4.511296660117879e-05, | |
| "loss": 0.0373, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1031434184675835, | |
| "grad_norm": 0.6171875, | |
| "learning_rate": 4.486738703339882e-05, | |
| "loss": 0.0446, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.10805500982318271, | |
| "grad_norm": 0.81640625, | |
| "learning_rate": 4.462180746561886e-05, | |
| "loss": 0.0414, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.11296660117878192, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 4.43762278978389e-05, | |
| "loss": 0.0429, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.11787819253438114, | |
| "grad_norm": 1.046875, | |
| "learning_rate": 4.413064833005894e-05, | |
| "loss": 0.0386, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.12278978388998035, | |
| "grad_norm": 1.015625, | |
| "learning_rate": 4.388506876227898e-05, | |
| "loss": 0.0404, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.12770137524557956, | |
| "grad_norm": 1.0390625, | |
| "learning_rate": 4.3639489194499023e-05, | |
| "loss": 0.037, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.13261296660117877, | |
| "grad_norm": 0.9140625, | |
| "learning_rate": 4.339390962671906e-05, | |
| "loss": 0.0389, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.137524557956778, | |
| "grad_norm": 0.97265625, | |
| "learning_rate": 4.31483300589391e-05, | |
| "loss": 0.0416, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.14243614931237722, | |
| "grad_norm": 1.7421875, | |
| "learning_rate": 4.290275049115914e-05, | |
| "loss": 0.0377, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.14734774066797643, | |
| "grad_norm": 0.6015625, | |
| "learning_rate": 4.265717092337918e-05, | |
| "loss": 0.0327, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.15225933202357564, | |
| "grad_norm": 1.015625, | |
| "learning_rate": 4.241159135559922e-05, | |
| "loss": 0.0377, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.15717092337917485, | |
| "grad_norm": 0.87109375, | |
| "learning_rate": 4.216601178781925e-05, | |
| "loss": 0.038, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.16208251473477406, | |
| "grad_norm": 0.65625, | |
| "learning_rate": 4.19204322200393e-05, | |
| "loss": 0.0381, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.16699410609037327, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 4.167485265225933e-05, | |
| "loss": 0.0331, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.1719056974459725, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 4.142927308447937e-05, | |
| "loss": 0.0374, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.17681728880157171, | |
| "grad_norm": 0.90234375, | |
| "learning_rate": 4.118369351669941e-05, | |
| "loss": 0.0389, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.18172888015717092, | |
| "grad_norm": 1.1015625, | |
| "learning_rate": 4.093811394891945e-05, | |
| "loss": 0.0443, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.18664047151277013, | |
| "grad_norm": 0.6484375, | |
| "learning_rate": 4.069253438113949e-05, | |
| "loss": 0.0374, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.19155206286836934, | |
| "grad_norm": 0.47265625, | |
| "learning_rate": 4.044695481335953e-05, | |
| "loss": 0.0232, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.19646365422396855, | |
| "grad_norm": 0.92578125, | |
| "learning_rate": 4.020137524557957e-05, | |
| "loss": 0.043, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.2013752455795678, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 3.995579567779961e-05, | |
| "loss": 0.0344, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.206286836935167, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 3.971021611001965e-05, | |
| "loss": 0.0271, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.2111984282907662, | |
| "grad_norm": 0.953125, | |
| "learning_rate": 3.946463654223969e-05, | |
| "loss": 0.0378, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.21611001964636542, | |
| "grad_norm": 1.0, | |
| "learning_rate": 3.921905697445973e-05, | |
| "loss": 0.0356, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.22102161100196463, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 3.897347740667976e-05, | |
| "loss": 0.0336, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.22593320235756384, | |
| "grad_norm": 0.89453125, | |
| "learning_rate": 3.872789783889981e-05, | |
| "loss": 0.0334, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.23084479371316308, | |
| "grad_norm": 0.5390625, | |
| "learning_rate": 3.848231827111984e-05, | |
| "loss": 0.0316, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.2357563850687623, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 3.823673870333989e-05, | |
| "loss": 0.037, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.2406679764243615, | |
| "grad_norm": 0.65234375, | |
| "learning_rate": 3.799115913555992e-05, | |
| "loss": 0.0336, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.2455795677799607, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 3.774557956777996e-05, | |
| "loss": 0.0316, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.2504911591355599, | |
| "grad_norm": 0.89453125, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "loss": 0.0287, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.2554027504911591, | |
| "grad_norm": 0.58984375, | |
| "learning_rate": 3.725442043222004e-05, | |
| "loss": 0.0353, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.26031434184675833, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 3.7008840864440084e-05, | |
| "loss": 0.0323, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.26522593320235754, | |
| "grad_norm": 0.640625, | |
| "learning_rate": 3.676326129666012e-05, | |
| "loss": 0.0372, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.27013752455795675, | |
| "grad_norm": 0.8671875, | |
| "learning_rate": 3.651768172888016e-05, | |
| "loss": 0.0267, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.275049115913556, | |
| "grad_norm": 1.0859375, | |
| "learning_rate": 3.62721021611002e-05, | |
| "loss": 0.0304, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.27996070726915523, | |
| "grad_norm": 0.5859375, | |
| "learning_rate": 3.602652259332024e-05, | |
| "loss": 0.0326, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.28487229862475444, | |
| "grad_norm": 0.62890625, | |
| "learning_rate": 3.578094302554028e-05, | |
| "loss": 0.0403, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.28978388998035365, | |
| "grad_norm": 0.8515625, | |
| "learning_rate": 3.553536345776032e-05, | |
| "loss": 0.0306, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.29469548133595286, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 3.528978388998035e-05, | |
| "loss": 0.028, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.29960707269155207, | |
| "grad_norm": 0.578125, | |
| "learning_rate": 3.50442043222004e-05, | |
| "loss": 0.029, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.3045186640471513, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 3.479862475442043e-05, | |
| "loss": 0.0299, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.3094302554027505, | |
| "grad_norm": 0.94140625, | |
| "learning_rate": 3.455304518664047e-05, | |
| "loss": 0.0293, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.3143418467583497, | |
| "grad_norm": 0.84375, | |
| "learning_rate": 3.4307465618860514e-05, | |
| "loss": 0.0288, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.3192534381139489, | |
| "grad_norm": 0.462890625, | |
| "learning_rate": 3.406188605108055e-05, | |
| "loss": 0.0272, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.3241650294695481, | |
| "grad_norm": 1.484375, | |
| "learning_rate": 3.3816306483300594e-05, | |
| "loss": 0.0392, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.3290766208251473, | |
| "grad_norm": 0.6640625, | |
| "learning_rate": 3.357072691552063e-05, | |
| "loss": 0.032, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.33398821218074654, | |
| "grad_norm": 1.109375, | |
| "learning_rate": 3.332514734774067e-05, | |
| "loss": 0.0388, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.3388998035363458, | |
| "grad_norm": 1.390625, | |
| "learning_rate": 3.307956777996071e-05, | |
| "loss": 0.03, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.343811394891945, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 3.283398821218075e-05, | |
| "loss": 0.0316, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.3487229862475442, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 3.258840864440079e-05, | |
| "loss": 0.037, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.35363457760314343, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 3.234282907662083e-05, | |
| "loss": 0.0302, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.35854616895874264, | |
| "grad_norm": 0.5546875, | |
| "learning_rate": 3.209724950884086e-05, | |
| "loss": 0.0294, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.36345776031434185, | |
| "grad_norm": 0.60546875, | |
| "learning_rate": 3.185166994106091e-05, | |
| "loss": 0.0303, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.36836935166994106, | |
| "grad_norm": 1.125, | |
| "learning_rate": 3.160609037328094e-05, | |
| "loss": 0.0304, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.37328094302554027, | |
| "grad_norm": 0.609375, | |
| "learning_rate": 3.1360510805500984e-05, | |
| "loss": 0.0303, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.3781925343811395, | |
| "grad_norm": 0.9609375, | |
| "learning_rate": 3.1114931237721024e-05, | |
| "loss": 0.0323, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.3831041257367387, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 3.086935166994106e-05, | |
| "loss": 0.0324, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.3880157170923379, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 3.0623772102161104e-05, | |
| "loss": 0.0318, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.3929273084479371, | |
| "grad_norm": 0.6484375, | |
| "learning_rate": 3.0378192534381138e-05, | |
| "loss": 0.0295, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.39783889980353637, | |
| "grad_norm": 0.6484375, | |
| "learning_rate": 3.013261296660118e-05, | |
| "loss": 0.0267, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.4027504911591356, | |
| "grad_norm": 1.0234375, | |
| "learning_rate": 2.988703339882122e-05, | |
| "loss": 0.0324, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.4076620825147348, | |
| "grad_norm": 0.81640625, | |
| "learning_rate": 2.964145383104126e-05, | |
| "loss": 0.027, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.412573673870334, | |
| "grad_norm": 0.609375, | |
| "learning_rate": 2.9395874263261296e-05, | |
| "loss": 0.0294, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.4174852652259332, | |
| "grad_norm": 0.625, | |
| "learning_rate": 2.915029469548134e-05, | |
| "loss": 0.0265, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.4223968565815324, | |
| "grad_norm": 0.6640625, | |
| "learning_rate": 2.8904715127701376e-05, | |
| "loss": 0.027, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.42730844793713163, | |
| "grad_norm": 0.4921875, | |
| "learning_rate": 2.865913555992142e-05, | |
| "loss": 0.0308, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.43222003929273084, | |
| "grad_norm": 0.57421875, | |
| "learning_rate": 2.8413555992141457e-05, | |
| "loss": 0.0312, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.43713163064833005, | |
| "grad_norm": 0.8984375, | |
| "learning_rate": 2.816797642436149e-05, | |
| "loss": 0.0297, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.44204322200392926, | |
| "grad_norm": 0.5703125, | |
| "learning_rate": 2.7922396856581534e-05, | |
| "loss": 0.0308, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.44695481335952847, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 2.767681728880157e-05, | |
| "loss": 0.0307, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.4518664047151277, | |
| "grad_norm": 0.6015625, | |
| "learning_rate": 2.7431237721021615e-05, | |
| "loss": 0.031, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.4567779960707269, | |
| "grad_norm": 0.609375, | |
| "learning_rate": 2.718565815324165e-05, | |
| "loss": 0.0388, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.46168958742632615, | |
| "grad_norm": 0.578125, | |
| "learning_rate": 2.6940078585461692e-05, | |
| "loss": 0.0293, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.46660117878192536, | |
| "grad_norm": 0.90234375, | |
| "learning_rate": 2.669449901768173e-05, | |
| "loss": 0.0339, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.4715127701375246, | |
| "grad_norm": 0.5625, | |
| "learning_rate": 2.6448919449901772e-05, | |
| "loss": 0.0322, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.4764243614931238, | |
| "grad_norm": 0.5234375, | |
| "learning_rate": 2.620333988212181e-05, | |
| "loss": 0.0316, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.481335952848723, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 2.595776031434185e-05, | |
| "loss": 0.0329, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.4862475442043222, | |
| "grad_norm": 0.58984375, | |
| "learning_rate": 2.5712180746561886e-05, | |
| "loss": 0.0251, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.4911591355599214, | |
| "grad_norm": 0.9921875, | |
| "learning_rate": 2.546660117878193e-05, | |
| "loss": 0.0256, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.4960707269155206, | |
| "grad_norm": 1.1015625, | |
| "learning_rate": 2.5221021611001967e-05, | |
| "loss": 0.0294, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.5009823182711198, | |
| "grad_norm": 0.6484375, | |
| "learning_rate": 2.4975442043222004e-05, | |
| "loss": 0.0251, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.5058939096267191, | |
| "grad_norm": 0.55859375, | |
| "learning_rate": 2.4729862475442044e-05, | |
| "loss": 0.0275, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.5108055009823183, | |
| "grad_norm": 0.5546875, | |
| "learning_rate": 2.4484282907662084e-05, | |
| "loss": 0.0261, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.5157170923379175, | |
| "grad_norm": 0.72265625, | |
| "learning_rate": 2.4238703339882125e-05, | |
| "loss": 0.0292, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.5206286836935167, | |
| "grad_norm": 0.640625, | |
| "learning_rate": 2.3993123772102165e-05, | |
| "loss": 0.0252, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.5255402750491159, | |
| "grad_norm": 0.671875, | |
| "learning_rate": 2.37475442043222e-05, | |
| "loss": 0.0305, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.5304518664047151, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 2.350196463654224e-05, | |
| "loss": 0.029, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.5353634577603144, | |
| "grad_norm": 0.62109375, | |
| "learning_rate": 2.325638506876228e-05, | |
| "loss": 0.0306, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.5402750491159135, | |
| "grad_norm": 0.5078125, | |
| "learning_rate": 2.301080550098232e-05, | |
| "loss": 0.0295, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.5451866404715128, | |
| "grad_norm": 0.55859375, | |
| "learning_rate": 2.276522593320236e-05, | |
| "loss": 0.0244, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.550098231827112, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 2.2519646365422397e-05, | |
| "loss": 0.0261, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.5550098231827112, | |
| "grad_norm": 0.59375, | |
| "learning_rate": 2.2274066797642437e-05, | |
| "loss": 0.0227, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.5599214145383105, | |
| "grad_norm": 0.44140625, | |
| "learning_rate": 2.2028487229862477e-05, | |
| "loss": 0.0244, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.5648330058939096, | |
| "grad_norm": 0.63671875, | |
| "learning_rate": 2.1782907662082517e-05, | |
| "loss": 0.0235, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.5697445972495089, | |
| "grad_norm": 0.60546875, | |
| "learning_rate": 2.1537328094302554e-05, | |
| "loss": 0.0297, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.574656188605108, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 2.1291748526522595e-05, | |
| "loss": 0.0267, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.5795677799607073, | |
| "grad_norm": 1.03125, | |
| "learning_rate": 2.1046168958742635e-05, | |
| "loss": 0.0276, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.5844793713163065, | |
| "grad_norm": 0.5390625, | |
| "learning_rate": 2.0800589390962675e-05, | |
| "loss": 0.0281, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.5893909626719057, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 2.0555009823182712e-05, | |
| "loss": 0.0306, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.5943025540275049, | |
| "grad_norm": 0.46875, | |
| "learning_rate": 2.030943025540275e-05, | |
| "loss": 0.0253, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.5992141453831041, | |
| "grad_norm": 0.625, | |
| "learning_rate": 2.006385068762279e-05, | |
| "loss": 0.0258, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.6041257367387033, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 1.981827111984283e-05, | |
| "loss": 0.0275, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.6090373280943026, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 1.957269155206287e-05, | |
| "loss": 0.0275, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.6139489194499018, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 1.932711198428291e-05, | |
| "loss": 0.0304, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.618860510805501, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 1.9081532416502947e-05, | |
| "loss": 0.0234, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.6237721021611002, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 1.8835952848722987e-05, | |
| "loss": 0.0315, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.6286836935166994, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 1.8590373280943028e-05, | |
| "loss": 0.0293, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.6335952848722987, | |
| "grad_norm": 0.4609375, | |
| "learning_rate": 1.8344793713163068e-05, | |
| "loss": 0.0261, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.6385068762278978, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 1.8099214145383105e-05, | |
| "loss": 0.0264, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.6434184675834971, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 1.7853634577603145e-05, | |
| "loss": 0.0283, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.6483300589390962, | |
| "grad_norm": 1.0625, | |
| "learning_rate": 1.7608055009823182e-05, | |
| "loss": 0.0244, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.6532416502946955, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 1.7362475442043222e-05, | |
| "loss": 0.0249, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.6581532416502947, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 1.7116895874263263e-05, | |
| "loss": 0.0246, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.6630648330058939, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 1.68713163064833e-05, | |
| "loss": 0.0268, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.6679764243614931, | |
| "grad_norm": 0.5390625, | |
| "learning_rate": 1.662573673870334e-05, | |
| "loss": 0.025, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.6728880157170923, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 1.638015717092338e-05, | |
| "loss": 0.0209, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.6777996070726916, | |
| "grad_norm": 0.5546875, | |
| "learning_rate": 1.613457760314342e-05, | |
| "loss": 0.0264, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.6827111984282908, | |
| "grad_norm": 1.0703125, | |
| "learning_rate": 1.5888998035363457e-05, | |
| "loss": 0.0227, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.68762278978389, | |
| "grad_norm": 0.66015625, | |
| "learning_rate": 1.5643418467583497e-05, | |
| "loss": 0.0313, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.6925343811394892, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 1.5397838899803538e-05, | |
| "loss": 0.0242, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.6974459724950884, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 1.5152259332023578e-05, | |
| "loss": 0.0274, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.7023575638506876, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 1.4906679764243617e-05, | |
| "loss": 0.0235, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.7072691552062869, | |
| "grad_norm": 0.66015625, | |
| "learning_rate": 1.4661100196463657e-05, | |
| "loss": 0.0259, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.712180746561886, | |
| "grad_norm": 0.453125, | |
| "learning_rate": 1.4415520628683694e-05, | |
| "loss": 0.0246, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.7170923379174853, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 1.4169941060903732e-05, | |
| "loss": 0.0275, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.7220039292730844, | |
| "grad_norm": 1.1015625, | |
| "learning_rate": 1.3924361493123773e-05, | |
| "loss": 0.0308, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.7269155206286837, | |
| "grad_norm": 0.9765625, | |
| "learning_rate": 1.3678781925343811e-05, | |
| "loss": 0.029, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.731827111984283, | |
| "grad_norm": 1.7578125, | |
| "learning_rate": 1.3433202357563852e-05, | |
| "loss": 0.0284, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.7367387033398821, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 1.318762278978389e-05, | |
| "loss": 0.0253, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.7416502946954814, | |
| "grad_norm": 0.8515625, | |
| "learning_rate": 1.294204322200393e-05, | |
| "loss": 0.028, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.7465618860510805, | |
| "grad_norm": 0.53515625, | |
| "learning_rate": 1.2696463654223969e-05, | |
| "loss": 0.0292, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.7514734774066798, | |
| "grad_norm": 0.62890625, | |
| "learning_rate": 1.245088408644401e-05, | |
| "loss": 0.0259, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.756385068762279, | |
| "grad_norm": 0.6171875, | |
| "learning_rate": 1.2205304518664048e-05, | |
| "loss": 0.0263, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.7612966601178782, | |
| "grad_norm": 0.72265625, | |
| "learning_rate": 1.1959724950884087e-05, | |
| "loss": 0.0279, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.7662082514734774, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 1.1714145383104127e-05, | |
| "loss": 0.0276, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.7711198428290766, | |
| "grad_norm": 0.98828125, | |
| "learning_rate": 1.1468565815324165e-05, | |
| "loss": 0.0314, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.7760314341846758, | |
| "grad_norm": 0.6640625, | |
| "learning_rate": 1.1222986247544206e-05, | |
| "loss": 0.0283, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.7809430255402751, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 1.0977406679764244e-05, | |
| "loss": 0.049, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.7858546168958742, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 1.0731827111984283e-05, | |
| "loss": 0.0294, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.7907662082514735, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 1.0486247544204323e-05, | |
| "loss": 0.0257, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.7956777996070727, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 1.0240667976424362e-05, | |
| "loss": 0.0316, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.8005893909626719, | |
| "grad_norm": 1.0234375, | |
| "learning_rate": 9.995088408644402e-06, | |
| "loss": 0.033, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.8055009823182712, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 9.74950884086444e-06, | |
| "loss": 0.0263, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.8104125736738703, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 9.503929273084481e-06, | |
| "loss": 0.0248, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.8153241650294696, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 9.258349705304518e-06, | |
| "loss": 0.0256, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.8202357563850687, | |
| "grad_norm": 0.498046875, | |
| "learning_rate": 9.012770137524558e-06, | |
| "loss": 0.0258, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.825147347740668, | |
| "grad_norm": 0.64453125, | |
| "learning_rate": 8.767190569744597e-06, | |
| "loss": 0.0271, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.8300589390962672, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 8.521611001964637e-06, | |
| "loss": 0.0334, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.8349705304518664, | |
| "grad_norm": 1.0078125, | |
| "learning_rate": 8.276031434184677e-06, | |
| "loss": 0.0245, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.8398821218074656, | |
| "grad_norm": 0.63671875, | |
| "learning_rate": 8.030451866404716e-06, | |
| "loss": 0.0262, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.8447937131630648, | |
| "grad_norm": 0.5390625, | |
| "learning_rate": 7.784872298624756e-06, | |
| "loss": 0.027, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.849705304518664, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 7.539292730844794e-06, | |
| "loss": 0.0253, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.8546168958742633, | |
| "grad_norm": 0.62109375, | |
| "learning_rate": 7.293713163064833e-06, | |
| "loss": 0.0247, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.8595284872298625, | |
| "grad_norm": 1.140625, | |
| "learning_rate": 7.048133595284873e-06, | |
| "loss": 0.0281, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.8644400785854617, | |
| "grad_norm": 0.59375, | |
| "learning_rate": 6.802554027504912e-06, | |
| "loss": 0.0269, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.869351669941061, | |
| "grad_norm": 1.2421875, | |
| "learning_rate": 6.556974459724952e-06, | |
| "loss": 0.0255, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.8742632612966601, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 6.311394891944991e-06, | |
| "loss": 0.0253, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.8791748526522594, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 6.06581532416503e-06, | |
| "loss": 0.024, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.8840864440078585, | |
| "grad_norm": 0.60546875, | |
| "learning_rate": 5.820235756385069e-06, | |
| "loss": 0.03, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.8889980353634578, | |
| "grad_norm": 0.5546875, | |
| "learning_rate": 5.5746561886051085e-06, | |
| "loss": 0.0246, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.8939096267190569, | |
| "grad_norm": 0.8671875, | |
| "learning_rate": 5.329076620825148e-06, | |
| "loss": 0.0249, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.8988212180746562, | |
| "grad_norm": 0.55859375, | |
| "learning_rate": 5.0834970530451866e-06, | |
| "loss": 0.0267, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.9037328094302554, | |
| "grad_norm": 0.5703125, | |
| "learning_rate": 4.837917485265226e-06, | |
| "loss": 0.0232, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.9086444007858546, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 4.5923379174852655e-06, | |
| "loss": 0.0242, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.9135559921414538, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 4.346758349705305e-06, | |
| "loss": 0.0256, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.918467583497053, | |
| "grad_norm": 1.125, | |
| "learning_rate": 4.1011787819253435e-06, | |
| "loss": 0.0318, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.9233791748526523, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 3.855599214145384e-06, | |
| "loss": 0.0268, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.9282907662082515, | |
| "grad_norm": 0.58203125, | |
| "learning_rate": 3.6100196463654228e-06, | |
| "loss": 0.0249, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.9332023575638507, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 3.364440078585462e-06, | |
| "loss": 0.0306, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.9381139489194499, | |
| "grad_norm": 0.59765625, | |
| "learning_rate": 3.1188605108055012e-06, | |
| "loss": 0.0301, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.9430255402750491, | |
| "grad_norm": 0.65625, | |
| "learning_rate": 2.8732809430255403e-06, | |
| "loss": 0.0248, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.9479371316306483, | |
| "grad_norm": 0.76953125, | |
| "learning_rate": 2.6277013752455797e-06, | |
| "loss": 0.0267, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.9528487229862476, | |
| "grad_norm": 0.61328125, | |
| "learning_rate": 2.382121807465619e-06, | |
| "loss": 0.0296, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.9577603143418467, | |
| "grad_norm": 0.59375, | |
| "learning_rate": 2.136542239685658e-06, | |
| "loss": 0.0264, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.962671905697446, | |
| "grad_norm": 0.51953125, | |
| "learning_rate": 1.8909626719056976e-06, | |
| "loss": 0.024, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.9675834970530451, | |
| "grad_norm": 0.56640625, | |
| "learning_rate": 1.6453831041257368e-06, | |
| "loss": 0.0315, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.9724950884086444, | |
| "grad_norm": 0.859375, | |
| "learning_rate": 1.3998035363457763e-06, | |
| "loss": 0.024, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.9774066797642437, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 1.1542239685658153e-06, | |
| "loss": 0.0282, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.9823182711198428, | |
| "grad_norm": 0.6328125, | |
| "learning_rate": 9.086444007858547e-07, | |
| "loss": 0.0231, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.9872298624754421, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 6.630648330058939e-07, | |
| "loss": 0.0247, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.9921414538310412, | |
| "grad_norm": 0.5546875, | |
| "learning_rate": 4.174852652259332e-07, | |
| "loss": 0.0233, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.9970530451866405, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 1.719056974459725e-07, | |
| "loss": 0.0366, | |
| "step": 2030 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2036, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.9084808495824896e+16, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |