| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.0, | |
| "eval_steps": 100, | |
| "global_step": 3860, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05181347150259067, | |
| "grad_norm": 0.1298828125, | |
| "learning_rate": 1.2953367875647668e-07, | |
| "loss": 1.0187, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.10362694300518134, | |
| "grad_norm": 0.11962890625, | |
| "learning_rate": 2.5906735751295336e-07, | |
| "loss": 1.016, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.15544041450777202, | |
| "grad_norm": 0.11962890625, | |
| "learning_rate": 3.886010362694301e-07, | |
| "loss": 1.0219, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.20725388601036268, | |
| "grad_norm": 0.11767578125, | |
| "learning_rate": 5.181347150259067e-07, | |
| "loss": 1.0164, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.25906735751295334, | |
| "grad_norm": 0.126953125, | |
| "learning_rate": 6.476683937823834e-07, | |
| "loss": 1.0119, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.25906735751295334, | |
| "eval_main_loss": 1.0200841426849365, | |
| "eval_main_runtime": 50.8504, | |
| "eval_main_samples_per_second": 30.206, | |
| "eval_main_steps_per_second": 3.776, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.25906735751295334, | |
| "eval_anatomy_loss": 2.9678244590759277, | |
| "eval_anatomy_runtime": 0.2681, | |
| "eval_anatomy_samples_per_second": 7.459, | |
| "eval_anatomy_steps_per_second": 3.729, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.25906735751295334, | |
| "eval_college_mathematics_loss": 2.1804275512695312, | |
| "eval_college_mathematics_runtime": 0.2684, | |
| "eval_college_mathematics_samples_per_second": 7.451, | |
| "eval_college_mathematics_steps_per_second": 3.726, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.25906735751295334, | |
| "eval_international_law_loss": 3.191988229751587, | |
| "eval_international_law_runtime": 0.2665, | |
| "eval_international_law_samples_per_second": 7.505, | |
| "eval_international_law_steps_per_second": 3.752, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.31088082901554404, | |
| "grad_norm": 0.12890625, | |
| "learning_rate": 7.772020725388602e-07, | |
| "loss": 1.0155, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.3626943005181347, | |
| "grad_norm": 0.134765625, | |
| "learning_rate": 9.067357512953369e-07, | |
| "loss": 1.022, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.41450777202072536, | |
| "grad_norm": 0.1328125, | |
| "learning_rate": 1.0362694300518134e-06, | |
| "loss": 1.0071, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.46632124352331605, | |
| "grad_norm": 0.13671875, | |
| "learning_rate": 1.1658031088082903e-06, | |
| "loss": 1.0226, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.5181347150259067, | |
| "grad_norm": 0.1494140625, | |
| "learning_rate": 1.2953367875647669e-06, | |
| "loss": 1.0175, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5181347150259067, | |
| "eval_main_loss": 1.017417311668396, | |
| "eval_main_runtime": 50.9406, | |
| "eval_main_samples_per_second": 30.153, | |
| "eval_main_steps_per_second": 3.769, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5181347150259067, | |
| "eval_anatomy_loss": 2.9614720344543457, | |
| "eval_anatomy_runtime": 0.2677, | |
| "eval_anatomy_samples_per_second": 7.471, | |
| "eval_anatomy_steps_per_second": 3.735, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5181347150259067, | |
| "eval_college_mathematics_loss": 2.1731653213500977, | |
| "eval_college_mathematics_runtime": 0.2679, | |
| "eval_college_mathematics_samples_per_second": 7.466, | |
| "eval_college_mathematics_steps_per_second": 3.733, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5181347150259067, | |
| "eval_international_law_loss": 3.1835579872131348, | |
| "eval_international_law_runtime": 0.2687, | |
| "eval_international_law_samples_per_second": 7.443, | |
| "eval_international_law_steps_per_second": 3.722, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5699481865284974, | |
| "grad_norm": 0.1591796875, | |
| "learning_rate": 1.4248704663212437e-06, | |
| "loss": 1.0125, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.6217616580310881, | |
| "grad_norm": 0.1767578125, | |
| "learning_rate": 1.5544041450777204e-06, | |
| "loss": 1.0142, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.6735751295336787, | |
| "grad_norm": 0.1728515625, | |
| "learning_rate": 1.683937823834197e-06, | |
| "loss": 1.0175, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.7253886010362695, | |
| "grad_norm": 0.205078125, | |
| "learning_rate": 1.8134715025906738e-06, | |
| "loss": 1.0147, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.7772020725388601, | |
| "grad_norm": 0.2294921875, | |
| "learning_rate": 1.9430051813471504e-06, | |
| "loss": 1.0072, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.7772020725388601, | |
| "eval_main_loss": 1.0116238594055176, | |
| "eval_main_runtime": 50.9159, | |
| "eval_main_samples_per_second": 30.167, | |
| "eval_main_steps_per_second": 3.771, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.7772020725388601, | |
| "eval_anatomy_loss": 2.944845676422119, | |
| "eval_anatomy_runtime": 0.2687, | |
| "eval_anatomy_samples_per_second": 7.443, | |
| "eval_anatomy_steps_per_second": 3.721, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.7772020725388601, | |
| "eval_college_mathematics_loss": 2.1590933799743652, | |
| "eval_college_mathematics_runtime": 0.2683, | |
| "eval_college_mathematics_samples_per_second": 7.455, | |
| "eval_college_mathematics_steps_per_second": 3.727, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.7772020725388601, | |
| "eval_international_law_loss": 3.1690821647644043, | |
| "eval_international_law_runtime": 0.2666, | |
| "eval_international_law_samples_per_second": 7.501, | |
| "eval_international_law_steps_per_second": 3.75, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.8290155440414507, | |
| "grad_norm": 0.26171875, | |
| "learning_rate": 2.072538860103627e-06, | |
| "loss": 1.017, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.8808290155440415, | |
| "grad_norm": 0.265625, | |
| "learning_rate": 2.2020725388601037e-06, | |
| "loss": 1.0084, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.9326424870466321, | |
| "grad_norm": 0.283203125, | |
| "learning_rate": 2.3316062176165805e-06, | |
| "loss": 1.0008, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.9844559585492227, | |
| "grad_norm": 0.283203125, | |
| "learning_rate": 2.461139896373057e-06, | |
| "loss": 1.0035, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.0362694300518134, | |
| "grad_norm": 0.35546875, | |
| "learning_rate": 2.5906735751295338e-06, | |
| "loss": 1.0034, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.0362694300518134, | |
| "eval_main_loss": 1.0014457702636719, | |
| "eval_main_runtime": 50.699, | |
| "eval_main_samples_per_second": 30.296, | |
| "eval_main_steps_per_second": 3.787, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.0362694300518134, | |
| "eval_anatomy_loss": 2.91424298286438, | |
| "eval_anatomy_runtime": 0.2675, | |
| "eval_anatomy_samples_per_second": 7.475, | |
| "eval_anatomy_steps_per_second": 3.738, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.0362694300518134, | |
| "eval_college_mathematics_loss": 2.136711597442627, | |
| "eval_college_mathematics_runtime": 0.2681, | |
| "eval_college_mathematics_samples_per_second": 7.46, | |
| "eval_college_mathematics_steps_per_second": 3.73, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.0362694300518134, | |
| "eval_international_law_loss": 3.133229970932007, | |
| "eval_international_law_runtime": 0.2663, | |
| "eval_international_law_samples_per_second": 7.511, | |
| "eval_international_law_steps_per_second": 3.755, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.0880829015544042, | |
| "grad_norm": 0.361328125, | |
| "learning_rate": 2.7202072538860106e-06, | |
| "loss": 0.9984, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.1398963730569949, | |
| "grad_norm": 0.373046875, | |
| "learning_rate": 2.8497409326424875e-06, | |
| "loss": 0.9973, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.1917098445595855, | |
| "grad_norm": 0.46484375, | |
| "learning_rate": 2.979274611398964e-06, | |
| "loss": 0.9898, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.2435233160621761, | |
| "grad_norm": 0.462890625, | |
| "learning_rate": 3.1088082901554407e-06, | |
| "loss": 0.9841, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.2953367875647668, | |
| "grad_norm": 0.498046875, | |
| "learning_rate": 3.238341968911917e-06, | |
| "loss": 0.9908, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.2953367875647668, | |
| "eval_main_loss": 0.9807333946228027, | |
| "eval_main_runtime": 50.6947, | |
| "eval_main_samples_per_second": 30.299, | |
| "eval_main_steps_per_second": 3.787, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.2953367875647668, | |
| "eval_anatomy_loss": 2.860694408416748, | |
| "eval_anatomy_runtime": 0.2674, | |
| "eval_anatomy_samples_per_second": 7.479, | |
| "eval_anatomy_steps_per_second": 3.739, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.2953367875647668, | |
| "eval_college_mathematics_loss": 2.0933423042297363, | |
| "eval_college_mathematics_runtime": 0.2663, | |
| "eval_college_mathematics_samples_per_second": 7.51, | |
| "eval_college_mathematics_steps_per_second": 3.755, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.2953367875647668, | |
| "eval_international_law_loss": 3.0646896362304688, | |
| "eval_international_law_runtime": 0.2664, | |
| "eval_international_law_samples_per_second": 7.506, | |
| "eval_international_law_steps_per_second": 3.753, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.3471502590673574, | |
| "grad_norm": 0.56640625, | |
| "learning_rate": 3.367875647668394e-06, | |
| "loss": 0.9792, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.3989637305699483, | |
| "grad_norm": 0.56640625, | |
| "learning_rate": 3.497409326424871e-06, | |
| "loss": 0.9697, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.450777202072539, | |
| "grad_norm": 0.5234375, | |
| "learning_rate": 3.6269430051813476e-06, | |
| "loss": 0.9604, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.5025906735751295, | |
| "grad_norm": 0.50390625, | |
| "learning_rate": 3.756476683937824e-06, | |
| "loss": 0.9549, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.5544041450777202, | |
| "grad_norm": 0.59375, | |
| "learning_rate": 3.886010362694301e-06, | |
| "loss": 0.9634, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.5544041450777202, | |
| "eval_main_loss": 0.9588530659675598, | |
| "eval_main_runtime": 50.8939, | |
| "eval_main_samples_per_second": 30.18, | |
| "eval_main_steps_per_second": 3.773, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.5544041450777202, | |
| "eval_anatomy_loss": 2.800969123840332, | |
| "eval_anatomy_runtime": 0.268, | |
| "eval_anatomy_samples_per_second": 7.464, | |
| "eval_anatomy_steps_per_second": 3.732, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.5544041450777202, | |
| "eval_college_mathematics_loss": 2.0425174236297607, | |
| "eval_college_mathematics_runtime": 0.2684, | |
| "eval_college_mathematics_samples_per_second": 7.452, | |
| "eval_college_mathematics_steps_per_second": 3.726, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.5544041450777202, | |
| "eval_international_law_loss": 2.9860198497772217, | |
| "eval_international_law_runtime": 0.2678, | |
| "eval_international_law_samples_per_second": 7.467, | |
| "eval_international_law_steps_per_second": 3.734, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.6062176165803108, | |
| "grad_norm": 0.59765625, | |
| "learning_rate": 4.015544041450777e-06, | |
| "loss": 0.9551, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.6580310880829017, | |
| "grad_norm": 0.640625, | |
| "learning_rate": 4.145077720207254e-06, | |
| "loss": 0.9462, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.709844559585492, | |
| "grad_norm": 0.51953125, | |
| "learning_rate": 4.274611398963731e-06, | |
| "loss": 0.9398, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.761658031088083, | |
| "grad_norm": 0.490234375, | |
| "learning_rate": 4.404145077720207e-06, | |
| "loss": 0.9289, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.8134715025906736, | |
| "grad_norm": 0.4921875, | |
| "learning_rate": 4.533678756476685e-06, | |
| "loss": 0.9209, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.8134715025906736, | |
| "eval_main_loss": 0.9181744456291199, | |
| "eval_main_runtime": 50.9128, | |
| "eval_main_samples_per_second": 30.169, | |
| "eval_main_steps_per_second": 3.771, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.8134715025906736, | |
| "eval_anatomy_loss": 2.6760823726654053, | |
| "eval_anatomy_runtime": 0.2694, | |
| "eval_anatomy_samples_per_second": 7.423, | |
| "eval_anatomy_steps_per_second": 3.711, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.8134715025906736, | |
| "eval_college_mathematics_loss": 1.9530092477798462, | |
| "eval_college_mathematics_runtime": 0.2669, | |
| "eval_college_mathematics_samples_per_second": 7.494, | |
| "eval_college_mathematics_steps_per_second": 3.747, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.8134715025906736, | |
| "eval_international_law_loss": 2.8536064624786377, | |
| "eval_international_law_runtime": 0.268, | |
| "eval_international_law_samples_per_second": 7.462, | |
| "eval_international_law_steps_per_second": 3.731, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.8652849740932642, | |
| "grad_norm": 0.435546875, | |
| "learning_rate": 4.663212435233161e-06, | |
| "loss": 0.9132, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.917098445595855, | |
| "grad_norm": 0.45703125, | |
| "learning_rate": 4.7927461139896375e-06, | |
| "loss": 0.8982, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.9689119170984455, | |
| "grad_norm": 0.42578125, | |
| "learning_rate": 4.922279792746114e-06, | |
| "loss": 0.898, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.0207253886010363, | |
| "grad_norm": 0.451171875, | |
| "learning_rate": 4.9999171995395824e-06, | |
| "loss": 0.8827, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.0725388601036268, | |
| "grad_norm": 0.41015625, | |
| "learning_rate": 4.9989857573474595e-06, | |
| "loss": 0.8667, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.0725388601036268, | |
| "eval_main_loss": 0.8773276209831238, | |
| "eval_main_runtime": 50.6915, | |
| "eval_main_samples_per_second": 30.301, | |
| "eval_main_steps_per_second": 3.788, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.0725388601036268, | |
| "eval_anatomy_loss": 2.5175046920776367, | |
| "eval_anatomy_runtime": 0.2667, | |
| "eval_anatomy_samples_per_second": 7.498, | |
| "eval_anatomy_steps_per_second": 3.749, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.0725388601036268, | |
| "eval_college_mathematics_loss": 1.8391788005828857, | |
| "eval_college_mathematics_runtime": 0.2681, | |
| "eval_college_mathematics_samples_per_second": 7.459, | |
| "eval_college_mathematics_steps_per_second": 3.73, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.0725388601036268, | |
| "eval_international_law_loss": 2.7238972187042236, | |
| "eval_international_law_runtime": 0.2657, | |
| "eval_international_law_samples_per_second": 7.529, | |
| "eval_international_law_steps_per_second": 3.764, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.1243523316062176, | |
| "grad_norm": 0.3984375, | |
| "learning_rate": 4.997019759281217e-06, | |
| "loss": 0.8706, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.1761658031088085, | |
| "grad_norm": 0.38671875, | |
| "learning_rate": 4.9940200192449906e-06, | |
| "loss": 0.8744, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.227979274611399, | |
| "grad_norm": 0.349609375, | |
| "learning_rate": 4.989987779102074e-06, | |
| "loss": 0.8583, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 2.2797927461139897, | |
| "grad_norm": 0.330078125, | |
| "learning_rate": 4.984924708160789e-06, | |
| "loss": 0.8637, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 2.33160621761658, | |
| "grad_norm": 0.296875, | |
| "learning_rate": 4.978832902483415e-06, | |
| "loss": 0.8592, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.33160621761658, | |
| "eval_main_loss": 0.8600960373878479, | |
| "eval_main_runtime": 50.6644, | |
| "eval_main_samples_per_second": 30.317, | |
| "eval_main_steps_per_second": 3.79, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.33160621761658, | |
| "eval_anatomy_loss": 2.4422662258148193, | |
| "eval_anatomy_runtime": 0.2671, | |
| "eval_anatomy_samples_per_second": 7.488, | |
| "eval_anatomy_steps_per_second": 3.744, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.33160621761658, | |
| "eval_college_mathematics_loss": 1.7872785329818726, | |
| "eval_college_mathematics_runtime": 0.267, | |
| "eval_college_mathematics_samples_per_second": 7.492, | |
| "eval_college_mathematics_steps_per_second": 3.746, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.33160621761658, | |
| "eval_international_law_loss": 2.6731436252593994, | |
| "eval_international_law_runtime": 0.2664, | |
| "eval_international_law_samples_per_second": 7.508, | |
| "eval_international_law_steps_per_second": 3.754, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.383419689119171, | |
| "grad_norm": 0.326171875, | |
| "learning_rate": 4.971714884018439e-06, | |
| "loss": 0.8623, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.4352331606217614, | |
| "grad_norm": 0.291015625, | |
| "learning_rate": 4.96357359955649e-06, | |
| "loss": 0.8496, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.4870466321243523, | |
| "grad_norm": 0.29296875, | |
| "learning_rate": 4.9544124195104015e-06, | |
| "loss": 0.8595, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.538860103626943, | |
| "grad_norm": 0.296875, | |
| "learning_rate": 4.944235136519888e-06, | |
| "loss": 0.8512, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.5906735751295336, | |
| "grad_norm": 0.265625, | |
| "learning_rate": 4.933045963881431e-06, | |
| "loss": 0.8522, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.5906735751295336, | |
| "eval_main_loss": 0.8523173928260803, | |
| "eval_main_runtime": 50.9143, | |
| "eval_main_samples_per_second": 30.168, | |
| "eval_main_steps_per_second": 3.771, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.5906735751295336, | |
| "eval_anatomy_loss": 2.420971632003784, | |
| "eval_anatomy_runtime": 0.2681, | |
| "eval_anatomy_samples_per_second": 7.459, | |
| "eval_anatomy_steps_per_second": 3.729, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.5906735751295336, | |
| "eval_college_mathematics_loss": 1.7655620574951172, | |
| "eval_college_mathematics_runtime": 0.2683, | |
| "eval_college_mathematics_samples_per_second": 7.454, | |
| "eval_college_mathematics_steps_per_second": 3.727, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.5906735751295336, | |
| "eval_international_law_loss": 2.6557209491729736, | |
| "eval_international_law_runtime": 0.2675, | |
| "eval_international_law_samples_per_second": 7.475, | |
| "eval_international_law_steps_per_second": 3.738, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.6424870466321244, | |
| "grad_norm": 0.263671875, | |
| "learning_rate": 4.920849533804017e-06, | |
| "loss": 0.8525, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.694300518134715, | |
| "grad_norm": 0.275390625, | |
| "learning_rate": 4.907650895491443e-06, | |
| "loss": 0.8554, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 2.7461139896373057, | |
| "grad_norm": 0.2421875, | |
| "learning_rate": 4.893455513052003e-06, | |
| "loss": 0.8489, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 2.7979274611398965, | |
| "grad_norm": 0.25, | |
| "learning_rate": 4.878269263236391e-06, | |
| "loss": 0.844, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.849740932642487, | |
| "grad_norm": 0.275390625, | |
| "learning_rate": 4.86209843300479e-06, | |
| "loss": 0.8456, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.849740932642487, | |
| "eval_main_loss": 0.8472149968147278, | |
| "eval_main_runtime": 50.8752, | |
| "eval_main_samples_per_second": 30.192, | |
| "eval_main_steps_per_second": 3.774, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.849740932642487, | |
| "eval_anatomy_loss": 2.407443046569824, | |
| "eval_anatomy_runtime": 0.2673, | |
| "eval_anatomy_samples_per_second": 7.481, | |
| "eval_anatomy_steps_per_second": 3.741, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.849740932642487, | |
| "eval_college_mathematics_loss": 1.7581956386566162, | |
| "eval_college_mathematics_runtime": 0.2664, | |
| "eval_college_mathematics_samples_per_second": 7.507, | |
| "eval_college_mathematics_steps_per_second": 3.753, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.849740932642487, | |
| "eval_international_law_loss": 2.643361806869507, | |
| "eval_international_law_runtime": 0.266, | |
| "eval_international_law_samples_per_second": 7.518, | |
| "eval_international_law_steps_per_second": 3.759, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.901554404145078, | |
| "grad_norm": 0.2412109375, | |
| "learning_rate": 4.8449497169241285e-06, | |
| "loss": 0.847, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.9533678756476682, | |
| "grad_norm": 0.248046875, | |
| "learning_rate": 4.826830214396594e-06, | |
| "loss": 0.8444, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 3.005181347150259, | |
| "grad_norm": 0.2490234375, | |
| "learning_rate": 4.807747426720553e-06, | |
| "loss": 0.8454, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 3.05699481865285, | |
| "grad_norm": 0.2431640625, | |
| "learning_rate": 4.78770925398508e-06, | |
| "loss": 0.8409, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 3.1088082901554404, | |
| "grad_norm": 0.25390625, | |
| "learning_rate": 4.766723991799407e-06, | |
| "loss": 0.8392, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.1088082901554404, | |
| "eval_main_loss": 0.8436682820320129, | |
| "eval_main_runtime": 50.6014, | |
| "eval_main_samples_per_second": 30.355, | |
| "eval_main_steps_per_second": 3.794, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.1088082901554404, | |
| "eval_anatomy_loss": 2.3988683223724365, | |
| "eval_anatomy_runtime": 0.2661, | |
| "eval_anatomy_samples_per_second": 7.517, | |
| "eval_anatomy_steps_per_second": 3.758, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.1088082901554404, | |
| "eval_college_mathematics_loss": 1.7500412464141846, | |
| "eval_college_mathematics_runtime": 0.2666, | |
| "eval_college_mathematics_samples_per_second": 7.502, | |
| "eval_college_mathematics_steps_per_second": 3.751, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.1088082901554404, | |
| "eval_international_law_loss": 2.639014482498169, | |
| "eval_international_law_runtime": 0.2661, | |
| "eval_international_law_samples_per_second": 7.517, | |
| "eval_international_law_steps_per_second": 3.759, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.160621761658031, | |
| "grad_norm": 0.2109375, | |
| "learning_rate": 4.744800327858608e-06, | |
| "loss": 0.8409, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 3.2124352331606216, | |
| "grad_norm": 0.255859375, | |
| "learning_rate": 4.721947338346993e-06, | |
| "loss": 0.8413, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 3.2642487046632125, | |
| "grad_norm": 0.2197265625, | |
| "learning_rate": 4.698174484180641e-06, | |
| "loss": 0.8364, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 3.3160621761658033, | |
| "grad_norm": 0.236328125, | |
| "learning_rate": 4.673491607090684e-06, | |
| "loss": 0.8351, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 3.3678756476683938, | |
| "grad_norm": 0.2158203125, | |
| "learning_rate": 4.647908925548918e-06, | |
| "loss": 0.8354, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 3.3678756476683938, | |
| "eval_main_loss": 0.8413074016571045, | |
| "eval_main_runtime": 50.8355, | |
| "eval_main_samples_per_second": 30.215, | |
| "eval_main_steps_per_second": 3.777, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 3.3678756476683938, | |
| "eval_anatomy_loss": 2.3927080631256104, | |
| "eval_anatomy_runtime": 0.2684, | |
| "eval_anatomy_samples_per_second": 7.451, | |
| "eval_anatomy_steps_per_second": 3.725, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 3.3678756476683938, | |
| "eval_college_mathematics_loss": 1.7482649087905884, | |
| "eval_college_mathematics_runtime": 0.2673, | |
| "eval_college_mathematics_samples_per_second": 7.482, | |
| "eval_college_mathematics_steps_per_second": 3.741, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 3.3678756476683938, | |
| "eval_international_law_loss": 2.6322338581085205, | |
| "eval_international_law_runtime": 0.2677, | |
| "eval_international_law_samples_per_second": 7.47, | |
| "eval_international_law_steps_per_second": 3.735, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 3.4196891191709846, | |
| "grad_norm": 0.236328125, | |
| "learning_rate": 4.621437030537461e-06, | |
| "loss": 0.8428, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 3.471502590673575, | |
| "grad_norm": 0.232421875, | |
| "learning_rate": 4.594086881164184e-06, | |
| "loss": 0.8495, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 3.523316062176166, | |
| "grad_norm": 0.244140625, | |
| "learning_rate": 4.565869800125747e-06, | |
| "loss": 0.8445, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 3.5751295336787567, | |
| "grad_norm": 0.2314453125, | |
| "learning_rate": 4.536797469020116e-06, | |
| "loss": 0.8441, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 3.626943005181347, | |
| "grad_norm": 0.220703125, | |
| "learning_rate": 4.506881923510493e-06, | |
| "loss": 0.8388, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.626943005181347, | |
| "eval_main_loss": 0.8397356867790222, | |
| "eval_main_runtime": 50.8597, | |
| "eval_main_samples_per_second": 30.201, | |
| "eval_main_steps_per_second": 3.775, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.626943005181347, | |
| "eval_anatomy_loss": 2.392620325088501, | |
| "eval_anatomy_runtime": 0.268, | |
| "eval_anatomy_samples_per_second": 7.464, | |
| "eval_anatomy_steps_per_second": 3.732, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.626943005181347, | |
| "eval_college_mathematics_loss": 1.746407151222229, | |
| "eval_college_mathematics_runtime": 0.2684, | |
| "eval_college_mathematics_samples_per_second": 7.451, | |
| "eval_college_mathematics_steps_per_second": 3.725, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.626943005181347, | |
| "eval_international_law_loss": 2.632657527923584, | |
| "eval_international_law_runtime": 0.2666, | |
| "eval_international_law_samples_per_second": 7.503, | |
| "eval_international_law_steps_per_second": 3.751, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.6787564766839376, | |
| "grad_norm": 0.2197265625, | |
| "learning_rate": 4.476135548342666e-06, | |
| "loss": 0.8405, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 3.7305699481865284, | |
| "grad_norm": 0.2119140625, | |
| "learning_rate": 4.444571072217848e-06, | |
| "loss": 0.8353, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 3.7823834196891193, | |
| "grad_norm": 0.2353515625, | |
| "learning_rate": 4.4122015625231125e-06, | |
| "loss": 0.838, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 3.8341968911917097, | |
| "grad_norm": 0.208984375, | |
| "learning_rate": 4.37904041992163e-06, | |
| "loss": 0.8339, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 3.8860103626943006, | |
| "grad_norm": 0.2099609375, | |
| "learning_rate": 4.345101372804917e-06, | |
| "loss": 0.8352, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.8860103626943006, | |
| "eval_main_loss": 0.8387607932090759, | |
| "eval_main_runtime": 50.9241, | |
| "eval_main_samples_per_second": 30.163, | |
| "eval_main_steps_per_second": 3.77, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.8860103626943006, | |
| "eval_anatomy_loss": 2.3933463096618652, | |
| "eval_anatomy_runtime": 0.2694, | |
| "eval_anatomy_samples_per_second": 7.424, | |
| "eval_anatomy_steps_per_second": 3.712, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.8860103626943006, | |
| "eval_college_mathematics_loss": 1.7452200651168823, | |
| "eval_college_mathematics_runtime": 0.2686, | |
| "eval_college_mathematics_samples_per_second": 7.447, | |
| "eval_college_mathematics_steps_per_second": 3.724, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.8860103626943006, | |
| "eval_international_law_loss": 2.631213903427124, | |
| "eval_international_law_runtime": 0.2675, | |
| "eval_international_law_samples_per_second": 7.475, | |
| "eval_international_law_steps_per_second": 3.738, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.937823834196891, | |
| "grad_norm": 0.2119140625, | |
| "learning_rate": 4.310398471609416e-06, | |
| "loss": 0.8379, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 3.989637305699482, | |
| "grad_norm": 0.228515625, | |
| "learning_rate": 4.274946082999753e-06, | |
| "loss": 0.8366, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 4.041450777202073, | |
| "grad_norm": 0.240234375, | |
| "learning_rate": 4.238758883921077e-06, | |
| "loss": 0.8351, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 4.0932642487046635, | |
| "grad_norm": 0.244140625, | |
| "learning_rate": 4.201851855522946e-06, | |
| "loss": 0.8427, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 4.1450777202072535, | |
| "grad_norm": 0.2197265625, | |
| "learning_rate": 4.1642402769572775e-06, | |
| "loss": 0.8375, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 4.1450777202072535, | |
| "eval_main_loss": 0.8381660580635071, | |
| "eval_main_runtime": 50.8823, | |
| "eval_main_samples_per_second": 30.187, | |
| "eval_main_steps_per_second": 3.773, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 4.1450777202072535, | |
| "eval_anatomy_loss": 2.3937363624572754, | |
| "eval_anatomy_runtime": 0.2711, | |
| "eval_anatomy_samples_per_second": 7.376, | |
| "eval_anatomy_steps_per_second": 3.688, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 4.1450777202072535, | |
| "eval_college_mathematics_loss": 1.7459180355072021, | |
| "eval_college_mathematics_runtime": 0.2669, | |
| "eval_college_mathematics_samples_per_second": 7.493, | |
| "eval_college_mathematics_steps_per_second": 3.747, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 4.1450777202072535, | |
| "eval_international_law_loss": 2.6354634761810303, | |
| "eval_international_law_runtime": 0.2663, | |
| "eval_international_law_samples_per_second": 7.509, | |
| "eval_international_law_steps_per_second": 3.755, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 4.196891191709844, | |
| "grad_norm": 0.2431640625, | |
| "learning_rate": 4.125939719052927e-06, | |
| "loss": 0.8405, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 4.248704663212435, | |
| "grad_norm": 0.193359375, | |
| "learning_rate": 4.086966037869515e-06, | |
| "loss": 0.8346, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 4.300518134715026, | |
| "grad_norm": 0.244140625, | |
| "learning_rate": 4.047335368133176e-06, | |
| "loss": 0.8388, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 4.352331606217617, | |
| "grad_norm": 0.21484375, | |
| "learning_rate": 4.0070641165569335e-06, | |
| "loss": 0.8301, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 4.404145077720207, | |
| "grad_norm": 0.2275390625, | |
| "learning_rate": 3.96616895504848e-06, | |
| "loss": 0.8366, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 4.404145077720207, | |
| "eval_main_loss": 0.8377915024757385, | |
| "eval_main_runtime": 50.9147, | |
| "eval_main_samples_per_second": 30.168, | |
| "eval_main_steps_per_second": 3.771, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 4.404145077720207, | |
| "eval_anatomy_loss": 2.394949197769165, | |
| "eval_anatomy_runtime": 0.2684, | |
| "eval_anatomy_samples_per_second": 7.451, | |
| "eval_anatomy_steps_per_second": 3.726, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 4.404145077720207, | |
| "eval_college_mathematics_loss": 1.7438111305236816, | |
| "eval_college_mathematics_runtime": 0.2676, | |
| "eval_college_mathematics_samples_per_second": 7.475, | |
| "eval_college_mathematics_steps_per_second": 3.737, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 4.404145077720207, | |
| "eval_international_law_loss": 2.6339313983917236, | |
| "eval_international_law_runtime": 0.2665, | |
| "eval_international_law_samples_per_second": 7.506, | |
| "eval_international_law_steps_per_second": 3.753, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 4.455958549222798, | |
| "grad_norm": 0.2197265625, | |
| "learning_rate": 3.924666813808176e-06, | |
| "loss": 0.833, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 4.507772020725389, | |
| "grad_norm": 0.234375, | |
| "learning_rate": 3.882574874320099e-06, | |
| "loss": 0.8381, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 4.5595854922279795, | |
| "grad_norm": 0.2177734375, | |
| "learning_rate": 3.839910562239088e-06, | |
| "loss": 0.8438, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 4.61139896373057, | |
| "grad_norm": 0.2138671875, | |
| "learning_rate": 3.7966915401766845e-06, | |
| "loss": 0.832, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 4.66321243523316, | |
| "grad_norm": 0.2177734375, | |
| "learning_rate": 3.752935700388982e-06, | |
| "loss": 0.843, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 4.66321243523316, | |
| "eval_main_loss": 0.8375835418701172, | |
| "eval_main_runtime": 50.8717, | |
| "eval_main_samples_per_second": 30.194, | |
| "eval_main_steps_per_second": 3.774, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 4.66321243523316, | |
| "eval_anatomy_loss": 2.3957273960113525, | |
| "eval_anatomy_runtime": 0.2674, | |
| "eval_anatomy_samples_per_second": 7.479, | |
| "eval_anatomy_steps_per_second": 3.74, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 4.66321243523316, | |
| "eval_college_mathematics_loss": 1.745951771736145, | |
| "eval_college_mathematics_runtime": 0.2678, | |
| "eval_college_mathematics_samples_per_second": 7.468, | |
| "eval_college_mathematics_steps_per_second": 3.734, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 4.66321243523316, | |
| "eval_international_law_loss": 2.635204315185547, | |
| "eval_international_law_runtime": 0.2682, | |
| "eval_international_law_samples_per_second": 7.458, | |
| "eval_international_law_steps_per_second": 3.729, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 4.715025906735751, | |
| "grad_norm": 0.22265625, | |
| "learning_rate": 3.7086611573694107e-06, | |
| "loss": 0.8254, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 4.766839378238342, | |
| "grad_norm": 0.232421875, | |
| "learning_rate": 3.663886240349507e-06, | |
| "loss": 0.8342, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 4.818652849740933, | |
| "grad_norm": 0.228515625, | |
| "learning_rate": 3.6186294857107933e-06, | |
| "loss": 0.8368, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 4.870466321243523, | |
| "grad_norm": 0.2255859375, | |
| "learning_rate": 3.5729096293108935e-06, | |
| "loss": 0.8406, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 4.922279792746114, | |
| "grad_norm": 0.23828125, | |
| "learning_rate": 3.526745598727071e-06, | |
| "loss": 0.8317, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 4.922279792746114, | |
| "eval_main_loss": 0.8374476432800293, | |
| "eval_main_runtime": 50.8861, | |
| "eval_main_samples_per_second": 30.185, | |
| "eval_main_steps_per_second": 3.773, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 4.922279792746114, | |
| "eval_anatomy_loss": 2.3970742225646973, | |
| "eval_anatomy_runtime": 0.268, | |
| "eval_anatomy_samples_per_second": 7.462, | |
| "eval_anatomy_steps_per_second": 3.731, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 4.922279792746114, | |
| "eval_college_mathematics_loss": 1.7456122636795044, | |
| "eval_college_mathematics_runtime": 0.2683, | |
| "eval_college_mathematics_samples_per_second": 7.454, | |
| "eval_college_mathematics_steps_per_second": 3.727, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 4.922279792746114, | |
| "eval_international_law_loss": 2.6364247798919678, | |
| "eval_international_law_runtime": 0.267, | |
| "eval_international_law_samples_per_second": 7.49, | |
| "eval_international_law_steps_per_second": 3.745, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 4.974093264248705, | |
| "grad_norm": 0.232421875, | |
| "learning_rate": 3.4801565054203962e-06, | |
| "loss": 0.834, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 5.025906735751295, | |
| "grad_norm": 0.228515625, | |
| "learning_rate": 3.433161636823782e-06, | |
| "loss": 0.8409, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 5.077720207253886, | |
| "grad_norm": 0.2099609375, | |
| "learning_rate": 3.3857804483571803e-06, | |
| "loss": 0.8324, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 5.129533678756476, | |
| "grad_norm": 0.236328125, | |
| "learning_rate": 3.3380325553732223e-06, | |
| "loss": 0.8433, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 5.181347150259067, | |
| "grad_norm": 0.21484375, | |
| "learning_rate": 3.2899377250366536e-06, | |
| "loss": 0.8307, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 5.181347150259067, | |
| "eval_main_loss": 0.8373920321464539, | |
| "eval_main_runtime": 50.7102, | |
| "eval_main_samples_per_second": 30.29, | |
| "eval_main_steps_per_second": 3.786, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 5.181347150259067, | |
| "eval_anatomy_loss": 2.3975374698638916, | |
| "eval_anatomy_runtime": 0.2673, | |
| "eval_anatomy_samples_per_second": 7.483, | |
| "eval_anatomy_steps_per_second": 3.741, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 5.181347150259067, | |
| "eval_college_mathematics_loss": 1.746949315071106, | |
| "eval_college_mathematics_runtime": 0.2679, | |
| "eval_college_mathematics_samples_per_second": 7.465, | |
| "eval_college_mathematics_steps_per_second": 3.732, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 5.181347150259067, | |
| "eval_international_law_loss": 2.637000799179077, | |
| "eval_international_law_runtime": 0.267, | |
| "eval_international_law_samples_per_second": 7.49, | |
| "eval_international_law_steps_per_second": 3.745, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 5.233160621761658, | |
| "grad_norm": 0.2138671875, | |
| "learning_rate": 3.2415158681409215e-06, | |
| "loss": 0.836, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 5.284974093264249, | |
| "grad_norm": 0.2099609375, | |
| "learning_rate": 3.1927870308652953e-06, | |
| "loss": 0.8447, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 5.33678756476684, | |
| "grad_norm": 0.220703125, | |
| "learning_rate": 3.1437713864759483e-06, | |
| "loss": 0.8383, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 5.38860103626943, | |
| "grad_norm": 0.2451171875, | |
| "learning_rate": 3.0944892269744155e-06, | |
| "loss": 0.8412, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 5.4404145077720205, | |
| "grad_norm": 0.23828125, | |
| "learning_rate": 3.044960954696906e-06, | |
| "loss": 0.837, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 5.4404145077720205, | |
| "eval_main_loss": 0.837336003780365, | |
| "eval_main_runtime": 50.8778, | |
| "eval_main_samples_per_second": 30.19, | |
| "eval_main_steps_per_second": 3.774, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 5.4404145077720205, | |
| "eval_anatomy_loss": 2.39707350730896, | |
| "eval_anatomy_runtime": 0.264, | |
| "eval_anatomy_samples_per_second": 7.575, | |
| "eval_anatomy_steps_per_second": 3.787, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 5.4404145077720205, | |
| "eval_college_mathematics_loss": 1.746044397354126, | |
| "eval_college_mathematics_runtime": 0.2668, | |
| "eval_college_mathematics_samples_per_second": 7.497, | |
| "eval_college_mathematics_steps_per_second": 3.748, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 5.4404145077720205, | |
| "eval_international_law_loss": 2.639299154281616, | |
| "eval_international_law_runtime": 0.2686, | |
| "eval_international_law_samples_per_second": 7.446, | |
| "eval_international_law_steps_per_second": 3.723, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 5.492227979274611, | |
| "grad_norm": 0.244140625, | |
| "learning_rate": 2.9952070738679312e-06, | |
| "loss": 0.8349, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 5.544041450777202, | |
| "grad_norm": 0.2421875, | |
| "learning_rate": 2.9452481821117544e-06, | |
| "loss": 0.8261, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 5.595854922279793, | |
| "grad_norm": 0.193359375, | |
| "learning_rate": 2.895104961925179e-06, | |
| "loss": 0.8305, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 5.647668393782383, | |
| "grad_norm": 0.2177734375, | |
| "learning_rate": 2.844798172115185e-06, | |
| "loss": 0.8316, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 5.699481865284974, | |
| "grad_norm": 0.220703125, | |
| "learning_rate": 2.7943486392049972e-06, | |
| "loss": 0.8334, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 5.699481865284974, | |
| "eval_main_loss": 0.8372721076011658, | |
| "eval_main_runtime": 50.8569, | |
| "eval_main_samples_per_second": 30.202, | |
| "eval_main_steps_per_second": 3.775, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 5.699481865284974, | |
| "eval_anatomy_loss": 2.3966822624206543, | |
| "eval_anatomy_runtime": 0.2669, | |
| "eval_anatomy_samples_per_second": 7.494, | |
| "eval_anatomy_steps_per_second": 3.747, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 5.699481865284974, | |
| "eval_college_mathematics_loss": 1.7497589588165283, | |
| "eval_college_mathematics_runtime": 0.2678, | |
| "eval_college_mathematics_samples_per_second": 7.468, | |
| "eval_college_mathematics_steps_per_second": 3.734, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 5.699481865284974, | |
| "eval_international_law_loss": 2.634140968322754, | |
| "eval_international_law_runtime": 0.267, | |
| "eval_international_law_samples_per_second": 7.492, | |
| "eval_international_law_steps_per_second": 3.746, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 5.751295336787565, | |
| "grad_norm": 0.212890625, | |
| "learning_rate": 2.7437772488120945e-06, | |
| "loss": 0.8305, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 5.803108808290156, | |
| "grad_norm": 0.2109375, | |
| "learning_rate": 2.6931049370017755e-06, | |
| "loss": 0.8383, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 5.8549222797927465, | |
| "grad_norm": 0.2333984375, | |
| "learning_rate": 2.6423526816198253e-06, | |
| "loss": 0.8445, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 5.9067357512953365, | |
| "grad_norm": 0.2236328125, | |
| "learning_rate": 2.5915414936078933e-06, | |
| "loss": 0.8385, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 5.958549222797927, | |
| "grad_norm": 0.2294921875, | |
| "learning_rate": 2.5406924083051683e-06, | |
| "loss": 0.8359, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 5.958549222797927, | |
| "eval_main_loss": 0.8372599482536316, | |
| "eval_main_runtime": 50.7835, | |
| "eval_main_samples_per_second": 30.246, | |
| "eval_main_steps_per_second": 3.781, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 5.958549222797927, | |
| "eval_anatomy_loss": 2.3953776359558105, | |
| "eval_anatomy_runtime": 0.2668, | |
| "eval_anatomy_samples_per_second": 7.496, | |
| "eval_anatomy_steps_per_second": 3.748, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 5.958549222797927, | |
| "eval_college_mathematics_loss": 1.7478266954421997, | |
| "eval_college_mathematics_runtime": 0.2698, | |
| "eval_college_mathematics_samples_per_second": 7.413, | |
| "eval_college_mathematics_steps_per_second": 3.707, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 5.958549222797927, | |
| "eval_international_law_loss": 2.6391143798828125, | |
| "eval_international_law_runtime": 0.2663, | |
| "eval_international_law_samples_per_second": 7.509, | |
| "eval_international_law_steps_per_second": 3.755, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 6.010362694300518, | |
| "grad_norm": 0.2392578125, | |
| "learning_rate": 2.4898264767399445e-06, | |
| "loss": 0.8316, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 6.062176165803109, | |
| "grad_norm": 0.2255859375, | |
| "learning_rate": 2.438964756914712e-06, | |
| "loss": 0.8412, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 6.1139896373057, | |
| "grad_norm": 0.2236328125, | |
| "learning_rate": 2.3881283050883368e-06, | |
| "loss": 0.8368, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 6.16580310880829, | |
| "grad_norm": 0.234375, | |
| "learning_rate": 2.337338167058981e-06, | |
| "loss": 0.8392, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 6.217616580310881, | |
| "grad_norm": 0.212890625, | |
| "learning_rate": 2.286615369451342e-06, | |
| "loss": 0.834, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 6.217616580310881, | |
| "eval_main_loss": 0.8373088836669922, | |
| "eval_main_runtime": 50.8482, | |
| "eval_main_samples_per_second": 30.208, | |
| "eval_main_steps_per_second": 3.776, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 6.217616580310881, | |
| "eval_anatomy_loss": 2.3978271484375, | |
| "eval_anatomy_runtime": 0.2671, | |
| "eval_anatomy_samples_per_second": 7.487, | |
| "eval_anatomy_steps_per_second": 3.744, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 6.217616580310881, | |
| "eval_college_mathematics_loss": 1.7477797269821167, | |
| "eval_college_mathematics_runtime": 0.2685, | |
| "eval_college_mathematics_samples_per_second": 7.449, | |
| "eval_college_mathematics_steps_per_second": 3.724, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 6.217616580310881, | |
| "eval_international_law_loss": 2.638612985610962, | |
| "eval_international_law_runtime": 0.2668, | |
| "eval_international_law_samples_per_second": 7.495, | |
| "eval_international_law_steps_per_second": 3.748, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 6.269430051813472, | |
| "grad_norm": 0.2275390625, | |
| "learning_rate": 2.2359809110118358e-06, | |
| "loss": 0.8391, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 6.321243523316062, | |
| "grad_norm": 0.2431640625, | |
| "learning_rate": 2.1854557539153203e-06, | |
| "loss": 0.8368, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 6.373056994818652, | |
| "grad_norm": 0.2294921875, | |
| "learning_rate": 2.1350608150869563e-06, | |
| "loss": 0.8321, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 6.424870466321243, | |
| "grad_norm": 0.23828125, | |
| "learning_rate": 2.0848169575428057e-06, | |
| "loss": 0.8375, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 6.476683937823834, | |
| "grad_norm": 0.232421875, | |
| "learning_rate": 2.034744981752741e-06, | |
| "loss": 0.835, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 6.476683937823834, | |
| "eval_main_loss": 0.8372688889503479, | |
| "eval_main_runtime": 50.7481, | |
| "eval_main_samples_per_second": 30.267, | |
| "eval_main_steps_per_second": 3.783, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 6.476683937823834, | |
| "eval_anatomy_loss": 2.397125244140625, | |
| "eval_anatomy_runtime": 0.2662, | |
| "eval_anatomy_samples_per_second": 7.513, | |
| "eval_anatomy_steps_per_second": 3.757, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 6.476683937823834, | |
| "eval_college_mathematics_loss": 1.7471617460250854, | |
| "eval_college_mathematics_runtime": 0.2671, | |
| "eval_college_mathematics_samples_per_second": 7.487, | |
| "eval_college_mathematics_steps_per_second": 3.744, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 6.476683937823834, | |
| "eval_international_law_loss": 2.6356821060180664, | |
| "eval_international_law_runtime": 0.2665, | |
| "eval_international_law_samples_per_second": 7.504, | |
| "eval_international_law_steps_per_second": 3.752, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 6.528497409326425, | |
| "grad_norm": 0.2490234375, | |
| "learning_rate": 1.9848656170292556e-06, | |
| "loss": 0.8291, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 6.580310880829016, | |
| "grad_norm": 0.2314453125, | |
| "learning_rate": 1.9351995129457305e-06, | |
| "loss": 0.8325, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 6.632124352331607, | |
| "grad_norm": 0.2294921875, | |
| "learning_rate": 1.88576723078771e-06, | |
| "loss": 0.8345, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 6.683937823834197, | |
| "grad_norm": 0.2412109375, | |
| "learning_rate": 1.8365892350407238e-06, | |
| "loss": 0.8387, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 6.7357512953367875, | |
| "grad_norm": 0.212890625, | |
| "learning_rate": 1.7876858849181982e-06, | |
| "loss": 0.8311, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 6.7357512953367875, | |
| "eval_main_loss": 0.8372130393981934, | |
| "eval_main_runtime": 50.6463, | |
| "eval_main_samples_per_second": 30.328, | |
| "eval_main_steps_per_second": 3.791, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 6.7357512953367875, | |
| "eval_anatomy_loss": 2.393521308898926, | |
| "eval_anatomy_runtime": 0.267, | |
| "eval_anatomy_samples_per_second": 7.49, | |
| "eval_anatomy_steps_per_second": 3.745, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 6.7357512953367875, | |
| "eval_college_mathematics_loss": 1.748924970626831, | |
| "eval_college_mathematics_runtime": 0.2664, | |
| "eval_college_mathematics_samples_per_second": 7.509, | |
| "eval_college_mathematics_steps_per_second": 3.754, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 6.7357512953367875, | |
| "eval_international_law_loss": 2.637521505355835, | |
| "eval_international_law_runtime": 0.2667, | |
| "eval_international_law_samples_per_second": 7.499, | |
| "eval_international_law_steps_per_second": 3.749, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 6.787564766839378, | |
| "grad_norm": 0.2421875, | |
| "learning_rate": 1.73907742593293e-06, | |
| "loss": 0.8395, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 6.839378238341969, | |
| "grad_norm": 0.220703125, | |
| "learning_rate": 1.690783981515648e-06, | |
| "loss": 0.8399, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 6.891191709844559, | |
| "grad_norm": 0.2470703125, | |
| "learning_rate": 1.642825544684101e-06, | |
| "loss": 0.8302, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 6.94300518134715, | |
| "grad_norm": 0.201171875, | |
| "learning_rate": 1.5952219697661455e-06, | |
| "loss": 0.8338, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 6.994818652849741, | |
| "grad_norm": 0.224609375, | |
| "learning_rate": 1.5479929641802492e-06, | |
| "loss": 0.837, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 6.994818652849741, | |
| "eval_main_loss": 0.8372467160224915, | |
| "eval_main_runtime": 50.6632, | |
| "eval_main_samples_per_second": 30.318, | |
| "eval_main_steps_per_second": 3.79, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 6.994818652849741, | |
| "eval_anatomy_loss": 2.399848222732544, | |
| "eval_anatomy_runtime": 0.2663, | |
| "eval_anatomy_samples_per_second": 7.512, | |
| "eval_anatomy_steps_per_second": 3.756, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 6.994818652849741, | |
| "eval_college_mathematics_loss": 1.7486381530761719, | |
| "eval_college_mathematics_runtime": 0.268, | |
| "eval_college_mathematics_samples_per_second": 7.464, | |
| "eval_college_mathematics_steps_per_second": 3.732, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 6.994818652849741, | |
| "eval_international_law_loss": 2.637385845184326, | |
| "eval_international_law_runtime": 0.2665, | |
| "eval_international_law_samples_per_second": 7.504, | |
| "eval_international_law_steps_per_second": 3.752, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 7.046632124352332, | |
| "grad_norm": 0.2431640625, | |
| "learning_rate": 1.5011580802768048e-06, | |
| "loss": 0.8392, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 7.098445595854923, | |
| "grad_norm": 0.2333984375, | |
| "learning_rate": 1.4547367072436519e-06, | |
| "loss": 0.8326, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 7.150259067357513, | |
| "grad_norm": 0.2275390625, | |
| "learning_rate": 1.4087480630791405e-06, | |
| "loss": 0.8324, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 7.2020725388601035, | |
| "grad_norm": 0.2333984375, | |
| "learning_rate": 1.3632111866360585e-06, | |
| "loss": 0.8309, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 7.253886010362694, | |
| "grad_norm": 0.2265625, | |
| "learning_rate": 1.318144929739743e-06, | |
| "loss": 0.8292, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 7.253886010362694, | |
| "eval_main_loss": 0.8372478485107422, | |
| "eval_main_runtime": 50.677, | |
| "eval_main_samples_per_second": 30.31, | |
| "eval_main_steps_per_second": 3.789, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 7.253886010362694, | |
| "eval_anatomy_loss": 2.3977112770080566, | |
| "eval_anatomy_runtime": 0.2672, | |
| "eval_anatomy_samples_per_second": 7.484, | |
| "eval_anatomy_steps_per_second": 3.742, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 7.253886010362694, | |
| "eval_college_mathematics_loss": 1.745300531387329, | |
| "eval_college_mathematics_runtime": 0.2659, | |
| "eval_college_mathematics_samples_per_second": 7.523, | |
| "eval_college_mathematics_steps_per_second": 3.761, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 7.253886010362694, | |
| "eval_international_law_loss": 2.6362087726593018, | |
| "eval_international_law_runtime": 0.2666, | |
| "eval_international_law_samples_per_second": 7.502, | |
| "eval_international_law_steps_per_second": 3.751, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 7.305699481865285, | |
| "grad_norm": 0.2451171875, | |
| "learning_rate": 1.273567949383601e-06, | |
| "loss": 0.8384, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 7.357512953367876, | |
| "grad_norm": 0.21875, | |
| "learning_rate": 1.229498700005295e-06, | |
| "loss": 0.8375, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 7.409326424870466, | |
| "grad_norm": 0.2314453125, | |
| "learning_rate": 1.1859554258467843e-06, | |
| "loss": 0.8416, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 7.461139896373057, | |
| "grad_norm": 0.2265625, | |
| "learning_rate": 1.1429561534013869e-06, | |
| "loss": 0.8367, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 7.512953367875648, | |
| "grad_norm": 0.2412109375, | |
| "learning_rate": 1.1005186839509887e-06, | |
| "loss": 0.8372, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 7.512953367875648, | |
| "eval_main_loss": 0.8372209668159485, | |
| "eval_main_runtime": 50.6545, | |
| "eval_main_samples_per_second": 30.323, | |
| "eval_main_steps_per_second": 3.79, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 7.512953367875648, | |
| "eval_anatomy_loss": 2.3963139057159424, | |
| "eval_anatomy_runtime": 0.2679, | |
| "eval_anatomy_samples_per_second": 7.464, | |
| "eval_anatomy_steps_per_second": 3.732, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 7.512953367875648, | |
| "eval_college_mathematics_loss": 1.7459056377410889, | |
| "eval_college_mathematics_runtime": 0.2655, | |
| "eval_college_mathematics_samples_per_second": 7.534, | |
| "eval_college_mathematics_steps_per_second": 3.767, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 7.512953367875648, | |
| "eval_international_law_loss": 2.6337826251983643, | |
| "eval_international_law_runtime": 0.2662, | |
| "eval_international_law_samples_per_second": 7.512, | |
| "eval_international_law_steps_per_second": 3.756, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 7.564766839378239, | |
| "grad_norm": 0.2138671875, | |
| "learning_rate": 1.0586605861964804e-06, | |
| "loss": 0.8313, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 7.616580310880829, | |
| "grad_norm": 0.2431640625, | |
| "learning_rate": 1.01739918898449e-06, | |
| "loss": 0.8346, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 7.668393782383419, | |
| "grad_norm": 0.21875, | |
| "learning_rate": 9.767515741334039e-07, | |
| "loss": 0.8372, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 7.72020725388601, | |
| "grad_norm": 0.2177734375, | |
| "learning_rate": 9.367345693616625e-07, | |
| "loss": 0.8343, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 7.772020725388601, | |
| "grad_norm": 0.24609375, | |
| "learning_rate": 8.973647413212494e-07, | |
| "loss": 0.8441, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 7.772020725388601, | |
| "eval_main_loss": 0.8372817039489746, | |
| "eval_main_runtime": 50.6245, | |
| "eval_main_samples_per_second": 30.341, | |
| "eval_main_steps_per_second": 3.793, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 7.772020725388601, | |
| "eval_anatomy_loss": 2.3954248428344727, | |
| "eval_anatomy_runtime": 0.2651, | |
| "eval_anatomy_samples_per_second": 7.545, | |
| "eval_anatomy_steps_per_second": 3.773, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 7.772020725388601, | |
| "eval_college_mathematics_loss": 1.744511365890503, | |
| "eval_college_mathematics_runtime": 0.2656, | |
| "eval_college_mathematics_samples_per_second": 7.53, | |
| "eval_college_mathematics_steps_per_second": 3.765, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 7.772020725388601, | |
| "eval_international_law_loss": 2.635951280593872, | |
| "eval_international_law_runtime": 0.2679, | |
| "eval_international_law_samples_per_second": 7.464, | |
| "eval_international_law_steps_per_second": 3.732, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 7.823834196891192, | |
| "grad_norm": 0.22265625, | |
| "learning_rate": 8.586583887392546e-07, | |
| "loss": 0.8383, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 7.875647668393782, | |
| "grad_norm": 0.2177734375, | |
| "learning_rate": 8.206315356703634e-07, | |
| "loss": 0.8312, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 7.927461139896373, | |
| "grad_norm": 0.2265625, | |
| "learning_rate": 7.832999248630479e-07, | |
| "loss": 0.8401, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 7.979274611398964, | |
| "grad_norm": 0.2490234375, | |
| "learning_rate": 7.466790112422257e-07, | |
| "loss": 0.8283, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 8.031088082901555, | |
| "grad_norm": 0.2216796875, | |
| "learning_rate": 7.107839555110707e-07, | |
| "loss": 0.8378, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 8.031088082901555, | |
| "eval_main_loss": 0.8372604846954346, | |
| "eval_main_runtime": 50.6329, | |
| "eval_main_samples_per_second": 30.336, | |
| "eval_main_steps_per_second": 3.792, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 8.031088082901555, | |
| "eval_anatomy_loss": 2.397829532623291, | |
| "eval_anatomy_runtime": 0.2664, | |
| "eval_anatomy_samples_per_second": 7.509, | |
| "eval_anatomy_steps_per_second": 3.754, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 8.031088082901555, | |
| "eval_college_mathematics_loss": 1.745050311088562, | |
| "eval_college_mathematics_runtime": 0.2662, | |
| "eval_college_mathematics_samples_per_second": 7.512, | |
| "eval_college_mathematics_steps_per_second": 3.756, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 8.031088082901555, | |
| "eval_international_law_loss": 2.635841131210327, | |
| "eval_international_law_runtime": 0.2664, | |
| "eval_international_law_samples_per_second": 7.508, | |
| "eval_international_law_steps_per_second": 3.754, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 8.082901554404145, | |
| "grad_norm": 0.20703125, | |
| "learning_rate": 6.756296178746282e-07, | |
| "loss": 0.8382, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 8.134715025906736, | |
| "grad_norm": 0.23828125, | |
| "learning_rate": 6.412305518878343e-07, | |
| "loss": 0.8411, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 8.186528497409327, | |
| "grad_norm": 0.2294921875, | |
| "learning_rate": 6.076009984304837e-07, | |
| "loss": 0.8411, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 8.238341968911918, | |
| "grad_norm": 0.2216796875, | |
| "learning_rate": 5.747548798116451e-07, | |
| "loss": 0.8384, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 8.290155440414507, | |
| "grad_norm": 0.2119140625, | |
| "learning_rate": 5.427057940059607e-07, | |
| "loss": 0.8304, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 8.290155440414507, | |
| "eval_main_loss": 0.8372419476509094, | |
| "eval_main_runtime": 50.616, | |
| "eval_main_samples_per_second": 30.346, | |
| "eval_main_steps_per_second": 3.793, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 8.290155440414507, | |
| "eval_anatomy_loss": 2.398320436477661, | |
| "eval_anatomy_runtime": 0.2666, | |
| "eval_anatomy_samples_per_second": 7.501, | |
| "eval_anatomy_steps_per_second": 3.751, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 8.290155440414507, | |
| "eval_college_mathematics_loss": 1.7492754459381104, | |
| "eval_college_mathematics_runtime": 0.2666, | |
| "eval_college_mathematics_samples_per_second": 7.501, | |
| "eval_college_mathematics_steps_per_second": 3.75, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 8.290155440414507, | |
| "eval_international_law_loss": 2.6363821029663086, | |
| "eval_international_law_runtime": 0.266, | |
| "eval_international_law_samples_per_second": 7.519, | |
| "eval_international_law_steps_per_second": 3.76, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 8.341968911917098, | |
| "grad_norm": 0.2255859375, | |
| "learning_rate": 5.11467009024216e-07, | |
| "loss": 0.8361, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 8.393782383419689, | |
| "grad_norm": 0.2294921875, | |
| "learning_rate": 4.810514574205125e-07, | |
| "loss": 0.8339, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 8.44559585492228, | |
| "grad_norm": 0.2294921875, | |
| "learning_rate": 4.5147173093831264e-07, | |
| "loss": 0.8345, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 8.49740932642487, | |
| "grad_norm": 0.2265625, | |
| "learning_rate": 4.227400752975835e-07, | |
| "loss": 0.8374, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 8.549222797927461, | |
| "grad_norm": 0.2294921875, | |
| "learning_rate": 3.9486838512518777e-07, | |
| "loss": 0.8374, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 8.549222797927461, | |
| "eval_main_loss": 0.8372399806976318, | |
| "eval_main_runtime": 50.5872, | |
| "eval_main_samples_per_second": 30.363, | |
| "eval_main_steps_per_second": 3.795, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 8.549222797927461, | |
| "eval_anatomy_loss": 2.3960189819335938, | |
| "eval_anatomy_runtime": 0.2667, | |
| "eval_anatomy_samples_per_second": 7.498, | |
| "eval_anatomy_steps_per_second": 3.749, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 8.549222797927461, | |
| "eval_college_mathematics_loss": 1.7492492198944092, | |
| "eval_college_mathematics_runtime": 0.2676, | |
| "eval_college_mathematics_samples_per_second": 7.475, | |
| "eval_college_mathematics_steps_per_second": 3.738, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 8.549222797927461, | |
| "eval_international_law_loss": 2.6380083560943604, | |
| "eval_international_law_runtime": 0.2658, | |
| "eval_international_law_samples_per_second": 7.525, | |
| "eval_international_law_steps_per_second": 3.762, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 8.601036269430052, | |
| "grad_norm": 0.255859375, | |
| "learning_rate": 3.678681990306207e-07, | |
| "loss": 0.8359, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 8.652849740932643, | |
| "grad_norm": 0.2412109375, | |
| "learning_rate": 3.4175069482914105e-07, | |
| "loss": 0.8284, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 8.704663212435234, | |
| "grad_norm": 0.2138671875, | |
| "learning_rate": 3.165266849142581e-07, | |
| "loss": 0.8334, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 8.756476683937823, | |
| "grad_norm": 0.232421875, | |
| "learning_rate": 2.9220661178151366e-07, | |
| "loss": 0.8337, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 8.808290155440414, | |
| "grad_norm": 0.20703125, | |
| "learning_rate": 2.688005437053845e-07, | |
| "loss": 0.8382, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 8.808290155440414, | |
| "eval_main_loss": 0.8372331261634827, | |
| "eval_main_runtime": 50.6285, | |
| "eval_main_samples_per_second": 30.339, | |
| "eval_main_steps_per_second": 3.792, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 8.808290155440414, | |
| "eval_anatomy_loss": 2.39411997795105, | |
| "eval_anatomy_runtime": 0.2664, | |
| "eval_anatomy_samples_per_second": 7.507, | |
| "eval_anatomy_steps_per_second": 3.753, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 8.808290155440414, | |
| "eval_college_mathematics_loss": 1.7451952695846558, | |
| "eval_college_mathematics_runtime": 0.2673, | |
| "eval_college_mathematics_samples_per_second": 7.482, | |
| "eval_college_mathematics_steps_per_second": 3.741, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 8.808290155440414, | |
| "eval_international_law_loss": 2.6384801864624023, | |
| "eval_international_law_runtime": 0.2666, | |
| "eval_international_law_samples_per_second": 7.502, | |
| "eval_international_law_steps_per_second": 3.751, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 8.860103626943005, | |
| "grad_norm": 0.21875, | |
| "learning_rate": 2.4631817057111597e-07, | |
| "loss": 0.8363, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 8.911917098445596, | |
| "grad_norm": 0.2255859375, | |
| "learning_rate": 2.247687998632031e-07, | |
| "loss": 0.8389, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 8.963730569948186, | |
| "grad_norm": 0.21875, | |
| "learning_rate": 2.0416135281218218e-07, | |
| "loss": 0.8336, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 9.015544041450777, | |
| "grad_norm": 0.236328125, | |
| "learning_rate": 1.8450436070132889e-07, | |
| "loss": 0.8333, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 9.067357512953368, | |
| "grad_norm": 0.23828125, | |
| "learning_rate": 1.6580596133478926e-07, | |
| "loss": 0.8373, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 9.067357512953368, | |
| "eval_main_loss": 0.837254524230957, | |
| "eval_main_runtime": 50.6558, | |
| "eval_main_samples_per_second": 30.322, | |
| "eval_main_steps_per_second": 3.79, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 9.067357512953368, | |
| "eval_anatomy_loss": 2.395867347717285, | |
| "eval_anatomy_runtime": 0.2665, | |
| "eval_anatomy_samples_per_second": 7.506, | |
| "eval_anatomy_steps_per_second": 3.753, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 9.067357512953368, | |
| "eval_college_mathematics_loss": 1.7460345029830933, | |
| "eval_college_mathematics_runtime": 0.267, | |
| "eval_college_mathematics_samples_per_second": 7.491, | |
| "eval_college_mathematics_steps_per_second": 3.746, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 9.067357512953368, | |
| "eval_international_law_loss": 2.6364336013793945, | |
| "eval_international_law_runtime": 0.2659, | |
| "eval_international_law_samples_per_second": 7.522, | |
| "eval_international_law_steps_per_second": 3.761, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 9.119170984455959, | |
| "grad_norm": 0.2431640625, | |
| "learning_rate": 1.4807389566860675e-07, | |
| "loss": 0.838, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 9.17098445595855, | |
| "grad_norm": 0.2158203125, | |
| "learning_rate": 1.3131550460604242e-07, | |
| "loss": 0.8266, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 9.22279792746114, | |
| "grad_norm": 0.224609375, | |
| "learning_rate": 1.1553772595851109e-07, | |
| "loss": 0.8412, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 9.27461139896373, | |
| "grad_norm": 0.2412109375, | |
| "learning_rate": 1.0074709157339657e-07, | |
| "loss": 0.8361, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 9.32642487046632, | |
| "grad_norm": 0.2197265625, | |
| "learning_rate": 8.694972462992918e-08, | |
| "loss": 0.8373, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 9.32642487046632, | |
| "eval_main_loss": 0.8372175693511963, | |
| "eval_main_runtime": 50.62, | |
| "eval_main_samples_per_second": 30.344, | |
| "eval_main_steps_per_second": 3.793, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 9.32642487046632, | |
| "eval_anatomy_loss": 2.394951105117798, | |
| "eval_anatomy_runtime": 0.2668, | |
| "eval_anatomy_samples_per_second": 7.496, | |
| "eval_anatomy_steps_per_second": 3.748, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 9.32642487046632, | |
| "eval_college_mathematics_loss": 1.7458065748214722, | |
| "eval_college_mathematics_runtime": 0.2669, | |
| "eval_college_mathematics_samples_per_second": 7.494, | |
| "eval_college_mathematics_steps_per_second": 3.747, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 9.32642487046632, | |
| "eval_international_law_loss": 2.639335870742798, | |
| "eval_international_law_runtime": 0.2665, | |
| "eval_international_law_samples_per_second": 7.505, | |
| "eval_international_law_steps_per_second": 3.752, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 9.378238341968911, | |
| "grad_norm": 0.2431640625, | |
| "learning_rate": 7.415133710424794e-08, | |
| "loss": 0.8335, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 9.430051813471502, | |
| "grad_norm": 0.248046875, | |
| "learning_rate": 6.235722740469936e-08, | |
| "loss": 0.8379, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 9.481865284974093, | |
| "grad_norm": 0.23046875, | |
| "learning_rate": 5.157227817834648e-08, | |
| "loss": 0.831, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 9.533678756476684, | |
| "grad_norm": 0.2138671875, | |
| "learning_rate": 4.180095428960168e-08, | |
| "loss": 0.8399, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 9.585492227979275, | |
| "grad_norm": 0.2216796875, | |
| "learning_rate": 3.304730097181463e-08, | |
| "loss": 0.8428, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 9.585492227979275, | |
| "eval_main_loss": 0.8372709155082703, | |
| "eval_main_runtime": 50.6434, | |
| "eval_main_samples_per_second": 30.33, | |
| "eval_main_steps_per_second": 3.791, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 9.585492227979275, | |
| "eval_anatomy_loss": 2.3959925174713135, | |
| "eval_anatomy_runtime": 0.2686, | |
| "eval_anatomy_samples_per_second": 7.447, | |
| "eval_anatomy_steps_per_second": 3.723, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 9.585492227979275, | |
| "eval_college_mathematics_loss": 1.7489873170852661, | |
| "eval_college_mathematics_runtime": 0.2656, | |
| "eval_college_mathematics_samples_per_second": 7.529, | |
| "eval_college_mathematics_steps_per_second": 3.765, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 9.585492227979275, | |
| "eval_international_law_loss": 2.6358718872070312, | |
| "eval_international_law_runtime": 0.2658, | |
| "eval_international_law_samples_per_second": 7.525, | |
| "eval_international_law_steps_per_second": 3.763, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 9.637305699481866, | |
| "grad_norm": 0.2373046875, | |
| "learning_rate": 2.5314942152586954e-08, | |
| "loss": 0.8366, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 9.689119170984457, | |
| "grad_norm": 0.208984375, | |
| "learning_rate": 1.8607078953498392e-08, | |
| "loss": 0.8386, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 9.740932642487046, | |
| "grad_norm": 0.2421875, | |
| "learning_rate": 1.292648836487609e-08, | |
| "loss": 0.8321, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 9.792746113989637, | |
| "grad_norm": 0.2412109375, | |
| "learning_rate": 8.275522096146404e-09, | |
| "loss": 0.8342, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 9.844559585492227, | |
| "grad_norm": 0.2275390625, | |
| "learning_rate": 4.656105602250382e-09, | |
| "loss": 0.8404, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 9.844559585492227, | |
| "eval_main_loss": 0.8372488021850586, | |
| "eval_main_runtime": 50.6539, | |
| "eval_main_samples_per_second": 30.323, | |
| "eval_main_steps_per_second": 3.79, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 9.844559585492227, | |
| "eval_anatomy_loss": 2.3990156650543213, | |
| "eval_anatomy_runtime": 0.2664, | |
| "eval_anatomy_samples_per_second": 7.508, | |
| "eval_anatomy_steps_per_second": 3.754, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 9.844559585492227, | |
| "eval_college_mathematics_loss": 1.7470096349716187, | |
| "eval_college_mathematics_runtime": 0.2654, | |
| "eval_college_mathematics_samples_per_second": 7.535, | |
| "eval_college_mathematics_steps_per_second": 3.767, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 9.844559585492227, | |
| "eval_international_law_loss": 2.6360206604003906, | |
| "eval_international_law_runtime": 0.2672, | |
| "eval_international_law_samples_per_second": 7.485, | |
| "eval_international_law_steps_per_second": 3.743, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 9.896373056994818, | |
| "grad_norm": 0.2138671875, | |
| "learning_rate": 2.0697372865235986e-09, | |
| "loss": 0.8301, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 9.94818652849741, | |
| "grad_norm": 0.2265625, | |
| "learning_rate": 5.174878803720917e-10, | |
| "loss": 0.8299, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "grad_norm": 0.201171875, | |
| "learning_rate": 0.0, | |
| "loss": 0.832, | |
| "step": 3860 | |
| } | |
| ], | |
| "logging_steps": 20, | |
| "max_steps": 3860, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.8346399786239263e+18, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |