| { | |
| "best_metric": 0.026551904156804085, | |
| "best_model_checkpoint": "./phishing-email-detection/checkpoint-825", | |
| "epoch": 3.0, | |
| "eval_steps": 1, | |
| "global_step": 825, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0036363636363636364, | |
| "grad_norm": 0.5588774085044861, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.6866, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.007272727272727273, | |
| "grad_norm": 1.193735957145691, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.7133, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01090909090909091, | |
| "grad_norm": 0.7555140256881714, | |
| "learning_rate": 3e-06, | |
| "loss": 0.6972, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.014545454545454545, | |
| "grad_norm": 0.7989323735237122, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.6927, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01818181818181818, | |
| "grad_norm": 0.608884334564209, | |
| "learning_rate": 5e-06, | |
| "loss": 0.6873, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02181818181818182, | |
| "grad_norm": 0.5146162509918213, | |
| "learning_rate": 6e-06, | |
| "loss": 0.6783, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.025454545454545455, | |
| "grad_norm": 1.2107092142105103, | |
| "learning_rate": 7.000000000000001e-06, | |
| "loss": 0.6978, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.02909090909090909, | |
| "grad_norm": 0.7423388957977295, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.6951, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03272727272727273, | |
| "grad_norm": 1.1845803260803223, | |
| "learning_rate": 9e-06, | |
| "loss": 0.6952, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.03636363636363636, | |
| "grad_norm": 0.6241262555122375, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6854, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 1.0969253778457642, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "loss": 0.6649, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.04363636363636364, | |
| "grad_norm": 0.7080792188644409, | |
| "learning_rate": 1.2e-05, | |
| "loss": 0.6827, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.04727272727272727, | |
| "grad_norm": 0.6348336935043335, | |
| "learning_rate": 1.3000000000000001e-05, | |
| "loss": 0.6681, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.05090909090909091, | |
| "grad_norm": 0.714078426361084, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 0.6719, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.05454545454545454, | |
| "grad_norm": 0.6718598008155823, | |
| "learning_rate": 1.5e-05, | |
| "loss": 0.6702, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.05818181818181818, | |
| "grad_norm": 1.0352996587753296, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 0.6689, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.06181818181818182, | |
| "grad_norm": 0.8572512269020081, | |
| "learning_rate": 1.7000000000000003e-05, | |
| "loss": 0.6459, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.06545454545454546, | |
| "grad_norm": 1.2123569250106812, | |
| "learning_rate": 1.8e-05, | |
| "loss": 0.6487, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.06909090909090909, | |
| "grad_norm": 2.0234522819519043, | |
| "learning_rate": 1.9e-05, | |
| "loss": 0.6572, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.07272727272727272, | |
| "grad_norm": 2.4744138717651367, | |
| "learning_rate": 2e-05, | |
| "loss": 0.586, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.07636363636363637, | |
| "grad_norm": 1.7895574569702148, | |
| "learning_rate": 2.1e-05, | |
| "loss": 0.6114, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 2.3886725902557373, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 0.6199, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.08363636363636363, | |
| "grad_norm": 2.5716464519500732, | |
| "learning_rate": 2.3000000000000003e-05, | |
| "loss": 0.5951, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.08727272727272728, | |
| "grad_norm": 1.6023527383804321, | |
| "learning_rate": 2.4e-05, | |
| "loss": 0.5584, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.09090909090909091, | |
| "grad_norm": 1.8739789724349976, | |
| "learning_rate": 2.5e-05, | |
| "loss": 0.5711, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.09454545454545454, | |
| "grad_norm": 1.912356972694397, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 0.5251, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.09818181818181818, | |
| "grad_norm": 2.616036891937256, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 0.4568, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.10181818181818182, | |
| "grad_norm": 2.2210693359375, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 0.4547, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.10545454545454545, | |
| "grad_norm": 4.0840325355529785, | |
| "learning_rate": 2.9e-05, | |
| "loss": 0.409, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.10909090909090909, | |
| "grad_norm": 3.1913392543792725, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3759, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.11272727272727273, | |
| "grad_norm": 2.67592191696167, | |
| "learning_rate": 3.1e-05, | |
| "loss": 0.3727, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.11636363636363636, | |
| "grad_norm": 3.8523147106170654, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 0.3395, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 3.1377878189086914, | |
| "learning_rate": 3.3e-05, | |
| "loss": 0.2602, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.12363636363636364, | |
| "grad_norm": 4.252383232116699, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 0.2641, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.12727272727272726, | |
| "grad_norm": 2.4112355709075928, | |
| "learning_rate": 3.5e-05, | |
| "loss": 0.2815, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.13090909090909092, | |
| "grad_norm": 2.1956121921539307, | |
| "learning_rate": 3.6e-05, | |
| "loss": 0.212, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.13454545454545455, | |
| "grad_norm": 1.6787267923355103, | |
| "learning_rate": 3.7e-05, | |
| "loss": 0.2461, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.13818181818181818, | |
| "grad_norm": 5.1338300704956055, | |
| "learning_rate": 3.8e-05, | |
| "loss": 0.2723, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.14181818181818182, | |
| "grad_norm": 1.7655150890350342, | |
| "learning_rate": 3.9000000000000006e-05, | |
| "loss": 0.2078, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.14545454545454545, | |
| "grad_norm": 2.906001329421997, | |
| "learning_rate": 4e-05, | |
| "loss": 0.3349, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.14909090909090908, | |
| "grad_norm": 4.4803900718688965, | |
| "learning_rate": 4.1e-05, | |
| "loss": 0.1256, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.15272727272727274, | |
| "grad_norm": 2.0880069732666016, | |
| "learning_rate": 4.2e-05, | |
| "loss": 0.1547, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.15636363636363637, | |
| "grad_norm": 3.3797402381896973, | |
| "learning_rate": 4.3e-05, | |
| "loss": 0.137, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 3.1822092533111572, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 0.1881, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.16363636363636364, | |
| "grad_norm": 1.7508845329284668, | |
| "learning_rate": 4.5e-05, | |
| "loss": 0.0657, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.16727272727272727, | |
| "grad_norm": 1.9360814094543457, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 0.0578, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.1709090909090909, | |
| "grad_norm": 5.421896934509277, | |
| "learning_rate": 4.7e-05, | |
| "loss": 0.2809, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.17454545454545456, | |
| "grad_norm": 3.034318685531616, | |
| "learning_rate": 4.8e-05, | |
| "loss": 0.1716, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.1781818181818182, | |
| "grad_norm": 8.016839027404785, | |
| "learning_rate": 4.9e-05, | |
| "loss": 0.261, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.18181818181818182, | |
| "grad_norm": 2.5083861351013184, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0848, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.18545454545454546, | |
| "grad_norm": 7.909552097320557, | |
| "learning_rate": 4.9935483870967744e-05, | |
| "loss": 0.0823, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.1890909090909091, | |
| "grad_norm": 8.652934074401855, | |
| "learning_rate": 4.9870967741935485e-05, | |
| "loss": 0.1812, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.19272727272727272, | |
| "grad_norm": 2.450180768966675, | |
| "learning_rate": 4.9806451612903226e-05, | |
| "loss": 0.065, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.19636363636363635, | |
| "grad_norm": 3.1136999130249023, | |
| "learning_rate": 4.9741935483870974e-05, | |
| "loss": 0.1683, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 6.53245735168457, | |
| "learning_rate": 4.967741935483871e-05, | |
| "loss": 0.1243, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.20363636363636364, | |
| "grad_norm": 3.6356089115142822, | |
| "learning_rate": 4.961290322580646e-05, | |
| "loss": 0.1899, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.20727272727272728, | |
| "grad_norm": 13.28806209564209, | |
| "learning_rate": 4.95483870967742e-05, | |
| "loss": 0.416, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.2109090909090909, | |
| "grad_norm": 6.773140907287598, | |
| "learning_rate": 4.948387096774193e-05, | |
| "loss": 0.1387, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.21454545454545454, | |
| "grad_norm": 4.9038848876953125, | |
| "learning_rate": 4.941935483870968e-05, | |
| "loss": 0.1795, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.21818181818181817, | |
| "grad_norm": 1.2570604085922241, | |
| "learning_rate": 4.935483870967742e-05, | |
| "loss": 0.0462, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.22181818181818183, | |
| "grad_norm": 4.595444202423096, | |
| "learning_rate": 4.929032258064516e-05, | |
| "loss": 0.2883, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.22545454545454546, | |
| "grad_norm": 5.1306891441345215, | |
| "learning_rate": 4.9225806451612904e-05, | |
| "loss": 0.1155, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.2290909090909091, | |
| "grad_norm": 3.7912144660949707, | |
| "learning_rate": 4.916129032258065e-05, | |
| "loss": 0.254, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.23272727272727273, | |
| "grad_norm": 5.713457107543945, | |
| "learning_rate": 4.9096774193548387e-05, | |
| "loss": 0.1327, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.23636363636363636, | |
| "grad_norm": 4.170778274536133, | |
| "learning_rate": 4.903225806451613e-05, | |
| "loss": 0.1039, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 1.75148344039917, | |
| "learning_rate": 4.8967741935483876e-05, | |
| "loss": 0.0852, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.24363636363636362, | |
| "grad_norm": 9.794595718383789, | |
| "learning_rate": 4.890322580645161e-05, | |
| "loss": 0.2108, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.24727272727272728, | |
| "grad_norm": 9.14588737487793, | |
| "learning_rate": 4.883870967741936e-05, | |
| "loss": 0.1652, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.2509090909090909, | |
| "grad_norm": 0.7871703505516052, | |
| "learning_rate": 4.87741935483871e-05, | |
| "loss": 0.0286, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.2545454545454545, | |
| "grad_norm": 3.963533401489258, | |
| "learning_rate": 4.870967741935484e-05, | |
| "loss": 0.1447, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.2581818181818182, | |
| "grad_norm": 1.9403924942016602, | |
| "learning_rate": 4.864516129032258e-05, | |
| "loss": 0.0346, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.26181818181818184, | |
| "grad_norm": 4.882287979125977, | |
| "learning_rate": 4.858064516129032e-05, | |
| "loss": 0.0858, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.26545454545454544, | |
| "grad_norm": 3.18686842918396, | |
| "learning_rate": 4.8516129032258065e-05, | |
| "loss": 0.1201, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.2690909090909091, | |
| "grad_norm": 1.2996779680252075, | |
| "learning_rate": 4.8451612903225806e-05, | |
| "loss": 0.0285, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.2727272727272727, | |
| "grad_norm": 4.498321056365967, | |
| "learning_rate": 4.8387096774193554e-05, | |
| "loss": 0.1272, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.27636363636363637, | |
| "grad_norm": 7.456954479217529, | |
| "learning_rate": 4.8322580645161295e-05, | |
| "loss": 0.1553, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 1.462753415107727, | |
| "learning_rate": 4.8258064516129036e-05, | |
| "loss": 0.0688, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.28363636363636363, | |
| "grad_norm": 2.150094985961914, | |
| "learning_rate": 4.819354838709678e-05, | |
| "loss": 0.0558, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.2872727272727273, | |
| "grad_norm": 4.6224284172058105, | |
| "learning_rate": 4.812903225806452e-05, | |
| "loss": 0.1185, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.2909090909090909, | |
| "grad_norm": 3.0150296688079834, | |
| "learning_rate": 4.806451612903226e-05, | |
| "loss": 0.2397, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.29454545454545455, | |
| "grad_norm": 13.630721092224121, | |
| "learning_rate": 4.8e-05, | |
| "loss": 0.2093, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.29818181818181816, | |
| "grad_norm": 8.345911026000977, | |
| "learning_rate": 4.793548387096774e-05, | |
| "loss": 0.1799, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.3018181818181818, | |
| "grad_norm": 4.615355491638184, | |
| "learning_rate": 4.7870967741935484e-05, | |
| "loss": 0.0951, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.3054545454545455, | |
| "grad_norm": 6.051864147186279, | |
| "learning_rate": 4.780645161290323e-05, | |
| "loss": 0.2423, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.3090909090909091, | |
| "grad_norm": 3.912353277206421, | |
| "learning_rate": 4.774193548387097e-05, | |
| "loss": 0.0923, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.31272727272727274, | |
| "grad_norm": 5.80419397354126, | |
| "learning_rate": 4.767741935483871e-05, | |
| "loss": 0.1406, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.31636363636363635, | |
| "grad_norm": 4.3275227546691895, | |
| "learning_rate": 4.7612903225806455e-05, | |
| "loss": 0.2217, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 3.613757610321045, | |
| "learning_rate": 4.75483870967742e-05, | |
| "loss": 0.1888, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.3236363636363636, | |
| "grad_norm": 0.619616687297821, | |
| "learning_rate": 4.748387096774194e-05, | |
| "loss": 0.0193, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.32727272727272727, | |
| "grad_norm": 3.2283294200897217, | |
| "learning_rate": 4.741935483870968e-05, | |
| "loss": 0.1036, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.33090909090909093, | |
| "grad_norm": 2.8031208515167236, | |
| "learning_rate": 4.735483870967742e-05, | |
| "loss": 0.1384, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.33454545454545453, | |
| "grad_norm": 2.338831901550293, | |
| "learning_rate": 4.729032258064516e-05, | |
| "loss": 0.0763, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.3381818181818182, | |
| "grad_norm": 1.7797901630401611, | |
| "learning_rate": 4.72258064516129e-05, | |
| "loss": 0.1343, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.3418181818181818, | |
| "grad_norm": 2.9699721336364746, | |
| "learning_rate": 4.716129032258065e-05, | |
| "loss": 0.124, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.34545454545454546, | |
| "grad_norm": 1.207624077796936, | |
| "learning_rate": 4.7096774193548385e-05, | |
| "loss": 0.0384, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.3490909090909091, | |
| "grad_norm": 1.7600481510162354, | |
| "learning_rate": 4.7032258064516133e-05, | |
| "loss": 0.1058, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.3527272727272727, | |
| "grad_norm": 1.9546951055526733, | |
| "learning_rate": 4.6967741935483875e-05, | |
| "loss": 0.0867, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.3563636363636364, | |
| "grad_norm": 1.6174895763397217, | |
| "learning_rate": 4.6903225806451616e-05, | |
| "loss": 0.0249, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 2.494776725769043, | |
| "learning_rate": 4.683870967741936e-05, | |
| "loss": 0.0813, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.36363636363636365, | |
| "grad_norm": 1.799558162689209, | |
| "learning_rate": 4.67741935483871e-05, | |
| "loss": 0.0698, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.36727272727272725, | |
| "grad_norm": 1.8396685123443604, | |
| "learning_rate": 4.670967741935484e-05, | |
| "loss": 0.1093, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.3709090909090909, | |
| "grad_norm": 1.1487964391708374, | |
| "learning_rate": 4.664516129032258e-05, | |
| "loss": 0.0199, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.37454545454545457, | |
| "grad_norm": 3.693464756011963, | |
| "learning_rate": 4.658064516129033e-05, | |
| "loss": 0.0468, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.3781818181818182, | |
| "grad_norm": 0.4796558916568756, | |
| "learning_rate": 4.651612903225806e-05, | |
| "loss": 0.0142, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.38181818181818183, | |
| "grad_norm": 1.3589314222335815, | |
| "learning_rate": 4.645161290322581e-05, | |
| "loss": 0.0197, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.38545454545454544, | |
| "grad_norm": 5.373131275177002, | |
| "learning_rate": 4.638709677419355e-05, | |
| "loss": 0.1297, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.3890909090909091, | |
| "grad_norm": 5.07847261428833, | |
| "learning_rate": 4.632258064516129e-05, | |
| "loss": 0.1152, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.3927272727272727, | |
| "grad_norm": 2.5810861587524414, | |
| "learning_rate": 4.6258064516129035e-05, | |
| "loss": 0.0499, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.39636363636363636, | |
| "grad_norm": 3.111640691757202, | |
| "learning_rate": 4.6193548387096776e-05, | |
| "loss": 0.0488, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 2.8592653274536133, | |
| "learning_rate": 4.612903225806452e-05, | |
| "loss": 0.0648, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.4036363636363636, | |
| "grad_norm": 4.070368766784668, | |
| "learning_rate": 4.606451612903226e-05, | |
| "loss": 0.1844, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.4072727272727273, | |
| "grad_norm": 2.12369441986084, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 0.0645, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.4109090909090909, | |
| "grad_norm": 2.3667800426483154, | |
| "learning_rate": 4.593548387096774e-05, | |
| "loss": 0.0471, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.41454545454545455, | |
| "grad_norm": 4.2587056159973145, | |
| "learning_rate": 4.587096774193548e-05, | |
| "loss": 0.062, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.41818181818181815, | |
| "grad_norm": 4.729971885681152, | |
| "learning_rate": 4.580645161290323e-05, | |
| "loss": 0.1277, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.4218181818181818, | |
| "grad_norm": 2.8815808296203613, | |
| "learning_rate": 4.5741935483870965e-05, | |
| "loss": 0.1424, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.4254545454545455, | |
| "grad_norm": 3.5163843631744385, | |
| "learning_rate": 4.567741935483871e-05, | |
| "loss": 0.1753, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.4290909090909091, | |
| "grad_norm": 0.9345911145210266, | |
| "learning_rate": 4.5612903225806454e-05, | |
| "loss": 0.0173, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.43272727272727274, | |
| "grad_norm": 2.4038431644439697, | |
| "learning_rate": 4.5548387096774196e-05, | |
| "loss": 0.1339, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.43636363636363634, | |
| "grad_norm": 1.086334228515625, | |
| "learning_rate": 4.548387096774194e-05, | |
| "loss": 0.06, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 3.3562028408050537, | |
| "learning_rate": 4.5419354838709685e-05, | |
| "loss": 0.2079, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.44363636363636366, | |
| "grad_norm": 3.6790003776550293, | |
| "learning_rate": 4.535483870967742e-05, | |
| "loss": 0.1622, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.44727272727272727, | |
| "grad_norm": 4.95260763168335, | |
| "learning_rate": 4.529032258064516e-05, | |
| "loss": 0.069, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.4509090909090909, | |
| "grad_norm": 1.6130903959274292, | |
| "learning_rate": 4.522580645161291e-05, | |
| "loss": 0.1197, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.45454545454545453, | |
| "grad_norm": 3.9144365787506104, | |
| "learning_rate": 4.516129032258064e-05, | |
| "loss": 0.063, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.4581818181818182, | |
| "grad_norm": 7.754761219024658, | |
| "learning_rate": 4.509677419354839e-05, | |
| "loss": 0.358, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.4618181818181818, | |
| "grad_norm": 4.277048110961914, | |
| "learning_rate": 4.503225806451613e-05, | |
| "loss": 0.1137, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.46545454545454545, | |
| "grad_norm": 4.798757553100586, | |
| "learning_rate": 4.4967741935483873e-05, | |
| "loss": 0.1365, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.4690909090909091, | |
| "grad_norm": 2.407785177230835, | |
| "learning_rate": 4.4903225806451615e-05, | |
| "loss": 0.0339, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.4727272727272727, | |
| "grad_norm": 2.8577773571014404, | |
| "learning_rate": 4.4838709677419356e-05, | |
| "loss": 0.0634, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.4763636363636364, | |
| "grad_norm": 2.8833367824554443, | |
| "learning_rate": 4.47741935483871e-05, | |
| "loss": 0.1274, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 3.3655707836151123, | |
| "learning_rate": 4.470967741935484e-05, | |
| "loss": 0.0845, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.48363636363636364, | |
| "grad_norm": 0.4524034559726715, | |
| "learning_rate": 4.4645161290322586e-05, | |
| "loss": 0.0201, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.48727272727272725, | |
| "grad_norm": 2.0502264499664307, | |
| "learning_rate": 4.458064516129032e-05, | |
| "loss": 0.0241, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.4909090909090909, | |
| "grad_norm": 3.171466588973999, | |
| "learning_rate": 4.451612903225807e-05, | |
| "loss": 0.0988, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.49454545454545457, | |
| "grad_norm": 3.592583179473877, | |
| "learning_rate": 4.445161290322581e-05, | |
| "loss": 0.0352, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.49818181818181817, | |
| "grad_norm": 2.234022617340088, | |
| "learning_rate": 4.438709677419355e-05, | |
| "loss": 0.1045, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.5018181818181818, | |
| "grad_norm": 1.0845825672149658, | |
| "learning_rate": 4.432258064516129e-05, | |
| "loss": 0.1009, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.5054545454545455, | |
| "grad_norm": 2.9205362796783447, | |
| "learning_rate": 4.4258064516129034e-05, | |
| "loss": 0.0515, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.509090909090909, | |
| "grad_norm": 3.975783109664917, | |
| "learning_rate": 4.4193548387096775e-05, | |
| "loss": 0.1285, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.5127272727272727, | |
| "grad_norm": 5.859277248382568, | |
| "learning_rate": 4.4129032258064516e-05, | |
| "loss": 0.0997, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.5163636363636364, | |
| "grad_norm": 4.712912559509277, | |
| "learning_rate": 4.4064516129032264e-05, | |
| "loss": 0.1876, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 3.739158868789673, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 0.0487, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.5236363636363637, | |
| "grad_norm": 0.8333390951156616, | |
| "learning_rate": 4.393548387096774e-05, | |
| "loss": 0.0169, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.5272727272727272, | |
| "grad_norm": 1.7556527853012085, | |
| "learning_rate": 4.387096774193549e-05, | |
| "loss": 0.0418, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.5309090909090909, | |
| "grad_norm": 0.7867715358734131, | |
| "learning_rate": 4.380645161290323e-05, | |
| "loss": 0.0097, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.5345454545454545, | |
| "grad_norm": 1.7784496545791626, | |
| "learning_rate": 4.374193548387097e-05, | |
| "loss": 0.0308, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.5381818181818182, | |
| "grad_norm": 3.104606866836548, | |
| "learning_rate": 4.367741935483871e-05, | |
| "loss": 0.0512, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.5418181818181819, | |
| "grad_norm": 1.6434437036514282, | |
| "learning_rate": 4.361290322580645e-05, | |
| "loss": 0.086, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.5454545454545454, | |
| "grad_norm": 3.0264408588409424, | |
| "learning_rate": 4.3548387096774194e-05, | |
| "loss": 0.0293, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.5490909090909091, | |
| "grad_norm": 0.4782671332359314, | |
| "learning_rate": 4.3483870967741936e-05, | |
| "loss": 0.0107, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.5527272727272727, | |
| "grad_norm": 3.8326480388641357, | |
| "learning_rate": 4.3419354838709684e-05, | |
| "loss": 0.1158, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.5563636363636364, | |
| "grad_norm": 2.803429126739502, | |
| "learning_rate": 4.335483870967742e-05, | |
| "loss": 0.0907, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 6.267507076263428, | |
| "learning_rate": 4.3290322580645166e-05, | |
| "loss": 0.0712, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.5636363636363636, | |
| "grad_norm": 2.4244847297668457, | |
| "learning_rate": 4.322580645161291e-05, | |
| "loss": 0.0567, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.5672727272727273, | |
| "grad_norm": 1.6119318008422852, | |
| "learning_rate": 4.316129032258065e-05, | |
| "loss": 0.166, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.5709090909090909, | |
| "grad_norm": 2.521883964538574, | |
| "learning_rate": 4.309677419354839e-05, | |
| "loss": 0.0308, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.5745454545454546, | |
| "grad_norm": 2.1113216876983643, | |
| "learning_rate": 4.303225806451613e-05, | |
| "loss": 0.0302, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.5781818181818181, | |
| "grad_norm": 1.94487726688385, | |
| "learning_rate": 4.296774193548387e-05, | |
| "loss": 0.0199, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.5818181818181818, | |
| "grad_norm": 3.142296552658081, | |
| "learning_rate": 4.2903225806451614e-05, | |
| "loss": 0.0744, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.5854545454545454, | |
| "grad_norm": 1.6849907636642456, | |
| "learning_rate": 4.283870967741936e-05, | |
| "loss": 0.0321, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.5890909090909091, | |
| "grad_norm": 5.332469463348389, | |
| "learning_rate": 4.2774193548387096e-05, | |
| "loss": 0.1344, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.5927272727272728, | |
| "grad_norm": 0.8968711495399475, | |
| "learning_rate": 4.2709677419354844e-05, | |
| "loss": 0.073, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.5963636363636363, | |
| "grad_norm": 5.70121955871582, | |
| "learning_rate": 4.2645161290322585e-05, | |
| "loss": 0.1092, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 1.3494465351104736, | |
| "learning_rate": 4.258064516129032e-05, | |
| "loss": 0.0153, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.6036363636363636, | |
| "grad_norm": 1.2071605920791626, | |
| "learning_rate": 4.251612903225807e-05, | |
| "loss": 0.0636, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.6072727272727273, | |
| "grad_norm": 2.2756154537200928, | |
| "learning_rate": 4.245161290322581e-05, | |
| "loss": 0.0523, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.610909090909091, | |
| "grad_norm": 2.9887783527374268, | |
| "learning_rate": 4.238709677419355e-05, | |
| "loss": 0.0991, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.6145454545454545, | |
| "grad_norm": 3.471949577331543, | |
| "learning_rate": 4.232258064516129e-05, | |
| "loss": 0.1146, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.6181818181818182, | |
| "grad_norm": 0.9178719520568848, | |
| "learning_rate": 4.225806451612904e-05, | |
| "loss": 0.0283, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.6218181818181818, | |
| "grad_norm": 1.7535454034805298, | |
| "learning_rate": 4.2193548387096774e-05, | |
| "loss": 0.1037, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.6254545454545455, | |
| "grad_norm": 0.523720383644104, | |
| "learning_rate": 4.2129032258064515e-05, | |
| "loss": 0.0188, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.6290909090909091, | |
| "grad_norm": 2.872236728668213, | |
| "learning_rate": 4.206451612903226e-05, | |
| "loss": 0.137, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.6327272727272727, | |
| "grad_norm": 1.1402463912963867, | |
| "learning_rate": 4.2e-05, | |
| "loss": 0.0469, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.6363636363636364, | |
| "grad_norm": 1.3822078704833984, | |
| "learning_rate": 4.1935483870967746e-05, | |
| "loss": 0.0299, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 1.6050485372543335, | |
| "learning_rate": 4.187096774193549e-05, | |
| "loss": 0.0313, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.6436363636363637, | |
| "grad_norm": 1.4865597486495972, | |
| "learning_rate": 4.180645161290323e-05, | |
| "loss": 0.0199, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.6472727272727272, | |
| "grad_norm": 1.0985281467437744, | |
| "learning_rate": 4.174193548387097e-05, | |
| "loss": 0.0161, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.6509090909090909, | |
| "grad_norm": 2.7019283771514893, | |
| "learning_rate": 4.167741935483871e-05, | |
| "loss": 0.1127, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.6545454545454545, | |
| "grad_norm": 0.20963142812252045, | |
| "learning_rate": 4.161290322580645e-05, | |
| "loss": 0.0069, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.6581818181818182, | |
| "grad_norm": 2.687619209289551, | |
| "learning_rate": 4.154838709677419e-05, | |
| "loss": 0.0189, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.6618181818181819, | |
| "grad_norm": 0.5330418348312378, | |
| "learning_rate": 4.148387096774194e-05, | |
| "loss": 0.0878, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.6654545454545454, | |
| "grad_norm": 6.277083873748779, | |
| "learning_rate": 4.1419354838709676e-05, | |
| "loss": 0.2121, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.6690909090909091, | |
| "grad_norm": 4.049676895141602, | |
| "learning_rate": 4.1354838709677424e-05, | |
| "loss": 0.069, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.6727272727272727, | |
| "grad_norm": 2.069470167160034, | |
| "learning_rate": 4.1290322580645165e-05, | |
| "loss": 0.0485, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.6763636363636364, | |
| "grad_norm": 5.907510280609131, | |
| "learning_rate": 4.1225806451612906e-05, | |
| "loss": 0.0886, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 2.1604321002960205, | |
| "learning_rate": 4.116129032258065e-05, | |
| "loss": 0.0116, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.6836363636363636, | |
| "grad_norm": 2.6261606216430664, | |
| "learning_rate": 4.109677419354839e-05, | |
| "loss": 0.0438, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.6872727272727273, | |
| "grad_norm": 2.597719669342041, | |
| "learning_rate": 4.103225806451613e-05, | |
| "loss": 0.1081, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.6909090909090909, | |
| "grad_norm": 0.45269355177879333, | |
| "learning_rate": 4.096774193548387e-05, | |
| "loss": 0.0057, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.6945454545454546, | |
| "grad_norm": 3.5486068725585938, | |
| "learning_rate": 4.090322580645162e-05, | |
| "loss": 0.0569, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.6981818181818182, | |
| "grad_norm": 2.485374927520752, | |
| "learning_rate": 4.0838709677419354e-05, | |
| "loss": 0.1183, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.7018181818181818, | |
| "grad_norm": 4.209328651428223, | |
| "learning_rate": 4.0774193548387095e-05, | |
| "loss": 0.1006, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.7054545454545454, | |
| "grad_norm": 1.0859043598175049, | |
| "learning_rate": 4.070967741935484e-05, | |
| "loss": 0.0117, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.7090909090909091, | |
| "grad_norm": 1.0647627115249634, | |
| "learning_rate": 4.0645161290322584e-05, | |
| "loss": 0.0618, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.7127272727272728, | |
| "grad_norm": 1.131689429283142, | |
| "learning_rate": 4.0580645161290325e-05, | |
| "loss": 0.0439, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.7163636363636363, | |
| "grad_norm": 3.031101942062378, | |
| "learning_rate": 4.0516129032258067e-05, | |
| "loss": 0.1021, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.11612915247678757, | |
| "learning_rate": 4.045161290322581e-05, | |
| "loss": 0.0048, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.7236363636363636, | |
| "grad_norm": 1.4697656631469727, | |
| "learning_rate": 4.038709677419355e-05, | |
| "loss": 0.0296, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.7272727272727273, | |
| "grad_norm": 2.3507072925567627, | |
| "learning_rate": 4.032258064516129e-05, | |
| "loss": 0.0975, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.730909090909091, | |
| "grad_norm": 3.2952606678009033, | |
| "learning_rate": 4.025806451612903e-05, | |
| "loss": 0.0741, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.7345454545454545, | |
| "grad_norm": 1.3989083766937256, | |
| "learning_rate": 4.019354838709677e-05, | |
| "loss": 0.0603, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.7381818181818182, | |
| "grad_norm": 0.7519178986549377, | |
| "learning_rate": 4.012903225806452e-05, | |
| "loss": 0.0195, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.7418181818181818, | |
| "grad_norm": 0.9893004298210144, | |
| "learning_rate": 4.006451612903226e-05, | |
| "loss": 0.0316, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.7454545454545455, | |
| "grad_norm": 2.2764501571655273, | |
| "learning_rate": 4e-05, | |
| "loss": 0.0347, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.7490909090909091, | |
| "grad_norm": 2.6084823608398438, | |
| "learning_rate": 3.9935483870967745e-05, | |
| "loss": 0.1071, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.7527272727272727, | |
| "grad_norm": 1.7547663450241089, | |
| "learning_rate": 3.9870967741935486e-05, | |
| "loss": 0.0181, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.7563636363636363, | |
| "grad_norm": 0.3479810953140259, | |
| "learning_rate": 3.980645161290323e-05, | |
| "loss": 0.0101, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 1.399048924446106, | |
| "learning_rate": 3.974193548387097e-05, | |
| "loss": 0.0244, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.7636363636363637, | |
| "grad_norm": 3.616453170776367, | |
| "learning_rate": 3.9677419354838716e-05, | |
| "loss": 0.0863, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.7672727272727272, | |
| "grad_norm": 2.61297345161438, | |
| "learning_rate": 3.961290322580645e-05, | |
| "loss": 0.0447, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.7709090909090909, | |
| "grad_norm": 1.899042010307312, | |
| "learning_rate": 3.95483870967742e-05, | |
| "loss": 0.0313, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.7745454545454545, | |
| "grad_norm": 0.9543875455856323, | |
| "learning_rate": 3.948387096774194e-05, | |
| "loss": 0.0243, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.7781818181818182, | |
| "grad_norm": 1.5872313976287842, | |
| "learning_rate": 3.9419354838709674e-05, | |
| "loss": 0.0122, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.7818181818181819, | |
| "grad_norm": 1.8551832437515259, | |
| "learning_rate": 3.935483870967742e-05, | |
| "loss": 0.0535, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.7854545454545454, | |
| "grad_norm": 3.93450927734375, | |
| "learning_rate": 3.9290322580645164e-05, | |
| "loss": 0.1321, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.7890909090909091, | |
| "grad_norm": 4.761463165283203, | |
| "learning_rate": 3.9225806451612905e-05, | |
| "loss": 0.142, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.7927272727272727, | |
| "grad_norm": 1.0365723371505737, | |
| "learning_rate": 3.9161290322580646e-05, | |
| "loss": 0.0197, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.7963636363636364, | |
| "grad_norm": 2.633030414581299, | |
| "learning_rate": 3.9096774193548394e-05, | |
| "loss": 0.0375, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 1.4770023822784424, | |
| "learning_rate": 3.903225806451613e-05, | |
| "loss": 0.0251, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.8036363636363636, | |
| "grad_norm": 2.1788926124572754, | |
| "learning_rate": 3.896774193548387e-05, | |
| "loss": 0.0146, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.8072727272727273, | |
| "grad_norm": 1.1818387508392334, | |
| "learning_rate": 3.890322580645162e-05, | |
| "loss": 0.0307, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.8109090909090909, | |
| "grad_norm": 1.7780365943908691, | |
| "learning_rate": 3.883870967741935e-05, | |
| "loss": 0.0436, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.8145454545454546, | |
| "grad_norm": 0.3603725731372833, | |
| "learning_rate": 3.87741935483871e-05, | |
| "loss": 0.0086, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.8181818181818182, | |
| "grad_norm": 2.094778060913086, | |
| "learning_rate": 3.870967741935484e-05, | |
| "loss": 0.0535, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.8218181818181818, | |
| "grad_norm": 1.7837364673614502, | |
| "learning_rate": 3.864516129032258e-05, | |
| "loss": 0.0324, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.8254545454545454, | |
| "grad_norm": 1.0126112699508667, | |
| "learning_rate": 3.8580645161290324e-05, | |
| "loss": 0.0195, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.8290909090909091, | |
| "grad_norm": 2.095411777496338, | |
| "learning_rate": 3.8516129032258065e-05, | |
| "loss": 0.0858, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.8327272727272728, | |
| "grad_norm": 2.12591290473938, | |
| "learning_rate": 3.845161290322581e-05, | |
| "loss": 0.0745, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.8363636363636363, | |
| "grad_norm": 4.18102502822876, | |
| "learning_rate": 3.838709677419355e-05, | |
| "loss": 0.0876, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 2.6359128952026367, | |
| "learning_rate": 3.8322580645161296e-05, | |
| "loss": 0.0261, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.8436363636363636, | |
| "grad_norm": 2.9271247386932373, | |
| "learning_rate": 3.825806451612903e-05, | |
| "loss": 0.0432, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.8472727272727273, | |
| "grad_norm": 4.78306245803833, | |
| "learning_rate": 3.819354838709678e-05, | |
| "loss": 0.1117, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.850909090909091, | |
| "grad_norm": 0.5705394148826599, | |
| "learning_rate": 3.812903225806452e-05, | |
| "loss": 0.0054, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.8545454545454545, | |
| "grad_norm": 2.0522520542144775, | |
| "learning_rate": 3.8064516129032254e-05, | |
| "loss": 0.0245, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.8581818181818182, | |
| "grad_norm": 2.7212295532226562, | |
| "learning_rate": 3.8e-05, | |
| "loss": 0.0252, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.8618181818181818, | |
| "grad_norm": 0.7977845072746277, | |
| "learning_rate": 3.793548387096774e-05, | |
| "loss": 0.0127, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.8654545454545455, | |
| "grad_norm": 3.720811605453491, | |
| "learning_rate": 3.7870967741935485e-05, | |
| "loss": 0.0465, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.8690909090909091, | |
| "grad_norm": 2.362733840942383, | |
| "learning_rate": 3.7806451612903226e-05, | |
| "loss": 0.0266, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.8727272727272727, | |
| "grad_norm": 2.5481631755828857, | |
| "learning_rate": 3.7741935483870974e-05, | |
| "loss": 0.0224, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.8763636363636363, | |
| "grad_norm": 2.1781375408172607, | |
| "learning_rate": 3.767741935483871e-05, | |
| "loss": 0.0228, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 1.9818776845932007, | |
| "learning_rate": 3.761290322580645e-05, | |
| "loss": 0.0231, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.8836363636363637, | |
| "grad_norm": 0.20954543352127075, | |
| "learning_rate": 3.75483870967742e-05, | |
| "loss": 0.0031, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.8872727272727273, | |
| "grad_norm": 2.769566535949707, | |
| "learning_rate": 3.748387096774193e-05, | |
| "loss": 0.039, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.8909090909090909, | |
| "grad_norm": 2.511801242828369, | |
| "learning_rate": 3.741935483870968e-05, | |
| "loss": 0.0154, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.8945454545454545, | |
| "grad_norm": 0.3851822018623352, | |
| "learning_rate": 3.735483870967742e-05, | |
| "loss": 0.0047, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.8981818181818182, | |
| "grad_norm": 1.6585170030593872, | |
| "learning_rate": 3.729032258064516e-05, | |
| "loss": 0.0882, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.9018181818181819, | |
| "grad_norm": 1.9124457836151123, | |
| "learning_rate": 3.7225806451612904e-05, | |
| "loss": 0.0138, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.9054545454545454, | |
| "grad_norm": 5.743299961090088, | |
| "learning_rate": 3.716129032258065e-05, | |
| "loss": 0.0494, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.9090909090909091, | |
| "grad_norm": 3.5978081226348877, | |
| "learning_rate": 3.7096774193548386e-05, | |
| "loss": 0.0182, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.9127272727272727, | |
| "grad_norm": 0.024638742208480835, | |
| "learning_rate": 3.703225806451613e-05, | |
| "loss": 0.0012, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.9163636363636364, | |
| "grad_norm": 0.04621901735663414, | |
| "learning_rate": 3.6967741935483876e-05, | |
| "loss": 0.0014, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 0.06507572531700134, | |
| "learning_rate": 3.690322580645162e-05, | |
| "loss": 0.0018, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.9236363636363636, | |
| "grad_norm": 0.683228075504303, | |
| "learning_rate": 3.683870967741936e-05, | |
| "loss": 0.0064, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.9272727272727272, | |
| "grad_norm": 9.123976707458496, | |
| "learning_rate": 3.67741935483871e-05, | |
| "loss": 0.087, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.9309090909090909, | |
| "grad_norm": 0.9182856678962708, | |
| "learning_rate": 3.670967741935484e-05, | |
| "loss": 0.0891, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.9345454545454546, | |
| "grad_norm": 4.344281196594238, | |
| "learning_rate": 3.664516129032258e-05, | |
| "loss": 0.0301, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.9381818181818182, | |
| "grad_norm": 0.22626306116580963, | |
| "learning_rate": 3.658064516129032e-05, | |
| "loss": 0.0027, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.9418181818181818, | |
| "grad_norm": 2.9443519115448, | |
| "learning_rate": 3.6516129032258064e-05, | |
| "loss": 0.0826, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.9454545454545454, | |
| "grad_norm": 0.35616394877433777, | |
| "learning_rate": 3.6451612903225805e-05, | |
| "loss": 0.003, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.9490909090909091, | |
| "grad_norm": 5.846389293670654, | |
| "learning_rate": 3.6387096774193553e-05, | |
| "loss": 0.1148, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.9527272727272728, | |
| "grad_norm": 0.351965069770813, | |
| "learning_rate": 3.6322580645161295e-05, | |
| "loss": 0.0054, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.9563636363636364, | |
| "grad_norm": 1.9264423847198486, | |
| "learning_rate": 3.6258064516129036e-05, | |
| "loss": 0.0241, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 7.070173263549805, | |
| "learning_rate": 3.619354838709678e-05, | |
| "loss": 0.1209, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.9636363636363636, | |
| "grad_norm": 4.125431537628174, | |
| "learning_rate": 3.612903225806452e-05, | |
| "loss": 0.0451, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.9672727272727273, | |
| "grad_norm": 0.08269821852445602, | |
| "learning_rate": 3.606451612903226e-05, | |
| "loss": 0.0024, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.9709090909090909, | |
| "grad_norm": 2.2969746589660645, | |
| "learning_rate": 3.6e-05, | |
| "loss": 0.03, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.9745454545454545, | |
| "grad_norm": 3.2518858909606934, | |
| "learning_rate": 3.593548387096774e-05, | |
| "loss": 0.0242, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.9781818181818182, | |
| "grad_norm": 0.2708548903465271, | |
| "learning_rate": 3.5870967741935483e-05, | |
| "loss": 0.0041, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.9818181818181818, | |
| "grad_norm": 3.211353063583374, | |
| "learning_rate": 3.580645161290323e-05, | |
| "loss": 0.0195, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.9854545454545455, | |
| "grad_norm": 2.5400710105895996, | |
| "learning_rate": 3.574193548387097e-05, | |
| "loss": 0.023, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.9890909090909091, | |
| "grad_norm": 4.467423915863037, | |
| "learning_rate": 3.567741935483871e-05, | |
| "loss": 0.1522, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.9927272727272727, | |
| "grad_norm": 0.29972556233406067, | |
| "learning_rate": 3.5612903225806455e-05, | |
| "loss": 0.0033, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.9963636363636363, | |
| "grad_norm": 2.243041753768921, | |
| "learning_rate": 3.5548387096774196e-05, | |
| "loss": 0.0159, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.03446698188781738, | |
| "learning_rate": 3.548387096774194e-05, | |
| "loss": 0.0015, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.9890710382513661, | |
| "eval_loss": 0.032855160534381866, | |
| "eval_runtime": 12.8051, | |
| "eval_samples_per_second": 342.989, | |
| "eval_steps_per_second": 5.388, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.0036363636363637, | |
| "grad_norm": 0.1759863644838333, | |
| "learning_rate": 3.541935483870968e-05, | |
| "loss": 0.0025, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 1.0072727272727273, | |
| "grad_norm": 0.30346181988716125, | |
| "learning_rate": 3.535483870967743e-05, | |
| "loss": 0.0035, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 1.010909090909091, | |
| "grad_norm": 0.5249446630477905, | |
| "learning_rate": 3.529032258064516e-05, | |
| "loss": 0.0071, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 1.0145454545454546, | |
| "grad_norm": 2.6401705741882324, | |
| "learning_rate": 3.52258064516129e-05, | |
| "loss": 0.0947, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 1.018181818181818, | |
| "grad_norm": 1.9211256504058838, | |
| "learning_rate": 3.516129032258065e-05, | |
| "loss": 0.0223, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.0218181818181817, | |
| "grad_norm": 0.13240855932235718, | |
| "learning_rate": 3.5096774193548385e-05, | |
| "loss": 0.0024, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 1.0254545454545454, | |
| "grad_norm": 1.043957233428955, | |
| "learning_rate": 3.503225806451613e-05, | |
| "loss": 0.0082, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 1.029090909090909, | |
| "grad_norm": 1.6595826148986816, | |
| "learning_rate": 3.4967741935483874e-05, | |
| "loss": 0.0711, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 1.0327272727272727, | |
| "grad_norm": 0.50986248254776, | |
| "learning_rate": 3.4903225806451616e-05, | |
| "loss": 0.0043, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 1.0363636363636364, | |
| "grad_norm": 2.219081401824951, | |
| "learning_rate": 3.483870967741936e-05, | |
| "loss": 0.0776, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "grad_norm": 0.7167057991027832, | |
| "learning_rate": 3.47741935483871e-05, | |
| "loss": 0.005, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 1.0436363636363637, | |
| "grad_norm": 0.13360421359539032, | |
| "learning_rate": 3.470967741935484e-05, | |
| "loss": 0.0023, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 1.0472727272727274, | |
| "grad_norm": 1.3663884401321411, | |
| "learning_rate": 3.464516129032258e-05, | |
| "loss": 0.0111, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 1.050909090909091, | |
| "grad_norm": 0.01989702135324478, | |
| "learning_rate": 3.458064516129033e-05, | |
| "loss": 0.001, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 1.0545454545454545, | |
| "grad_norm": 1.4087766408920288, | |
| "learning_rate": 3.451612903225806e-05, | |
| "loss": 0.014, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.0581818181818181, | |
| "grad_norm": 0.47182703018188477, | |
| "learning_rate": 3.445161290322581e-05, | |
| "loss": 0.0047, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 1.0618181818181818, | |
| "grad_norm": 0.7205504775047302, | |
| "learning_rate": 3.438709677419355e-05, | |
| "loss": 0.0045, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 1.0654545454545454, | |
| "grad_norm": 0.07702212780714035, | |
| "learning_rate": 3.432258064516129e-05, | |
| "loss": 0.002, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 1.069090909090909, | |
| "grad_norm": 0.44914644956588745, | |
| "learning_rate": 3.4258064516129035e-05, | |
| "loss": 0.0044, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 1.0727272727272728, | |
| "grad_norm": 0.26627811789512634, | |
| "learning_rate": 3.4193548387096776e-05, | |
| "loss": 0.0035, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.0763636363636364, | |
| "grad_norm": 1.8016518354415894, | |
| "learning_rate": 3.412903225806452e-05, | |
| "loss": 0.012, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "grad_norm": 2.806971311569214, | |
| "learning_rate": 3.406451612903226e-05, | |
| "loss": 0.051, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 1.0836363636363637, | |
| "grad_norm": 0.03818153962492943, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 0.0014, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 1.0872727272727274, | |
| "grad_norm": 0.04983190819621086, | |
| "learning_rate": 3.393548387096774e-05, | |
| "loss": 0.0016, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 1.0909090909090908, | |
| "grad_norm": 0.05590420216321945, | |
| "learning_rate": 3.387096774193548e-05, | |
| "loss": 0.0013, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.0945454545454545, | |
| "grad_norm": 0.3262059986591339, | |
| "learning_rate": 3.380645161290323e-05, | |
| "loss": 0.0024, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 1.0981818181818181, | |
| "grad_norm": 0.10000384598970413, | |
| "learning_rate": 3.3741935483870965e-05, | |
| "loss": 0.0016, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 1.1018181818181818, | |
| "grad_norm": 1.12041437625885, | |
| "learning_rate": 3.367741935483871e-05, | |
| "loss": 0.0116, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 1.1054545454545455, | |
| "grad_norm": 3.5186989307403564, | |
| "learning_rate": 3.3612903225806454e-05, | |
| "loss": 0.0688, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 1.1090909090909091, | |
| "grad_norm": 0.599452018737793, | |
| "learning_rate": 3.3548387096774195e-05, | |
| "loss": 0.004, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.1127272727272728, | |
| "grad_norm": 4.202869415283203, | |
| "learning_rate": 3.3483870967741936e-05, | |
| "loss": 0.0139, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 1.1163636363636364, | |
| "grad_norm": 0.8411999344825745, | |
| "learning_rate": 3.341935483870968e-05, | |
| "loss": 0.0053, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "grad_norm": 0.222166046500206, | |
| "learning_rate": 3.335483870967742e-05, | |
| "loss": 0.0016, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 1.1236363636363635, | |
| "grad_norm": 0.7113834023475647, | |
| "learning_rate": 3.329032258064516e-05, | |
| "loss": 0.0049, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 1.1272727272727272, | |
| "grad_norm": 0.022323666140437126, | |
| "learning_rate": 3.322580645161291e-05, | |
| "loss": 0.0008, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.1309090909090909, | |
| "grad_norm": 0.7203249335289001, | |
| "learning_rate": 3.316129032258064e-05, | |
| "loss": 0.0047, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 1.1345454545454545, | |
| "grad_norm": 0.20508606731891632, | |
| "learning_rate": 3.309677419354839e-05, | |
| "loss": 0.0016, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 1.1381818181818182, | |
| "grad_norm": 0.1757669299840927, | |
| "learning_rate": 3.303225806451613e-05, | |
| "loss": 0.0012, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 1.1418181818181818, | |
| "grad_norm": 0.6888485550880432, | |
| "learning_rate": 3.296774193548387e-05, | |
| "loss": 0.1091, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 1.1454545454545455, | |
| "grad_norm": 3.370866298675537, | |
| "learning_rate": 3.2903225806451614e-05, | |
| "loss": 0.0273, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.1490909090909092, | |
| "grad_norm": 2.7193377017974854, | |
| "learning_rate": 3.2838709677419356e-05, | |
| "loss": 0.1505, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 1.1527272727272728, | |
| "grad_norm": 0.01697608083486557, | |
| "learning_rate": 3.27741935483871e-05, | |
| "loss": 0.0007, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 1.1563636363636363, | |
| "grad_norm": 1.244520902633667, | |
| "learning_rate": 3.270967741935484e-05, | |
| "loss": 0.0076, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "grad_norm": 0.18265090882778168, | |
| "learning_rate": 3.2645161290322586e-05, | |
| "loss": 0.0018, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 1.1636363636363636, | |
| "grad_norm": 0.03304216265678406, | |
| "learning_rate": 3.258064516129033e-05, | |
| "loss": 0.0008, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.1672727272727272, | |
| "grad_norm": 1.4162017107009888, | |
| "learning_rate": 3.251612903225806e-05, | |
| "loss": 0.0071, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 1.170909090909091, | |
| "grad_norm": 0.0734863430261612, | |
| "learning_rate": 3.245161290322581e-05, | |
| "loss": 0.0012, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 1.1745454545454546, | |
| "grad_norm": 4.381263256072998, | |
| "learning_rate": 3.238709677419355e-05, | |
| "loss": 0.0188, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 1.1781818181818182, | |
| "grad_norm": 2.71781325340271, | |
| "learning_rate": 3.232258064516129e-05, | |
| "loss": 0.0278, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 1.1818181818181819, | |
| "grad_norm": 0.1261809766292572, | |
| "learning_rate": 3.2258064516129034e-05, | |
| "loss": 0.0014, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.1854545454545455, | |
| "grad_norm": 2.4839842319488525, | |
| "learning_rate": 3.2193548387096775e-05, | |
| "loss": 0.072, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 1.189090909090909, | |
| "grad_norm": 0.1383955031633377, | |
| "learning_rate": 3.2129032258064516e-05, | |
| "loss": 0.0016, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 1.1927272727272726, | |
| "grad_norm": 3.2481300830841064, | |
| "learning_rate": 3.206451612903226e-05, | |
| "loss": 0.0301, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 1.1963636363636363, | |
| "grad_norm": 4.403379440307617, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 0.0323, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 0.015226184390485287, | |
| "learning_rate": 3.193548387096774e-05, | |
| "loss": 0.0007, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.2036363636363636, | |
| "grad_norm": 0.37973034381866455, | |
| "learning_rate": 3.187096774193549e-05, | |
| "loss": 0.0028, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 1.2072727272727273, | |
| "grad_norm": 6.127589225769043, | |
| "learning_rate": 3.180645161290323e-05, | |
| "loss": 0.0691, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 1.210909090909091, | |
| "grad_norm": 0.8358224034309387, | |
| "learning_rate": 3.174193548387097e-05, | |
| "loss": 0.1013, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 1.2145454545454546, | |
| "grad_norm": 0.036338452249765396, | |
| "learning_rate": 3.167741935483871e-05, | |
| "loss": 0.0011, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 1.2181818181818183, | |
| "grad_norm": 0.02295631729066372, | |
| "learning_rate": 3.161290322580645e-05, | |
| "loss": 0.0009, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.221818181818182, | |
| "grad_norm": 0.05319954827427864, | |
| "learning_rate": 3.1548387096774194e-05, | |
| "loss": 0.0012, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 1.2254545454545456, | |
| "grad_norm": 6.075497150421143, | |
| "learning_rate": 3.1483870967741935e-05, | |
| "loss": 0.1689, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 1.229090909090909, | |
| "grad_norm": 0.02964276447892189, | |
| "learning_rate": 3.141935483870968e-05, | |
| "loss": 0.0012, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 1.2327272727272727, | |
| "grad_norm": 3.0539228916168213, | |
| "learning_rate": 3.135483870967742e-05, | |
| "loss": 0.0269, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 1.2363636363636363, | |
| "grad_norm": 0.04519687220454216, | |
| "learning_rate": 3.1290322580645166e-05, | |
| "loss": 0.0017, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "grad_norm": 0.10507076978683472, | |
| "learning_rate": 3.122580645161291e-05, | |
| "loss": 0.0027, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 1.2436363636363637, | |
| "grad_norm": 3.865663766860962, | |
| "learning_rate": 3.116129032258064e-05, | |
| "loss": 0.0435, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 1.2472727272727273, | |
| "grad_norm": 1.0297927856445312, | |
| "learning_rate": 3.109677419354839e-05, | |
| "loss": 0.0093, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 1.250909090909091, | |
| "grad_norm": 1.1631922721862793, | |
| "learning_rate": 3.103225806451613e-05, | |
| "loss": 0.0073, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.2545454545454544, | |
| "grad_norm": 0.11844321340322495, | |
| "learning_rate": 3.096774193548387e-05, | |
| "loss": 0.0022, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.2581818181818183, | |
| "grad_norm": 0.6237489581108093, | |
| "learning_rate": 3.090322580645161e-05, | |
| "loss": 0.004, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.2618181818181817, | |
| "grad_norm": 0.883538544178009, | |
| "learning_rate": 3.083870967741936e-05, | |
| "loss": 0.0056, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 1.2654545454545454, | |
| "grad_norm": 0.5708529949188232, | |
| "learning_rate": 3.0774193548387096e-05, | |
| "loss": 0.0051, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.269090909090909, | |
| "grad_norm": 0.3859453797340393, | |
| "learning_rate": 3.070967741935484e-05, | |
| "loss": 0.0026, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 1.2727272727272727, | |
| "grad_norm": 0.07389602810144424, | |
| "learning_rate": 3.0645161290322585e-05, | |
| "loss": 0.0018, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.2763636363636364, | |
| "grad_norm": 5.179856300354004, | |
| "learning_rate": 3.058064516129032e-05, | |
| "loss": 0.1931, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "grad_norm": 1.7609326839447021, | |
| "learning_rate": 3.0516129032258067e-05, | |
| "loss": 0.0627, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.2836363636363637, | |
| "grad_norm": 2.443671464920044, | |
| "learning_rate": 3.0451612903225805e-05, | |
| "loss": 0.0098, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.2872727272727273, | |
| "grad_norm": 0.024096990004181862, | |
| "learning_rate": 3.0387096774193553e-05, | |
| "loss": 0.001, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.290909090909091, | |
| "grad_norm": 0.03544619679450989, | |
| "learning_rate": 3.032258064516129e-05, | |
| "loss": 0.0013, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.2945454545454544, | |
| "grad_norm": 0.8081026673316956, | |
| "learning_rate": 3.0258064516129032e-05, | |
| "loss": 0.0049, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.298181818181818, | |
| "grad_norm": 0.13682828843593597, | |
| "learning_rate": 3.0193548387096777e-05, | |
| "loss": 0.0025, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.3018181818181818, | |
| "grad_norm": 0.04892565682530403, | |
| "learning_rate": 3.0129032258064515e-05, | |
| "loss": 0.0012, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.3054545454545454, | |
| "grad_norm": 0.04417359083890915, | |
| "learning_rate": 3.006451612903226e-05, | |
| "loss": 0.0009, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.309090909090909, | |
| "grad_norm": 1.1065845489501953, | |
| "learning_rate": 3e-05, | |
| "loss": 0.0099, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.3127272727272727, | |
| "grad_norm": 1.6691566705703735, | |
| "learning_rate": 2.9935483870967745e-05, | |
| "loss": 0.0704, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.3163636363636364, | |
| "grad_norm": 0.744577944278717, | |
| "learning_rate": 2.9870967741935487e-05, | |
| "loss": 0.0546, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "grad_norm": 1.7217384576797485, | |
| "learning_rate": 2.9806451612903224e-05, | |
| "loss": 0.11, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.3236363636363637, | |
| "grad_norm": 0.28095120191574097, | |
| "learning_rate": 2.974193548387097e-05, | |
| "loss": 0.0035, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.3272727272727272, | |
| "grad_norm": 0.25743359327316284, | |
| "learning_rate": 2.967741935483871e-05, | |
| "loss": 0.0026, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.330909090909091, | |
| "grad_norm": 0.04120393097400665, | |
| "learning_rate": 2.9612903225806455e-05, | |
| "loss": 0.0012, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.3345454545454545, | |
| "grad_norm": 0.08902228623628616, | |
| "learning_rate": 2.9548387096774193e-05, | |
| "loss": 0.0026, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.3381818181818181, | |
| "grad_norm": 2.2871615886688232, | |
| "learning_rate": 2.9483870967741937e-05, | |
| "loss": 0.0267, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.3418181818181818, | |
| "grad_norm": 0.9016657471656799, | |
| "learning_rate": 2.941935483870968e-05, | |
| "loss": 0.127, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.3454545454545455, | |
| "grad_norm": 3.976222276687622, | |
| "learning_rate": 2.9354838709677417e-05, | |
| "loss": 0.0278, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.3490909090909091, | |
| "grad_norm": 0.11579588800668716, | |
| "learning_rate": 2.9290322580645165e-05, | |
| "loss": 0.0027, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.3527272727272728, | |
| "grad_norm": 2.413121223449707, | |
| "learning_rate": 2.9225806451612902e-05, | |
| "loss": 0.0107, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.3563636363636364, | |
| "grad_norm": 1.4316357374191284, | |
| "learning_rate": 2.9161290322580647e-05, | |
| "loss": 0.0093, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.3599999999999999, | |
| "grad_norm": 0.041745781898498535, | |
| "learning_rate": 2.909677419354839e-05, | |
| "loss": 0.0015, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.3636363636363638, | |
| "grad_norm": 0.59097820520401, | |
| "learning_rate": 2.9032258064516133e-05, | |
| "loss": 0.0034, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.3672727272727272, | |
| "grad_norm": 0.06494897603988647, | |
| "learning_rate": 2.896774193548387e-05, | |
| "loss": 0.0015, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.3709090909090909, | |
| "grad_norm": 3.1881821155548096, | |
| "learning_rate": 2.8903225806451615e-05, | |
| "loss": 0.0568, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.3745454545454545, | |
| "grad_norm": 0.9627525210380554, | |
| "learning_rate": 2.8838709677419357e-05, | |
| "loss": 0.0067, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.3781818181818182, | |
| "grad_norm": 2.6984145641326904, | |
| "learning_rate": 2.8774193548387095e-05, | |
| "loss": 0.0655, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.3818181818181818, | |
| "grad_norm": 1.9964426755905151, | |
| "learning_rate": 2.8709677419354843e-05, | |
| "loss": 0.0546, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.3854545454545455, | |
| "grad_norm": 0.019922640174627304, | |
| "learning_rate": 2.864516129032258e-05, | |
| "loss": 0.001, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.3890909090909092, | |
| "grad_norm": 3.0499842166900635, | |
| "learning_rate": 2.8580645161290325e-05, | |
| "loss": 0.0726, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.3927272727272726, | |
| "grad_norm": 0.20027759671211243, | |
| "learning_rate": 2.8516129032258066e-05, | |
| "loss": 0.003, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.3963636363636365, | |
| "grad_norm": 0.549941897392273, | |
| "learning_rate": 2.845161290322581e-05, | |
| "loss": 0.1096, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "grad_norm": 5.2639946937561035, | |
| "learning_rate": 2.838709677419355e-05, | |
| "loss": 0.0766, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.4036363636363636, | |
| "grad_norm": 0.5863090753555298, | |
| "learning_rate": 2.832258064516129e-05, | |
| "loss": 0.0076, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.4072727272727272, | |
| "grad_norm": 0.020899731665849686, | |
| "learning_rate": 2.8258064516129035e-05, | |
| "loss": 0.001, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.410909090909091, | |
| "grad_norm": 0.2003995031118393, | |
| "learning_rate": 2.8193548387096776e-05, | |
| "loss": 0.0031, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.4145454545454546, | |
| "grad_norm": 2.7366487979888916, | |
| "learning_rate": 2.812903225806452e-05, | |
| "loss": 0.052, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.4181818181818182, | |
| "grad_norm": 0.03661293536424637, | |
| "learning_rate": 2.806451612903226e-05, | |
| "loss": 0.0014, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.4218181818181819, | |
| "grad_norm": 1.4454936981201172, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 0.0122, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.4254545454545455, | |
| "grad_norm": 0.03340213745832443, | |
| "learning_rate": 2.7935483870967744e-05, | |
| "loss": 0.0015, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.4290909090909092, | |
| "grad_norm": 0.654367983341217, | |
| "learning_rate": 2.7870967741935482e-05, | |
| "loss": 0.0069, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.4327272727272726, | |
| "grad_norm": 0.08472099900245667, | |
| "learning_rate": 2.7806451612903227e-05, | |
| "loss": 0.0018, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.4363636363636363, | |
| "grad_norm": 3.4286415576934814, | |
| "learning_rate": 2.7741935483870968e-05, | |
| "loss": 0.0662, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "grad_norm": 1.312857747077942, | |
| "learning_rate": 2.7677419354838713e-05, | |
| "loss": 0.014, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.4436363636363636, | |
| "grad_norm": 0.035416845232248306, | |
| "learning_rate": 2.7612903225806454e-05, | |
| "loss": 0.0013, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.4472727272727273, | |
| "grad_norm": 4.134556770324707, | |
| "learning_rate": 2.75483870967742e-05, | |
| "loss": 0.0354, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.450909090909091, | |
| "grad_norm": 0.43342649936676025, | |
| "learning_rate": 2.7483870967741936e-05, | |
| "loss": 0.0053, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.4545454545454546, | |
| "grad_norm": 0.270112007856369, | |
| "learning_rate": 2.7419354838709678e-05, | |
| "loss": 0.0029, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.4581818181818182, | |
| "grad_norm": 0.039803147315979004, | |
| "learning_rate": 2.7354838709677422e-05, | |
| "loss": 0.0017, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.461818181818182, | |
| "grad_norm": 2.7924137115478516, | |
| "learning_rate": 2.729032258064516e-05, | |
| "loss": 0.0366, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.4654545454545453, | |
| "grad_norm": 0.29749271273612976, | |
| "learning_rate": 2.7225806451612905e-05, | |
| "loss": 0.0026, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.4690909090909092, | |
| "grad_norm": 1.2988048791885376, | |
| "learning_rate": 2.7161290322580646e-05, | |
| "loss": 0.0083, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.4727272727272727, | |
| "grad_norm": 0.07334749400615692, | |
| "learning_rate": 2.709677419354839e-05, | |
| "loss": 0.0019, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.4763636363636363, | |
| "grad_norm": 0.04504287615418434, | |
| "learning_rate": 2.7032258064516132e-05, | |
| "loss": 0.0018, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "grad_norm": 0.08913391828536987, | |
| "learning_rate": 2.696774193548387e-05, | |
| "loss": 0.0018, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.4836363636363636, | |
| "grad_norm": 0.20501653850078583, | |
| "learning_rate": 2.6903225806451614e-05, | |
| "loss": 0.0022, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.4872727272727273, | |
| "grad_norm": 1.9350942373275757, | |
| "learning_rate": 2.6838709677419355e-05, | |
| "loss": 0.0097, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.490909090909091, | |
| "grad_norm": 4.018691062927246, | |
| "learning_rate": 2.67741935483871e-05, | |
| "loss": 0.0256, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.4945454545454546, | |
| "grad_norm": 0.02390647679567337, | |
| "learning_rate": 2.6709677419354838e-05, | |
| "loss": 0.001, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.498181818181818, | |
| "grad_norm": 2.684476375579834, | |
| "learning_rate": 2.6645161290322586e-05, | |
| "loss": 0.0146, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.501818181818182, | |
| "grad_norm": 0.021060334518551826, | |
| "learning_rate": 2.6580645161290324e-05, | |
| "loss": 0.0009, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.5054545454545454, | |
| "grad_norm": 0.4146246910095215, | |
| "learning_rate": 2.6516129032258065e-05, | |
| "loss": 0.0025, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.509090909090909, | |
| "grad_norm": 3.564082384109497, | |
| "learning_rate": 2.645161290322581e-05, | |
| "loss": 0.0188, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.5127272727272727, | |
| "grad_norm": 0.027183400467038155, | |
| "learning_rate": 2.6387096774193548e-05, | |
| "loss": 0.0012, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.5163636363636364, | |
| "grad_norm": 0.4653225541114807, | |
| "learning_rate": 2.6322580645161292e-05, | |
| "loss": 0.0055, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "grad_norm": 0.2512191832065582, | |
| "learning_rate": 2.6258064516129033e-05, | |
| "loss": 0.0015, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.5236363636363637, | |
| "grad_norm": 2.2450575828552246, | |
| "learning_rate": 2.6193548387096778e-05, | |
| "loss": 0.0209, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.5272727272727273, | |
| "grad_norm": 2.3167543411254883, | |
| "learning_rate": 2.6129032258064516e-05, | |
| "loss": 0.0544, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.5309090909090908, | |
| "grad_norm": 0.03268599510192871, | |
| "learning_rate": 2.6064516129032257e-05, | |
| "loss": 0.0011, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.5345454545454547, | |
| "grad_norm": 0.028883187100291252, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 0.001, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.538181818181818, | |
| "grad_norm": 0.1544177383184433, | |
| "learning_rate": 2.5935483870967743e-05, | |
| "loss": 0.0022, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.541818181818182, | |
| "grad_norm": 2.9123668670654297, | |
| "learning_rate": 2.5870967741935488e-05, | |
| "loss": 0.0187, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.5454545454545454, | |
| "grad_norm": 0.15737684071063995, | |
| "learning_rate": 2.5806451612903226e-05, | |
| "loss": 0.0021, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.549090909090909, | |
| "grad_norm": 0.023125503212213516, | |
| "learning_rate": 2.574193548387097e-05, | |
| "loss": 0.0009, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.5527272727272727, | |
| "grad_norm": 5.130437850952148, | |
| "learning_rate": 2.567741935483871e-05, | |
| "loss": 0.1732, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.5563636363636364, | |
| "grad_norm": 0.017136206850409508, | |
| "learning_rate": 2.561290322580645e-05, | |
| "loss": 0.0008, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "grad_norm": 2.5280985832214355, | |
| "learning_rate": 2.5548387096774197e-05, | |
| "loss": 0.0143, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.5636363636363635, | |
| "grad_norm": 0.2261432558298111, | |
| "learning_rate": 2.5483870967741935e-05, | |
| "loss": 0.0029, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.5672727272727274, | |
| "grad_norm": 0.018230870366096497, | |
| "learning_rate": 2.541935483870968e-05, | |
| "loss": 0.0008, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 1.5709090909090908, | |
| "grad_norm": 0.12864048779010773, | |
| "learning_rate": 2.535483870967742e-05, | |
| "loss": 0.0016, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.5745454545454547, | |
| "grad_norm": 0.03463654965162277, | |
| "learning_rate": 2.5290322580645166e-05, | |
| "loss": 0.0009, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 1.5781818181818181, | |
| "grad_norm": 0.04040815308690071, | |
| "learning_rate": 2.5225806451612903e-05, | |
| "loss": 0.0009, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.5818181818181818, | |
| "grad_norm": 6.010333061218262, | |
| "learning_rate": 2.5161290322580645e-05, | |
| "loss": 0.0476, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.5854545454545454, | |
| "grad_norm": 2.538048505783081, | |
| "learning_rate": 2.509677419354839e-05, | |
| "loss": 0.0706, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.589090909090909, | |
| "grad_norm": 0.9795745015144348, | |
| "learning_rate": 2.5032258064516127e-05, | |
| "loss": 0.0043, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 1.5927272727272728, | |
| "grad_norm": 0.07065194100141525, | |
| "learning_rate": 2.4967741935483872e-05, | |
| "loss": 0.0011, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.5963636363636362, | |
| "grad_norm": 2.8576443195343018, | |
| "learning_rate": 2.4903225806451613e-05, | |
| "loss": 0.0471, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "grad_norm": 2.197402238845825, | |
| "learning_rate": 2.4838709677419354e-05, | |
| "loss": 0.0809, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.6036363636363635, | |
| "grad_norm": 0.3648858368396759, | |
| "learning_rate": 2.47741935483871e-05, | |
| "loss": 0.0038, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 1.6072727272727274, | |
| "grad_norm": 0.1489875763654709, | |
| "learning_rate": 2.470967741935484e-05, | |
| "loss": 0.0017, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.6109090909090908, | |
| "grad_norm": 0.07092121988534927, | |
| "learning_rate": 2.464516129032258e-05, | |
| "loss": 0.0014, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.6145454545454545, | |
| "grad_norm": 2.7964096069335938, | |
| "learning_rate": 2.4580645161290326e-05, | |
| "loss": 0.0509, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.6181818181818182, | |
| "grad_norm": 1.7272242307662964, | |
| "learning_rate": 2.4516129032258064e-05, | |
| "loss": 0.0067, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.6218181818181818, | |
| "grad_norm": 0.06453117728233337, | |
| "learning_rate": 2.4451612903225805e-05, | |
| "loss": 0.0016, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.6254545454545455, | |
| "grad_norm": 3.195802927017212, | |
| "learning_rate": 2.438709677419355e-05, | |
| "loss": 0.0547, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 1.6290909090909091, | |
| "grad_norm": 0.01947391778230667, | |
| "learning_rate": 2.432258064516129e-05, | |
| "loss": 0.0008, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.6327272727272728, | |
| "grad_norm": 0.5001751780509949, | |
| "learning_rate": 2.4258064516129032e-05, | |
| "loss": 0.004, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.6363636363636362, | |
| "grad_norm": 0.4999580383300781, | |
| "learning_rate": 2.4193548387096777e-05, | |
| "loss": 0.0057, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.6400000000000001, | |
| "grad_norm": 1.5922938585281372, | |
| "learning_rate": 2.4129032258064518e-05, | |
| "loss": 0.0116, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.6436363636363636, | |
| "grad_norm": 2.461899518966675, | |
| "learning_rate": 2.406451612903226e-05, | |
| "loss": 0.018, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.6472727272727272, | |
| "grad_norm": 0.07266916334629059, | |
| "learning_rate": 2.4e-05, | |
| "loss": 0.0014, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.6509090909090909, | |
| "grad_norm": 1.8282543420791626, | |
| "learning_rate": 2.3935483870967742e-05, | |
| "loss": 0.0841, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.6545454545454545, | |
| "grad_norm": 0.06280002743005753, | |
| "learning_rate": 2.3870967741935486e-05, | |
| "loss": 0.0013, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.6581818181818182, | |
| "grad_norm": 0.06250961124897003, | |
| "learning_rate": 2.3806451612903228e-05, | |
| "loss": 0.0016, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.6618181818181819, | |
| "grad_norm": 0.35135146975517273, | |
| "learning_rate": 2.374193548387097e-05, | |
| "loss": 0.0253, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.6654545454545455, | |
| "grad_norm": 0.0423726923763752, | |
| "learning_rate": 2.367741935483871e-05, | |
| "loss": 0.0009, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.669090909090909, | |
| "grad_norm": 1.339455246925354, | |
| "learning_rate": 2.361290322580645e-05, | |
| "loss": 0.0794, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.6727272727272728, | |
| "grad_norm": 0.1556975394487381, | |
| "learning_rate": 2.3548387096774193e-05, | |
| "loss": 0.0018, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.6763636363636363, | |
| "grad_norm": 1.5904016494750977, | |
| "learning_rate": 2.3483870967741937e-05, | |
| "loss": 0.0036, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.6800000000000002, | |
| "grad_norm": 0.6247179508209229, | |
| "learning_rate": 2.341935483870968e-05, | |
| "loss": 0.0037, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.6836363636363636, | |
| "grad_norm": 1.879459023475647, | |
| "learning_rate": 2.335483870967742e-05, | |
| "loss": 0.0099, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.6872727272727273, | |
| "grad_norm": 0.07403961569070816, | |
| "learning_rate": 2.3290322580645164e-05, | |
| "loss": 0.0016, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.690909090909091, | |
| "grad_norm": 0.05183988809585571, | |
| "learning_rate": 2.3225806451612906e-05, | |
| "loss": 0.0012, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.6945454545454546, | |
| "grad_norm": 5.193166255950928, | |
| "learning_rate": 2.3161290322580644e-05, | |
| "loss": 0.0709, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.6981818181818182, | |
| "grad_norm": 1.2823094129562378, | |
| "learning_rate": 2.3096774193548388e-05, | |
| "loss": 0.0195, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.7018181818181817, | |
| "grad_norm": 2.308457851409912, | |
| "learning_rate": 2.303225806451613e-05, | |
| "loss": 0.0082, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.7054545454545456, | |
| "grad_norm": 0.39891281723976135, | |
| "learning_rate": 2.296774193548387e-05, | |
| "loss": 0.0036, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.709090909090909, | |
| "grad_norm": 0.03392359986901283, | |
| "learning_rate": 2.2903225806451615e-05, | |
| "loss": 0.0012, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.7127272727272729, | |
| "grad_norm": 0.01698954403400421, | |
| "learning_rate": 2.2838709677419357e-05, | |
| "loss": 0.0007, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.7163636363636363, | |
| "grad_norm": 0.059636473655700684, | |
| "learning_rate": 2.2774193548387098e-05, | |
| "loss": 0.0016, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "grad_norm": 0.07056300342082977, | |
| "learning_rate": 2.2709677419354842e-05, | |
| "loss": 0.0017, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.7236363636363636, | |
| "grad_norm": 0.1932046115398407, | |
| "learning_rate": 2.264516129032258e-05, | |
| "loss": 0.0029, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.7272727272727273, | |
| "grad_norm": 2.354381561279297, | |
| "learning_rate": 2.258064516129032e-05, | |
| "loss": 0.0141, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.730909090909091, | |
| "grad_norm": 0.17981848120689392, | |
| "learning_rate": 2.2516129032258066e-05, | |
| "loss": 0.002, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.7345454545454544, | |
| "grad_norm": 0.03439001739025116, | |
| "learning_rate": 2.2451612903225807e-05, | |
| "loss": 0.0008, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.7381818181818183, | |
| "grad_norm": 1.2802492380142212, | |
| "learning_rate": 2.238709677419355e-05, | |
| "loss": 0.0143, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.7418181818181817, | |
| "grad_norm": 0.36729562282562256, | |
| "learning_rate": 2.2322580645161293e-05, | |
| "loss": 0.0085, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.7454545454545456, | |
| "grad_norm": 0.03446509316563606, | |
| "learning_rate": 2.2258064516129034e-05, | |
| "loss": 0.0009, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.749090909090909, | |
| "grad_norm": 0.05642567202448845, | |
| "learning_rate": 2.2193548387096776e-05, | |
| "loss": 0.0014, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.7527272727272727, | |
| "grad_norm": 0.12370045483112335, | |
| "learning_rate": 2.2129032258064517e-05, | |
| "loss": 0.0016, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.7563636363636363, | |
| "grad_norm": 0.038627080619335175, | |
| "learning_rate": 2.2064516129032258e-05, | |
| "loss": 0.0011, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "grad_norm": 0.037693917751312256, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 0.0007, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.7636363636363637, | |
| "grad_norm": 1.4172790050506592, | |
| "learning_rate": 2.1935483870967744e-05, | |
| "loss": 0.0865, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.767272727272727, | |
| "grad_norm": 2.643702268600464, | |
| "learning_rate": 2.1870967741935485e-05, | |
| "loss": 0.0205, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.770909090909091, | |
| "grad_norm": 3.630894899368286, | |
| "learning_rate": 2.1806451612903227e-05, | |
| "loss": 0.0227, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.7745454545454544, | |
| "grad_norm": 0.11117129772901535, | |
| "learning_rate": 2.1741935483870968e-05, | |
| "loss": 0.0014, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.7781818181818183, | |
| "grad_norm": 0.041525308042764664, | |
| "learning_rate": 2.167741935483871e-05, | |
| "loss": 0.001, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.7818181818181817, | |
| "grad_norm": 0.8061334490776062, | |
| "learning_rate": 2.1612903225806454e-05, | |
| "loss": 0.1222, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.7854545454545454, | |
| "grad_norm": 0.03137822821736336, | |
| "learning_rate": 2.1548387096774195e-05, | |
| "loss": 0.0008, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.789090909090909, | |
| "grad_norm": 0.44713956117630005, | |
| "learning_rate": 2.1483870967741936e-05, | |
| "loss": 0.0023, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.7927272727272727, | |
| "grad_norm": 0.43540823459625244, | |
| "learning_rate": 2.141935483870968e-05, | |
| "loss": 0.0035, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 1.7963636363636364, | |
| "grad_norm": 0.1677086055278778, | |
| "learning_rate": 2.1354838709677422e-05, | |
| "loss": 0.002, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "grad_norm": 0.30730533599853516, | |
| "learning_rate": 2.129032258064516e-05, | |
| "loss": 0.0044, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.8036363636363637, | |
| "grad_norm": 0.14470332860946655, | |
| "learning_rate": 2.1225806451612904e-05, | |
| "loss": 0.0018, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.8072727272727271, | |
| "grad_norm": 0.05039800703525543, | |
| "learning_rate": 2.1161290322580646e-05, | |
| "loss": 0.0011, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 1.810909090909091, | |
| "grad_norm": 0.15073451399803162, | |
| "learning_rate": 2.1096774193548387e-05, | |
| "loss": 0.0018, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.8145454545454545, | |
| "grad_norm": 2.081968307495117, | |
| "learning_rate": 2.103225806451613e-05, | |
| "loss": 0.0765, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 1.8181818181818183, | |
| "grad_norm": 0.030230529606342316, | |
| "learning_rate": 2.0967741935483873e-05, | |
| "loss": 0.001, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.8218181818181818, | |
| "grad_norm": 0.2655714750289917, | |
| "learning_rate": 2.0903225806451614e-05, | |
| "loss": 0.0021, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 1.8254545454545454, | |
| "grad_norm": 0.15943261981010437, | |
| "learning_rate": 2.0838709677419355e-05, | |
| "loss": 0.0015, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 1.829090909090909, | |
| "grad_norm": 4.7471489906311035, | |
| "learning_rate": 2.0774193548387097e-05, | |
| "loss": 0.1086, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 1.8327272727272728, | |
| "grad_norm": 0.1162559986114502, | |
| "learning_rate": 2.0709677419354838e-05, | |
| "loss": 0.0016, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 1.8363636363636364, | |
| "grad_norm": 0.0627504363656044, | |
| "learning_rate": 2.0645161290322582e-05, | |
| "loss": 0.0013, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.8399999999999999, | |
| "grad_norm": 0.5027517080307007, | |
| "learning_rate": 2.0580645161290324e-05, | |
| "loss": 0.0035, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 1.8436363636363637, | |
| "grad_norm": 0.02140502817928791, | |
| "learning_rate": 2.0516129032258065e-05, | |
| "loss": 0.0008, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 1.8472727272727272, | |
| "grad_norm": 0.07203751057386398, | |
| "learning_rate": 2.045161290322581e-05, | |
| "loss": 0.0018, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 1.850909090909091, | |
| "grad_norm": 1.8638368844985962, | |
| "learning_rate": 2.0387096774193547e-05, | |
| "loss": 0.0594, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 1.8545454545454545, | |
| "grad_norm": 0.12549816071987152, | |
| "learning_rate": 2.0322580645161292e-05, | |
| "loss": 0.0019, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.8581818181818182, | |
| "grad_norm": 0.5917963981628418, | |
| "learning_rate": 2.0258064516129033e-05, | |
| "loss": 0.0044, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 1.8618181818181818, | |
| "grad_norm": 0.013607682660222054, | |
| "learning_rate": 2.0193548387096775e-05, | |
| "loss": 0.0007, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 1.8654545454545455, | |
| "grad_norm": 0.01925772987306118, | |
| "learning_rate": 2.0129032258064516e-05, | |
| "loss": 0.0007, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 1.8690909090909091, | |
| "grad_norm": 0.023444948717951775, | |
| "learning_rate": 2.006451612903226e-05, | |
| "loss": 0.0008, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 1.8727272727272726, | |
| "grad_norm": 0.19968028366565704, | |
| "learning_rate": 2e-05, | |
| "loss": 0.0018, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.8763636363636365, | |
| "grad_norm": 1.308447003364563, | |
| "learning_rate": 1.9935483870967743e-05, | |
| "loss": 0.0883, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 0.20825034379959106, | |
| "learning_rate": 1.9870967741935484e-05, | |
| "loss": 0.0022, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 1.8836363636363638, | |
| "grad_norm": 0.06526435911655426, | |
| "learning_rate": 1.9806451612903225e-05, | |
| "loss": 0.0014, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 1.8872727272727272, | |
| "grad_norm": 0.1210133358836174, | |
| "learning_rate": 1.974193548387097e-05, | |
| "loss": 0.0009, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 1.8909090909090909, | |
| "grad_norm": 1.0739092826843262, | |
| "learning_rate": 1.967741935483871e-05, | |
| "loss": 0.1127, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.8945454545454545, | |
| "grad_norm": 0.018074801191687584, | |
| "learning_rate": 1.9612903225806452e-05, | |
| "loss": 0.0008, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 1.8981818181818182, | |
| "grad_norm": 2.0916597843170166, | |
| "learning_rate": 1.9548387096774197e-05, | |
| "loss": 0.0115, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 1.9018181818181819, | |
| "grad_norm": 0.015548643656075, | |
| "learning_rate": 1.9483870967741935e-05, | |
| "loss": 0.0006, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 1.9054545454545453, | |
| "grad_norm": 0.10145322978496552, | |
| "learning_rate": 1.9419354838709676e-05, | |
| "loss": 0.002, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 1.9090909090909092, | |
| "grad_norm": 0.24982163310050964, | |
| "learning_rate": 1.935483870967742e-05, | |
| "loss": 0.0024, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.9127272727272726, | |
| "grad_norm": 0.14598214626312256, | |
| "learning_rate": 1.9290322580645162e-05, | |
| "loss": 0.0017, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.9163636363636365, | |
| "grad_norm": 4.5397629737854, | |
| "learning_rate": 1.9225806451612903e-05, | |
| "loss": 0.1227, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "grad_norm": 0.5093303322792053, | |
| "learning_rate": 1.9161290322580648e-05, | |
| "loss": 0.0323, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.9236363636363636, | |
| "grad_norm": 0.14823508262634277, | |
| "learning_rate": 1.909677419354839e-05, | |
| "loss": 0.0017, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 1.9272727272727272, | |
| "grad_norm": 0.6760469675064087, | |
| "learning_rate": 1.9032258064516127e-05, | |
| "loss": 0.0054, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.930909090909091, | |
| "grad_norm": 0.05310118570923805, | |
| "learning_rate": 1.896774193548387e-05, | |
| "loss": 0.0012, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 1.9345454545454546, | |
| "grad_norm": 0.2781686782836914, | |
| "learning_rate": 1.8903225806451613e-05, | |
| "loss": 0.0026, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.9381818181818182, | |
| "grad_norm": 0.0464974045753479, | |
| "learning_rate": 1.8838709677419354e-05, | |
| "loss": 0.0015, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 1.9418181818181819, | |
| "grad_norm": 0.5551739931106567, | |
| "learning_rate": 1.87741935483871e-05, | |
| "loss": 0.0065, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.9454545454545453, | |
| "grad_norm": 0.08245756477117538, | |
| "learning_rate": 1.870967741935484e-05, | |
| "loss": 0.0015, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.9490909090909092, | |
| "grad_norm": 1.5298570394515991, | |
| "learning_rate": 1.864516129032258e-05, | |
| "loss": 0.0814, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.9527272727272726, | |
| "grad_norm": 0.028485940769314766, | |
| "learning_rate": 1.8580645161290326e-05, | |
| "loss": 0.0009, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 1.9563636363636365, | |
| "grad_norm": 0.12668201327323914, | |
| "learning_rate": 1.8516129032258064e-05, | |
| "loss": 0.0015, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "grad_norm": 0.11611904203891754, | |
| "learning_rate": 1.845161290322581e-05, | |
| "loss": 0.0023, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 1.9636363636363636, | |
| "grad_norm": 1.0167148113250732, | |
| "learning_rate": 1.838709677419355e-05, | |
| "loss": 0.0055, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.9672727272727273, | |
| "grad_norm": 4.344989776611328, | |
| "learning_rate": 1.832258064516129e-05, | |
| "loss": 0.0237, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 1.970909090909091, | |
| "grad_norm": 1.655159831047058, | |
| "learning_rate": 1.8258064516129032e-05, | |
| "loss": 0.01, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.9745454545454546, | |
| "grad_norm": 0.03663047030568123, | |
| "learning_rate": 1.8193548387096777e-05, | |
| "loss": 0.0012, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.978181818181818, | |
| "grad_norm": 0.4332762062549591, | |
| "learning_rate": 1.8129032258064518e-05, | |
| "loss": 0.0039, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.981818181818182, | |
| "grad_norm": 3.9883310794830322, | |
| "learning_rate": 1.806451612903226e-05, | |
| "loss": 0.0309, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.9854545454545454, | |
| "grad_norm": 2.0200157165527344, | |
| "learning_rate": 1.8e-05, | |
| "loss": 0.0127, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.9890909090909092, | |
| "grad_norm": 1.3924773931503296, | |
| "learning_rate": 1.7935483870967742e-05, | |
| "loss": 0.0069, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 1.9927272727272727, | |
| "grad_norm": 0.2380281239748001, | |
| "learning_rate": 1.7870967741935486e-05, | |
| "loss": 0.0196, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.9963636363636363, | |
| "grad_norm": 0.9676334261894226, | |
| "learning_rate": 1.7806451612903228e-05, | |
| "loss": 0.1096, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.012336778454482555, | |
| "learning_rate": 1.774193548387097e-05, | |
| "loss": 0.0006, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9918032786885246, | |
| "eval_loss": 0.03281828388571739, | |
| "eval_runtime": 12.743, | |
| "eval_samples_per_second": 344.661, | |
| "eval_steps_per_second": 5.415, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.0036363636363634, | |
| "grad_norm": 0.019314678385853767, | |
| "learning_rate": 1.7677419354838713e-05, | |
| "loss": 0.0008, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 2.0072727272727273, | |
| "grad_norm": 0.11856160312891006, | |
| "learning_rate": 1.761290322580645e-05, | |
| "loss": 0.0016, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 2.0109090909090908, | |
| "grad_norm": 0.019284116104245186, | |
| "learning_rate": 1.7548387096774193e-05, | |
| "loss": 0.0008, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 2.0145454545454546, | |
| "grad_norm": 0.701896607875824, | |
| "learning_rate": 1.7483870967741937e-05, | |
| "loss": 0.0056, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 2.018181818181818, | |
| "grad_norm": 0.1820172518491745, | |
| "learning_rate": 1.741935483870968e-05, | |
| "loss": 0.0016, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 2.021818181818182, | |
| "grad_norm": 0.014085445553064346, | |
| "learning_rate": 1.735483870967742e-05, | |
| "loss": 0.0007, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 2.0254545454545454, | |
| "grad_norm": 0.01478899922221899, | |
| "learning_rate": 1.7290322580645164e-05, | |
| "loss": 0.0007, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 2.0290909090909093, | |
| "grad_norm": 0.01645870879292488, | |
| "learning_rate": 1.7225806451612906e-05, | |
| "loss": 0.0007, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 2.0327272727272727, | |
| "grad_norm": 0.09879927337169647, | |
| "learning_rate": 1.7161290322580643e-05, | |
| "loss": 0.0013, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 2.036363636363636, | |
| "grad_norm": 0.016561178490519524, | |
| "learning_rate": 1.7096774193548388e-05, | |
| "loss": 0.0008, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "grad_norm": 0.030180329456925392, | |
| "learning_rate": 1.703225806451613e-05, | |
| "loss": 0.0011, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 2.0436363636363635, | |
| "grad_norm": 0.023942433297634125, | |
| "learning_rate": 1.696774193548387e-05, | |
| "loss": 0.0009, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 2.0472727272727274, | |
| "grad_norm": 0.5132169127464294, | |
| "learning_rate": 1.6903225806451615e-05, | |
| "loss": 0.0216, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 2.050909090909091, | |
| "grad_norm": 0.01695244014263153, | |
| "learning_rate": 1.6838709677419356e-05, | |
| "loss": 0.0008, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 2.0545454545454547, | |
| "grad_norm": 0.019326528534293175, | |
| "learning_rate": 1.6774193548387098e-05, | |
| "loss": 0.0008, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 2.058181818181818, | |
| "grad_norm": 0.018215378746390343, | |
| "learning_rate": 1.670967741935484e-05, | |
| "loss": 0.0007, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 2.061818181818182, | |
| "grad_norm": 0.02021806314587593, | |
| "learning_rate": 1.664516129032258e-05, | |
| "loss": 0.0008, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 2.0654545454545454, | |
| "grad_norm": 0.030195925384759903, | |
| "learning_rate": 1.658064516129032e-05, | |
| "loss": 0.0009, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 2.0690909090909093, | |
| "grad_norm": 0.2582416236400604, | |
| "learning_rate": 1.6516129032258066e-05, | |
| "loss": 0.0014, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 2.0727272727272728, | |
| "grad_norm": 0.052597131580114365, | |
| "learning_rate": 1.6451612903225807e-05, | |
| "loss": 0.0013, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 2.076363636363636, | |
| "grad_norm": 4.027952194213867, | |
| "learning_rate": 1.638709677419355e-05, | |
| "loss": 0.0367, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "grad_norm": 3.2471768856048584, | |
| "learning_rate": 1.6322580645161293e-05, | |
| "loss": 0.0131, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 2.0836363636363635, | |
| "grad_norm": 0.013683440163731575, | |
| "learning_rate": 1.625806451612903e-05, | |
| "loss": 0.0006, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 2.0872727272727274, | |
| "grad_norm": 0.9055846929550171, | |
| "learning_rate": 1.6193548387096776e-05, | |
| "loss": 0.0045, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 2.090909090909091, | |
| "grad_norm": 0.06875227391719818, | |
| "learning_rate": 1.6129032258064517e-05, | |
| "loss": 0.0011, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.0945454545454547, | |
| "grad_norm": 0.08661270886659622, | |
| "learning_rate": 1.6064516129032258e-05, | |
| "loss": 0.0018, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 2.098181818181818, | |
| "grad_norm": 0.3580359220504761, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 0.0208, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 2.101818181818182, | |
| "grad_norm": 3.5332114696502686, | |
| "learning_rate": 1.5935483870967744e-05, | |
| "loss": 0.0246, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 2.1054545454545455, | |
| "grad_norm": 0.23546698689460754, | |
| "learning_rate": 1.5870967741935485e-05, | |
| "loss": 0.0058, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 2.109090909090909, | |
| "grad_norm": 0.0922674611210823, | |
| "learning_rate": 1.5806451612903226e-05, | |
| "loss": 0.001, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 2.112727272727273, | |
| "grad_norm": 0.03179372474551201, | |
| "learning_rate": 1.5741935483870968e-05, | |
| "loss": 0.0009, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 2.1163636363636362, | |
| "grad_norm": 0.08128567039966583, | |
| "learning_rate": 1.567741935483871e-05, | |
| "loss": 0.0021, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "grad_norm": 0.03141499683260918, | |
| "learning_rate": 1.5612903225806454e-05, | |
| "loss": 0.0009, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 2.1236363636363635, | |
| "grad_norm": 0.056340087205171585, | |
| "learning_rate": 1.5548387096774195e-05, | |
| "loss": 0.0012, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 2.1272727272727274, | |
| "grad_norm": 0.011782072484493256, | |
| "learning_rate": 1.5483870967741936e-05, | |
| "loss": 0.0005, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 2.130909090909091, | |
| "grad_norm": 0.017823919653892517, | |
| "learning_rate": 1.541935483870968e-05, | |
| "loss": 0.0007, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 2.1345454545454547, | |
| "grad_norm": 0.19493459165096283, | |
| "learning_rate": 1.535483870967742e-05, | |
| "loss": 0.0015, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 2.138181818181818, | |
| "grad_norm": 3.2469050884246826, | |
| "learning_rate": 1.529032258064516e-05, | |
| "loss": 0.0172, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 2.1418181818181816, | |
| "grad_norm": 0.05841919407248497, | |
| "learning_rate": 1.5225806451612903e-05, | |
| "loss": 0.0008, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 2.1454545454545455, | |
| "grad_norm": 1.5849275588989258, | |
| "learning_rate": 1.5161290322580646e-05, | |
| "loss": 0.0064, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 2.149090909090909, | |
| "grad_norm": 0.0897936001420021, | |
| "learning_rate": 1.5096774193548389e-05, | |
| "loss": 0.0012, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 2.152727272727273, | |
| "grad_norm": 0.030406808480620384, | |
| "learning_rate": 1.503225806451613e-05, | |
| "loss": 0.0008, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 2.1563636363636363, | |
| "grad_norm": 0.42377015948295593, | |
| "learning_rate": 1.4967741935483873e-05, | |
| "loss": 0.0014, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "grad_norm": 0.037775181233882904, | |
| "learning_rate": 1.4903225806451612e-05, | |
| "loss": 0.0009, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 2.1636363636363636, | |
| "grad_norm": 0.01757960021495819, | |
| "learning_rate": 1.4838709677419355e-05, | |
| "loss": 0.0007, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 2.1672727272727275, | |
| "grad_norm": 0.13892705738544464, | |
| "learning_rate": 1.4774193548387096e-05, | |
| "loss": 0.0013, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 2.170909090909091, | |
| "grad_norm": 0.24957135319709778, | |
| "learning_rate": 1.470967741935484e-05, | |
| "loss": 0.0021, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 2.174545454545455, | |
| "grad_norm": 0.042269039899110794, | |
| "learning_rate": 1.4645161290322582e-05, | |
| "loss": 0.0007, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 2.178181818181818, | |
| "grad_norm": 0.027613814920186996, | |
| "learning_rate": 1.4580645161290324e-05, | |
| "loss": 0.0006, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 2.1818181818181817, | |
| "grad_norm": 0.09081171452999115, | |
| "learning_rate": 1.4516129032258066e-05, | |
| "loss": 0.0007, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.1854545454545455, | |
| "grad_norm": 0.014807288534939289, | |
| "learning_rate": 1.4451612903225808e-05, | |
| "loss": 0.0006, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 2.189090909090909, | |
| "grad_norm": 0.016062721610069275, | |
| "learning_rate": 1.4387096774193547e-05, | |
| "loss": 0.0006, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 2.192727272727273, | |
| "grad_norm": 0.011297466233372688, | |
| "learning_rate": 1.432258064516129e-05, | |
| "loss": 0.0005, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 2.1963636363636363, | |
| "grad_norm": 0.018607553094625473, | |
| "learning_rate": 1.4258064516129033e-05, | |
| "loss": 0.0006, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "grad_norm": 1.5661729574203491, | |
| "learning_rate": 1.4193548387096774e-05, | |
| "loss": 0.0571, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 2.2036363636363636, | |
| "grad_norm": 0.018998922780156136, | |
| "learning_rate": 1.4129032258064517e-05, | |
| "loss": 0.0007, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 2.207272727272727, | |
| "grad_norm": 0.00895577110350132, | |
| "learning_rate": 1.406451612903226e-05, | |
| "loss": 0.0004, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 2.210909090909091, | |
| "grad_norm": 0.023930538445711136, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 0.0006, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 2.2145454545454544, | |
| "grad_norm": 0.31831488013267517, | |
| "learning_rate": 1.3935483870967741e-05, | |
| "loss": 0.002, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 2.2181818181818183, | |
| "grad_norm": 0.46921107172966003, | |
| "learning_rate": 1.3870967741935484e-05, | |
| "loss": 0.0026, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 2.2218181818181817, | |
| "grad_norm": 1.8380354642868042, | |
| "learning_rate": 1.3806451612903227e-05, | |
| "loss": 0.0341, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 2.2254545454545456, | |
| "grad_norm": 0.016989752650260925, | |
| "learning_rate": 1.3741935483870968e-05, | |
| "loss": 0.0006, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 2.229090909090909, | |
| "grad_norm": 0.04719064012169838, | |
| "learning_rate": 1.3677419354838711e-05, | |
| "loss": 0.0009, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 2.232727272727273, | |
| "grad_norm": 0.45522865653038025, | |
| "learning_rate": 1.3612903225806452e-05, | |
| "loss": 0.004, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 2.2363636363636363, | |
| "grad_norm": 4.254357814788818, | |
| "learning_rate": 1.3548387096774195e-05, | |
| "loss": 0.0321, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "grad_norm": 0.42286837100982666, | |
| "learning_rate": 1.3483870967741935e-05, | |
| "loss": 0.0112, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 2.2436363636363637, | |
| "grad_norm": 0.011312934570014477, | |
| "learning_rate": 1.3419354838709678e-05, | |
| "loss": 0.0005, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 2.247272727272727, | |
| "grad_norm": 0.056383103132247925, | |
| "learning_rate": 1.3354838709677419e-05, | |
| "loss": 0.0007, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 2.250909090909091, | |
| "grad_norm": 0.008479413576424122, | |
| "learning_rate": 1.3290322580645162e-05, | |
| "loss": 0.0004, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 2.2545454545454544, | |
| "grad_norm": 0.2737773358821869, | |
| "learning_rate": 1.3225806451612905e-05, | |
| "loss": 0.017, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.2581818181818183, | |
| "grad_norm": 0.021578149870038033, | |
| "learning_rate": 1.3161290322580646e-05, | |
| "loss": 0.0005, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 2.2618181818181817, | |
| "grad_norm": 0.017386844381690025, | |
| "learning_rate": 1.3096774193548389e-05, | |
| "loss": 0.0005, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 2.2654545454545456, | |
| "grad_norm": 0.010488376021385193, | |
| "learning_rate": 1.3032258064516129e-05, | |
| "loss": 0.0005, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 2.269090909090909, | |
| "grad_norm": 0.2101997584104538, | |
| "learning_rate": 1.2967741935483872e-05, | |
| "loss": 0.0067, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 2.2727272727272725, | |
| "grad_norm": 0.01739770732820034, | |
| "learning_rate": 1.2903225806451613e-05, | |
| "loss": 0.0006, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.2763636363636364, | |
| "grad_norm": 4.312204837799072, | |
| "learning_rate": 1.2838709677419356e-05, | |
| "loss": 0.0568, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 2.2800000000000002, | |
| "grad_norm": 0.03970419242978096, | |
| "learning_rate": 1.2774193548387099e-05, | |
| "loss": 0.0008, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 2.2836363636363637, | |
| "grad_norm": 0.06886734068393707, | |
| "learning_rate": 1.270967741935484e-05, | |
| "loss": 0.001, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 2.287272727272727, | |
| "grad_norm": 0.020005859434604645, | |
| "learning_rate": 1.2645161290322583e-05, | |
| "loss": 0.0006, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 2.290909090909091, | |
| "grad_norm": 0.042734235525131226, | |
| "learning_rate": 1.2580645161290322e-05, | |
| "loss": 0.0006, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 2.2945454545454544, | |
| "grad_norm": 0.1663779318332672, | |
| "learning_rate": 1.2516129032258064e-05, | |
| "loss": 0.0013, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 2.2981818181818183, | |
| "grad_norm": 0.8976339101791382, | |
| "learning_rate": 1.2451612903225807e-05, | |
| "loss": 0.0025, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 2.3018181818181818, | |
| "grad_norm": 0.013320104219019413, | |
| "learning_rate": 1.238709677419355e-05, | |
| "loss": 0.0005, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 2.3054545454545456, | |
| "grad_norm": 0.20691031217575073, | |
| "learning_rate": 1.232258064516129e-05, | |
| "loss": 0.001, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 2.309090909090909, | |
| "grad_norm": 0.05960312858223915, | |
| "learning_rate": 1.2258064516129032e-05, | |
| "loss": 0.0008, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 2.3127272727272725, | |
| "grad_norm": 3.190467596054077, | |
| "learning_rate": 1.2193548387096775e-05, | |
| "loss": 0.0132, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 2.3163636363636364, | |
| "grad_norm": 0.9548889994621277, | |
| "learning_rate": 1.2129032258064516e-05, | |
| "loss": 0.0046, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "grad_norm": 0.037591926753520966, | |
| "learning_rate": 1.2064516129032259e-05, | |
| "loss": 0.0006, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 2.3236363636363637, | |
| "grad_norm": 0.025826606899499893, | |
| "learning_rate": 1.2e-05, | |
| "loss": 0.0006, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 2.327272727272727, | |
| "grad_norm": 0.13162964582443237, | |
| "learning_rate": 1.1935483870967743e-05, | |
| "loss": 0.0013, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 2.330909090909091, | |
| "grad_norm": 0.009305083192884922, | |
| "learning_rate": 1.1870967741935484e-05, | |
| "loss": 0.0004, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 2.3345454545454545, | |
| "grad_norm": 0.00890254881232977, | |
| "learning_rate": 1.1806451612903226e-05, | |
| "loss": 0.0004, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 2.3381818181818184, | |
| "grad_norm": 0.00882215891033411, | |
| "learning_rate": 1.1741935483870969e-05, | |
| "loss": 0.0004, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 2.341818181818182, | |
| "grad_norm": 0.11077472567558289, | |
| "learning_rate": 1.167741935483871e-05, | |
| "loss": 0.0008, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 2.3454545454545457, | |
| "grad_norm": 0.016244694590568542, | |
| "learning_rate": 1.1612903225806453e-05, | |
| "loss": 0.0005, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 2.349090909090909, | |
| "grad_norm": 0.038961056619882584, | |
| "learning_rate": 1.1548387096774194e-05, | |
| "loss": 0.0007, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 2.3527272727272726, | |
| "grad_norm": 0.008070679381489754, | |
| "learning_rate": 1.1483870967741935e-05, | |
| "loss": 0.0004, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 2.3563636363636364, | |
| "grad_norm": 0.03419802337884903, | |
| "learning_rate": 1.1419354838709678e-05, | |
| "loss": 0.0007, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "grad_norm": 0.1610228568315506, | |
| "learning_rate": 1.1354838709677421e-05, | |
| "loss": 0.0018, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 2.3636363636363638, | |
| "grad_norm": 0.21933913230895996, | |
| "learning_rate": 1.129032258064516e-05, | |
| "loss": 0.0012, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.367272727272727, | |
| "grad_norm": 0.024096714332699776, | |
| "learning_rate": 1.1225806451612904e-05, | |
| "loss": 0.0005, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 2.370909090909091, | |
| "grad_norm": 0.013976830057799816, | |
| "learning_rate": 1.1161290322580647e-05, | |
| "loss": 0.0005, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 2.3745454545454545, | |
| "grad_norm": 0.009037399664521217, | |
| "learning_rate": 1.1096774193548388e-05, | |
| "loss": 0.0004, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 2.378181818181818, | |
| "grad_norm": 4.028433799743652, | |
| "learning_rate": 1.1032258064516129e-05, | |
| "loss": 0.0074, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 2.381818181818182, | |
| "grad_norm": 0.9609191417694092, | |
| "learning_rate": 1.0967741935483872e-05, | |
| "loss": 0.0054, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 2.3854545454545453, | |
| "grad_norm": 1.4599242210388184, | |
| "learning_rate": 1.0903225806451613e-05, | |
| "loss": 0.0105, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 2.389090909090909, | |
| "grad_norm": 1.0803523063659668, | |
| "learning_rate": 1.0838709677419355e-05, | |
| "loss": 0.0037, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 2.3927272727272726, | |
| "grad_norm": 0.006454968359321356, | |
| "learning_rate": 1.0774193548387097e-05, | |
| "loss": 0.0003, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 2.3963636363636365, | |
| "grad_norm": 0.043990444391965866, | |
| "learning_rate": 1.070967741935484e-05, | |
| "loss": 0.0008, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "grad_norm": 0.04728386178612709, | |
| "learning_rate": 1.064516129032258e-05, | |
| "loss": 0.0006, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.403636363636364, | |
| "grad_norm": 0.012434919364750385, | |
| "learning_rate": 1.0580645161290323e-05, | |
| "loss": 0.0005, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 2.4072727272727272, | |
| "grad_norm": 0.010447041131556034, | |
| "learning_rate": 1.0516129032258066e-05, | |
| "loss": 0.0003, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 2.410909090909091, | |
| "grad_norm": 0.007828759960830212, | |
| "learning_rate": 1.0451612903225807e-05, | |
| "loss": 0.0004, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 2.4145454545454546, | |
| "grad_norm": 0.00887393206357956, | |
| "learning_rate": 1.0387096774193548e-05, | |
| "loss": 0.0004, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 2.418181818181818, | |
| "grad_norm": 0.8037237524986267, | |
| "learning_rate": 1.0322580645161291e-05, | |
| "loss": 0.0983, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 2.421818181818182, | |
| "grad_norm": 0.014007828198373318, | |
| "learning_rate": 1.0258064516129032e-05, | |
| "loss": 0.0004, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 2.4254545454545453, | |
| "grad_norm": 0.010563932359218597, | |
| "learning_rate": 1.0193548387096774e-05, | |
| "loss": 0.0004, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 2.429090909090909, | |
| "grad_norm": 0.009581638500094414, | |
| "learning_rate": 1.0129032258064517e-05, | |
| "loss": 0.0004, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 2.4327272727272726, | |
| "grad_norm": 0.014039217494428158, | |
| "learning_rate": 1.0064516129032258e-05, | |
| "loss": 0.0004, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 2.4363636363636365, | |
| "grad_norm": 0.035384513437747955, | |
| "learning_rate": 1e-05, | |
| "loss": 0.0005, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "grad_norm": 0.011463082395493984, | |
| "learning_rate": 9.935483870967742e-06, | |
| "loss": 0.0004, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 2.443636363636364, | |
| "grad_norm": 0.008181700482964516, | |
| "learning_rate": 9.870967741935485e-06, | |
| "loss": 0.0004, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 2.4472727272727273, | |
| "grad_norm": 0.7373052835464478, | |
| "learning_rate": 9.806451612903226e-06, | |
| "loss": 0.006, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 2.450909090909091, | |
| "grad_norm": 0.018753718584775925, | |
| "learning_rate": 9.741935483870967e-06, | |
| "loss": 0.0004, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 2.4545454545454546, | |
| "grad_norm": 0.009037282317876816, | |
| "learning_rate": 9.67741935483871e-06, | |
| "loss": 0.0004, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.458181818181818, | |
| "grad_norm": 0.016004854813218117, | |
| "learning_rate": 9.612903225806452e-06, | |
| "loss": 0.0005, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 2.461818181818182, | |
| "grad_norm": 0.037706032395362854, | |
| "learning_rate": 9.548387096774195e-06, | |
| "loss": 0.0006, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 2.4654545454545453, | |
| "grad_norm": 0.05756361410021782, | |
| "learning_rate": 9.483870967741936e-06, | |
| "loss": 0.0008, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 2.4690909090909092, | |
| "grad_norm": 0.0136951869353652, | |
| "learning_rate": 9.419354838709677e-06, | |
| "loss": 0.0005, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 2.4727272727272727, | |
| "grad_norm": 0.011940555647015572, | |
| "learning_rate": 9.35483870967742e-06, | |
| "loss": 0.0006, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 2.4763636363636365, | |
| "grad_norm": 0.25020283460617065, | |
| "learning_rate": 9.290322580645163e-06, | |
| "loss": 0.0088, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "grad_norm": 0.008918453007936478, | |
| "learning_rate": 9.225806451612904e-06, | |
| "loss": 0.0004, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 2.4836363636363634, | |
| "grad_norm": 0.0643925741314888, | |
| "learning_rate": 9.161290322580645e-06, | |
| "loss": 0.0008, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 2.4872727272727273, | |
| "grad_norm": 0.08033094555139542, | |
| "learning_rate": 9.096774193548388e-06, | |
| "loss": 0.0007, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 2.4909090909090907, | |
| "grad_norm": 0.009339767508208752, | |
| "learning_rate": 9.03225806451613e-06, | |
| "loss": 0.0004, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 2.4945454545454546, | |
| "grad_norm": 0.012149178422987461, | |
| "learning_rate": 8.967741935483871e-06, | |
| "loss": 0.0005, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 2.498181818181818, | |
| "grad_norm": 0.028004921972751617, | |
| "learning_rate": 8.903225806451614e-06, | |
| "loss": 0.0004, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 2.501818181818182, | |
| "grad_norm": 0.013502350077033043, | |
| "learning_rate": 8.838709677419357e-06, | |
| "loss": 0.0004, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 2.5054545454545454, | |
| "grad_norm": 0.008973742835223675, | |
| "learning_rate": 8.774193548387096e-06, | |
| "loss": 0.0004, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 2.509090909090909, | |
| "grad_norm": 0.017967596650123596, | |
| "learning_rate": 8.70967741935484e-06, | |
| "loss": 0.0005, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.5127272727272727, | |
| "grad_norm": 0.009786793030798435, | |
| "learning_rate": 8.645161290322582e-06, | |
| "loss": 0.0004, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 2.5163636363636366, | |
| "grad_norm": 0.03004172444343567, | |
| "learning_rate": 8.580645161290322e-06, | |
| "loss": 0.0007, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "grad_norm": 0.014010576531291008, | |
| "learning_rate": 8.516129032258065e-06, | |
| "loss": 0.0005, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 2.5236363636363635, | |
| "grad_norm": 0.10186956822872162, | |
| "learning_rate": 8.451612903225808e-06, | |
| "loss": 0.0007, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 2.5272727272727273, | |
| "grad_norm": 0.007655604742467403, | |
| "learning_rate": 8.387096774193549e-06, | |
| "loss": 0.0003, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 2.5309090909090908, | |
| "grad_norm": 0.026972953230142593, | |
| "learning_rate": 8.32258064516129e-06, | |
| "loss": 0.0006, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 2.5345454545454547, | |
| "grad_norm": 0.00847043376415968, | |
| "learning_rate": 8.258064516129033e-06, | |
| "loss": 0.0004, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 2.538181818181818, | |
| "grad_norm": 3.0761332511901855, | |
| "learning_rate": 8.193548387096774e-06, | |
| "loss": 0.045, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 2.541818181818182, | |
| "grad_norm": 0.010722169652581215, | |
| "learning_rate": 8.129032258064515e-06, | |
| "loss": 0.0004, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 2.5454545454545454, | |
| "grad_norm": 0.024674193933606148, | |
| "learning_rate": 8.064516129032258e-06, | |
| "loss": 0.0004, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.549090909090909, | |
| "grad_norm": 0.010132328607141972, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.0004, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 2.5527272727272727, | |
| "grad_norm": 0.007771102711558342, | |
| "learning_rate": 7.935483870967743e-06, | |
| "loss": 0.0003, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 2.5563636363636366, | |
| "grad_norm": 0.009345349855720997, | |
| "learning_rate": 7.870967741935484e-06, | |
| "loss": 0.0004, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 0.061426129192113876, | |
| "learning_rate": 7.806451612903227e-06, | |
| "loss": 0.0012, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 2.5636363636363635, | |
| "grad_norm": 0.006356612779200077, | |
| "learning_rate": 7.741935483870968e-06, | |
| "loss": 0.0003, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 2.5672727272727274, | |
| "grad_norm": 0.5145014524459839, | |
| "learning_rate": 7.67741935483871e-06, | |
| "loss": 0.0159, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 2.570909090909091, | |
| "grad_norm": 0.01627645082771778, | |
| "learning_rate": 7.612903225806451e-06, | |
| "loss": 0.0005, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 2.5745454545454547, | |
| "grad_norm": 0.009603966027498245, | |
| "learning_rate": 7.548387096774194e-06, | |
| "loss": 0.0005, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 2.578181818181818, | |
| "grad_norm": 0.008407434448599815, | |
| "learning_rate": 7.483870967741936e-06, | |
| "loss": 0.0004, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 2.581818181818182, | |
| "grad_norm": 0.007048910949379206, | |
| "learning_rate": 7.419354838709678e-06, | |
| "loss": 0.0003, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.5854545454545454, | |
| "grad_norm": 0.007168797310441732, | |
| "learning_rate": 7.35483870967742e-06, | |
| "loss": 0.0003, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 2.589090909090909, | |
| "grad_norm": 0.017820533365011215, | |
| "learning_rate": 7.290322580645162e-06, | |
| "loss": 0.0006, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 2.5927272727272728, | |
| "grad_norm": 1.8030993938446045, | |
| "learning_rate": 7.225806451612904e-06, | |
| "loss": 0.1001, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 2.596363636363636, | |
| "grad_norm": 0.006081653293222189, | |
| "learning_rate": 7.161290322580645e-06, | |
| "loss": 0.0003, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "grad_norm": 0.013014406897127628, | |
| "learning_rate": 7.096774193548387e-06, | |
| "loss": 0.0005, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 2.6036363636363635, | |
| "grad_norm": 0.5529889464378357, | |
| "learning_rate": 7.03225806451613e-06, | |
| "loss": 0.0032, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 2.6072727272727274, | |
| "grad_norm": 0.010706817731261253, | |
| "learning_rate": 6.9677419354838705e-06, | |
| "loss": 0.0004, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 2.610909090909091, | |
| "grad_norm": 0.009283789433538914, | |
| "learning_rate": 6.9032258064516135e-06, | |
| "loss": 0.0004, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 2.6145454545454543, | |
| "grad_norm": 0.006831625942140818, | |
| "learning_rate": 6.8387096774193555e-06, | |
| "loss": 0.0003, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 2.618181818181818, | |
| "grad_norm": 0.010544957593083382, | |
| "learning_rate": 6.774193548387098e-06, | |
| "loss": 0.0004, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.621818181818182, | |
| "grad_norm": 0.573939323425293, | |
| "learning_rate": 6.709677419354839e-06, | |
| "loss": 0.1136, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 2.6254545454545455, | |
| "grad_norm": 0.007392039522528648, | |
| "learning_rate": 6.645161290322581e-06, | |
| "loss": 0.0003, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 2.629090909090909, | |
| "grad_norm": 0.03405938670039177, | |
| "learning_rate": 6.580645161290323e-06, | |
| "loss": 0.0007, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 2.632727272727273, | |
| "grad_norm": 0.018447572365403175, | |
| "learning_rate": 6.516129032258064e-06, | |
| "loss": 0.0005, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 2.6363636363636362, | |
| "grad_norm": 0.00638270378112793, | |
| "learning_rate": 6.451612903225806e-06, | |
| "loss": 0.0003, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "grad_norm": 0.017991170287132263, | |
| "learning_rate": 6.387096774193549e-06, | |
| "loss": 0.0004, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 2.6436363636363636, | |
| "grad_norm": 0.00989875290542841, | |
| "learning_rate": 6.322580645161291e-06, | |
| "loss": 0.0004, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 2.6472727272727274, | |
| "grad_norm": 3.8099822998046875, | |
| "learning_rate": 6.258064516129032e-06, | |
| "loss": 0.0512, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 2.650909090909091, | |
| "grad_norm": 0.012733093462884426, | |
| "learning_rate": 6.193548387096775e-06, | |
| "loss": 0.0005, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 2.6545454545454543, | |
| "grad_norm": 0.04748505726456642, | |
| "learning_rate": 6.129032258064516e-06, | |
| "loss": 0.0007, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.658181818181818, | |
| "grad_norm": 4.789937973022461, | |
| "learning_rate": 6.064516129032258e-06, | |
| "loss": 0.048, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 2.661818181818182, | |
| "grad_norm": 0.03021158277988434, | |
| "learning_rate": 6e-06, | |
| "loss": 0.0006, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 2.6654545454545455, | |
| "grad_norm": 0.017927074804902077, | |
| "learning_rate": 5.935483870967742e-06, | |
| "loss": 0.0005, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 2.669090909090909, | |
| "grad_norm": 0.01334038283675909, | |
| "learning_rate": 5.870967741935484e-06, | |
| "loss": 0.0005, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 2.672727272727273, | |
| "grad_norm": 5.3513970375061035, | |
| "learning_rate": 5.806451612903226e-06, | |
| "loss": 0.0182, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 2.6763636363636363, | |
| "grad_norm": 0.012456363067030907, | |
| "learning_rate": 5.741935483870968e-06, | |
| "loss": 0.0004, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "grad_norm": 0.024449974298477173, | |
| "learning_rate": 5.677419354838711e-06, | |
| "loss": 0.0006, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 2.6836363636363636, | |
| "grad_norm": 0.015060571022331715, | |
| "learning_rate": 5.612903225806452e-06, | |
| "loss": 0.0005, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 2.6872727272727275, | |
| "grad_norm": 4.723052024841309, | |
| "learning_rate": 5.548387096774194e-06, | |
| "loss": 0.0212, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 2.690909090909091, | |
| "grad_norm": 0.020275188609957695, | |
| "learning_rate": 5.483870967741936e-06, | |
| "loss": 0.0005, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 2.6945454545454544, | |
| "grad_norm": 0.00891982950270176, | |
| "learning_rate": 5.419354838709677e-06, | |
| "loss": 0.0004, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 2.6981818181818182, | |
| "grad_norm": 0.006951956544071436, | |
| "learning_rate": 5.35483870967742e-06, | |
| "loss": 0.0003, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 2.7018181818181817, | |
| "grad_norm": 1.1897835731506348, | |
| "learning_rate": 5.2903225806451614e-06, | |
| "loss": 0.1006, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 2.7054545454545456, | |
| "grad_norm": 0.0193193256855011, | |
| "learning_rate": 5.2258064516129035e-06, | |
| "loss": 0.0005, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 2.709090909090909, | |
| "grad_norm": 0.006243540905416012, | |
| "learning_rate": 5.161290322580646e-06, | |
| "loss": 0.0003, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 2.712727272727273, | |
| "grad_norm": 0.011104391887784004, | |
| "learning_rate": 5.096774193548387e-06, | |
| "loss": 0.0005, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 2.7163636363636363, | |
| "grad_norm": 0.015710929408669472, | |
| "learning_rate": 5.032258064516129e-06, | |
| "loss": 0.0005, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 2.7199999999999998, | |
| "grad_norm": 0.008195963688194752, | |
| "learning_rate": 4.967741935483871e-06, | |
| "loss": 0.0004, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 2.7236363636363636, | |
| "grad_norm": 0.5103004574775696, | |
| "learning_rate": 4.903225806451613e-06, | |
| "loss": 0.0028, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 2.7272727272727275, | |
| "grad_norm": 0.009401354938745499, | |
| "learning_rate": 4.838709677419355e-06, | |
| "loss": 0.0004, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.730909090909091, | |
| "grad_norm": 0.010534017346799374, | |
| "learning_rate": 4.774193548387097e-06, | |
| "loss": 0.0004, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 2.7345454545454544, | |
| "grad_norm": 0.02028539776802063, | |
| "learning_rate": 4.7096774193548385e-06, | |
| "loss": 0.0005, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 2.7381818181818183, | |
| "grad_norm": 0.10612978786230087, | |
| "learning_rate": 4.6451612903225815e-06, | |
| "loss": 0.0013, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 2.7418181818181817, | |
| "grad_norm": 0.008999668061733246, | |
| "learning_rate": 4.580645161290323e-06, | |
| "loss": 0.0004, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 2.7454545454545456, | |
| "grad_norm": 0.020546872168779373, | |
| "learning_rate": 4.516129032258065e-06, | |
| "loss": 0.0006, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 2.749090909090909, | |
| "grad_norm": 0.04722006618976593, | |
| "learning_rate": 4.451612903225807e-06, | |
| "loss": 0.0006, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 2.752727272727273, | |
| "grad_norm": 0.6383763551712036, | |
| "learning_rate": 4.387096774193548e-06, | |
| "loss": 0.0017, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 2.7563636363636363, | |
| "grad_norm": 0.013335658237338066, | |
| "learning_rate": 4.322580645161291e-06, | |
| "loss": 0.0005, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "grad_norm": 0.01042084489017725, | |
| "learning_rate": 4.258064516129032e-06, | |
| "loss": 0.0004, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 2.7636363636363637, | |
| "grad_norm": 0.0967690572142601, | |
| "learning_rate": 4.193548387096774e-06, | |
| "loss": 0.0011, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.767272727272727, | |
| "grad_norm": 0.02381141297519207, | |
| "learning_rate": 4.1290322580645165e-06, | |
| "loss": 0.0007, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 2.770909090909091, | |
| "grad_norm": 0.02535305730998516, | |
| "learning_rate": 4.064516129032258e-06, | |
| "loss": 0.0006, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 2.7745454545454544, | |
| "grad_norm": 0.0302995927631855, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.0007, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 2.7781818181818183, | |
| "grad_norm": 0.01775578036904335, | |
| "learning_rate": 3.935483870967742e-06, | |
| "loss": 0.0005, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 2.7818181818181817, | |
| "grad_norm": 0.011038933880627155, | |
| "learning_rate": 3.870967741935484e-06, | |
| "loss": 0.0004, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 2.785454545454545, | |
| "grad_norm": 0.010119972750544548, | |
| "learning_rate": 3.8064516129032257e-06, | |
| "loss": 0.0004, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 2.789090909090909, | |
| "grad_norm": 0.015269882045686245, | |
| "learning_rate": 3.741935483870968e-06, | |
| "loss": 0.0005, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 2.792727272727273, | |
| "grad_norm": 0.048511989414691925, | |
| "learning_rate": 3.67741935483871e-06, | |
| "loss": 0.0009, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 2.7963636363636364, | |
| "grad_norm": 0.08957032114267349, | |
| "learning_rate": 3.612903225806452e-06, | |
| "loss": 0.0009, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "grad_norm": 0.012502241879701614, | |
| "learning_rate": 3.5483870967741936e-06, | |
| "loss": 0.0005, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 2.8036363636363637, | |
| "grad_norm": 0.017656538635492325, | |
| "learning_rate": 3.4838709677419353e-06, | |
| "loss": 0.0004, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 2.807272727272727, | |
| "grad_norm": 0.03845641762018204, | |
| "learning_rate": 3.4193548387096778e-06, | |
| "loss": 0.001, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 2.810909090909091, | |
| "grad_norm": 0.6619153618812561, | |
| "learning_rate": 3.3548387096774194e-06, | |
| "loss": 0.0026, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 2.8145454545454545, | |
| "grad_norm": 0.012663335539400578, | |
| "learning_rate": 3.2903225806451615e-06, | |
| "loss": 0.0005, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 2.8181818181818183, | |
| "grad_norm": 0.2195906788110733, | |
| "learning_rate": 3.225806451612903e-06, | |
| "loss": 0.0085, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 2.821818181818182, | |
| "grad_norm": 0.02970001846551895, | |
| "learning_rate": 3.1612903225806457e-06, | |
| "loss": 0.0009, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 2.825454545454545, | |
| "grad_norm": 0.025815211236476898, | |
| "learning_rate": 3.0967741935483874e-06, | |
| "loss": 0.0007, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 2.829090909090909, | |
| "grad_norm": 0.026814987882971764, | |
| "learning_rate": 3.032258064516129e-06, | |
| "loss": 0.0006, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 2.832727272727273, | |
| "grad_norm": 0.7879754900932312, | |
| "learning_rate": 2.967741935483871e-06, | |
| "loss": 0.1117, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 2.8363636363636364, | |
| "grad_norm": 0.0365438349545002, | |
| "learning_rate": 2.903225806451613e-06, | |
| "loss": 0.0009, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "grad_norm": 0.1561412513256073, | |
| "learning_rate": 2.8387096774193553e-06, | |
| "loss": 0.0012, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 2.8436363636363637, | |
| "grad_norm": 0.010296767577528954, | |
| "learning_rate": 2.774193548387097e-06, | |
| "loss": 0.0004, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 2.847272727272727, | |
| "grad_norm": 0.16209469735622406, | |
| "learning_rate": 2.7096774193548386e-06, | |
| "loss": 0.0013, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 2.850909090909091, | |
| "grad_norm": 0.014227217994630337, | |
| "learning_rate": 2.6451612903225807e-06, | |
| "loss": 0.0004, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 2.8545454545454545, | |
| "grad_norm": 0.06868361681699753, | |
| "learning_rate": 2.580645161290323e-06, | |
| "loss": 0.0006, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 2.8581818181818184, | |
| "grad_norm": 0.02630774676799774, | |
| "learning_rate": 2.5161290322580645e-06, | |
| "loss": 0.0006, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 2.861818181818182, | |
| "grad_norm": 0.007779018487781286, | |
| "learning_rate": 2.4516129032258066e-06, | |
| "loss": 0.0004, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 2.8654545454545453, | |
| "grad_norm": 0.03792530298233032, | |
| "learning_rate": 2.3870967741935486e-06, | |
| "loss": 0.0006, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 2.869090909090909, | |
| "grad_norm": 0.009642829187214375, | |
| "learning_rate": 2.3225806451612907e-06, | |
| "loss": 0.0004, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 2.8727272727272726, | |
| "grad_norm": 0.014530934393405914, | |
| "learning_rate": 2.2580645161290324e-06, | |
| "loss": 0.0005, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 2.8763636363636365, | |
| "grad_norm": 0.09846967458724976, | |
| "learning_rate": 2.193548387096774e-06, | |
| "loss": 0.001, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "grad_norm": 0.02130993641912937, | |
| "learning_rate": 2.129032258064516e-06, | |
| "loss": 0.0006, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 2.8836363636363638, | |
| "grad_norm": 0.6667435765266418, | |
| "learning_rate": 2.0645161290322582e-06, | |
| "loss": 0.0023, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 2.887272727272727, | |
| "grad_norm": 0.10888272523880005, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.0009, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 2.8909090909090907, | |
| "grad_norm": 0.048606500029563904, | |
| "learning_rate": 1.935483870967742e-06, | |
| "loss": 0.0005, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 2.8945454545454545, | |
| "grad_norm": 0.23399078845977783, | |
| "learning_rate": 1.870967741935484e-06, | |
| "loss": 0.0027, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 2.8981818181818184, | |
| "grad_norm": 0.01046321727335453, | |
| "learning_rate": 1.806451612903226e-06, | |
| "loss": 0.0004, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 2.901818181818182, | |
| "grad_norm": 0.015533102676272392, | |
| "learning_rate": 1.7419354838709676e-06, | |
| "loss": 0.0006, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 2.9054545454545453, | |
| "grad_norm": 0.016338596120476723, | |
| "learning_rate": 1.6774193548387097e-06, | |
| "loss": 0.0006, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 2.909090909090909, | |
| "grad_norm": 0.00850651878863573, | |
| "learning_rate": 1.6129032258064516e-06, | |
| "loss": 0.0004, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.9127272727272726, | |
| "grad_norm": 0.7118433713912964, | |
| "learning_rate": 1.5483870967741937e-06, | |
| "loss": 0.0027, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 2.9163636363636365, | |
| "grad_norm": 1.7679587602615356, | |
| "learning_rate": 1.4838709677419356e-06, | |
| "loss": 0.0496, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "grad_norm": 0.010516179725527763, | |
| "learning_rate": 1.4193548387096776e-06, | |
| "loss": 0.0004, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 2.923636363636364, | |
| "grad_norm": 0.0124241653829813, | |
| "learning_rate": 1.3548387096774193e-06, | |
| "loss": 0.0004, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 2.9272727272727272, | |
| "grad_norm": 0.00775744765996933, | |
| "learning_rate": 1.2903225806451614e-06, | |
| "loss": 0.0003, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 2.9309090909090907, | |
| "grad_norm": 1.7472180128097534, | |
| "learning_rate": 1.2258064516129033e-06, | |
| "loss": 0.0587, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 2.9345454545454546, | |
| "grad_norm": 0.014171603135764599, | |
| "learning_rate": 1.1612903225806454e-06, | |
| "loss": 0.0005, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 2.9381818181818184, | |
| "grad_norm": 0.011718512512743473, | |
| "learning_rate": 1.096774193548387e-06, | |
| "loss": 0.0004, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 2.941818181818182, | |
| "grad_norm": 0.011687002144753933, | |
| "learning_rate": 1.0322580645161291e-06, | |
| "loss": 0.0004, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 2.9454545454545453, | |
| "grad_norm": 0.010272631421685219, | |
| "learning_rate": 9.67741935483871e-07, | |
| "loss": 0.0004, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.949090909090909, | |
| "grad_norm": 0.4066472351551056, | |
| "learning_rate": 9.03225806451613e-07, | |
| "loss": 0.0023, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 2.9527272727272726, | |
| "grad_norm": 1.5127947330474854, | |
| "learning_rate": 8.387096774193549e-07, | |
| "loss": 0.0037, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 2.9563636363636365, | |
| "grad_norm": 0.017161400988698006, | |
| "learning_rate": 7.741935483870968e-07, | |
| "loss": 0.0005, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "grad_norm": 0.009147647768259048, | |
| "learning_rate": 7.096774193548388e-07, | |
| "loss": 0.0004, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 2.963636363636364, | |
| "grad_norm": 0.010304290801286697, | |
| "learning_rate": 6.451612903225807e-07, | |
| "loss": 0.0005, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 2.9672727272727273, | |
| "grad_norm": 0.027671974152326584, | |
| "learning_rate": 5.806451612903227e-07, | |
| "loss": 0.0008, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 2.9709090909090907, | |
| "grad_norm": 3.4109323024749756, | |
| "learning_rate": 5.161290322580646e-07, | |
| "loss": 0.0096, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 2.9745454545454546, | |
| "grad_norm": 0.1351071149110794, | |
| "learning_rate": 4.516129032258065e-07, | |
| "loss": 0.0012, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 2.978181818181818, | |
| "grad_norm": 0.01651870645582676, | |
| "learning_rate": 3.870967741935484e-07, | |
| "loss": 0.0005, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 2.981818181818182, | |
| "grad_norm": 0.012661241926252842, | |
| "learning_rate": 3.2258064516129035e-07, | |
| "loss": 0.0004, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.9854545454545454, | |
| "grad_norm": 0.007508403621613979, | |
| "learning_rate": 2.580645161290323e-07, | |
| "loss": 0.0003, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 2.9890909090909092, | |
| "grad_norm": 0.2275754064321518, | |
| "learning_rate": 1.935483870967742e-07, | |
| "loss": 0.001, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 2.9927272727272727, | |
| "grad_norm": 0.012626181356608868, | |
| "learning_rate": 1.2903225806451614e-07, | |
| "loss": 0.0005, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 2.996363636363636, | |
| "grad_norm": 0.016164276748895645, | |
| "learning_rate": 6.451612903225807e-08, | |
| "loss": 0.0006, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 0.007542683742940426, | |
| "learning_rate": 0.0, | |
| "loss": 0.0003, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9938524590163934, | |
| "eval_loss": 0.026551904156804085, | |
| "eval_runtime": 12.7394, | |
| "eval_samples_per_second": 344.756, | |
| "eval_steps_per_second": 5.416, | |
| "step": 825 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 825, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 6981561778765824.0, | |
| "train_batch_size": 64, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |