| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.11904761904761904, | |
| "eval_steps": 500, | |
| "global_step": 3000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 23.495065689086914, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 8.6232, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 24.87381935119629, | |
| "learning_rate": 9e-07, | |
| "loss": 8.4514, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 22.63846778869629, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 8.3152, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 20.65507698059082, | |
| "learning_rate": 1.9e-06, | |
| "loss": 7.9732, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 16.147356033325195, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 7.6879, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 14.252687454223633, | |
| "learning_rate": 2.9e-06, | |
| "loss": 7.4152, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 12.011382102966309, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 7.2472, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 10.081536293029785, | |
| "learning_rate": 3.9e-06, | |
| "loss": 7.2433, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 9.65556526184082, | |
| "learning_rate": 4.4e-06, | |
| "loss": 7.2298, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 8.830799102783203, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 7.19, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 9.019427299499512, | |
| "learning_rate": 5.4e-06, | |
| "loss": 7.2362, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 9.13768482208252, | |
| "learning_rate": 5.9e-06, | |
| "loss": 7.2804, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 9.02840805053711, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 7.2345, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 8.178731918334961, | |
| "learning_rate": 6.900000000000001e-06, | |
| "loss": 7.3014, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 8.83882999420166, | |
| "learning_rate": 7.4e-06, | |
| "loss": 7.3875, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 8.3687162399292, | |
| "learning_rate": 7.9e-06, | |
| "loss": 7.4546, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 8.09121036529541, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 7.407, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 12.731450080871582, | |
| "learning_rate": 8.9e-06, | |
| "loss": 7.4321, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 7.601162910461426, | |
| "learning_rate": 9.4e-06, | |
| "loss": 7.4155, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 8.006213188171387, | |
| "learning_rate": 9.900000000000002e-06, | |
| "loss": 7.3705, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 7.79794454574585, | |
| "learning_rate": 1.04e-05, | |
| "loss": 7.3759, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 7.281346321105957, | |
| "learning_rate": 1.09e-05, | |
| "loss": 7.3798, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 7.605266094207764, | |
| "learning_rate": 1.1400000000000001e-05, | |
| "loss": 7.3878, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 7.446412563323975, | |
| "learning_rate": 1.19e-05, | |
| "loss": 7.3098, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 7.6324334144592285, | |
| "learning_rate": 1.24e-05, | |
| "loss": 7.2995, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 8.330737113952637, | |
| "learning_rate": 1.29e-05, | |
| "loss": 7.1887, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 7.283708572387695, | |
| "learning_rate": 1.3400000000000002e-05, | |
| "loss": 7.1184, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 6.950069904327393, | |
| "learning_rate": 1.3900000000000002e-05, | |
| "loss": 6.9876, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 6.844958782196045, | |
| "learning_rate": 1.44e-05, | |
| "loss": 6.9319, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 7.499362468719482, | |
| "learning_rate": 1.49e-05, | |
| "loss": 6.9253, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 8.45572280883789, | |
| "learning_rate": 1.54e-05, | |
| "loss": 6.7999, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 8.108625411987305, | |
| "learning_rate": 1.59e-05, | |
| "loss": 6.7667, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 8.131596565246582, | |
| "learning_rate": 1.6400000000000002e-05, | |
| "loss": 6.7599, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 7.2159247398376465, | |
| "learning_rate": 1.69e-05, | |
| "loss": 6.7379, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 7.335048675537109, | |
| "learning_rate": 1.74e-05, | |
| "loss": 6.6553, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 7.361081600189209, | |
| "learning_rate": 1.79e-05, | |
| "loss": 6.6905, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 7.59442138671875, | |
| "learning_rate": 1.84e-05, | |
| "loss": 6.6356, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 7.31695032119751, | |
| "learning_rate": 1.8900000000000002e-05, | |
| "loss": 6.6538, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 7.541380882263184, | |
| "learning_rate": 1.94e-05, | |
| "loss": 6.6873, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 8.056180000305176, | |
| "learning_rate": 1.9900000000000003e-05, | |
| "loss": 6.6332, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 6.74859094619751, | |
| "learning_rate": 2.04e-05, | |
| "loss": 6.5488, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 7.700525760650635, | |
| "learning_rate": 2.09e-05, | |
| "loss": 6.611, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 7.347009181976318, | |
| "learning_rate": 2.1400000000000002e-05, | |
| "loss": 6.6621, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 6.607337951660156, | |
| "learning_rate": 2.19e-05, | |
| "loss": 6.568, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 6.816825866699219, | |
| "learning_rate": 2.2400000000000002e-05, | |
| "loss": 6.6293, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 6.984443664550781, | |
| "learning_rate": 2.29e-05, | |
| "loss": 6.5667, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 6.995021820068359, | |
| "learning_rate": 2.3400000000000003e-05, | |
| "loss": 6.5062, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 6.876286029815674, | |
| "learning_rate": 2.39e-05, | |
| "loss": 6.5034, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 6.953365802764893, | |
| "learning_rate": 2.44e-05, | |
| "loss": 6.459, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 7.246520519256592, | |
| "learning_rate": 2.4900000000000002e-05, | |
| "loss": 6.4359, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "eval_loss": 6.133457660675049, | |
| "eval_runtime": 422.8798, | |
| "eval_samples_per_second": 50.182, | |
| "eval_steps_per_second": 0.393, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 7.1657819747924805, | |
| "learning_rate": 2.54e-05, | |
| "loss": 6.4372, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 7.203813552856445, | |
| "learning_rate": 2.5900000000000003e-05, | |
| "loss": 6.3684, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 6.829228401184082, | |
| "learning_rate": 2.64e-05, | |
| "loss": 6.3275, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 6.641738414764404, | |
| "learning_rate": 2.6900000000000003e-05, | |
| "loss": 6.2923, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 7.824960708618164, | |
| "learning_rate": 2.7400000000000002e-05, | |
| "loss": 6.3685, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 6.7754364013671875, | |
| "learning_rate": 2.7900000000000004e-05, | |
| "loss": 6.3069, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 7.3608903884887695, | |
| "learning_rate": 2.84e-05, | |
| "loss": 6.2915, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 6.551064968109131, | |
| "learning_rate": 2.8899999999999998e-05, | |
| "loss": 6.1676, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 6.588657379150391, | |
| "learning_rate": 2.94e-05, | |
| "loss": 6.2311, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 6.631275177001953, | |
| "learning_rate": 2.9900000000000002e-05, | |
| "loss": 6.1824, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 6.2990288734436035, | |
| "learning_rate": 3.04e-05, | |
| "loss": 6.1641, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 6.6439714431762695, | |
| "learning_rate": 3.09e-05, | |
| "loss": 6.0125, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.402431964874268, | |
| "learning_rate": 3.1400000000000004e-05, | |
| "loss": 6.1314, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 5.883937358856201, | |
| "learning_rate": 3.19e-05, | |
| "loss": 5.9858, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 5.988973140716553, | |
| "learning_rate": 3.24e-05, | |
| "loss": 5.9886, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.186301231384277, | |
| "learning_rate": 3.29e-05, | |
| "loss": 5.9732, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.2982354164123535, | |
| "learning_rate": 3.3400000000000005e-05, | |
| "loss": 6.0391, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.037730693817139, | |
| "learning_rate": 3.3900000000000004e-05, | |
| "loss": 6.0547, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.280949592590332, | |
| "learning_rate": 3.4399999999999996e-05, | |
| "loss": 5.8734, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.861902236938477, | |
| "learning_rate": 3.49e-05, | |
| "loss": 5.8997, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.4033589363098145, | |
| "learning_rate": 3.54e-05, | |
| "loss": 5.8111, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.376819133758545, | |
| "learning_rate": 3.59e-05, | |
| "loss": 5.9027, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 5.777069568634033, | |
| "learning_rate": 3.6400000000000004e-05, | |
| "loss": 5.727, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.183162689208984, | |
| "learning_rate": 3.69e-05, | |
| "loss": 5.733, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.658604621887207, | |
| "learning_rate": 3.74e-05, | |
| "loss": 5.7018, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.263491153717041, | |
| "learning_rate": 3.79e-05, | |
| "loss": 5.7315, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 5.86756706237793, | |
| "learning_rate": 3.8400000000000005e-05, | |
| "loss": 5.6637, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.122657299041748, | |
| "learning_rate": 3.8900000000000004e-05, | |
| "loss": 5.6849, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 5.6024580001831055, | |
| "learning_rate": 3.94e-05, | |
| "loss": 5.5835, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 5.619604587554932, | |
| "learning_rate": 3.99e-05, | |
| "loss": 5.7, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 5.7342352867126465, | |
| "learning_rate": 4.0400000000000006e-05, | |
| "loss": 5.4825, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.796237468719482, | |
| "learning_rate": 4.09e-05, | |
| "loss": 5.6496, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 5.601135730743408, | |
| "learning_rate": 4.14e-05, | |
| "loss": 5.3526, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 5.576679229736328, | |
| "learning_rate": 4.19e-05, | |
| "loss": 5.3757, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 5.774380683898926, | |
| "learning_rate": 4.24e-05, | |
| "loss": 5.5164, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.140038967132568, | |
| "learning_rate": 4.29e-05, | |
| "loss": 5.3124, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 5.971868515014648, | |
| "learning_rate": 4.3400000000000005e-05, | |
| "loss": 5.3387, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.431824207305908, | |
| "learning_rate": 4.39e-05, | |
| "loss": 5.4363, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.174604892730713, | |
| "learning_rate": 4.44e-05, | |
| "loss": 5.282, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.313474178314209, | |
| "learning_rate": 4.49e-05, | |
| "loss": 5.3754, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.066281318664551, | |
| "learning_rate": 4.5400000000000006e-05, | |
| "loss": 5.3197, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 7.4433817863464355, | |
| "learning_rate": 4.5900000000000004e-05, | |
| "loss": 5.3277, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 5.884599208831787, | |
| "learning_rate": 4.64e-05, | |
| "loss": 5.0681, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.17655611038208, | |
| "learning_rate": 4.69e-05, | |
| "loss": 5.1159, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 7.090456485748291, | |
| "learning_rate": 4.74e-05, | |
| "loss": 5.2264, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.508240699768066, | |
| "learning_rate": 4.79e-05, | |
| "loss": 5.2113, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.5788445472717285, | |
| "learning_rate": 4.8400000000000004e-05, | |
| "loss": 4.9996, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 5.722221374511719, | |
| "learning_rate": 4.89e-05, | |
| "loss": 4.9688, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 5.813933849334717, | |
| "learning_rate": 4.94e-05, | |
| "loss": 4.8691, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.425748348236084, | |
| "learning_rate": 4.99e-05, | |
| "loss": 4.9883, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "eval_loss": 4.425854682922363, | |
| "eval_runtime": 422.6592, | |
| "eval_samples_per_second": 50.208, | |
| "eval_steps_per_second": 0.393, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.170848369598389, | |
| "learning_rate": 4.9999996764541565e-05, | |
| "loss": 4.8794, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.247808933258057, | |
| "learning_rate": 4.99999836204931e-05, | |
| "loss": 4.9664, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.27077054977417, | |
| "learning_rate": 4.9999960365643774e-05, | |
| "loss": 4.7753, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.284353733062744, | |
| "learning_rate": 4.999992700000298e-05, | |
| "loss": 4.8909, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 5.716318607330322, | |
| "learning_rate": 4.999988352358421e-05, | |
| "loss": 4.861, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.448729991912842, | |
| "learning_rate": 4.999982993640506e-05, | |
| "loss": 4.8776, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 5.401464939117432, | |
| "learning_rate": 4.9999766238487204e-05, | |
| "loss": 4.8507, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.219631195068359, | |
| "learning_rate": 4.999969242985639e-05, | |
| "loss": 4.8234, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 5.225848197937012, | |
| "learning_rate": 4.9999608510542485e-05, | |
| "loss": 4.6093, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.117152690887451, | |
| "learning_rate": 4.999951448057941e-05, | |
| "loss": 4.786, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 5.788968086242676, | |
| "learning_rate": 4.999941034000521e-05, | |
| "loss": 4.8389, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.037156581878662, | |
| "learning_rate": 4.999929608886198e-05, | |
| "loss": 4.7674, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 5.567563056945801, | |
| "learning_rate": 4.999917172719596e-05, | |
| "loss": 4.8247, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 5.557987213134766, | |
| "learning_rate": 4.999903725505742e-05, | |
| "loss": 4.7517, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 6.490599632263184, | |
| "learning_rate": 4.999889267250075e-05, | |
| "loss": 4.6037, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 6.000855922698975, | |
| "learning_rate": 4.999873797958443e-05, | |
| "loss": 4.6077, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 5.378462314605713, | |
| "learning_rate": 4.999857317637102e-05, | |
| "loss": 4.3662, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 7.4290924072265625, | |
| "learning_rate": 4.999839826292718e-05, | |
| "loss": 4.5339, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 5.727224826812744, | |
| "learning_rate": 4.999821323932363e-05, | |
| "loss": 4.5744, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 5.893082618713379, | |
| "learning_rate": 4.9998018105635214e-05, | |
| "loss": 4.5247, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 5.759618759155273, | |
| "learning_rate": 4.999781286194085e-05, | |
| "loss": 4.5744, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 5.082376003265381, | |
| "learning_rate": 4.999759750832355e-05, | |
| "loss": 4.6011, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 6.305749416351318, | |
| "learning_rate": 4.999737204487039e-05, | |
| "loss": 4.4587, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 6.710574626922607, | |
| "learning_rate": 4.999713647167258e-05, | |
| "loss": 4.46, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 6.753530502319336, | |
| "learning_rate": 4.999689078882537e-05, | |
| "loss": 4.5674, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 5.949398040771484, | |
| "learning_rate": 4.999663499642814e-05, | |
| "loss": 4.3999, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 6.530541896820068, | |
| "learning_rate": 4.999636909458434e-05, | |
| "loss": 4.533, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 5.9646453857421875, | |
| "learning_rate": 4.9996093083401495e-05, | |
| "loss": 4.374, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 6.346386432647705, | |
| "learning_rate": 4.9995806962991246e-05, | |
| "loss": 4.4758, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 5.797152519226074, | |
| "learning_rate": 4.9995510733469295e-05, | |
| "loss": 4.6275, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 6.426751136779785, | |
| "learning_rate": 4.999520439495547e-05, | |
| "loss": 4.3561, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 5.3839030265808105, | |
| "learning_rate": 4.9994887947573646e-05, | |
| "loss": 4.3134, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 6.01127290725708, | |
| "learning_rate": 4.9994594501953555e-05, | |
| "loss": 4.4336, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 5.228607654571533, | |
| "learning_rate": 4.999425884807851e-05, | |
| "loss": 4.1543, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 5.958486080169678, | |
| "learning_rate": 4.999391308571788e-05, | |
| "loss": 4.3806, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 5.816386699676514, | |
| "learning_rate": 4.99935572150115e-05, | |
| "loss": 4.3759, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 6.345638751983643, | |
| "learning_rate": 4.9993191236103295e-05, | |
| "loss": 4.5759, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 5.738880634307861, | |
| "learning_rate": 4.9992815149141294e-05, | |
| "loss": 4.3274, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 6.694160461425781, | |
| "learning_rate": 4.9992428954277573e-05, | |
| "loss": 4.376, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.050902843475342, | |
| "learning_rate": 4.999203265166834e-05, | |
| "loss": 4.1629, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 7.09380578994751, | |
| "learning_rate": 4.9991626241473876e-05, | |
| "loss": 4.3638, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.538166046142578, | |
| "learning_rate": 4.999120972385853e-05, | |
| "loss": 4.1318, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 6.052664279937744, | |
| "learning_rate": 4.9990783098990765e-05, | |
| "loss": 4.1856, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 6.707901477813721, | |
| "learning_rate": 4.9990346367043114e-05, | |
| "loss": 4.3361, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.3822150230407715, | |
| "learning_rate": 4.998989952819223e-05, | |
| "loss": 4.301, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.586170196533203, | |
| "learning_rate": 4.998944258261879e-05, | |
| "loss": 4.3131, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.62904691696167, | |
| "learning_rate": 4.998897553050763e-05, | |
| "loss": 4.3357, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.669075965881348, | |
| "learning_rate": 4.9988498372047633e-05, | |
| "loss": 4.2434, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.776549339294434, | |
| "learning_rate": 4.998801110743177e-05, | |
| "loss": 4.0441, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.442574977874756, | |
| "learning_rate": 4.9987513736857106e-05, | |
| "loss": 4.0931, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "eval_loss": 3.676598310470581, | |
| "eval_runtime": 423.4536, | |
| "eval_samples_per_second": 50.114, | |
| "eval_steps_per_second": 0.392, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.529382705688477, | |
| "learning_rate": 4.998700626052481e-05, | |
| "loss": 4.1739, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.5605387687683105, | |
| "learning_rate": 4.99864886786401e-05, | |
| "loss": 4.1861, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.168206214904785, | |
| "learning_rate": 4.998596099141231e-05, | |
| "loss": 4.0245, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.634309768676758, | |
| "learning_rate": 4.998542319905486e-05, | |
| "loss": 4.2131, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 6.191194534301758, | |
| "learning_rate": 4.998487530178525e-05, | |
| "loss": 4.163, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 6.52882719039917, | |
| "learning_rate": 4.998431729982506e-05, | |
| "loss": 4.0668, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.614863395690918, | |
| "learning_rate": 4.998374919339998e-05, | |
| "loss": 4.2501, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 4.7852373123168945, | |
| "learning_rate": 4.9983170982739745e-05, | |
| "loss": 4.207, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 7.439071178436279, | |
| "learning_rate": 4.998258266807822e-05, | |
| "loss": 4.0462, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.510477542877197, | |
| "learning_rate": 4.998198424965333e-05, | |
| "loss": 4.277, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.421988487243652, | |
| "learning_rate": 4.9981375727707114e-05, | |
| "loss": 4.0289, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.632848739624023, | |
| "learning_rate": 4.998075710248565e-05, | |
| "loss": 4.2922, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 6.031602382659912, | |
| "learning_rate": 4.9980128374239156e-05, | |
| "loss": 3.9845, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.3867669105529785, | |
| "learning_rate": 4.997948954322189e-05, | |
| "loss": 4.1095, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.150663375854492, | |
| "learning_rate": 4.997884060969222e-05, | |
| "loss": 3.9247, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.717851638793945, | |
| "learning_rate": 4.99781815739126e-05, | |
| "loss": 4.1911, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 6.766101837158203, | |
| "learning_rate": 4.9977512436149563e-05, | |
| "loss": 3.9194, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 6.316621780395508, | |
| "learning_rate": 4.9976833196673735e-05, | |
| "loss": 4.101, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.107331275939941, | |
| "learning_rate": 4.9976143855759814e-05, | |
| "loss": 4.0706, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.533346652984619, | |
| "learning_rate": 4.99754444136866e-05, | |
| "loss": 4.0164, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.967229843139648, | |
| "learning_rate": 4.9974734870736975e-05, | |
| "loss": 4.1229, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.502392768859863, | |
| "learning_rate": 4.997401522719788e-05, | |
| "loss": 4.0456, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.6358962059021, | |
| "learning_rate": 4.997328548336038e-05, | |
| "loss": 3.9432, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 6.525423049926758, | |
| "learning_rate": 4.99725456395196e-05, | |
| "loss": 4.05, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 4.675175666809082, | |
| "learning_rate": 4.997179569597475e-05, | |
| "loss": 3.8145, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.409067630767822, | |
| "learning_rate": 4.9971035653029153e-05, | |
| "loss": 3.914, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.812373161315918, | |
| "learning_rate": 4.997026551099017e-05, | |
| "loss": 3.9817, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.165591239929199, | |
| "learning_rate": 4.996948527016929e-05, | |
| "loss": 3.9414, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.0605034828186035, | |
| "learning_rate": 4.996869493088205e-05, | |
| "loss": 3.8139, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 6.348527908325195, | |
| "learning_rate": 4.996789449344812e-05, | |
| "loss": 3.7905, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.135634422302246, | |
| "learning_rate": 4.9967083958191187e-05, | |
| "loss": 3.7834, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 8.48004150390625, | |
| "learning_rate": 4.996626332543907e-05, | |
| "loss": 3.7551, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.518482208251953, | |
| "learning_rate": 4.996543259552367e-05, | |
| "loss": 3.8619, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.7253737449646, | |
| "learning_rate": 4.996459176878095e-05, | |
| "loss": 3.7978, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 6.030267715454102, | |
| "learning_rate": 4.9963740845550965e-05, | |
| "loss": 3.9309, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.561445713043213, | |
| "learning_rate": 4.996287982617787e-05, | |
| "loss": 3.9761, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.555076599121094, | |
| "learning_rate": 4.996200871100988e-05, | |
| "loss": 3.8587, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.502940654754639, | |
| "learning_rate": 4.9961127500399295e-05, | |
| "loss": 3.847, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 5.557964324951172, | |
| "learning_rate": 4.996023619470252e-05, | |
| "loss": 3.6117, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 6.37682580947876, | |
| "learning_rate": 4.9959334794280014e-05, | |
| "loss": 3.7581, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 6.152722358703613, | |
| "learning_rate": 4.9958423299496344e-05, | |
| "loss": 3.8454, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 5.4067301750183105, | |
| "learning_rate": 4.995750171072014e-05, | |
| "loss": 3.9353, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 5.564638614654541, | |
| "learning_rate": 4.9956570028324124e-05, | |
| "loss": 3.738, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 5.320713520050049, | |
| "learning_rate": 4.99556282526851e-05, | |
| "loss": 3.5365, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 6.384575843811035, | |
| "learning_rate": 4.995467638418395e-05, | |
| "loss": 3.8153, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 6.424959659576416, | |
| "learning_rate": 4.995371442320565e-05, | |
| "loss": 3.779, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 5.928106784820557, | |
| "learning_rate": 4.9952742370139224e-05, | |
| "loss": 3.7395, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 6.09404182434082, | |
| "learning_rate": 4.995176022537784e-05, | |
| "loss": 3.757, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 8.332275390625, | |
| "learning_rate": 4.995076798931867e-05, | |
| "loss": 3.8866, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 6.44631814956665, | |
| "learning_rate": 4.994976566236303e-05, | |
| "loss": 3.8395, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "eval_loss": 3.2655727863311768, | |
| "eval_runtime": 423.1622, | |
| "eval_samples_per_second": 50.149, | |
| "eval_steps_per_second": 0.392, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 5.428518295288086, | |
| "learning_rate": 4.994875324491629e-05, | |
| "loss": 3.7103, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 6.11751651763916, | |
| "learning_rate": 4.9947730737387885e-05, | |
| "loss": 3.8224, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 5.404431343078613, | |
| "learning_rate": 4.994669814019138e-05, | |
| "loss": 3.9208, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 5.487029075622559, | |
| "learning_rate": 4.994565545374438e-05, | |
| "loss": 3.7412, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 4.979548454284668, | |
| "learning_rate": 4.9944602678468575e-05, | |
| "loss": 3.6008, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 5.379942417144775, | |
| "learning_rate": 4.994353981478974e-05, | |
| "loss": 3.6077, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 5.666346549987793, | |
| "learning_rate": 4.994246686313774e-05, | |
| "loss": 3.8157, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 5.252011299133301, | |
| "learning_rate": 4.99413838239465e-05, | |
| "loss": 3.6929, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 6.579212188720703, | |
| "learning_rate": 4.9940290697654045e-05, | |
| "loss": 3.9075, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 5.747458457946777, | |
| "learning_rate": 4.993918748470246e-05, | |
| "loss": 3.5528, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 5.356452465057373, | |
| "learning_rate": 4.993818596932118e-05, | |
| "loss": 3.4987, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 5.761812686920166, | |
| "learning_rate": 4.993706359294986e-05, | |
| "loss": 3.7627, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 5.732122898101807, | |
| "learning_rate": 4.993593113122456e-05, | |
| "loss": 3.6106, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 11.450267791748047, | |
| "learning_rate": 4.993478858460329e-05, | |
| "loss": 3.4716, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.774083614349365, | |
| "learning_rate": 4.993363595354813e-05, | |
| "loss": 3.5741, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.652318000793457, | |
| "learning_rate": 4.9932473238525247e-05, | |
| "loss": 3.676, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.903106689453125, | |
| "learning_rate": 4.993130044000487e-05, | |
| "loss": 3.6334, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.358557224273682, | |
| "learning_rate": 4.993011755846132e-05, | |
| "loss": 3.5592, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 6.509594440460205, | |
| "learning_rate": 4.9928924594372995e-05, | |
| "loss": 3.6512, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 6.318192958831787, | |
| "learning_rate": 4.9927721548222374e-05, | |
| "loss": 3.3691, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.848221778869629, | |
| "learning_rate": 4.9926508420496e-05, | |
| "loss": 3.3721, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.9782490730285645, | |
| "learning_rate": 4.992528521168449e-05, | |
| "loss": 3.5859, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 6.406031608581543, | |
| "learning_rate": 4.992405192228258e-05, | |
| "loss": 3.6206, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 6.102032661437988, | |
| "learning_rate": 4.992280855278903e-05, | |
| "loss": 3.4615, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.5005035400390625, | |
| "learning_rate": 4.992155510370669e-05, | |
| "loss": 3.507, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 8.336762428283691, | |
| "learning_rate": 4.9920291575542515e-05, | |
| "loss": 3.3897, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 7.851747989654541, | |
| "learning_rate": 4.9919017968807514e-05, | |
| "loss": 3.5232, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.816250324249268, | |
| "learning_rate": 4.991773428401676e-05, | |
| "loss": 3.5638, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 4.746279716491699, | |
| "learning_rate": 4.9916440521689433e-05, | |
| "loss": 3.5446, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.8899922370910645, | |
| "learning_rate": 4.9915136682348774e-05, | |
| "loss": 3.3533, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 6.160922050476074, | |
| "learning_rate": 4.9913822766522076e-05, | |
| "loss": 3.4222, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 6.936460494995117, | |
| "learning_rate": 4.991249877474075e-05, | |
| "loss": 3.5763, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 16.517559051513672, | |
| "learning_rate": 4.991116470754025e-05, | |
| "loss": 3.5324, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 4.840503692626953, | |
| "learning_rate": 4.9909820565460116e-05, | |
| "loss": 3.5764, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.952908515930176, | |
| "learning_rate": 4.990846634904398e-05, | |
| "loss": 3.4369, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.108431339263916, | |
| "learning_rate": 4.9907102058839504e-05, | |
| "loss": 3.4518, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.068904399871826, | |
| "learning_rate": 4.990572769539848e-05, | |
| "loss": 3.5594, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.512117385864258, | |
| "learning_rate": 4.990434325927671e-05, | |
| "loss": 3.6366, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.916763782501221, | |
| "learning_rate": 4.990294875103413e-05, | |
| "loss": 3.4344, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 6.244757175445557, | |
| "learning_rate": 4.990154417123472e-05, | |
| "loss": 3.5858, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 6.071867942810059, | |
| "learning_rate": 4.990012952044654e-05, | |
| "loss": 3.5888, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 5.984373092651367, | |
| "learning_rate": 4.98987047992417e-05, | |
| "loss": 3.5084, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 7.062913417816162, | |
| "learning_rate": 4.989727000819644e-05, | |
| "loss": 3.4169, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 4.671230792999268, | |
| "learning_rate": 4.9895825147891e-05, | |
| "loss": 3.3354, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 5.235222339630127, | |
| "learning_rate": 4.989437021890975e-05, | |
| "loss": 3.4756, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 5.888786792755127, | |
| "learning_rate": 4.98929052218411e-05, | |
| "loss": 3.3409, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 6.014183044433594, | |
| "learning_rate": 4.989143015727754e-05, | |
| "loss": 3.4246, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 8.75373649597168, | |
| "learning_rate": 4.988994502581565e-05, | |
| "loss": 3.2431, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 5.800814628601074, | |
| "learning_rate": 4.988844982805605e-05, | |
| "loss": 3.4702, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 5.22312068939209, | |
| "learning_rate": 4.9886944564603444e-05, | |
| "loss": 3.2994, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_loss": 2.9097418785095215, | |
| "eval_runtime": 423.0547, | |
| "eval_samples_per_second": 50.161, | |
| "eval_steps_per_second": 0.392, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 5.41413688659668, | |
| "learning_rate": 4.988542923606663e-05, | |
| "loss": 3.5138, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 5.5836968421936035, | |
| "learning_rate": 4.988390384305843e-05, | |
| "loss": 3.4161, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 5.957060813903809, | |
| "learning_rate": 4.9882368386195776e-05, | |
| "loss": 3.4227, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 5.718446731567383, | |
| "learning_rate": 4.988082286609965e-05, | |
| "loss": 3.3632, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 6.273780822753906, | |
| "learning_rate": 4.9879267283395125e-05, | |
| "loss": 3.4018, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 5.000144004821777, | |
| "learning_rate": 4.9877701638711314e-05, | |
| "loss": 3.2865, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 6.369386672973633, | |
| "learning_rate": 4.987612593268142e-05, | |
| "loss": 3.2916, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 5.275691032409668, | |
| "learning_rate": 4.98745401659427e-05, | |
| "loss": 3.3678, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 6.77783727645874, | |
| "learning_rate": 4.9872944339136503e-05, | |
| "loss": 3.2567, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 6.3765435218811035, | |
| "learning_rate": 4.987133845290822e-05, | |
| "loss": 3.3649, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 6.956243991851807, | |
| "learning_rate": 4.9869722507907343e-05, | |
| "loss": 3.428, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 4.933529376983643, | |
| "learning_rate": 4.9868096504787395e-05, | |
| "loss": 3.0448, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 6.319052696228027, | |
| "learning_rate": 4.9866460444205983e-05, | |
| "loss": 3.3518, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 5.592201232910156, | |
| "learning_rate": 4.986481432682479e-05, | |
| "loss": 3.1236, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.243892192840576, | |
| "learning_rate": 4.986315815330956e-05, | |
| "loss": 3.2447, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 6.197099208831787, | |
| "learning_rate": 4.98614919243301e-05, | |
| "loss": 3.3031, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 6.438327312469482, | |
| "learning_rate": 4.98598156405603e-05, | |
| "loss": 3.5433, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.134957313537598, | |
| "learning_rate": 4.985812930267808e-05, | |
| "loss": 3.3124, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.727309703826904, | |
| "learning_rate": 4.985643291136546e-05, | |
| "loss": 3.2169, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.680230617523193, | |
| "learning_rate": 4.985472646730852e-05, | |
| "loss": 3.0445, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.9644598960876465, | |
| "learning_rate": 4.985300997119739e-05, | |
| "loss": 3.1588, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.430914402008057, | |
| "learning_rate": 4.985128342372629e-05, | |
| "loss": 3.2385, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.565753936767578, | |
| "learning_rate": 4.9849546825593476e-05, | |
| "loss": 3.2654, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 6.126991271972656, | |
| "learning_rate": 4.9847800177501305e-05, | |
| "loss": 3.3024, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.406001567840576, | |
| "learning_rate": 4.984604348015616e-05, | |
| "loss": 3.219, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.55892276763916, | |
| "learning_rate": 4.984427673426851e-05, | |
| "loss": 3.3882, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.547699451446533, | |
| "learning_rate": 4.9842499940552886e-05, | |
| "loss": 3.2334, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 6.514155864715576, | |
| "learning_rate": 4.9840713099727885e-05, | |
| "loss": 3.2041, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.4210991859436035, | |
| "learning_rate": 4.983891621251616e-05, | |
| "loss": 3.1746, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.854736328125, | |
| "learning_rate": 4.983710927964442e-05, | |
| "loss": 3.1961, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.277487754821777, | |
| "learning_rate": 4.9835292301843476e-05, | |
| "loss": 2.9688, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.925424098968506, | |
| "learning_rate": 4.983346527984813e-05, | |
| "loss": 3.2315, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 7.818734169006348, | |
| "learning_rate": 4.983162821439733e-05, | |
| "loss": 3.325, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 6.312294006347656, | |
| "learning_rate": 4.982978110623402e-05, | |
| "loss": 3.2141, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 6.804452419281006, | |
| "learning_rate": 4.982792395610524e-05, | |
| "loss": 3.1214, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 6.256209850311279, | |
| "learning_rate": 4.9826056764762074e-05, | |
| "loss": 3.2357, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.713292121887207, | |
| "learning_rate": 4.982417953295968e-05, | |
| "loss": 3.0448, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.762032985687256, | |
| "learning_rate": 4.982229226145727e-05, | |
| "loss": 3.1855, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.245543956756592, | |
| "learning_rate": 4.982039495101812e-05, | |
| "loss": 3.1024, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 5.456753730773926, | |
| "learning_rate": 4.9818487602409556e-05, | |
| "loss": 3.2703, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 5.831927299499512, | |
| "learning_rate": 4.981657021640298e-05, | |
| "loss": 3.2172, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 6.4401350021362305, | |
| "learning_rate": 4.981464279377384e-05, | |
| "loss": 3.2972, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 5.1732892990112305, | |
| "learning_rate": 4.981270533530164e-05, | |
| "loss": 3.1305, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 5.929070472717285, | |
| "learning_rate": 4.9810757841769976e-05, | |
| "loss": 3.0932, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 5.50985860824585, | |
| "learning_rate": 4.980880031396645e-05, | |
| "loss": 3.1417, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 6.071355819702148, | |
| "learning_rate": 4.980683275268277e-05, | |
| "loss": 3.1796, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 5.182791709899902, | |
| "learning_rate": 4.980485515871467e-05, | |
| "loss": 3.268, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 6.4410786628723145, | |
| "learning_rate": 4.980286753286195e-05, | |
| "loss": 3.2922, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 22.151193618774414, | |
| "learning_rate": 4.9800869875928475e-05, | |
| "loss": 3.0482, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 6.808634281158447, | |
| "learning_rate": 4.9798862188722164e-05, | |
| "loss": 3.1515, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "eval_loss": 2.6660187244415283, | |
| "eval_runtime": 423.2518, | |
| "eval_samples_per_second": 50.138, | |
| "eval_steps_per_second": 0.392, | |
| "step": 3000 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 50400, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 500, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |