| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 68.02721088435374, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.3401360544217687, | |
| "grad_norm": 7.277314186096191, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 1.2508, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.6802721088435374, | |
| "grad_norm": 4.874781131744385, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.8343, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.0204081632653061, | |
| "grad_norm": 3.6367509365081787, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.5016, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.3605442176870748, | |
| "grad_norm": 3.666282892227173, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.3941, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.7006802721088436, | |
| "grad_norm": 3.6144936084747314, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.3451, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 2.0408163265306123, | |
| "grad_norm": 2.4187328815460205, | |
| "learning_rate": 3e-06, | |
| "loss": 0.3123, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.380952380952381, | |
| "grad_norm": 3.255674362182617, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.2263, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 2.7210884353741496, | |
| "grad_norm": 3.5080747604370117, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.2111, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 3.061224489795918, | |
| "grad_norm": 2.3079421520233154, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.201, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 3.4013605442176873, | |
| "grad_norm": 2.2684943675994873, | |
| "learning_rate": 5e-06, | |
| "loss": 0.1196, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 3.741496598639456, | |
| "grad_norm": 2.2747371196746826, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.1258, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 4.081632653061225, | |
| "grad_norm": 2.270559787750244, | |
| "learning_rate": 6e-06, | |
| "loss": 0.1133, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 4.421768707482993, | |
| "grad_norm": 2.20074200630188, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.0682, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 4.761904761904762, | |
| "grad_norm": 2.804894208908081, | |
| "learning_rate": 7e-06, | |
| "loss": 0.0731, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 5.1020408163265305, | |
| "grad_norm": 2.6068227291107178, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.0672, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 5.442176870748299, | |
| "grad_norm": 3.2111849784851074, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.0478, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 5.782312925170068, | |
| "grad_norm": 1.5456165075302124, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.0505, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 6.122448979591836, | |
| "grad_norm": 1.4710639715194702, | |
| "learning_rate": 9e-06, | |
| "loss": 0.0473, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 6.462585034013605, | |
| "grad_norm": 2.914450168609619, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.0368, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 6.802721088435375, | |
| "grad_norm": 2.29419207572937, | |
| "learning_rate": 1e-05, | |
| "loss": 0.0441, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 7.142857142857143, | |
| "grad_norm": 0.8864838480949402, | |
| "learning_rate": 9.944444444444445e-06, | |
| "loss": 0.0412, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 7.482993197278912, | |
| "grad_norm": 1.3316959142684937, | |
| "learning_rate": 9.88888888888889e-06, | |
| "loss": 0.0347, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 7.8231292517006805, | |
| "grad_norm": 2.046053647994995, | |
| "learning_rate": 9.833333333333333e-06, | |
| "loss": 0.0299, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 8.16326530612245, | |
| "grad_norm": 1.579432725906372, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 0.0268, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 8.503401360544217, | |
| "grad_norm": 1.7325966358184814, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 0.019, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 8.843537414965986, | |
| "grad_norm": 1.6740517616271973, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 0.0262, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 9.183673469387756, | |
| "grad_norm": 0.65809166431427, | |
| "learning_rate": 9.611111111111112e-06, | |
| "loss": 0.0183, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 9.523809523809524, | |
| "grad_norm": 2.288142442703247, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 0.0177, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 9.863945578231293, | |
| "grad_norm": 1.3612937927246094, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.0179, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 10.204081632653061, | |
| "grad_norm": 1.2634042501449585, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.0142, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 10.54421768707483, | |
| "grad_norm": 0.8320721387863159, | |
| "learning_rate": 9.38888888888889e-06, | |
| "loss": 0.0131, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 10.884353741496598, | |
| "grad_norm": 1.1112641096115112, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 0.0148, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 11.224489795918368, | |
| "grad_norm": 0.4855366349220276, | |
| "learning_rate": 9.277777777777778e-06, | |
| "loss": 0.0124, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 11.564625850340136, | |
| "grad_norm": 1.5221891403198242, | |
| "learning_rate": 9.222222222222224e-06, | |
| "loss": 0.0116, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 11.904761904761905, | |
| "grad_norm": 1.4894431829452515, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.0144, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 12.244897959183673, | |
| "grad_norm": 1.2897030115127563, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 0.0099, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 12.585034013605442, | |
| "grad_norm": 0.5321520566940308, | |
| "learning_rate": 9.055555555555556e-06, | |
| "loss": 0.0068, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 12.92517006802721, | |
| "grad_norm": 1.2740304470062256, | |
| "learning_rate": 9e-06, | |
| "loss": 0.01, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 13.26530612244898, | |
| "grad_norm": 0.6653080582618713, | |
| "learning_rate": 8.944444444444446e-06, | |
| "loss": 0.0072, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 13.60544217687075, | |
| "grad_norm": 1.0347912311553955, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.0091, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 13.60544217687075, | |
| "eval_loss": 0.4027358293533325, | |
| "eval_runtime": 102.3269, | |
| "eval_samples_per_second": 2.541, | |
| "eval_steps_per_second": 0.166, | |
| "eval_wer": 0.21894815367400225, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 13.945578231292517, | |
| "grad_norm": 0.6724150776863098, | |
| "learning_rate": 8.833333333333334e-06, | |
| "loss": 0.0119, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 14.285714285714286, | |
| "grad_norm": 0.44545045495033264, | |
| "learning_rate": 8.777777777777778e-06, | |
| "loss": 0.0069, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 14.625850340136054, | |
| "grad_norm": 1.5290805101394653, | |
| "learning_rate": 8.722222222222224e-06, | |
| "loss": 0.0078, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 14.965986394557824, | |
| "grad_norm": 0.5309610962867737, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 0.0053, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 15.306122448979592, | |
| "grad_norm": 0.1730959564447403, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.0031, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 15.646258503401361, | |
| "grad_norm": 0.06295394897460938, | |
| "learning_rate": 8.555555555555556e-06, | |
| "loss": 0.0044, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 15.986394557823129, | |
| "grad_norm": 0.4541076421737671, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.0036, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 16.3265306122449, | |
| "grad_norm": 0.24278602004051208, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 0.0045, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 16.666666666666668, | |
| "grad_norm": 0.3119202256202698, | |
| "learning_rate": 8.38888888888889e-06, | |
| "loss": 0.0026, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 17.006802721088434, | |
| "grad_norm": 0.7003461718559265, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.0031, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 17.346938775510203, | |
| "grad_norm": 0.3013887107372284, | |
| "learning_rate": 8.277777777777778e-06, | |
| "loss": 0.0024, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 17.687074829931973, | |
| "grad_norm": 0.054453521966934204, | |
| "learning_rate": 8.222222222222222e-06, | |
| "loss": 0.0018, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 18.027210884353742, | |
| "grad_norm": 0.46942847967147827, | |
| "learning_rate": 8.166666666666668e-06, | |
| "loss": 0.0016, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 18.367346938775512, | |
| "grad_norm": 0.051955267786979675, | |
| "learning_rate": 8.111111111111112e-06, | |
| "loss": 0.0014, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 18.707482993197278, | |
| "grad_norm": 0.022960973903536797, | |
| "learning_rate": 8.055555555555557e-06, | |
| "loss": 0.0007, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 19.047619047619047, | |
| "grad_norm": 0.060347653925418854, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.0023, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 19.387755102040817, | |
| "grad_norm": 0.045802563428878784, | |
| "learning_rate": 7.944444444444445e-06, | |
| "loss": 0.0009, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 19.727891156462587, | |
| "grad_norm": 0.03706446662545204, | |
| "learning_rate": 7.88888888888889e-06, | |
| "loss": 0.0009, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 20.068027210884352, | |
| "grad_norm": 0.014720437116920948, | |
| "learning_rate": 7.833333333333333e-06, | |
| "loss": 0.0008, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 20.408163265306122, | |
| "grad_norm": 0.01641876809298992, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.0005, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 20.74829931972789, | |
| "grad_norm": 0.051012102514505386, | |
| "learning_rate": 7.722222222222223e-06, | |
| "loss": 0.0006, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 21.08843537414966, | |
| "grad_norm": 0.07526645809412003, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 0.0006, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 21.428571428571427, | |
| "grad_norm": 0.014325820840895176, | |
| "learning_rate": 7.611111111111111e-06, | |
| "loss": 0.0013, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 21.768707482993197, | |
| "grad_norm": 0.013493416830897331, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 0.0008, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 22.108843537414966, | |
| "grad_norm": 0.09651411324739456, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.0007, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 22.448979591836736, | |
| "grad_norm": 0.00849719438701868, | |
| "learning_rate": 7.444444444444445e-06, | |
| "loss": 0.0004, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 22.7891156462585, | |
| "grad_norm": 0.12997107207775116, | |
| "learning_rate": 7.38888888888889e-06, | |
| "loss": 0.0004, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 23.12925170068027, | |
| "grad_norm": 0.017772994935512543, | |
| "learning_rate": 7.333333333333333e-06, | |
| "loss": 0.0004, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 23.46938775510204, | |
| "grad_norm": 0.015189074911177158, | |
| "learning_rate": 7.277777777777778e-06, | |
| "loss": 0.0005, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 23.80952380952381, | |
| "grad_norm": 0.015632469207048416, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.0005, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 24.14965986394558, | |
| "grad_norm": 0.01429369393736124, | |
| "learning_rate": 7.166666666666667e-06, | |
| "loss": 0.0008, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 24.489795918367346, | |
| "grad_norm": 0.009108461439609528, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 0.0004, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 24.829931972789115, | |
| "grad_norm": 0.00893578864634037, | |
| "learning_rate": 7.055555555555557e-06, | |
| "loss": 0.0008, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 25.170068027210885, | |
| "grad_norm": 0.010999013669788837, | |
| "learning_rate": 7e-06, | |
| "loss": 0.0017, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 25.510204081632654, | |
| "grad_norm": 0.039021287113428116, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 0.001, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 25.85034013605442, | |
| "grad_norm": 0.8390358686447144, | |
| "learning_rate": 6.88888888888889e-06, | |
| "loss": 0.0009, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 26.19047619047619, | |
| "grad_norm": 0.02965502440929413, | |
| "learning_rate": 6.833333333333334e-06, | |
| "loss": 0.0013, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 26.53061224489796, | |
| "grad_norm": 0.21533174812793732, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 0.0029, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 26.87074829931973, | |
| "grad_norm": 0.06805889308452606, | |
| "learning_rate": 6.7222222222222235e-06, | |
| "loss": 0.0023, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 27.2108843537415, | |
| "grad_norm": 0.13776488602161407, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.0026, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 27.2108843537415, | |
| "eval_loss": 0.4282212555408478, | |
| "eval_runtime": 102.1629, | |
| "eval_samples_per_second": 2.545, | |
| "eval_steps_per_second": 0.166, | |
| "eval_wer": 0.22603506154419992, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 27.551020408163264, | |
| "grad_norm": 0.7586637735366821, | |
| "learning_rate": 6.6111111111111115e-06, | |
| "loss": 0.0045, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 27.891156462585034, | |
| "grad_norm": 0.19451989233493805, | |
| "learning_rate": 6.555555555555556e-06, | |
| "loss": 0.0051, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 28.231292517006803, | |
| "grad_norm": 0.1951601356267929, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.0037, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 28.571428571428573, | |
| "grad_norm": 0.3530046045780182, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 0.0037, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 28.91156462585034, | |
| "grad_norm": 0.26006898283958435, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 0.0045, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 29.25170068027211, | |
| "grad_norm": 1.6727913618087769, | |
| "learning_rate": 6.333333333333333e-06, | |
| "loss": 0.0034, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 29.591836734693878, | |
| "grad_norm": 0.23586586117744446, | |
| "learning_rate": 6.277777777777778e-06, | |
| "loss": 0.0033, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 29.931972789115648, | |
| "grad_norm": 2.080326557159424, | |
| "learning_rate": 6.222222222222223e-06, | |
| "loss": 0.0041, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 30.272108843537413, | |
| "grad_norm": 0.43551284074783325, | |
| "learning_rate": 6.166666666666667e-06, | |
| "loss": 0.0056, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 30.612244897959183, | |
| "grad_norm": 0.40780606865882874, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.0034, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 30.952380952380953, | |
| "grad_norm": 0.2716953754425049, | |
| "learning_rate": 6.055555555555555e-06, | |
| "loss": 0.0059, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 31.292517006802722, | |
| "grad_norm": 0.7849836945533752, | |
| "learning_rate": 6e-06, | |
| "loss": 0.004, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 31.632653061224488, | |
| "grad_norm": 0.0561537891626358, | |
| "learning_rate": 5.944444444444445e-06, | |
| "loss": 0.0031, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 31.972789115646258, | |
| "grad_norm": 0.2189192771911621, | |
| "learning_rate": 5.88888888888889e-06, | |
| "loss": 0.0025, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 32.31292517006803, | |
| "grad_norm": 0.7395544052124023, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.0035, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 32.6530612244898, | |
| "grad_norm": 0.8322564959526062, | |
| "learning_rate": 5.777777777777778e-06, | |
| "loss": 0.0029, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 32.993197278911566, | |
| "grad_norm": 0.1809442937374115, | |
| "learning_rate": 5.722222222222222e-06, | |
| "loss": 0.004, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 33.333333333333336, | |
| "grad_norm": 0.09690059721469879, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 0.0019, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 33.673469387755105, | |
| "grad_norm": 0.29557883739471436, | |
| "learning_rate": 5.611111111111112e-06, | |
| "loss": 0.0023, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 34.01360544217687, | |
| "grad_norm": 0.026952777057886124, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.001, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 34.35374149659864, | |
| "grad_norm": 0.04151106998324394, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.0008, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 34.69387755102041, | |
| "grad_norm": 0.01914023421704769, | |
| "learning_rate": 5.444444444444445e-06, | |
| "loss": 0.0012, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 35.034013605442176, | |
| "grad_norm": 0.15637150406837463, | |
| "learning_rate": 5.388888888888889e-06, | |
| "loss": 0.0011, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 35.374149659863946, | |
| "grad_norm": 0.18300969898700714, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 0.0013, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 35.714285714285715, | |
| "grad_norm": 0.019685134291648865, | |
| "learning_rate": 5.2777777777777785e-06, | |
| "loss": 0.0009, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 36.054421768707485, | |
| "grad_norm": 0.012847774662077427, | |
| "learning_rate": 5.2222222222222226e-06, | |
| "loss": 0.0004, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 36.394557823129254, | |
| "grad_norm": 0.011110267601907253, | |
| "learning_rate": 5.1666666666666675e-06, | |
| "loss": 0.0003, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 36.734693877551024, | |
| "grad_norm": 0.010116774588823318, | |
| "learning_rate": 5.1111111111111115e-06, | |
| "loss": 0.0003, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 37.074829931972786, | |
| "grad_norm": 0.00764122698456049, | |
| "learning_rate": 5.0555555555555555e-06, | |
| "loss": 0.0003, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 37.414965986394556, | |
| "grad_norm": 0.02383231371641159, | |
| "learning_rate": 5e-06, | |
| "loss": 0.0002, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 37.755102040816325, | |
| "grad_norm": 0.006380359176546335, | |
| "learning_rate": 4.944444444444445e-06, | |
| "loss": 0.0002, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 38.095238095238095, | |
| "grad_norm": 0.0054608238860964775, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 38.435374149659864, | |
| "grad_norm": 0.005097352433949709, | |
| "learning_rate": 4.833333333333333e-06, | |
| "loss": 0.0002, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 38.775510204081634, | |
| "grad_norm": 0.005154197569936514, | |
| "learning_rate": 4.777777777777778e-06, | |
| "loss": 0.0002, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 39.1156462585034, | |
| "grad_norm": 0.005586385726928711, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 0.0002, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 39.45578231292517, | |
| "grad_norm": 0.005172679666429758, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 0.0002, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 39.795918367346935, | |
| "grad_norm": 0.004589286167174578, | |
| "learning_rate": 4.611111111111112e-06, | |
| "loss": 0.0002, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 40.136054421768705, | |
| "grad_norm": 0.005452868994325399, | |
| "learning_rate": 4.555555555555556e-06, | |
| "loss": 0.0002, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 40.476190476190474, | |
| "grad_norm": 0.004887877497822046, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.0002, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 40.816326530612244, | |
| "grad_norm": 0.00491141015663743, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.0002, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 40.816326530612244, | |
| "eval_loss": 0.4444006681442261, | |
| "eval_runtime": 102.1694, | |
| "eval_samples_per_second": 2.545, | |
| "eval_steps_per_second": 0.166, | |
| "eval_wer": 0.21782916822081314, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 41.156462585034014, | |
| "grad_norm": 0.005289255175739527, | |
| "learning_rate": 4.388888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 41.49659863945578, | |
| "grad_norm": 0.004311501048505306, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 0.0001, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 41.83673469387755, | |
| "grad_norm": 0.004461329896003008, | |
| "learning_rate": 4.277777777777778e-06, | |
| "loss": 0.0002, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 42.17687074829932, | |
| "grad_norm": 0.0043508694507181644, | |
| "learning_rate": 4.222222222222223e-06, | |
| "loss": 0.0001, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 42.51700680272109, | |
| "grad_norm": 0.004281959030777216, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 42.857142857142854, | |
| "grad_norm": 0.004064427223056555, | |
| "learning_rate": 4.111111111111111e-06, | |
| "loss": 0.0001, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 43.197278911564624, | |
| "grad_norm": 0.00377645343542099, | |
| "learning_rate": 4.055555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 43.53741496598639, | |
| "grad_norm": 0.004163255449384451, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.0001, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 43.87755102040816, | |
| "grad_norm": 0.004156498704105616, | |
| "learning_rate": 3.944444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 44.21768707482993, | |
| "grad_norm": 0.002806807868182659, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 44.5578231292517, | |
| "grad_norm": 0.0035042332019656897, | |
| "learning_rate": 3.833333333333334e-06, | |
| "loss": 0.0001, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 44.89795918367347, | |
| "grad_norm": 0.003332278924062848, | |
| "learning_rate": 3.777777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 45.23809523809524, | |
| "grad_norm": 0.003195433644577861, | |
| "learning_rate": 3.7222222222222225e-06, | |
| "loss": 0.0001, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 45.578231292517, | |
| "grad_norm": 0.0031893581617623568, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 0.0001, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 45.91836734693877, | |
| "grad_norm": 0.0031172255985438824, | |
| "learning_rate": 3.6111111111111115e-06, | |
| "loss": 0.0001, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 46.25850340136054, | |
| "grad_norm": 0.0032927945721894503, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 46.59863945578231, | |
| "grad_norm": 0.003365216078236699, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.0001, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 46.93877551020408, | |
| "grad_norm": 0.003197253216058016, | |
| "learning_rate": 3.444444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 47.27891156462585, | |
| "grad_norm": 0.0033013387583196163, | |
| "learning_rate": 3.3888888888888893e-06, | |
| "loss": 0.0001, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 47.61904761904762, | |
| "grad_norm": 0.0030345292761921883, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.0001, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 47.95918367346939, | |
| "grad_norm": 0.002814779756590724, | |
| "learning_rate": 3.277777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 48.29931972789116, | |
| "grad_norm": 0.00269457814283669, | |
| "learning_rate": 3.2222222222222227e-06, | |
| "loss": 0.0001, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 48.63945578231292, | |
| "grad_norm": 0.0031415882986038923, | |
| "learning_rate": 3.1666666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 48.97959183673469, | |
| "grad_norm": 0.0029843435622751713, | |
| "learning_rate": 3.1111111111111116e-06, | |
| "loss": 0.0001, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 49.31972789115646, | |
| "grad_norm": 0.002468927064910531, | |
| "learning_rate": 3.055555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 49.65986394557823, | |
| "grad_norm": 0.002835317747667432, | |
| "learning_rate": 3e-06, | |
| "loss": 0.0001, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "grad_norm": 0.003936604596674442, | |
| "learning_rate": 2.944444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 50.34013605442177, | |
| "grad_norm": 0.0028534727171063423, | |
| "learning_rate": 2.888888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 50.68027210884354, | |
| "grad_norm": 0.0029034256003797054, | |
| "learning_rate": 2.8333333333333335e-06, | |
| "loss": 0.0001, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 51.02040816326531, | |
| "grad_norm": 0.0032608446199446917, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 0.0001, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 51.36054421768708, | |
| "grad_norm": 0.0026580116245895624, | |
| "learning_rate": 2.7222222222222224e-06, | |
| "loss": 0.0001, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 51.70068027210884, | |
| "grad_norm": 0.0024348797742277384, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 52.04081632653061, | |
| "grad_norm": 0.002745755948126316, | |
| "learning_rate": 2.6111111111111113e-06, | |
| "loss": 0.0001, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 52.38095238095238, | |
| "grad_norm": 0.0023754702415317297, | |
| "learning_rate": 2.5555555555555557e-06, | |
| "loss": 0.0001, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 52.72108843537415, | |
| "grad_norm": 0.0026109826285392046, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.0001, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 53.06122448979592, | |
| "grad_norm": 0.0025893962010741234, | |
| "learning_rate": 2.4444444444444447e-06, | |
| "loss": 0.0001, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 53.40136054421769, | |
| "grad_norm": 0.0023246912751346827, | |
| "learning_rate": 2.388888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 53.74149659863946, | |
| "grad_norm": 0.002337114419788122, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 0.0001, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 54.08163265306123, | |
| "grad_norm": 0.002635551616549492, | |
| "learning_rate": 2.277777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 54.421768707483, | |
| "grad_norm": 0.002480172784999013, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.0001, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 54.421768707483, | |
| "eval_loss": 0.4667165279388428, | |
| "eval_runtime": 102.3194, | |
| "eval_samples_per_second": 2.541, | |
| "eval_steps_per_second": 0.166, | |
| "eval_wer": 0.21596419246549795, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 54.76190476190476, | |
| "grad_norm": 0.00236712908372283, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 55.10204081632653, | |
| "grad_norm": 0.002560981782153249, | |
| "learning_rate": 2.1111111111111114e-06, | |
| "loss": 0.0001, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 55.4421768707483, | |
| "grad_norm": 0.002310627605766058, | |
| "learning_rate": 2.0555555555555555e-06, | |
| "loss": 0.0001, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 55.78231292517007, | |
| "grad_norm": 0.0023427396081387997, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.0001, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 56.12244897959184, | |
| "grad_norm": 0.002552021760493517, | |
| "learning_rate": 1.944444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 56.46258503401361, | |
| "grad_norm": 0.0021091082599014044, | |
| "learning_rate": 1.888888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 56.802721088435376, | |
| "grad_norm": 0.002378846053034067, | |
| "learning_rate": 1.8333333333333333e-06, | |
| "loss": 0.0001, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 57.142857142857146, | |
| "grad_norm": 0.0022475309669971466, | |
| "learning_rate": 1.777777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 57.48299319727891, | |
| "grad_norm": 0.0024991375394165516, | |
| "learning_rate": 1.7222222222222224e-06, | |
| "loss": 0.0001, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 57.82312925170068, | |
| "grad_norm": 0.0026695560663938522, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 58.16326530612245, | |
| "grad_norm": 0.0023107912857085466, | |
| "learning_rate": 1.6111111111111113e-06, | |
| "loss": 0.0001, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 58.50340136054422, | |
| "grad_norm": 0.0022581815719604492, | |
| "learning_rate": 1.5555555555555558e-06, | |
| "loss": 0.0001, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 58.843537414965986, | |
| "grad_norm": 0.0021738242357969284, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.0001, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 59.183673469387756, | |
| "grad_norm": 0.0025273971259593964, | |
| "learning_rate": 1.4444444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 59.523809523809526, | |
| "grad_norm": 0.0021680588833987713, | |
| "learning_rate": 1.3888888888888892e-06, | |
| "loss": 0.0001, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 59.863945578231295, | |
| "grad_norm": 0.0023293830454349518, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 0.0001, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 60.204081632653065, | |
| "grad_norm": 0.0019948012195527554, | |
| "learning_rate": 1.2777777777777779e-06, | |
| "loss": 0.0001, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 60.54421768707483, | |
| "grad_norm": 0.00274069607257843, | |
| "learning_rate": 1.2222222222222223e-06, | |
| "loss": 0.0001, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 60.8843537414966, | |
| "grad_norm": 0.002070850459858775, | |
| "learning_rate": 1.1666666666666668e-06, | |
| "loss": 0.0001, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 61.224489795918366, | |
| "grad_norm": 0.002111976034939289, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 0.0001, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 61.564625850340136, | |
| "grad_norm": 0.002158733317628503, | |
| "learning_rate": 1.0555555555555557e-06, | |
| "loss": 0.0001, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 61.904761904761905, | |
| "grad_norm": 0.002263331552967429, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.0001, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 62.244897959183675, | |
| "grad_norm": 0.0020453669130802155, | |
| "learning_rate": 9.444444444444445e-07, | |
| "loss": 0.0001, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 62.585034013605444, | |
| "grad_norm": 0.0018243337981402874, | |
| "learning_rate": 8.88888888888889e-07, | |
| "loss": 0.0001, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 62.925170068027214, | |
| "grad_norm": 0.0023292056284844875, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 0.0001, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 63.265306122448976, | |
| "grad_norm": 0.0019000651082023978, | |
| "learning_rate": 7.777777777777779e-07, | |
| "loss": 0.0001, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 63.605442176870746, | |
| "grad_norm": 0.002300037071108818, | |
| "learning_rate": 7.222222222222222e-07, | |
| "loss": 0.0001, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 63.945578231292515, | |
| "grad_norm": 0.0020359556656330824, | |
| "learning_rate": 6.666666666666667e-07, | |
| "loss": 0.0001, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 64.28571428571429, | |
| "grad_norm": 0.0019380120793357491, | |
| "learning_rate": 6.111111111111112e-07, | |
| "loss": 0.0001, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 64.62585034013605, | |
| "grad_norm": 0.0018156260484829545, | |
| "learning_rate": 5.555555555555555e-07, | |
| "loss": 0.0001, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 64.96598639455782, | |
| "grad_norm": 0.0021546927746385336, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.0001, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 65.3061224489796, | |
| "grad_norm": 0.0019719544798135757, | |
| "learning_rate": 4.444444444444445e-07, | |
| "loss": 0.0001, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 65.64625850340136, | |
| "grad_norm": 0.0018750294111669064, | |
| "learning_rate": 3.8888888888888895e-07, | |
| "loss": 0.0001, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 65.98639455782313, | |
| "grad_norm": 0.002267962321639061, | |
| "learning_rate": 3.3333333333333335e-07, | |
| "loss": 0.0001, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 66.3265306122449, | |
| "grad_norm": 0.0019681702833622694, | |
| "learning_rate": 2.7777777777777776e-07, | |
| "loss": 0.0001, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 66.66666666666667, | |
| "grad_norm": 0.0022395187988877296, | |
| "learning_rate": 2.2222222222222224e-07, | |
| "loss": 0.0001, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 67.00680272108843, | |
| "grad_norm": 0.0033925268799066544, | |
| "learning_rate": 1.6666666666666668e-07, | |
| "loss": 0.0001, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 67.34693877551021, | |
| "grad_norm": 0.0019715563394129276, | |
| "learning_rate": 1.1111111111111112e-07, | |
| "loss": 0.0001, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 67.68707482993197, | |
| "grad_norm": 0.00179803348146379, | |
| "learning_rate": 5.555555555555556e-08, | |
| "loss": 0.0001, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 68.02721088435374, | |
| "grad_norm": 0.0019662685226649046, | |
| "learning_rate": 0.0, | |
| "loss": 0.0001, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 68.02721088435374, | |
| "eval_loss": 0.47325974702835083, | |
| "eval_runtime": 102.3255, | |
| "eval_samples_per_second": 2.541, | |
| "eval_steps_per_second": 0.166, | |
| "eval_wer": 0.21372622155911974, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 68.02721088435374, | |
| "step": 5000, | |
| "total_flos": 5.4059634054660096e+20, | |
| "train_loss": 0.028010966634354553, | |
| "train_runtime": 44897.3386, | |
| "train_samples_per_second": 3.564, | |
| "train_steps_per_second": 0.111 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 69, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.4059634054660096e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |