| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 7.072135785007072, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03536067892503536, | |
| "grad_norm": 5.942479610443115, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 1.024, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07072135785007072, | |
| "grad_norm": 4.6511549949646, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.7269, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.10608203677510608, | |
| "grad_norm": 3.9613099098205566, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.5419, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.14144271570014144, | |
| "grad_norm": 4.0183186531066895, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.4772, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1768033946251768, | |
| "grad_norm": 4.522441864013672, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.449, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.21216407355021216, | |
| "grad_norm": 3.861356258392334, | |
| "learning_rate": 3e-06, | |
| "loss": 0.4327, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.24752475247524752, | |
| "grad_norm": 4.831862449645996, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.4107, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.2828854314002829, | |
| "grad_norm": 3.336674690246582, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.4043, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.31824611032531824, | |
| "grad_norm": 4.305329322814941, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.3914, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.3536067892503536, | |
| "grad_norm": 3.83046817779541, | |
| "learning_rate": 5e-06, | |
| "loss": 0.3685, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.38896746817538896, | |
| "grad_norm": 3.484257698059082, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.351, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.4243281471004243, | |
| "grad_norm": 3.844430923461914, | |
| "learning_rate": 6e-06, | |
| "loss": 0.3629, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.4596888260254597, | |
| "grad_norm": 4.026344299316406, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.3662, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.49504950495049505, | |
| "grad_norm": 3.9913763999938965, | |
| "learning_rate": 7e-06, | |
| "loss": 0.3473, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.5304101838755304, | |
| "grad_norm": 4.128275394439697, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.3476, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.5657708628005658, | |
| "grad_norm": 3.474440813064575, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.3357, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.6011315417256011, | |
| "grad_norm": 2.85972261428833, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.3081, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.6364922206506365, | |
| "grad_norm": 3.8829457759857178, | |
| "learning_rate": 9e-06, | |
| "loss": 0.3129, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.6718528995756718, | |
| "grad_norm": 3.661864757537842, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.315, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.7072135785007072, | |
| "grad_norm": 3.2057082653045654, | |
| "learning_rate": 1e-05, | |
| "loss": 0.324, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.7425742574257426, | |
| "grad_norm": 3.3009073734283447, | |
| "learning_rate": 9.944444444444445e-06, | |
| "loss": 0.3201, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.7779349363507779, | |
| "grad_norm": 3.073880910873413, | |
| "learning_rate": 9.88888888888889e-06, | |
| "loss": 0.2754, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.8132956152758133, | |
| "grad_norm": 2.913012742996216, | |
| "learning_rate": 9.833333333333333e-06, | |
| "loss": 0.2737, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.8486562942008486, | |
| "grad_norm": 2.8291831016540527, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 0.2823, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.884016973125884, | |
| "grad_norm": 3.002702474594116, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 0.2639, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.9193776520509194, | |
| "grad_norm": 3.184739112854004, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 0.2689, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.9547383309759547, | |
| "grad_norm": 3.179779291152954, | |
| "learning_rate": 9.611111111111112e-06, | |
| "loss": 0.268, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.9900990099009901, | |
| "grad_norm": 3.430415391921997, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 0.2657, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.0254596888260255, | |
| "grad_norm": 2.4761476516723633, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.2076, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.0608203677510608, | |
| "grad_norm": 2.250551700592041, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.177, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.0961810466760962, | |
| "grad_norm": 2.776700258255005, | |
| "learning_rate": 9.38888888888889e-06, | |
| "loss": 0.1664, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.1315417256011315, | |
| "grad_norm": 2.360950469970703, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 0.1576, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.166902404526167, | |
| "grad_norm": 2.858429193496704, | |
| "learning_rate": 9.277777777777778e-06, | |
| "loss": 0.1706, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.2022630834512023, | |
| "grad_norm": 2.6928622722625732, | |
| "learning_rate": 9.222222222222224e-06, | |
| "loss": 0.1733, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.2376237623762376, | |
| "grad_norm": 2.4788143634796143, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.1653, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.272984441301273, | |
| "grad_norm": 2.5625805854797363, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 0.1669, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.3083451202263083, | |
| "grad_norm": 2.179570198059082, | |
| "learning_rate": 9.055555555555556e-06, | |
| "loss": 0.1594, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.3437057991513437, | |
| "grad_norm": 2.751049280166626, | |
| "learning_rate": 9e-06, | |
| "loss": 0.161, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.379066478076379, | |
| "grad_norm": 2.8607306480407715, | |
| "learning_rate": 8.944444444444446e-06, | |
| "loss": 0.1563, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.4144271570014144, | |
| "grad_norm": 2.380326271057129, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.1583, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.4144271570014144, | |
| "eval_loss": 0.25619471073150635, | |
| "eval_runtime": 2242.3857, | |
| "eval_samples_per_second": 2.4, | |
| "eval_steps_per_second": 0.15, | |
| "eval_wer": 0.20623086405026045, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.4497878359264498, | |
| "grad_norm": 2.163681983947754, | |
| "learning_rate": 8.833333333333334e-06, | |
| "loss": 0.1759, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.4851485148514851, | |
| "grad_norm": 2.3342068195343018, | |
| "learning_rate": 8.777777777777778e-06, | |
| "loss": 0.1553, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.5205091937765205, | |
| "grad_norm": 1.9928505420684814, | |
| "learning_rate": 8.722222222222224e-06, | |
| "loss": 0.1647, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.5558698727015559, | |
| "grad_norm": 2.837106227874756, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 0.153, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.5912305516265912, | |
| "grad_norm": 2.0526952743530273, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.1621, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.6265912305516266, | |
| "grad_norm": 2.482564926147461, | |
| "learning_rate": 8.555555555555556e-06, | |
| "loss": 0.1556, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.661951909476662, | |
| "grad_norm": 1.8798569440841675, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.1387, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.6973125884016973, | |
| "grad_norm": 2.249302864074707, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 0.141, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.7326732673267327, | |
| "grad_norm": 2.683685064315796, | |
| "learning_rate": 8.38888888888889e-06, | |
| "loss": 0.1454, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.768033946251768, | |
| "grad_norm": 2.46101450920105, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.1522, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.8033946251768034, | |
| "grad_norm": 2.1939165592193604, | |
| "learning_rate": 8.277777777777778e-06, | |
| "loss": 0.1384, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.8387553041018387, | |
| "grad_norm": 1.9950499534606934, | |
| "learning_rate": 8.222222222222222e-06, | |
| "loss": 0.1487, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.874115983026874, | |
| "grad_norm": 2.307433605194092, | |
| "learning_rate": 8.166666666666668e-06, | |
| "loss": 0.1429, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.9094766619519095, | |
| "grad_norm": 2.7583506107330322, | |
| "learning_rate": 8.111111111111112e-06, | |
| "loss": 0.1372, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.9448373408769448, | |
| "grad_norm": 2.037691593170166, | |
| "learning_rate": 8.055555555555557e-06, | |
| "loss": 0.1438, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.9801980198019802, | |
| "grad_norm": 3.130493640899658, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.1399, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.0155586987270158, | |
| "grad_norm": 1.4508907794952393, | |
| "learning_rate": 7.944444444444445e-06, | |
| "loss": 0.0945, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.050919377652051, | |
| "grad_norm": 1.6655480861663818, | |
| "learning_rate": 7.88888888888889e-06, | |
| "loss": 0.0695, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.0862800565770865, | |
| "grad_norm": 1.3497799634933472, | |
| "learning_rate": 7.833333333333333e-06, | |
| "loss": 0.0653, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.1216407355021216, | |
| "grad_norm": 1.9043290615081787, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.0685, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.157001414427157, | |
| "grad_norm": 1.7058606147766113, | |
| "learning_rate": 7.722222222222223e-06, | |
| "loss": 0.0686, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.1923620933521923, | |
| "grad_norm": 1.5799323320388794, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 0.0713, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.227722772277228, | |
| "grad_norm": 2.1463682651519775, | |
| "learning_rate": 7.611111111111111e-06, | |
| "loss": 0.0742, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.263083451202263, | |
| "grad_norm": 1.647290587425232, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 0.0658, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.298444130127298, | |
| "grad_norm": 2.096013307571411, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.0647, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.333804809052334, | |
| "grad_norm": 1.7993731498718262, | |
| "learning_rate": 7.444444444444445e-06, | |
| "loss": 0.064, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.3691654879773694, | |
| "grad_norm": 1.6189881563186646, | |
| "learning_rate": 7.38888888888889e-06, | |
| "loss": 0.071, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.4045261669024045, | |
| "grad_norm": 1.9548283815383911, | |
| "learning_rate": 7.333333333333333e-06, | |
| "loss": 0.065, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.4398868458274396, | |
| "grad_norm": 1.7484291791915894, | |
| "learning_rate": 7.277777777777778e-06, | |
| "loss": 0.0621, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.4752475247524752, | |
| "grad_norm": 1.8069815635681152, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.0752, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.510608203677511, | |
| "grad_norm": 1.3003634214401245, | |
| "learning_rate": 7.166666666666667e-06, | |
| "loss": 0.0719, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.545968882602546, | |
| "grad_norm": 1.7700819969177246, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 0.0723, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.581329561527581, | |
| "grad_norm": 2.041518211364746, | |
| "learning_rate": 7.055555555555557e-06, | |
| "loss": 0.0659, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.6166902404526167, | |
| "grad_norm": 1.4527554512023926, | |
| "learning_rate": 7e-06, | |
| "loss": 0.0669, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.6520509193776522, | |
| "grad_norm": 1.6484756469726562, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 0.0624, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 2.6874115983026874, | |
| "grad_norm": 1.9906888008117676, | |
| "learning_rate": 6.88888888888889e-06, | |
| "loss": 0.0614, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.7227722772277225, | |
| "grad_norm": 1.4477077722549438, | |
| "learning_rate": 6.833333333333334e-06, | |
| "loss": 0.0651, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 2.758132956152758, | |
| "grad_norm": 1.5185719728469849, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 0.0626, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.7934936350777937, | |
| "grad_norm": 2.1326749324798584, | |
| "learning_rate": 6.7222222222222235e-06, | |
| "loss": 0.0582, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 2.828854314002829, | |
| "grad_norm": 1.7811297178268433, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.0675, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.828854314002829, | |
| "eval_loss": 0.23944300413131714, | |
| "eval_runtime": 2146.1705, | |
| "eval_samples_per_second": 2.507, | |
| "eval_steps_per_second": 0.157, | |
| "eval_wer": 0.1849377708855223, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.864214992927864, | |
| "grad_norm": 2.299253463745117, | |
| "learning_rate": 6.6111111111111115e-06, | |
| "loss": 0.0632, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 2.8995756718528995, | |
| "grad_norm": 1.6922551393508911, | |
| "learning_rate": 6.555555555555556e-06, | |
| "loss": 0.0624, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.934936350777935, | |
| "grad_norm": 1.5919361114501953, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.0621, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 2.9702970297029703, | |
| "grad_norm": 1.9371802806854248, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 0.0665, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.005657708628006, | |
| "grad_norm": 1.0974698066711426, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 0.0577, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 3.041018387553041, | |
| "grad_norm": 0.9451077580451965, | |
| "learning_rate": 6.333333333333333e-06, | |
| "loss": 0.026, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.0763790664780766, | |
| "grad_norm": 1.0205219984054565, | |
| "learning_rate": 6.277777777777778e-06, | |
| "loss": 0.0271, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 3.1117397454031117, | |
| "grad_norm": 1.12224543094635, | |
| "learning_rate": 6.222222222222223e-06, | |
| "loss": 0.0277, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.1471004243281473, | |
| "grad_norm": 1.326258897781372, | |
| "learning_rate": 6.166666666666667e-06, | |
| "loss": 0.0245, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 3.1824611032531824, | |
| "grad_norm": 1.4001460075378418, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.0268, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.217821782178218, | |
| "grad_norm": 1.4342401027679443, | |
| "learning_rate": 6.055555555555555e-06, | |
| "loss": 0.0348, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 3.253182461103253, | |
| "grad_norm": 0.9824921488761902, | |
| "learning_rate": 6e-06, | |
| "loss": 0.0287, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.2885431400282887, | |
| "grad_norm": 1.2897576093673706, | |
| "learning_rate": 5.944444444444445e-06, | |
| "loss": 0.0278, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 3.323903818953324, | |
| "grad_norm": 0.8747806549072266, | |
| "learning_rate": 5.88888888888889e-06, | |
| "loss": 0.0284, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.3592644978783595, | |
| "grad_norm": 1.4630858898162842, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.0248, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 3.3946251768033946, | |
| "grad_norm": 1.220078945159912, | |
| "learning_rate": 5.777777777777778e-06, | |
| "loss": 0.0278, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.42998585572843, | |
| "grad_norm": 1.1768743991851807, | |
| "learning_rate": 5.722222222222222e-06, | |
| "loss": 0.0249, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 3.4653465346534653, | |
| "grad_norm": 1.5660589933395386, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 0.0307, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.500707213578501, | |
| "grad_norm": 0.8128176331520081, | |
| "learning_rate": 5.611111111111112e-06, | |
| "loss": 0.0321, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 3.536067892503536, | |
| "grad_norm": 1.3139251470565796, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.0295, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.571428571428571, | |
| "grad_norm": 0.9214044809341431, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.0279, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 3.6067892503536068, | |
| "grad_norm": 1.11368989944458, | |
| "learning_rate": 5.444444444444445e-06, | |
| "loss": 0.031, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.6421499292786423, | |
| "grad_norm": 1.238482117652893, | |
| "learning_rate": 5.388888888888889e-06, | |
| "loss": 0.0283, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 3.6775106082036775, | |
| "grad_norm": 1.528247594833374, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 0.0285, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.7128712871287126, | |
| "grad_norm": 1.4220536947250366, | |
| "learning_rate": 5.2777777777777785e-06, | |
| "loss": 0.0254, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 3.748231966053748, | |
| "grad_norm": 0.7737128734588623, | |
| "learning_rate": 5.2222222222222226e-06, | |
| "loss": 0.0275, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 3.783592644978784, | |
| "grad_norm": 1.0860861539840698, | |
| "learning_rate": 5.1666666666666675e-06, | |
| "loss": 0.0332, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 3.818953323903819, | |
| "grad_norm": 1.240889072418213, | |
| "learning_rate": 5.1111111111111115e-06, | |
| "loss": 0.0238, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 3.854314002828854, | |
| "grad_norm": 1.815717101097107, | |
| "learning_rate": 5.0555555555555555e-06, | |
| "loss": 0.0288, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 3.8896746817538896, | |
| "grad_norm": 0.997189998626709, | |
| "learning_rate": 5e-06, | |
| "loss": 0.0291, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 3.9250353606789252, | |
| "grad_norm": 0.7937121987342834, | |
| "learning_rate": 4.944444444444445e-06, | |
| "loss": 0.0289, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 3.9603960396039604, | |
| "grad_norm": 0.74869704246521, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 0.0294, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 3.9957567185289955, | |
| "grad_norm": 0.791512131690979, | |
| "learning_rate": 4.833333333333333e-06, | |
| "loss": 0.0278, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 4.0311173974540315, | |
| "grad_norm": 0.7676866054534912, | |
| "learning_rate": 4.777777777777778e-06, | |
| "loss": 0.0128, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 4.066478076379067, | |
| "grad_norm": 0.8113923668861389, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 0.012, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 4.101838755304102, | |
| "grad_norm": 0.9902219176292419, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 0.0104, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 4.137199434229137, | |
| "grad_norm": 0.6298081874847412, | |
| "learning_rate": 4.611111111111112e-06, | |
| "loss": 0.0101, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 4.172560113154173, | |
| "grad_norm": 0.4004449248313904, | |
| "learning_rate": 4.555555555555556e-06, | |
| "loss": 0.0124, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 4.207920792079208, | |
| "grad_norm": 1.081607699394226, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.0131, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 4.243281471004243, | |
| "grad_norm": 1.244435429573059, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.0113, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.243281471004243, | |
| "eval_loss": 0.2729232609272003, | |
| "eval_runtime": 2135.6924, | |
| "eval_samples_per_second": 2.52, | |
| "eval_steps_per_second": 0.158, | |
| "eval_wer": 0.17223348840908187, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.278642149929278, | |
| "grad_norm": 0.7269119620323181, | |
| "learning_rate": 4.388888888888889e-06, | |
| "loss": 0.0141, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 4.314002828854314, | |
| "grad_norm": 0.7338502407073975, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 0.0094, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 4.3493635077793495, | |
| "grad_norm": 1.575302004814148, | |
| "learning_rate": 4.277777777777778e-06, | |
| "loss": 0.0117, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 4.384724186704385, | |
| "grad_norm": 0.8755286931991577, | |
| "learning_rate": 4.222222222222223e-06, | |
| "loss": 0.0096, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 4.42008486562942, | |
| "grad_norm": 0.606636106967926, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.0108, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 4.455445544554456, | |
| "grad_norm": 0.4599132835865021, | |
| "learning_rate": 4.111111111111111e-06, | |
| "loss": 0.0109, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 4.490806223479491, | |
| "grad_norm": 0.9773505330085754, | |
| "learning_rate": 4.055555555555556e-06, | |
| "loss": 0.0095, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 4.526166902404526, | |
| "grad_norm": 1.153731107711792, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.0102, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 4.561527581329561, | |
| "grad_norm": 0.38161006569862366, | |
| "learning_rate": 3.944444444444445e-06, | |
| "loss": 0.01, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 4.596888260254596, | |
| "grad_norm": 0.4255661368370056, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.0098, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 4.632248939179632, | |
| "grad_norm": 0.47027960419654846, | |
| "learning_rate": 3.833333333333334e-06, | |
| "loss": 0.0082, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 4.667609618104668, | |
| "grad_norm": 0.7094388604164124, | |
| "learning_rate": 3.777777777777778e-06, | |
| "loss": 0.013, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 4.702970297029703, | |
| "grad_norm": 1.0208382606506348, | |
| "learning_rate": 3.7222222222222225e-06, | |
| "loss": 0.01, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 4.738330975954739, | |
| "grad_norm": 0.41139042377471924, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 0.0115, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 4.773691654879774, | |
| "grad_norm": 0.2961082458496094, | |
| "learning_rate": 3.6111111111111115e-06, | |
| "loss": 0.0102, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 4.809052333804809, | |
| "grad_norm": 0.7990303039550781, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 0.0085, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 4.844413012729844, | |
| "grad_norm": 0.7530162930488586, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.0098, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 4.879773691654879, | |
| "grad_norm": 0.44302016496658325, | |
| "learning_rate": 3.444444444444445e-06, | |
| "loss": 0.0078, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 4.915134370579915, | |
| "grad_norm": 0.43590739369392395, | |
| "learning_rate": 3.3888888888888893e-06, | |
| "loss": 0.008, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 4.9504950495049505, | |
| "grad_norm": 1.0824764966964722, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.0133, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 4.985855728429986, | |
| "grad_norm": 1.0049275159835815, | |
| "learning_rate": 3.277777777777778e-06, | |
| "loss": 0.0098, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 5.021216407355022, | |
| "grad_norm": 0.13272249698638916, | |
| "learning_rate": 3.2222222222222227e-06, | |
| "loss": 0.0093, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 5.056577086280057, | |
| "grad_norm": 0.42071208357810974, | |
| "learning_rate": 3.1666666666666667e-06, | |
| "loss": 0.0045, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 5.091937765205092, | |
| "grad_norm": 1.430225133895874, | |
| "learning_rate": 3.1111111111111116e-06, | |
| "loss": 0.0044, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 5.127298444130127, | |
| "grad_norm": 0.8059938549995422, | |
| "learning_rate": 3.055555555555556e-06, | |
| "loss": 0.0042, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 5.162659123055163, | |
| "grad_norm": 0.2698410153388977, | |
| "learning_rate": 3e-06, | |
| "loss": 0.0031, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 5.198019801980198, | |
| "grad_norm": 0.4564538598060608, | |
| "learning_rate": 2.944444444444445e-06, | |
| "loss": 0.0038, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 5.233380480905233, | |
| "grad_norm": 2.555163621902466, | |
| "learning_rate": 2.888888888888889e-06, | |
| "loss": 0.0035, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 5.2687411598302685, | |
| "grad_norm": 0.6030502915382385, | |
| "learning_rate": 2.8333333333333335e-06, | |
| "loss": 0.0041, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 5.3041018387553045, | |
| "grad_norm": 0.2795655131340027, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 0.0039, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 5.33946251768034, | |
| "grad_norm": 0.23795121908187866, | |
| "learning_rate": 2.7222222222222224e-06, | |
| "loss": 0.0038, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 5.374823196605375, | |
| "grad_norm": 0.08186432719230652, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 0.0023, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 5.41018387553041, | |
| "grad_norm": 0.7226074934005737, | |
| "learning_rate": 2.6111111111111113e-06, | |
| "loss": 0.0034, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 5.445544554455446, | |
| "grad_norm": 0.4257412850856781, | |
| "learning_rate": 2.5555555555555557e-06, | |
| "loss": 0.004, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 5.480905233380481, | |
| "grad_norm": 0.48502105474472046, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.003, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 5.516265912305516, | |
| "grad_norm": 0.2505144774913788, | |
| "learning_rate": 2.4444444444444447e-06, | |
| "loss": 0.0047, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 5.551626591230551, | |
| "grad_norm": 0.24553366005420685, | |
| "learning_rate": 2.388888888888889e-06, | |
| "loss": 0.0027, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 5.586987270155587, | |
| "grad_norm": 0.1790177971124649, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 0.0031, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 5.6223479490806225, | |
| "grad_norm": 0.625106155872345, | |
| "learning_rate": 2.277777777777778e-06, | |
| "loss": 0.0045, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 5.657708628005658, | |
| "grad_norm": 0.3036724925041199, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.0036, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.657708628005658, | |
| "eval_loss": 0.3003575801849365, | |
| "eval_runtime": 2115.084, | |
| "eval_samples_per_second": 2.544, | |
| "eval_steps_per_second": 0.159, | |
| "eval_wer": 0.1705435603801344, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.693069306930693, | |
| "grad_norm": 0.5264594554901123, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 0.0038, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 5.728429985855728, | |
| "grad_norm": 0.43084102869033813, | |
| "learning_rate": 2.1111111111111114e-06, | |
| "loss": 0.003, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 5.763790664780764, | |
| "grad_norm": 0.3291011452674866, | |
| "learning_rate": 2.0555555555555555e-06, | |
| "loss": 0.0032, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 5.799151343705799, | |
| "grad_norm": 0.11041796952486038, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.0041, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 5.834512022630834, | |
| "grad_norm": 0.40376558899879456, | |
| "learning_rate": 1.944444444444445e-06, | |
| "loss": 0.0029, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 5.86987270155587, | |
| "grad_norm": 1.008151650428772, | |
| "learning_rate": 1.888888888888889e-06, | |
| "loss": 0.0023, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 5.905233380480905, | |
| "grad_norm": 0.29702553153038025, | |
| "learning_rate": 1.8333333333333333e-06, | |
| "loss": 0.0039, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 5.9405940594059405, | |
| "grad_norm": 0.0774470716714859, | |
| "learning_rate": 1.777777777777778e-06, | |
| "loss": 0.0025, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 5.975954738330976, | |
| "grad_norm": 0.2981080114841461, | |
| "learning_rate": 1.7222222222222224e-06, | |
| "loss": 0.0025, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 6.011315417256012, | |
| "grad_norm": 0.08830191940069199, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.002, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 6.046676096181047, | |
| "grad_norm": 0.06720926612615585, | |
| "learning_rate": 1.6111111111111113e-06, | |
| "loss": 0.0014, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 6.082036775106082, | |
| "grad_norm": 0.07184426486492157, | |
| "learning_rate": 1.5555555555555558e-06, | |
| "loss": 0.0017, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 6.117397454031117, | |
| "grad_norm": 0.06138487532734871, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.0012, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 6.152758132956153, | |
| "grad_norm": 0.11943061649799347, | |
| "learning_rate": 1.4444444444444445e-06, | |
| "loss": 0.0014, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 6.188118811881188, | |
| "grad_norm": 0.05326369032263756, | |
| "learning_rate": 1.3888888888888892e-06, | |
| "loss": 0.0017, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 6.223479490806223, | |
| "grad_norm": 0.05557446554303169, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 0.0011, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 6.258840169731259, | |
| "grad_norm": 0.05990071967244148, | |
| "learning_rate": 1.2777777777777779e-06, | |
| "loss": 0.0015, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 6.294200848656295, | |
| "grad_norm": 0.06366278976202011, | |
| "learning_rate": 1.2222222222222223e-06, | |
| "loss": 0.0012, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 6.32956152758133, | |
| "grad_norm": 0.0781664326786995, | |
| "learning_rate": 1.1666666666666668e-06, | |
| "loss": 0.0016, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 6.364922206506365, | |
| "grad_norm": 0.19605182111263275, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 0.0016, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 6.4002828854314, | |
| "grad_norm": 0.07928458601236343, | |
| "learning_rate": 1.0555555555555557e-06, | |
| "loss": 0.0015, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 6.435643564356436, | |
| "grad_norm": 0.2883352041244507, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.0018, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 6.471004243281471, | |
| "grad_norm": 0.037699662148952484, | |
| "learning_rate": 9.444444444444445e-07, | |
| "loss": 0.0015, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 6.506364922206506, | |
| "grad_norm": 0.2606990933418274, | |
| "learning_rate": 8.88888888888889e-07, | |
| "loss": 0.0012, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 6.5417256011315414, | |
| "grad_norm": 0.04962063580751419, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 0.0016, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 6.5770862800565775, | |
| "grad_norm": 0.11313042789697647, | |
| "learning_rate": 7.777777777777779e-07, | |
| "loss": 0.0014, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 6.612446958981613, | |
| "grad_norm": 0.10243412852287292, | |
| "learning_rate": 7.222222222222222e-07, | |
| "loss": 0.0011, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 6.647807637906648, | |
| "grad_norm": 0.2408544421195984, | |
| "learning_rate": 6.666666666666667e-07, | |
| "loss": 0.0015, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 6.683168316831683, | |
| "grad_norm": 0.08557415008544922, | |
| "learning_rate": 6.111111111111112e-07, | |
| "loss": 0.0013, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 6.718528995756719, | |
| "grad_norm": 0.05530049279332161, | |
| "learning_rate": 5.555555555555555e-07, | |
| "loss": 0.0012, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 6.753889674681754, | |
| "grad_norm": 0.1299530267715454, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.0018, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 6.789250353606789, | |
| "grad_norm": 0.03794874995946884, | |
| "learning_rate": 4.444444444444445e-07, | |
| "loss": 0.0011, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 6.824611032531824, | |
| "grad_norm": 0.051171157509088516, | |
| "learning_rate": 3.8888888888888895e-07, | |
| "loss": 0.0016, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 6.85997171145686, | |
| "grad_norm": 0.05626167729496956, | |
| "learning_rate": 3.3333333333333335e-07, | |
| "loss": 0.0014, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 6.8953323903818955, | |
| "grad_norm": 0.05049213021993637, | |
| "learning_rate": 2.7777777777777776e-07, | |
| "loss": 0.0017, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 6.930693069306931, | |
| "grad_norm": 0.2871930003166199, | |
| "learning_rate": 2.2222222222222224e-07, | |
| "loss": 0.0014, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 6.966053748231966, | |
| "grad_norm": 0.05043856427073479, | |
| "learning_rate": 1.6666666666666668e-07, | |
| "loss": 0.0014, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 7.001414427157002, | |
| "grad_norm": 0.052644409239292145, | |
| "learning_rate": 1.1111111111111112e-07, | |
| "loss": 0.0018, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 7.036775106082037, | |
| "grad_norm": 0.04488613083958626, | |
| "learning_rate": 5.555555555555556e-08, | |
| "loss": 0.0013, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 7.072135785007072, | |
| "grad_norm": 0.05924491211771965, | |
| "learning_rate": 0.0, | |
| "loss": 0.0012, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.072135785007072, | |
| "eval_loss": 0.3279673457145691, | |
| "eval_runtime": 2105.881, | |
| "eval_samples_per_second": 2.555, | |
| "eval_steps_per_second": 0.16, | |
| "eval_wer": 0.1676010974591435, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.072135785007072, | |
| "step": 5000, | |
| "total_flos": 5.4329055670370304e+20, | |
| "train_loss": 0.09207180016152561, | |
| "train_runtime": 58226.4708, | |
| "train_samples_per_second": 2.748, | |
| "train_steps_per_second": 0.086 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 8, | |
| "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": 5.4329055670370304e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |