| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 20.0, | |
| "eval_steps": 1000, | |
| "global_step": 4320, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.046296296296296294, | |
| "grad_norm": 9.519277572631836, | |
| "learning_rate": 1.8254252421884168e-05, | |
| "loss": 2.9957, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.09259259259259259, | |
| "grad_norm": 8.448235511779785, | |
| "learning_rate": 2.3749329949293174e-05, | |
| "loss": 0.7782, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.1388888888888889, | |
| "grad_norm": 5.370626449584961, | |
| "learning_rate": 2.6963744241382978e-05, | |
| "loss": 0.6191, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.18518518518518517, | |
| "grad_norm": 5.509120464324951, | |
| "learning_rate": 2.924440747670218e-05, | |
| "loss": 0.6412, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.23148148148148148, | |
| "grad_norm": 4.508421897888184, | |
| "learning_rate": 3e-05, | |
| "loss": 0.528, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2777777777777778, | |
| "grad_norm": 3.5512583255767822, | |
| "learning_rate": 3e-05, | |
| "loss": 0.4154, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.32407407407407407, | |
| "grad_norm": 4.571855545043945, | |
| "learning_rate": 3e-05, | |
| "loss": 0.4413, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.37037037037037035, | |
| "grad_norm": 3.775933265686035, | |
| "learning_rate": 3e-05, | |
| "loss": 0.5107, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.4166666666666667, | |
| "grad_norm": 6.0832414627075195, | |
| "learning_rate": 3e-05, | |
| "loss": 0.4848, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.46296296296296297, | |
| "grad_norm": 5.910586357116699, | |
| "learning_rate": 3e-05, | |
| "loss": 0.5073, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.5092592592592593, | |
| "grad_norm": 3.3788774013519287, | |
| "learning_rate": 3e-05, | |
| "loss": 0.383, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.5555555555555556, | |
| "grad_norm": 2.238117218017578, | |
| "learning_rate": 3e-05, | |
| "loss": 0.4012, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.6018518518518519, | |
| "grad_norm": 3.789665937423706, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3533, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.6481481481481481, | |
| "grad_norm": 3.239994764328003, | |
| "learning_rate": 3e-05, | |
| "loss": 0.4858, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.6944444444444444, | |
| "grad_norm": 3.5634827613830566, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3845, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.7407407407407407, | |
| "grad_norm": 4.085621356964111, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3996, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.7870370370370371, | |
| "grad_norm": 2.4866838455200195, | |
| "learning_rate": 3e-05, | |
| "loss": 0.4485, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 2.403611421585083, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3869, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.8796296296296297, | |
| "grad_norm": 24.364118576049805, | |
| "learning_rate": 3e-05, | |
| "loss": 0.4605, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.9259259259259259, | |
| "grad_norm": 3.141599655151367, | |
| "learning_rate": 3e-05, | |
| "loss": 0.4298, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.9722222222222222, | |
| "grad_norm": 2.0329315662384033, | |
| "learning_rate": 3e-05, | |
| "loss": 0.4584, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.0185185185185186, | |
| "grad_norm": 2.6605312824249268, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3293, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.0648148148148149, | |
| "grad_norm": 2.2286181449890137, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3676, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.1111111111111112, | |
| "grad_norm": 1.9300888776779175, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3423, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.1574074074074074, | |
| "grad_norm": 2.018017053604126, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3378, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.2037037037037037, | |
| "grad_norm": 2.760201930999756, | |
| "learning_rate": 3e-05, | |
| "loss": 0.334, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 3.0350053310394287, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3699, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.2962962962962963, | |
| "grad_norm": 2.3771510124206543, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3583, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.3425925925925926, | |
| "grad_norm": 3.1241819858551025, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3588, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.3888888888888888, | |
| "grad_norm": 1.825257658958435, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3715, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.4351851851851851, | |
| "grad_norm": 2.6988837718963623, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3707, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.4814814814814814, | |
| "grad_norm": 4.2009406089782715, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3721, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.5277777777777777, | |
| "grad_norm": 3.3356688022613525, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3076, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.574074074074074, | |
| "grad_norm": 2.5690464973449707, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3025, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.6203703703703702, | |
| "grad_norm": 1.67151939868927, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3098, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "grad_norm": 1.9243090152740479, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3651, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.7129629629629628, | |
| "grad_norm": 3.038501739501953, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3634, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.7592592592592593, | |
| "grad_norm": 2.4423258304595947, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3351, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.8055555555555556, | |
| "grad_norm": 3.9061005115509033, | |
| "learning_rate": 3e-05, | |
| "loss": 0.356, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.8518518518518519, | |
| "grad_norm": 2.0409352779388428, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3341, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.8981481481481481, | |
| "grad_norm": 1.5015701055526733, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3501, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.9444444444444444, | |
| "grad_norm": 1.3792345523834229, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3143, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.9907407407407407, | |
| "grad_norm": 1.581567406654358, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3473, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 2.037037037037037, | |
| "grad_norm": 1.8894872665405273, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2751, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 2.0833333333333335, | |
| "grad_norm": 2.4780213832855225, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2993, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.1296296296296298, | |
| "grad_norm": 3.4187729358673096, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3286, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 2.175925925925926, | |
| "grad_norm": 1.52755606174469, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2814, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 2.2222222222222223, | |
| "grad_norm": 1.6060500144958496, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2999, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 2.2685185185185186, | |
| "grad_norm": 2.4506280422210693, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2929, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 2.314814814814815, | |
| "grad_norm": 1.7601087093353271, | |
| "learning_rate": 3e-05, | |
| "loss": 0.351, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.361111111111111, | |
| "grad_norm": 1.7349791526794434, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2827, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.4074074074074074, | |
| "grad_norm": 1.8765602111816406, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3244, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.4537037037037037, | |
| "grad_norm": 2.0878164768218994, | |
| "learning_rate": 3e-05, | |
| "loss": 0.323, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 2.7491815090179443, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2458, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 2.5462962962962963, | |
| "grad_norm": 1.950523018836975, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2599, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.5925925925925926, | |
| "grad_norm": 2.113060712814331, | |
| "learning_rate": 3e-05, | |
| "loss": 0.309, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.638888888888889, | |
| "grad_norm": 2.221663475036621, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3166, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 2.685185185185185, | |
| "grad_norm": 2.251364231109619, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2997, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 2.7314814814814814, | |
| "grad_norm": 3.0125315189361572, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2885, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 2.7777777777777777, | |
| "grad_norm": 1.529222846031189, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2885, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.824074074074074, | |
| "grad_norm": 1.5727730989456177, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3328, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 2.8703703703703702, | |
| "grad_norm": 1.7602397203445435, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3089, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.9166666666666665, | |
| "grad_norm": 1.2951428890228271, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2912, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 2.962962962962963, | |
| "grad_norm": 2.3238091468811035, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3007, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 3.009259259259259, | |
| "grad_norm": 1.14413321018219, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2939, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 3.0555555555555554, | |
| "grad_norm": 1.6297560930252075, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2275, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 3.1018518518518516, | |
| "grad_norm": 1.8607372045516968, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2504, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 3.148148148148148, | |
| "grad_norm": 1.9002240896224976, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2693, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 3.1944444444444446, | |
| "grad_norm": 2.417104959487915, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2609, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 3.240740740740741, | |
| "grad_norm": 1.9233345985412598, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2669, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.287037037037037, | |
| "grad_norm": 1.330132246017456, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2776, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 3.3333333333333335, | |
| "grad_norm": 2.0991740226745605, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2365, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 3.3796296296296298, | |
| "grad_norm": 1.7612348794937134, | |
| "learning_rate": 3e-05, | |
| "loss": 0.229, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 3.425925925925926, | |
| "grad_norm": 2.1555356979370117, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3083, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 3.4722222222222223, | |
| "grad_norm": 1.8027970790863037, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2834, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.5185185185185186, | |
| "grad_norm": 1.9787973165512085, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2771, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 3.564814814814815, | |
| "grad_norm": 1.3383827209472656, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3039, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 3.611111111111111, | |
| "grad_norm": 1.6759270429611206, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2637, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 3.6574074074074074, | |
| "grad_norm": 1.937604308128357, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2702, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 3.7037037037037037, | |
| "grad_norm": 1.4416754245758057, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2522, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "grad_norm": 1.7105003595352173, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2751, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 3.7962962962962963, | |
| "grad_norm": 1.843482255935669, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2753, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 3.8425925925925926, | |
| "grad_norm": 1.3820255994796753, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2542, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 3.888888888888889, | |
| "grad_norm": 1.7006720304489136, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2612, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 3.935185185185185, | |
| "grad_norm": 1.5114960670471191, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2849, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.9814814814814814, | |
| "grad_norm": 2.0061888694763184, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2932, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 4.027777777777778, | |
| "grad_norm": 1.8746589422225952, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2782, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 4.074074074074074, | |
| "grad_norm": 3.473623037338257, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2532, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 4.12037037037037, | |
| "grad_norm": 1.4170212745666504, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2259, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 4.166666666666667, | |
| "grad_norm": 24.153331756591797, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2273, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 4.212962962962963, | |
| "grad_norm": 1.2688499689102173, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2204, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 4.2592592592592595, | |
| "grad_norm": 1.7967044115066528, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2248, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 4.305555555555555, | |
| "grad_norm": 1.8120778799057007, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2409, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 4.351851851851852, | |
| "grad_norm": 1.1904025077819824, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2173, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 4.398148148148148, | |
| "grad_norm": 1.1123757362365723, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2418, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 4.444444444444445, | |
| "grad_norm": 1.3130488395690918, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2456, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 4.4907407407407405, | |
| "grad_norm": 2.091806173324585, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2219, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 4.537037037037037, | |
| "grad_norm": 3.246419906616211, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2512, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 4.583333333333333, | |
| "grad_norm": 1.6817961931228638, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2635, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 4.62962962962963, | |
| "grad_norm": 1.7699165344238281, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2489, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.62962962962963, | |
| "eval_loss": 0.39369502663612366, | |
| "eval_runtime": 88.9571, | |
| "eval_samples_per_second": 6.07, | |
| "eval_steps_per_second": 0.304, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.675925925925926, | |
| "grad_norm": 1.4021247625350952, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2433, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 4.722222222222222, | |
| "grad_norm": 1.8957483768463135, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2328, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 4.768518518518518, | |
| "grad_norm": 1.1866583824157715, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2477, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 4.814814814814815, | |
| "grad_norm": 1.4801381826400757, | |
| "learning_rate": 3e-05, | |
| "loss": 0.255, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 4.861111111111111, | |
| "grad_norm": 1.803208351135254, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2645, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 4.907407407407407, | |
| "grad_norm": 1.838924527168274, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2039, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 4.953703703703704, | |
| "grad_norm": 1.457726001739502, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2714, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 1.394857406616211, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2533, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 5.046296296296297, | |
| "grad_norm": 1.6803613901138306, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2275, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 5.092592592592593, | |
| "grad_norm": 0.976388156414032, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2169, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 5.138888888888889, | |
| "grad_norm": 1.692513346672058, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2132, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 5.185185185185185, | |
| "grad_norm": 1.3766963481903076, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1997, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 5.231481481481482, | |
| "grad_norm": 2.006950616836548, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2142, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 5.277777777777778, | |
| "grad_norm": 1.7959928512573242, | |
| "learning_rate": 3e-05, | |
| "loss": 0.213, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 5.324074074074074, | |
| "grad_norm": 1.3566243648529053, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1878, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 5.37037037037037, | |
| "grad_norm": 2.6776769161224365, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2325, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 5.416666666666667, | |
| "grad_norm": 2.4411234855651855, | |
| "learning_rate": 3e-05, | |
| "loss": 0.213, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 5.462962962962963, | |
| "grad_norm": 1.2669930458068848, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2174, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 5.5092592592592595, | |
| "grad_norm": 1.9736443758010864, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2309, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 5.555555555555555, | |
| "grad_norm": 1.8092678785324097, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2403, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 5.601851851851852, | |
| "grad_norm": 1.1341086626052856, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2329, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 5.648148148148148, | |
| "grad_norm": 1.8351995944976807, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2112, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 5.694444444444445, | |
| "grad_norm": 1.6290438175201416, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2387, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 5.7407407407407405, | |
| "grad_norm": 1.9721460342407227, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2191, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 5.787037037037037, | |
| "grad_norm": 1.7176364660263062, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2358, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 5.833333333333333, | |
| "grad_norm": 1.7181181907653809, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2532, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 5.87962962962963, | |
| "grad_norm": 1.7120261192321777, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2199, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 5.925925925925926, | |
| "grad_norm": 1.7863956689834595, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2425, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 5.972222222222222, | |
| "grad_norm": 1.8824771642684937, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2404, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 6.018518518518518, | |
| "grad_norm": 1.471390724182129, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2045, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 6.064814814814815, | |
| "grad_norm": 1.220346212387085, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1946, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 6.111111111111111, | |
| "grad_norm": 1.6119085550308228, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2073, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 6.157407407407407, | |
| "grad_norm": 2.3939287662506104, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1928, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 6.203703703703703, | |
| "grad_norm": 1.7139561176300049, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2011, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "grad_norm": 1.6871051788330078, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2034, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 6.296296296296296, | |
| "grad_norm": 1.7590142488479614, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1869, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 6.342592592592593, | |
| "grad_norm": 0.823594868183136, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 6.388888888888889, | |
| "grad_norm": 2.3577373027801514, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2333, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 6.435185185185185, | |
| "grad_norm": 1.2093870639801025, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2143, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 6.481481481481482, | |
| "grad_norm": 1.7904540300369263, | |
| "learning_rate": 3e-05, | |
| "loss": 0.23, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 6.527777777777778, | |
| "grad_norm": 1.5087484121322632, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2576, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 6.574074074074074, | |
| "grad_norm": 1.7448809146881104, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2294, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 6.62037037037037, | |
| "grad_norm": 1.4423376321792603, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2051, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 6.666666666666667, | |
| "grad_norm": 3.799774646759033, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2014, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 6.712962962962963, | |
| "grad_norm": 2.199467420578003, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2321, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 6.7592592592592595, | |
| "grad_norm": 9.484046936035156, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1942, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 6.805555555555555, | |
| "grad_norm": 1.1799571514129639, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2162, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 6.851851851851852, | |
| "grad_norm": 1.902580738067627, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2172, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 6.898148148148148, | |
| "grad_norm": 1.4826103448867798, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2115, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 6.944444444444445, | |
| "grad_norm": 1.7186169624328613, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2578, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 6.9907407407407405, | |
| "grad_norm": 1.441232442855835, | |
| "learning_rate": 3e-05, | |
| "loss": 0.215, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 7.037037037037037, | |
| "grad_norm": 1.0220807790756226, | |
| "learning_rate": 3e-05, | |
| "loss": 0.181, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 7.083333333333333, | |
| "grad_norm": 1.1033027172088623, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1875, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 7.12962962962963, | |
| "grad_norm": 1.4569389820098877, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2119, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 7.175925925925926, | |
| "grad_norm": 1.2294921875, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1978, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 7.222222222222222, | |
| "grad_norm": 1.2544004917144775, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1752, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 7.268518518518518, | |
| "grad_norm": 1.6726148128509521, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1952, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 7.314814814814815, | |
| "grad_norm": 1.556419849395752, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2052, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 7.361111111111111, | |
| "grad_norm": 1.9425777196884155, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1974, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 7.407407407407407, | |
| "grad_norm": 1.915613055229187, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2083, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 7.453703703703704, | |
| "grad_norm": 2.244108200073242, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2169, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 7.5, | |
| "grad_norm": 1.9481151103973389, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1952, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 7.546296296296296, | |
| "grad_norm": 1.2878506183624268, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2063, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 7.592592592592593, | |
| "grad_norm": 0.863036572933197, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 7.638888888888889, | |
| "grad_norm": 1.7689223289489746, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1879, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 7.685185185185185, | |
| "grad_norm": 1.5072979927062988, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2032, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 7.731481481481482, | |
| "grad_norm": 1.675126314163208, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2017, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 7.777777777777778, | |
| "grad_norm": 1.5218942165374756, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2051, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 7.824074074074074, | |
| "grad_norm": 1.1887863874435425, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1767, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 7.87037037037037, | |
| "grad_norm": 3.391000270843506, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2019, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 7.916666666666667, | |
| "grad_norm": 1.23732328414917, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2008, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 7.962962962962963, | |
| "grad_norm": 1.2429234981536865, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1795, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 8.00925925925926, | |
| "grad_norm": 1.0785852670669556, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1828, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 8.055555555555555, | |
| "grad_norm": 1.2642838954925537, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1909, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 8.101851851851851, | |
| "grad_norm": 1.251452922821045, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1807, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 8.148148148148149, | |
| "grad_norm": 1.4957802295684814, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1877, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 8.194444444444445, | |
| "grad_norm": 2.1930840015411377, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1785, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 8.24074074074074, | |
| "grad_norm": 1.210434913635254, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1713, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 8.287037037037036, | |
| "grad_norm": 0.8785271048545837, | |
| "learning_rate": 3e-05, | |
| "loss": 0.181, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 8.333333333333334, | |
| "grad_norm": 1.0910536050796509, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1773, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 8.37962962962963, | |
| "grad_norm": 1.1678903102874756, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1847, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 8.425925925925926, | |
| "grad_norm": 1.546592354774475, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1938, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 8.472222222222221, | |
| "grad_norm": 0.8986390829086304, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1872, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 8.518518518518519, | |
| "grad_norm": 1.2651722431182861, | |
| "learning_rate": 3e-05, | |
| "loss": 0.181, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 8.564814814814815, | |
| "grad_norm": 1.5655752420425415, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1932, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 8.61111111111111, | |
| "grad_norm": 1.7608935832977295, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1857, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 8.657407407407408, | |
| "grad_norm": 2.4741933345794678, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1766, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 8.703703703703704, | |
| "grad_norm": 1.4740314483642578, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1755, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 8.75, | |
| "grad_norm": 1.4977920055389404, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1867, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 8.796296296296296, | |
| "grad_norm": 1.9219239950180054, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1881, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 8.842592592592592, | |
| "grad_norm": 1.5815542936325073, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2056, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 8.88888888888889, | |
| "grad_norm": 1.3508894443511963, | |
| "learning_rate": 3e-05, | |
| "loss": 0.192, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 8.935185185185185, | |
| "grad_norm": 1.4464216232299805, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1848, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 8.981481481481481, | |
| "grad_norm": 1.9274829626083374, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2004, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 9.027777777777779, | |
| "grad_norm": 1.2753411531448364, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1787, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 9.074074074074074, | |
| "grad_norm": 1.0816010236740112, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1579, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 9.12037037037037, | |
| "grad_norm": 1.283239483833313, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1713, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 9.166666666666666, | |
| "grad_norm": 1.5639042854309082, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1651, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 9.212962962962964, | |
| "grad_norm": 1.0681788921356201, | |
| "learning_rate": 3e-05, | |
| "loss": 0.163, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 9.25925925925926, | |
| "grad_norm": 1.6595847606658936, | |
| "learning_rate": 3e-05, | |
| "loss": 0.173, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 9.25925925925926, | |
| "eval_loss": 0.4976138770580292, | |
| "eval_runtime": 89.2657, | |
| "eval_samples_per_second": 6.049, | |
| "eval_steps_per_second": 0.302, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 9.305555555555555, | |
| "grad_norm": 1.3802722692489624, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1827, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 9.351851851851851, | |
| "grad_norm": 1.4063881635665894, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1766, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 9.398148148148149, | |
| "grad_norm": 1.460650086402893, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1745, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 9.444444444444445, | |
| "grad_norm": 1.2076504230499268, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1604, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 9.49074074074074, | |
| "grad_norm": 2.070692539215088, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1565, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 9.537037037037036, | |
| "grad_norm": 1.0405620336532593, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1772, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 9.583333333333334, | |
| "grad_norm": 1.3936161994934082, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1644, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 9.62962962962963, | |
| "grad_norm": 1.1049543619155884, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1716, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 9.675925925925926, | |
| "grad_norm": 0.986007034778595, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1733, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 9.722222222222221, | |
| "grad_norm": 0.7544007897377014, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1845, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 9.768518518518519, | |
| "grad_norm": 1.322129249572754, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1555, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 9.814814814814815, | |
| "grad_norm": 1.0886545181274414, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1904, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 9.86111111111111, | |
| "grad_norm": 1.0224428176879883, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1917, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 9.907407407407408, | |
| "grad_norm": 1.9668025970458984, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1887, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 9.953703703703704, | |
| "grad_norm": 1.6308960914611816, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2052, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "grad_norm": 1.2425373792648315, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1955, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 10.046296296296296, | |
| "grad_norm": 1.3138960599899292, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1622, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 10.092592592592593, | |
| "grad_norm": 1.3728431463241577, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1688, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 10.13888888888889, | |
| "grad_norm": 1.1341276168823242, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1711, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 10.185185185185185, | |
| "grad_norm": 1.199234962463379, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1638, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 10.231481481481481, | |
| "grad_norm": 1.1935800313949585, | |
| "learning_rate": 3e-05, | |
| "loss": 0.148, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 10.277777777777779, | |
| "grad_norm": 1.2901479005813599, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1577, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 10.324074074074074, | |
| "grad_norm": 1.572486400604248, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1511, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 10.37037037037037, | |
| "grad_norm": 1.2550723552703857, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1523, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 10.416666666666666, | |
| "grad_norm": 1.5212897062301636, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1701, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 10.462962962962964, | |
| "grad_norm": 1.4204086065292358, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1695, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 10.50925925925926, | |
| "grad_norm": 1.7510631084442139, | |
| "learning_rate": 3e-05, | |
| "loss": 0.179, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 10.555555555555555, | |
| "grad_norm": 1.1196744441986084, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1713, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 10.601851851851851, | |
| "grad_norm": 1.7313737869262695, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1625, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 10.648148148148149, | |
| "grad_norm": 1.1136900186538696, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1584, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 10.694444444444445, | |
| "grad_norm": 0.900171160697937, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1746, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 10.74074074074074, | |
| "grad_norm": 1.5992180109024048, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1766, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 10.787037037037036, | |
| "grad_norm": 1.1199150085449219, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1641, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 10.833333333333334, | |
| "grad_norm": 1.3033232688903809, | |
| "learning_rate": 3e-05, | |
| "loss": 0.164, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 10.87962962962963, | |
| "grad_norm": 1.589557409286499, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1629, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 10.925925925925926, | |
| "grad_norm": 1.715973973274231, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1709, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 10.972222222222221, | |
| "grad_norm": 0.9544859528541565, | |
| "learning_rate": 3e-05, | |
| "loss": 0.169, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 11.018518518518519, | |
| "grad_norm": 2.256932497024536, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1683, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 11.064814814814815, | |
| "grad_norm": 0.7787851691246033, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1428, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 11.11111111111111, | |
| "grad_norm": 0.9551990628242493, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1587, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 11.157407407407407, | |
| "grad_norm": 1.0669728517532349, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1507, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 11.203703703703704, | |
| "grad_norm": 1.5834068059921265, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1479, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 11.25, | |
| "grad_norm": 1.749540090560913, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1605, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 11.296296296296296, | |
| "grad_norm": 1.4417942762374878, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1636, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 11.342592592592593, | |
| "grad_norm": 1.16630220413208, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1717, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 11.38888888888889, | |
| "grad_norm": 1.6200324296951294, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1502, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 11.435185185185185, | |
| "grad_norm": 1.5884885787963867, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1569, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 11.481481481481481, | |
| "grad_norm": 1.0750160217285156, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1741, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 11.527777777777779, | |
| "grad_norm": 1.3603105545043945, | |
| "learning_rate": 3e-05, | |
| "loss": 0.14, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 11.574074074074074, | |
| "grad_norm": 0.7031620740890503, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1589, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 11.62037037037037, | |
| "grad_norm": 0.9268414378166199, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1679, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 11.666666666666666, | |
| "grad_norm": 1.3924790620803833, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1681, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 11.712962962962964, | |
| "grad_norm": 1.586488962173462, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1553, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 11.75925925925926, | |
| "grad_norm": 1.3625974655151367, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1665, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 11.805555555555555, | |
| "grad_norm": 1.6760375499725342, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1826, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 11.851851851851851, | |
| "grad_norm": 1.3086282014846802, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1689, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 11.898148148148149, | |
| "grad_norm": 1.2318929433822632, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1575, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 11.944444444444445, | |
| "grad_norm": 1.7143497467041016, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1632, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 11.99074074074074, | |
| "grad_norm": 1.5415269136428833, | |
| "learning_rate": 3e-05, | |
| "loss": 0.177, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 12.037037037037036, | |
| "grad_norm": 1.0246485471725464, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1289, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 12.083333333333334, | |
| "grad_norm": 0.7156587243080139, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1552, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 12.12962962962963, | |
| "grad_norm": 1.3618501424789429, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1599, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 12.175925925925926, | |
| "grad_norm": 1.4899464845657349, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1421, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 12.222222222222221, | |
| "grad_norm": 0.9834685325622559, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1471, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 12.268518518518519, | |
| "grad_norm": 1.0517306327819824, | |
| "learning_rate": 3e-05, | |
| "loss": 0.164, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 12.314814814814815, | |
| "grad_norm": 1.0001568794250488, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1571, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 12.36111111111111, | |
| "grad_norm": 1.084662914276123, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1585, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 12.407407407407407, | |
| "grad_norm": 1.2706600427627563, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1452, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 12.453703703703704, | |
| "grad_norm": 1.2018481492996216, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1556, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "grad_norm": 1.043424367904663, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1665, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 12.546296296296296, | |
| "grad_norm": 1.4996309280395508, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1445, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 12.592592592592592, | |
| "grad_norm": 0.9113991856575012, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1553, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 12.63888888888889, | |
| "grad_norm": 1.2703651189804077, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1683, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 12.685185185185185, | |
| "grad_norm": 1.1912919282913208, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1526, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 12.731481481481481, | |
| "grad_norm": 1.1190593242645264, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1627, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 12.777777777777779, | |
| "grad_norm": 1.0739121437072754, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1495, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 12.824074074074074, | |
| "grad_norm": 1.53770112991333, | |
| "learning_rate": 3e-05, | |
| "loss": 0.157, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 12.87037037037037, | |
| "grad_norm": 1.3282749652862549, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1593, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 12.916666666666666, | |
| "grad_norm": 1.5690109729766846, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1568, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 12.962962962962964, | |
| "grad_norm": 1.1614137887954712, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1664, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 13.00925925925926, | |
| "grad_norm": 1.3638834953308105, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1443, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 13.055555555555555, | |
| "grad_norm": 1.4731112718582153, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1415, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 13.101851851851851, | |
| "grad_norm": 1.0417250394821167, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1315, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 13.148148148148149, | |
| "grad_norm": 1.570486307144165, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1473, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 13.194444444444445, | |
| "grad_norm": 1.0269405841827393, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1541, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 13.24074074074074, | |
| "grad_norm": 1.0735430717468262, | |
| "learning_rate": 3e-05, | |
| "loss": 0.148, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 13.287037037037036, | |
| "grad_norm": 1.6693620681762695, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1543, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 13.333333333333334, | |
| "grad_norm": 1.2953250408172607, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1495, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 13.37962962962963, | |
| "grad_norm": 1.1142494678497314, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1484, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 13.425925925925926, | |
| "grad_norm": 1.1380842924118042, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1548, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 13.472222222222221, | |
| "grad_norm": 1.7548078298568726, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1468, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 13.518518518518519, | |
| "grad_norm": 0.8444296717643738, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1348, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 13.564814814814815, | |
| "grad_norm": 0.9462345242500305, | |
| "learning_rate": 3e-05, | |
| "loss": 0.156, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 13.61111111111111, | |
| "grad_norm": 1.155483365058899, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1419, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 13.657407407407408, | |
| "grad_norm": 0.9843894243240356, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1418, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 13.703703703703704, | |
| "grad_norm": 0.7480261325836182, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1465, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 13.75, | |
| "grad_norm": 1.5990567207336426, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1447, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 13.796296296296296, | |
| "grad_norm": 1.809402346611023, | |
| "learning_rate": 3e-05, | |
| "loss": 0.175, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 13.842592592592592, | |
| "grad_norm": 1.865115761756897, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1418, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 13.88888888888889, | |
| "grad_norm": 1.2112107276916504, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1715, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 13.88888888888889, | |
| "eval_loss": 0.5447816252708435, | |
| "eval_runtime": 89.1984, | |
| "eval_samples_per_second": 6.054, | |
| "eval_steps_per_second": 0.303, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 13.935185185185185, | |
| "grad_norm": 1.0728169679641724, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1596, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 13.981481481481481, | |
| "grad_norm": 1.8573318719863892, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1665, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 14.027777777777779, | |
| "grad_norm": 1.1071803569793701, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1544, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 14.074074074074074, | |
| "grad_norm": 1.5413600206375122, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1432, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 14.12037037037037, | |
| "grad_norm": 1.1424851417541504, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1366, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 14.166666666666666, | |
| "grad_norm": 0.8150787353515625, | |
| "learning_rate": 3e-05, | |
| "loss": 0.154, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 14.212962962962964, | |
| "grad_norm": 1.217228889465332, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1287, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 14.25925925925926, | |
| "grad_norm": 0.901224672794342, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1401, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 14.305555555555555, | |
| "grad_norm": 1.5435835123062134, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1436, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 14.351851851851851, | |
| "grad_norm": 1.2040764093399048, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1432, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 14.398148148148149, | |
| "grad_norm": 0.8125296235084534, | |
| "learning_rate": 3e-05, | |
| "loss": 0.138, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 14.444444444444445, | |
| "grad_norm": 1.1610842943191528, | |
| "learning_rate": 3e-05, | |
| "loss": 0.145, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 14.49074074074074, | |
| "grad_norm": 1.8344600200653076, | |
| "learning_rate": 3e-05, | |
| "loss": 0.159, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 14.537037037037036, | |
| "grad_norm": 0.9230263829231262, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1482, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 14.583333333333334, | |
| "grad_norm": 1.1558572053909302, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1515, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 14.62962962962963, | |
| "grad_norm": 1.1165474653244019, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1398, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 14.675925925925926, | |
| "grad_norm": 1.8796123266220093, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1477, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 14.722222222222221, | |
| "grad_norm": 1.0877175331115723, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1537, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 14.768518518518519, | |
| "grad_norm": 1.361430048942566, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1467, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 14.814814814814815, | |
| "grad_norm": 0.8188061714172363, | |
| "learning_rate": 3e-05, | |
| "loss": 0.149, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 14.86111111111111, | |
| "grad_norm": 1.4913760423660278, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1586, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 14.907407407407408, | |
| "grad_norm": 0.8685842156410217, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1513, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 14.953703703703704, | |
| "grad_norm": 1.0157376527786255, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1499, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "grad_norm": 1.0665968656539917, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1673, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 15.046296296296296, | |
| "grad_norm": 1.3810080289840698, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1307, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 15.092592592592593, | |
| "grad_norm": 0.9249776601791382, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1389, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 15.13888888888889, | |
| "grad_norm": 1.1835044622421265, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1337, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 15.185185185185185, | |
| "grad_norm": 1.341606616973877, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1596, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 15.231481481481481, | |
| "grad_norm": 1.2246119976043701, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1368, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 15.277777777777779, | |
| "grad_norm": 1.2647781372070312, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1442, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 15.324074074074074, | |
| "grad_norm": 1.158076524734497, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1366, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 15.37037037037037, | |
| "grad_norm": 1.2419984340667725, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1429, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 15.416666666666666, | |
| "grad_norm": 1.4107789993286133, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1565, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 15.462962962962964, | |
| "grad_norm": 1.0196819305419922, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1477, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 15.50925925925926, | |
| "grad_norm": 1.286967158317566, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1418, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 15.555555555555555, | |
| "grad_norm": 0.8533580899238586, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1438, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 15.601851851851851, | |
| "grad_norm": 3.57099986076355, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1628, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 15.648148148148149, | |
| "grad_norm": 0.7320424914360046, | |
| "learning_rate": 3e-05, | |
| "loss": 0.152, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 15.694444444444445, | |
| "grad_norm": 1.4328055381774902, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1394, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 15.74074074074074, | |
| "grad_norm": 1.4889895915985107, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1419, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 15.787037037037036, | |
| "grad_norm": 1.5108295679092407, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1519, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 15.833333333333334, | |
| "grad_norm": 0.8050726652145386, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1447, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 15.87962962962963, | |
| "grad_norm": 0.9805731177330017, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1489, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 15.925925925925926, | |
| "grad_norm": 0.8113260865211487, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1542, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 15.972222222222221, | |
| "grad_norm": 1.0790350437164307, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1551, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 16.01851851851852, | |
| "grad_norm": 0.9748013615608215, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1509, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 16.064814814814813, | |
| "grad_norm": 0.8632538914680481, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1272, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 16.11111111111111, | |
| "grad_norm": 0.9647319912910461, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1283, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 16.15740740740741, | |
| "grad_norm": 1.337175726890564, | |
| "learning_rate": 3e-05, | |
| "loss": 0.133, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 16.203703703703702, | |
| "grad_norm": 1.582036018371582, | |
| "learning_rate": 3e-05, | |
| "loss": 0.139, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 16.25, | |
| "grad_norm": 0.8627964854240417, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1382, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 16.296296296296298, | |
| "grad_norm": 0.9686309099197388, | |
| "learning_rate": 3e-05, | |
| "loss": 0.132, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 16.34259259259259, | |
| "grad_norm": 1.2272250652313232, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1503, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 16.38888888888889, | |
| "grad_norm": 1.213373064994812, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1428, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 16.435185185185187, | |
| "grad_norm": 1.6020861864089966, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1392, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 16.48148148148148, | |
| "grad_norm": 0.9201906323432922, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1421, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 16.52777777777778, | |
| "grad_norm": 0.764667272567749, | |
| "learning_rate": 3e-05, | |
| "loss": 0.14, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 16.574074074074073, | |
| "grad_norm": 0.983479917049408, | |
| "learning_rate": 3e-05, | |
| "loss": 0.141, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 16.62037037037037, | |
| "grad_norm": 1.4715135097503662, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1447, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 16.666666666666668, | |
| "grad_norm": 0.8298927545547485, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1466, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 16.712962962962962, | |
| "grad_norm": 13.994155883789062, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1519, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 16.75925925925926, | |
| "grad_norm": 0.9934417605400085, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1476, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 16.805555555555557, | |
| "grad_norm": 0.7088779211044312, | |
| "learning_rate": 3e-05, | |
| "loss": 0.149, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 16.85185185185185, | |
| "grad_norm": 0.7401979565620422, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1295, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 16.89814814814815, | |
| "grad_norm": 1.1984256505966187, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1532, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 16.944444444444443, | |
| "grad_norm": 0.974200963973999, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1461, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 16.99074074074074, | |
| "grad_norm": 0.9635884761810303, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1406, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 17.037037037037038, | |
| "grad_norm": 1.3428035974502563, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1376, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 17.083333333333332, | |
| "grad_norm": 5.439419269561768, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1282, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 17.12962962962963, | |
| "grad_norm": 0.9634020924568176, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1331, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 17.175925925925927, | |
| "grad_norm": 1.4761830568313599, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1303, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 17.22222222222222, | |
| "grad_norm": 0.9295877814292908, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1395, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 17.26851851851852, | |
| "grad_norm": 0.9482799172401428, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1363, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 17.314814814814813, | |
| "grad_norm": 1.4431120157241821, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1396, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 17.36111111111111, | |
| "grad_norm": 1.708990454673767, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1376, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 17.40740740740741, | |
| "grad_norm": 1.1707541942596436, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1482, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 17.453703703703702, | |
| "grad_norm": 1.2253369092941284, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1424, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 17.5, | |
| "grad_norm": 1.2547239065170288, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1426, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 17.546296296296298, | |
| "grad_norm": 0.9535191655158997, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1467, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 17.59259259259259, | |
| "grad_norm": 1.2108802795410156, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1417, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 17.63888888888889, | |
| "grad_norm": 0.9786614179611206, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1341, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 17.685185185185187, | |
| "grad_norm": 1.202910304069519, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1424, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 17.73148148148148, | |
| "grad_norm": 0.8020798563957214, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1351, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 17.77777777777778, | |
| "grad_norm": 1.3408881425857544, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1423, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 17.824074074074073, | |
| "grad_norm": 1.0835148096084595, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1278, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 17.87037037037037, | |
| "grad_norm": 0.8985652327537537, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1543, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 17.916666666666668, | |
| "grad_norm": 1.2214702367782593, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1471, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 17.962962962962962, | |
| "grad_norm": 0.89570152759552, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1403, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 18.00925925925926, | |
| "grad_norm": 1.1677643060684204, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1481, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 18.055555555555557, | |
| "grad_norm": 1.2765920162200928, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1281, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 18.10185185185185, | |
| "grad_norm": 1.1932995319366455, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1319, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 18.14814814814815, | |
| "grad_norm": 0.8899739384651184, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1315, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 18.194444444444443, | |
| "grad_norm": 0.9039879441261292, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1367, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 18.24074074074074, | |
| "grad_norm": 1.6004914045333862, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1373, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 18.287037037037038, | |
| "grad_norm": 0.8183736205101013, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1353, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 18.333333333333332, | |
| "grad_norm": 0.9930373430252075, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1485, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 18.37962962962963, | |
| "grad_norm": 1.204205870628357, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1436, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 18.425925925925927, | |
| "grad_norm": 0.8955985903739929, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1307, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 18.47222222222222, | |
| "grad_norm": 0.5730966329574585, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1405, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 18.51851851851852, | |
| "grad_norm": 0.8282036781311035, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1352, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 18.51851851851852, | |
| "eval_loss": 0.57914799451828, | |
| "eval_runtime": 89.2393, | |
| "eval_samples_per_second": 6.051, | |
| "eval_steps_per_second": 0.303, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 18.564814814814813, | |
| "grad_norm": 1.324410319328308, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1455, | |
| "step": 4010 | |
| }, | |
| { | |
| "epoch": 18.61111111111111, | |
| "grad_norm": 1.626372218132019, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1249, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 18.65740740740741, | |
| "grad_norm": 0.9152678847312927, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1318, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 18.703703703703702, | |
| "grad_norm": 0.8717418909072876, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1274, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 18.75, | |
| "grad_norm": 1.3345868587493896, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1534, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 18.796296296296298, | |
| "grad_norm": 0.9496395587921143, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1384, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 18.84259259259259, | |
| "grad_norm": 1.159990906715393, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1428, | |
| "step": 4070 | |
| }, | |
| { | |
| "epoch": 18.88888888888889, | |
| "grad_norm": 0.9783802032470703, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1454, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 18.935185185185187, | |
| "grad_norm": 1.1276887655258179, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1342, | |
| "step": 4090 | |
| }, | |
| { | |
| "epoch": 18.98148148148148, | |
| "grad_norm": 1.6755024194717407, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1441, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 19.02777777777778, | |
| "grad_norm": 0.9260687828063965, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1236, | |
| "step": 4110 | |
| }, | |
| { | |
| "epoch": 19.074074074074073, | |
| "grad_norm": 1.0526010990142822, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1209, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 19.12037037037037, | |
| "grad_norm": 1.0216201543807983, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1217, | |
| "step": 4130 | |
| }, | |
| { | |
| "epoch": 19.166666666666668, | |
| "grad_norm": 1.3933924436569214, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1361, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 19.212962962962962, | |
| "grad_norm": 1.1197868585586548, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1342, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 19.25925925925926, | |
| "grad_norm": 0.781484842300415, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1331, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 19.305555555555557, | |
| "grad_norm": 1.4528825283050537, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1388, | |
| "step": 4170 | |
| }, | |
| { | |
| "epoch": 19.35185185185185, | |
| "grad_norm": 0.972441554069519, | |
| "learning_rate": 3e-05, | |
| "loss": 0.139, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 19.39814814814815, | |
| "grad_norm": 1.5952485799789429, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1388, | |
| "step": 4190 | |
| }, | |
| { | |
| "epoch": 19.444444444444443, | |
| "grad_norm": 1.5413490533828735, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1361, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 19.49074074074074, | |
| "grad_norm": 1.033690094947815, | |
| "learning_rate": 3e-05, | |
| "loss": 0.131, | |
| "step": 4210 | |
| }, | |
| { | |
| "epoch": 19.537037037037038, | |
| "grad_norm": 0.702571451663971, | |
| "learning_rate": 3e-05, | |
| "loss": 0.15, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 19.583333333333332, | |
| "grad_norm": 1.326290488243103, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1325, | |
| "step": 4230 | |
| }, | |
| { | |
| "epoch": 19.62962962962963, | |
| "grad_norm": 0.7209446430206299, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1346, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 19.675925925925927, | |
| "grad_norm": 0.8662400841712952, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1507, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 19.72222222222222, | |
| "grad_norm": 1.1025499105453491, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1498, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 19.76851851851852, | |
| "grad_norm": 1.2367596626281738, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1469, | |
| "step": 4270 | |
| }, | |
| { | |
| "epoch": 19.814814814814813, | |
| "grad_norm": 1.2120156288146973, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1384, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 19.86111111111111, | |
| "grad_norm": 1.5859310626983643, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1296, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 19.90740740740741, | |
| "grad_norm": 1.9393326044082642, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1517, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 19.953703703703702, | |
| "grad_norm": 0.6213334202766418, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1368, | |
| "step": 4310 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "grad_norm": 1.004947543144226, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1345, | |
| "step": 4320 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 4320, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 20, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 65323465064448.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |