| { | |
| "best_metric": 0.10248288219107954, | |
| "best_model_checkpoint": "d:\\\\whisper-medium-pt-cv19-fleurs2-lr-wu\\checkpoint-5000", | |
| "epoch": 11.441647597254004, | |
| "eval_steps": 5000, | |
| "global_step": 25000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.011441647597254004, | |
| "grad_norm": 16.260902404785156, | |
| "learning_rate": 2.875e-07, | |
| "loss": 0.7432, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02288329519450801, | |
| "grad_norm": 28.320634841918945, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 1.1561, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.034324942791762014, | |
| "grad_norm": 9.76030158996582, | |
| "learning_rate": 9.124999999999999e-07, | |
| "loss": 0.479, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.04576659038901602, | |
| "grad_norm": 17.32778549194336, | |
| "learning_rate": 1.2250000000000001e-06, | |
| "loss": 0.3224, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.057208237986270026, | |
| "grad_norm": 7.594138145446777, | |
| "learning_rate": 1.5375e-06, | |
| "loss": 0.2161, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.06864988558352403, | |
| "grad_norm": 18.392087936401367, | |
| "learning_rate": 1.85e-06, | |
| "loss": 0.262, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08009153318077804, | |
| "grad_norm": 10.459943771362305, | |
| "learning_rate": 2.1625e-06, | |
| "loss": 0.1765, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.09153318077803203, | |
| "grad_norm": 17.077810287475586, | |
| "learning_rate": 2.4750000000000004e-06, | |
| "loss": 0.2262, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.10297482837528604, | |
| "grad_norm": 6.674950122833252, | |
| "learning_rate": 2.7875e-06, | |
| "loss": 0.1917, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.11441647597254005, | |
| "grad_norm": 10.16779613494873, | |
| "learning_rate": 3.1e-06, | |
| "loss": 0.2162, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.12585812356979406, | |
| "grad_norm": 6.614413738250732, | |
| "learning_rate": 3.4125000000000004e-06, | |
| "loss": 0.2, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.13729977116704806, | |
| "grad_norm": 13.418252944946289, | |
| "learning_rate": 3.725e-06, | |
| "loss": 0.2007, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.14874141876430205, | |
| "grad_norm": 6.218944549560547, | |
| "learning_rate": 4.037500000000001e-06, | |
| "loss": 0.1574, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.16018306636155608, | |
| "grad_norm": 15.351899147033691, | |
| "learning_rate": 4.35e-06, | |
| "loss": 0.1894, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.17162471395881007, | |
| "grad_norm": 7.518232822418213, | |
| "learning_rate": 4.6625e-06, | |
| "loss": 0.1861, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.18306636155606407, | |
| "grad_norm": 16.16284942626953, | |
| "learning_rate": 4.975000000000001e-06, | |
| "loss": 0.2085, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.1945080091533181, | |
| "grad_norm": 8.712101936340332, | |
| "learning_rate": 5.2875e-06, | |
| "loss": 0.1664, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.20594965675057209, | |
| "grad_norm": 15.569042205810547, | |
| "learning_rate": 5.600000000000001e-06, | |
| "loss": 0.2054, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.21739130434782608, | |
| "grad_norm": 7.28713846206665, | |
| "learning_rate": 5.9125e-06, | |
| "loss": 0.1822, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.2288329519450801, | |
| "grad_norm": 14.67645263671875, | |
| "learning_rate": 6.2250000000000005e-06, | |
| "loss": 0.2018, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.2402745995423341, | |
| "grad_norm": 9.035391807556152, | |
| "learning_rate": 6.244132653061225e-06, | |
| "loss": 0.175, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.2517162471395881, | |
| "grad_norm": 13.814667701721191, | |
| "learning_rate": 6.237755102040817e-06, | |
| "loss": 0.2023, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.2631578947368421, | |
| "grad_norm": 5.595588684082031, | |
| "learning_rate": 6.231377551020408e-06, | |
| "loss": 0.1745, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.2745995423340961, | |
| "grad_norm": 14.676183700561523, | |
| "learning_rate": 6.2250000000000005e-06, | |
| "loss": 0.2066, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.28604118993135014, | |
| "grad_norm": 7.4100213050842285, | |
| "learning_rate": 6.218622448979592e-06, | |
| "loss": 0.1968, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.2974828375286041, | |
| "grad_norm": 17.322240829467773, | |
| "learning_rate": 6.212244897959184e-06, | |
| "loss": 0.1916, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.30892448512585813, | |
| "grad_norm": 7.691104412078857, | |
| "learning_rate": 6.205867346938775e-06, | |
| "loss": 0.1725, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.32036613272311215, | |
| "grad_norm": 16.282310485839844, | |
| "learning_rate": 6.1994897959183676e-06, | |
| "loss": 0.2122, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.3318077803203661, | |
| "grad_norm": 8.981350898742676, | |
| "learning_rate": 6.19311224489796e-06, | |
| "loss": 0.1638, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.34324942791762014, | |
| "grad_norm": 16.19715690612793, | |
| "learning_rate": 6.1867346938775515e-06, | |
| "loss": 0.2046, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.35469107551487417, | |
| "grad_norm": 5.906866073608398, | |
| "learning_rate": 6.180357142857143e-06, | |
| "loss": 0.1916, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.36613272311212813, | |
| "grad_norm": 10.55771255493164, | |
| "learning_rate": 6.1739795918367354e-06, | |
| "loss": 0.2104, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.37757437070938216, | |
| "grad_norm": 6.290849685668945, | |
| "learning_rate": 6.167602040816327e-06, | |
| "loss": 0.1618, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.3890160183066362, | |
| "grad_norm": 15.78228759765625, | |
| "learning_rate": 6.1612244897959185e-06, | |
| "loss": 0.1897, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.40045766590389015, | |
| "grad_norm": 4.934595108032227, | |
| "learning_rate": 6.15484693877551e-06, | |
| "loss": 0.1808, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.41189931350114417, | |
| "grad_norm": 8.691235542297363, | |
| "learning_rate": 6.1484693877551024e-06, | |
| "loss": 0.1913, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.4233409610983982, | |
| "grad_norm": 9.48690414428711, | |
| "learning_rate": 6.142091836734694e-06, | |
| "loss": 0.167, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.43478260869565216, | |
| "grad_norm": 17.154409408569336, | |
| "learning_rate": 6.135714285714286e-06, | |
| "loss": 0.1863, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.4462242562929062, | |
| "grad_norm": 5.628190517425537, | |
| "learning_rate": 6.129336734693878e-06, | |
| "loss": 0.1632, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.4576659038901602, | |
| "grad_norm": 14.407511711120605, | |
| "learning_rate": 6.1229591836734695e-06, | |
| "loss": 0.1877, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.4691075514874142, | |
| "grad_norm": 4.591071128845215, | |
| "learning_rate": 6.116581632653062e-06, | |
| "loss": 0.1698, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.4805491990846682, | |
| "grad_norm": 10.605140686035156, | |
| "learning_rate": 6.110204081632653e-06, | |
| "loss": 0.1764, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.4919908466819222, | |
| "grad_norm": 9.22734260559082, | |
| "learning_rate": 6.103826530612245e-06, | |
| "loss": 0.1632, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.5034324942791762, | |
| "grad_norm": 13.262040138244629, | |
| "learning_rate": 6.0974489795918365e-06, | |
| "loss": 0.1734, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.5148741418764302, | |
| "grad_norm": 9.931065559387207, | |
| "learning_rate": 6.091071428571429e-06, | |
| "loss": 0.1628, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.5263157894736842, | |
| "grad_norm": 17.013771057128906, | |
| "learning_rate": 6.08469387755102e-06, | |
| "loss": 0.1902, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.5377574370709383, | |
| "grad_norm": 4.440840244293213, | |
| "learning_rate": 6.078316326530613e-06, | |
| "loss": 0.1729, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.5491990846681922, | |
| "grad_norm": 6.86649751663208, | |
| "learning_rate": 6.071938775510204e-06, | |
| "loss": 0.1728, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.5606407322654462, | |
| "grad_norm": 5.384552478790283, | |
| "learning_rate": 6.065561224489797e-06, | |
| "loss": 0.1661, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.5720823798627003, | |
| "grad_norm": 16.237516403198242, | |
| "learning_rate": 6.059183673469388e-06, | |
| "loss": 0.1866, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.5835240274599542, | |
| "grad_norm": 9.271600723266602, | |
| "learning_rate": 6.05280612244898e-06, | |
| "loss": 0.1572, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.5949656750572082, | |
| "grad_norm": 12.505192756652832, | |
| "learning_rate": 6.046428571428571e-06, | |
| "loss": 0.1793, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.6064073226544623, | |
| "grad_norm": 4.431430816650391, | |
| "learning_rate": 6.040051020408164e-06, | |
| "loss": 0.1742, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.6178489702517163, | |
| "grad_norm": 13.353468894958496, | |
| "learning_rate": 6.033673469387755e-06, | |
| "loss": 0.1813, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.6292906178489702, | |
| "grad_norm": 10.686460494995117, | |
| "learning_rate": 6.027295918367347e-06, | |
| "loss": 0.1789, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.6407322654462243, | |
| "grad_norm": 8.10379409790039, | |
| "learning_rate": 6.020918367346939e-06, | |
| "loss": 0.1584, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.6521739130434783, | |
| "grad_norm": 5.418963432312012, | |
| "learning_rate": 6.014540816326531e-06, | |
| "loss": 0.1668, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.6636155606407322, | |
| "grad_norm": 12.503841400146484, | |
| "learning_rate": 6.008163265306123e-06, | |
| "loss": 0.1827, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.6750572082379863, | |
| "grad_norm": 6.517795562744141, | |
| "learning_rate": 6.001785714285715e-06, | |
| "loss": 0.171, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.6864988558352403, | |
| "grad_norm": 13.4345703125, | |
| "learning_rate": 5.995408163265306e-06, | |
| "loss": 0.1854, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.6979405034324943, | |
| "grad_norm": 5.514252185821533, | |
| "learning_rate": 5.989030612244898e-06, | |
| "loss": 0.1751, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.7093821510297483, | |
| "grad_norm": 16.108661651611328, | |
| "learning_rate": 5.98265306122449e-06, | |
| "loss": 0.2066, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.7208237986270023, | |
| "grad_norm": 8.58029842376709, | |
| "learning_rate": 5.976275510204082e-06, | |
| "loss": 0.1787, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.7322654462242563, | |
| "grad_norm": 10.38893985748291, | |
| "learning_rate": 5.969897959183673e-06, | |
| "loss": 0.1831, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.7437070938215103, | |
| "grad_norm": 6.40317964553833, | |
| "learning_rate": 5.963520408163266e-06, | |
| "loss": 0.177, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.7551487414187643, | |
| "grad_norm": 14.541539192199707, | |
| "learning_rate": 5.957142857142858e-06, | |
| "loss": 0.1835, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.7665903890160183, | |
| "grad_norm": 7.053554058074951, | |
| "learning_rate": 5.95076530612245e-06, | |
| "loss": 0.158, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.7780320366132724, | |
| "grad_norm": 12.54846477508545, | |
| "learning_rate": 5.944387755102041e-06, | |
| "loss": 0.1742, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.7894736842105263, | |
| "grad_norm": 5.44398307800293, | |
| "learning_rate": 5.938010204081633e-06, | |
| "loss": 0.15, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.8009153318077803, | |
| "grad_norm": 19.205368041992188, | |
| "learning_rate": 5.931632653061225e-06, | |
| "loss": 0.1578, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.8123569794050344, | |
| "grad_norm": 5.504051685333252, | |
| "learning_rate": 5.925255102040817e-06, | |
| "loss": 0.1675, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.8237986270022883, | |
| "grad_norm": 11.270848274230957, | |
| "learning_rate": 5.918877551020408e-06, | |
| "loss": 0.1861, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.8352402745995423, | |
| "grad_norm": 4.46786642074585, | |
| "learning_rate": 5.9125e-06, | |
| "loss": 0.1471, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.8466819221967964, | |
| "grad_norm": 16.594099044799805, | |
| "learning_rate": 5.906122448979592e-06, | |
| "loss": 0.2044, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.8581235697940504, | |
| "grad_norm": 5.97224760055542, | |
| "learning_rate": 5.8997448979591845e-06, | |
| "loss": 0.1671, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.8695652173913043, | |
| "grad_norm": 18.787992477416992, | |
| "learning_rate": 5.893367346938776e-06, | |
| "loss": 0.1896, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.8810068649885584, | |
| "grad_norm": 5.963818550109863, | |
| "learning_rate": 5.886989795918368e-06, | |
| "loss": 0.1563, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.8924485125858124, | |
| "grad_norm": 10.634485244750977, | |
| "learning_rate": 5.880612244897959e-06, | |
| "loss": 0.1709, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.9038901601830663, | |
| "grad_norm": 3.953845739364624, | |
| "learning_rate": 5.8742346938775515e-06, | |
| "loss": 0.1466, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.9153318077803204, | |
| "grad_norm": 10.291520118713379, | |
| "learning_rate": 5.867857142857143e-06, | |
| "loss": 0.1995, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.9267734553775744, | |
| "grad_norm": 7.01460599899292, | |
| "learning_rate": 5.861479591836735e-06, | |
| "loss": 0.1554, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.9382151029748284, | |
| "grad_norm": 9.408185005187988, | |
| "learning_rate": 5.855357142857143e-06, | |
| "loss": 0.2045, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.9496567505720824, | |
| "grad_norm": 7.104637622833252, | |
| "learning_rate": 5.848979591836735e-06, | |
| "loss": 0.1728, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.9610983981693364, | |
| "grad_norm": 16.595233917236328, | |
| "learning_rate": 5.842602040816327e-06, | |
| "loss": 0.2121, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.9725400457665904, | |
| "grad_norm": 3.5816707611083984, | |
| "learning_rate": 5.836224489795919e-06, | |
| "loss": 0.1576, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.9839816933638444, | |
| "grad_norm": 16.117435455322266, | |
| "learning_rate": 5.829846938775511e-06, | |
| "loss": 0.1948, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.9954233409610984, | |
| "grad_norm": 7.5333251953125, | |
| "learning_rate": 5.823469387755102e-06, | |
| "loss": 0.1679, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.0068649885583525, | |
| "grad_norm": 1.987963080406189, | |
| "learning_rate": 5.8170918367346946e-06, | |
| "loss": 0.1037, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.0183066361556063, | |
| "grad_norm": 5.73256778717041, | |
| "learning_rate": 5.810714285714286e-06, | |
| "loss": 0.0692, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.0297482837528604, | |
| "grad_norm": 3.2069363594055176, | |
| "learning_rate": 5.804336734693878e-06, | |
| "loss": 0.0989, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.0411899313501145, | |
| "grad_norm": 5.0149946212768555, | |
| "learning_rate": 5.797959183673469e-06, | |
| "loss": 0.0731, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.0526315789473684, | |
| "grad_norm": 5.605563163757324, | |
| "learning_rate": 5.791581632653062e-06, | |
| "loss": 0.091, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.0640732265446224, | |
| "grad_norm": 4.094449043273926, | |
| "learning_rate": 5.785204081632653e-06, | |
| "loss": 0.0617, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.0755148741418765, | |
| "grad_norm": 3.5027787685394287, | |
| "learning_rate": 5.7788265306122455e-06, | |
| "loss": 0.0905, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.0869565217391304, | |
| "grad_norm": 3.6455764770507812, | |
| "learning_rate": 5.772448979591837e-06, | |
| "loss": 0.0597, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.0983981693363845, | |
| "grad_norm": 3.261543035507202, | |
| "learning_rate": 5.766071428571429e-06, | |
| "loss": 0.097, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.1098398169336385, | |
| "grad_norm": 5.9047698974609375, | |
| "learning_rate": 5.759693877551021e-06, | |
| "loss": 0.0803, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.1212814645308924, | |
| "grad_norm": 1.9865431785583496, | |
| "learning_rate": 5.7533163265306125e-06, | |
| "loss": 0.0913, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.1327231121281465, | |
| "grad_norm": 5.4243316650390625, | |
| "learning_rate": 5.746938775510204e-06, | |
| "loss": 0.0652, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.1441647597254005, | |
| "grad_norm": 5.0221452713012695, | |
| "learning_rate": 5.7405612244897965e-06, | |
| "loss": 0.0825, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.1556064073226544, | |
| "grad_norm": 6.293609619140625, | |
| "learning_rate": 5.734183673469388e-06, | |
| "loss": 0.0819, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.1670480549199085, | |
| "grad_norm": 4.149010181427002, | |
| "learning_rate": 5.7278061224489796e-06, | |
| "loss": 0.1188, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.1784897025171626, | |
| "grad_norm": 2.6588358879089355, | |
| "learning_rate": 5.721428571428572e-06, | |
| "loss": 0.0723, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.1899313501144164, | |
| "grad_norm": 3.407505750656128, | |
| "learning_rate": 5.7150510204081635e-06, | |
| "loss": 0.0948, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.2013729977116705, | |
| "grad_norm": 6.228948593139648, | |
| "learning_rate": 5.708673469387756e-06, | |
| "loss": 0.0777, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.2128146453089246, | |
| "grad_norm": 2.339284896850586, | |
| "learning_rate": 5.7022959183673474e-06, | |
| "loss": 0.0896, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.2242562929061784, | |
| "grad_norm": 7.423865795135498, | |
| "learning_rate": 5.695918367346939e-06, | |
| "loss": 0.0721, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.2356979405034325, | |
| "grad_norm": 4.202932834625244, | |
| "learning_rate": 5.6895408163265305e-06, | |
| "loss": 0.096, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.2471395881006866, | |
| "grad_norm": 5.0601725578308105, | |
| "learning_rate": 5.683163265306123e-06, | |
| "loss": 0.0829, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.2585812356979404, | |
| "grad_norm": 4.4953460693359375, | |
| "learning_rate": 5.6767857142857144e-06, | |
| "loss": 0.1027, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.2700228832951945, | |
| "grad_norm": 6.2997612953186035, | |
| "learning_rate": 5.670408163265306e-06, | |
| "loss": 0.0789, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.2814645308924484, | |
| "grad_norm": 2.9170215129852295, | |
| "learning_rate": 5.664030612244898e-06, | |
| "loss": 0.1007, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.2929061784897025, | |
| "grad_norm": 6.78530740737915, | |
| "learning_rate": 5.65765306122449e-06, | |
| "loss": 0.0787, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.3043478260869565, | |
| "grad_norm": 3.2991139888763428, | |
| "learning_rate": 5.651275510204082e-06, | |
| "loss": 0.1118, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.3157894736842106, | |
| "grad_norm": 8.449702262878418, | |
| "learning_rate": 5.644897959183674e-06, | |
| "loss": 0.0742, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.3272311212814645, | |
| "grad_norm": 5.295764923095703, | |
| "learning_rate": 5.638520408163265e-06, | |
| "loss": 0.0951, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.3386727688787186, | |
| "grad_norm": 5.989330291748047, | |
| "learning_rate": 5.632142857142857e-06, | |
| "loss": 0.0758, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.3501144164759724, | |
| "grad_norm": 5.153733253479004, | |
| "learning_rate": 5.625765306122449e-06, | |
| "loss": 0.0984, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.3615560640732265, | |
| "grad_norm": 6.334990501403809, | |
| "learning_rate": 5.619387755102041e-06, | |
| "loss": 0.0678, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.3729977116704806, | |
| "grad_norm": 4.231469631195068, | |
| "learning_rate": 5.613010204081632e-06, | |
| "loss": 0.098, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.3844393592677346, | |
| "grad_norm": 4.833641529083252, | |
| "learning_rate": 5.606632653061225e-06, | |
| "loss": 0.0624, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.3958810068649885, | |
| "grad_norm": 3.244565725326538, | |
| "learning_rate": 5.600255102040817e-06, | |
| "loss": 0.0894, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.4073226544622426, | |
| "grad_norm": 1.8014638423919678, | |
| "learning_rate": 5.593877551020409e-06, | |
| "loss": 0.0709, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.4187643020594964, | |
| "grad_norm": 4.825936794281006, | |
| "learning_rate": 5.5875e-06, | |
| "loss": 0.1008, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.4302059496567505, | |
| "grad_norm": 5.295871734619141, | |
| "learning_rate": 5.581122448979592e-06, | |
| "loss": 0.0663, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.4416475972540046, | |
| "grad_norm": 4.34277868270874, | |
| "learning_rate": 5.574744897959184e-06, | |
| "loss": 0.0854, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.4530892448512587, | |
| "grad_norm": 6.447768688201904, | |
| "learning_rate": 5.568367346938776e-06, | |
| "loss": 0.0813, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.4645308924485125, | |
| "grad_norm": 3.3338069915771484, | |
| "learning_rate": 5.561989795918367e-06, | |
| "loss": 0.0921, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.4759725400457666, | |
| "grad_norm": 4.244719505310059, | |
| "learning_rate": 5.555612244897959e-06, | |
| "loss": 0.0683, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.4874141876430205, | |
| "grad_norm": 3.7263808250427246, | |
| "learning_rate": 5.549234693877551e-06, | |
| "loss": 0.095, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.4988558352402745, | |
| "grad_norm": 6.328766822814941, | |
| "learning_rate": 5.542857142857144e-06, | |
| "loss": 0.0837, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.5102974828375286, | |
| "grad_norm": 2.921656370162964, | |
| "learning_rate": 5.536479591836735e-06, | |
| "loss": 0.1021, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.5217391304347827, | |
| "grad_norm": 8.238923072814941, | |
| "learning_rate": 5.530102040816327e-06, | |
| "loss": 0.0697, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.5331807780320366, | |
| "grad_norm": 3.801389455795288, | |
| "learning_rate": 5.523724489795918e-06, | |
| "loss": 0.0995, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.5446224256292906, | |
| "grad_norm": 6.373222351074219, | |
| "learning_rate": 5.517346938775511e-06, | |
| "loss": 0.0693, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.5560640732265445, | |
| "grad_norm": 5.936155319213867, | |
| "learning_rate": 5.510969387755102e-06, | |
| "loss": 0.0989, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.5675057208237986, | |
| "grad_norm": 3.9559526443481445, | |
| "learning_rate": 5.504591836734694e-06, | |
| "loss": 0.0799, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.5789473684210527, | |
| "grad_norm": 2.2364418506622314, | |
| "learning_rate": 5.498214285714285e-06, | |
| "loss": 0.1069, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.5903890160183067, | |
| "grad_norm": 5.343243598937988, | |
| "learning_rate": 5.4918367346938785e-06, | |
| "loss": 0.0711, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.6018306636155606, | |
| "grad_norm": 6.238850116729736, | |
| "learning_rate": 5.48545918367347e-06, | |
| "loss": 0.0833, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.6132723112128147, | |
| "grad_norm": 7.247325897216797, | |
| "learning_rate": 5.479081632653062e-06, | |
| "loss": 0.0772, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.6247139588100685, | |
| "grad_norm": 5.513607025146484, | |
| "learning_rate": 5.472704081632653e-06, | |
| "loss": 0.102, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.6361556064073226, | |
| "grad_norm": 4.2667365074157715, | |
| "learning_rate": 5.4663265306122455e-06, | |
| "loss": 0.084, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.6475972540045767, | |
| "grad_norm": 3.866903066635132, | |
| "learning_rate": 5.459948979591837e-06, | |
| "loss": 0.0984, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.6590389016018308, | |
| "grad_norm": 6.93393611907959, | |
| "learning_rate": 5.453571428571429e-06, | |
| "loss": 0.0793, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.6704805491990846, | |
| "grad_norm": 5.100342750549316, | |
| "learning_rate": 5.44719387755102e-06, | |
| "loss": 0.0999, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.6819221967963387, | |
| "grad_norm": 8.363016128540039, | |
| "learning_rate": 5.4408163265306126e-06, | |
| "loss": 0.0709, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.6933638443935926, | |
| "grad_norm": 4.775864124298096, | |
| "learning_rate": 5.434438775510204e-06, | |
| "loss": 0.1039, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.7048054919908466, | |
| "grad_norm": 6.310733795166016, | |
| "learning_rate": 5.4280612244897965e-06, | |
| "loss": 0.08, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.7162471395881007, | |
| "grad_norm": 3.2022969722747803, | |
| "learning_rate": 5.421683673469388e-06, | |
| "loss": 0.0969, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.7276887871853548, | |
| "grad_norm": 1.5878781080245972, | |
| "learning_rate": 5.41530612244898e-06, | |
| "loss": 0.0807, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.7391304347826086, | |
| "grad_norm": 4.442117691040039, | |
| "learning_rate": 5.408928571428572e-06, | |
| "loss": 0.0978, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.7505720823798627, | |
| "grad_norm": 6.980683326721191, | |
| "learning_rate": 5.4025510204081635e-06, | |
| "loss": 0.0804, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.7620137299771166, | |
| "grad_norm": 1.5422896146774292, | |
| "learning_rate": 5.396173469387755e-06, | |
| "loss": 0.0971, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.7734553775743707, | |
| "grad_norm": 5.739109516143799, | |
| "learning_rate": 5.389795918367347e-06, | |
| "loss": 0.0946, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.7848970251716247, | |
| "grad_norm": 3.076174020767212, | |
| "learning_rate": 5.383418367346939e-06, | |
| "loss": 0.1183, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.7963386727688788, | |
| "grad_norm": 7.785946846008301, | |
| "learning_rate": 5.3770408163265305e-06, | |
| "loss": 0.0806, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.8077803203661327, | |
| "grad_norm": 3.3288745880126953, | |
| "learning_rate": 5.370663265306123e-06, | |
| "loss": 0.0924, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.8192219679633868, | |
| "grad_norm": 3.700629472732544, | |
| "learning_rate": 5.3642857142857145e-06, | |
| "loss": 0.0827, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.8306636155606406, | |
| "grad_norm": 5.729650020599365, | |
| "learning_rate": 5.357908163265307e-06, | |
| "loss": 0.1036, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.8421052631578947, | |
| "grad_norm": 6.0177202224731445, | |
| "learning_rate": 5.351530612244898e-06, | |
| "loss": 0.085, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.8535469107551488, | |
| "grad_norm": 4.52460241317749, | |
| "learning_rate": 5.34515306122449e-06, | |
| "loss": 0.0985, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.8649885583524028, | |
| "grad_norm": 11.662372589111328, | |
| "learning_rate": 5.3387755102040815e-06, | |
| "loss": 0.0845, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.8764302059496567, | |
| "grad_norm": 5.090856552124023, | |
| "learning_rate": 5.332397959183674e-06, | |
| "loss": 0.0884, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.8878718535469108, | |
| "grad_norm": 6.701780319213867, | |
| "learning_rate": 5.3260204081632654e-06, | |
| "loss": 0.0708, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.8993135011441646, | |
| "grad_norm": 2.6517765522003174, | |
| "learning_rate": 5.319897959183674e-06, | |
| "loss": 0.0956, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.9107551487414187, | |
| "grad_norm": 4.654801368713379, | |
| "learning_rate": 5.313520408163265e-06, | |
| "loss": 0.073, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.9221967963386728, | |
| "grad_norm": 4.046535968780518, | |
| "learning_rate": 5.307142857142857e-06, | |
| "loss": 0.0984, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.9336384439359269, | |
| "grad_norm": 5.534332752227783, | |
| "learning_rate": 5.30076530612245e-06, | |
| "loss": 0.0764, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.9450800915331807, | |
| "grad_norm": 4.218522071838379, | |
| "learning_rate": 5.2943877551020414e-06, | |
| "loss": 0.0941, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.9565217391304348, | |
| "grad_norm": 5.872758865356445, | |
| "learning_rate": 5.288010204081633e-06, | |
| "loss": 0.069, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.9679633867276887, | |
| "grad_norm": 4.280802249908447, | |
| "learning_rate": 5.2816326530612245e-06, | |
| "loss": 0.0986, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.9794050343249427, | |
| "grad_norm": 6.042983531951904, | |
| "learning_rate": 5.275255102040817e-06, | |
| "loss": 0.0711, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.9908466819221968, | |
| "grad_norm": 3.573798179626465, | |
| "learning_rate": 5.2688775510204085e-06, | |
| "loss": 0.0891, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.002288329519451, | |
| "grad_norm": 2.0011675357818604, | |
| "learning_rate": 5.2625e-06, | |
| "loss": 0.0834, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 2.013729977116705, | |
| "grad_norm": 0.759939432144165, | |
| "learning_rate": 5.2561224489795916e-06, | |
| "loss": 0.037, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.0251716247139586, | |
| "grad_norm": 3.735200881958008, | |
| "learning_rate": 5.249744897959184e-06, | |
| "loss": 0.0448, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 2.0366132723112127, | |
| "grad_norm": 2.4551169872283936, | |
| "learning_rate": 5.243367346938776e-06, | |
| "loss": 0.038, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.0480549199084668, | |
| "grad_norm": 1.7801021337509155, | |
| "learning_rate": 5.236989795918368e-06, | |
| "loss": 0.05, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 2.059496567505721, | |
| "grad_norm": 2.692507743835449, | |
| "learning_rate": 5.2306122448979594e-06, | |
| "loss": 0.042, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.070938215102975, | |
| "grad_norm": 2.6997761726379395, | |
| "learning_rate": 5.224234693877551e-06, | |
| "loss": 0.0459, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 2.082379862700229, | |
| "grad_norm": 3.892010450363159, | |
| "learning_rate": 5.217857142857143e-06, | |
| "loss": 0.0364, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 2.0938215102974826, | |
| "grad_norm": 3.354294776916504, | |
| "learning_rate": 5.211479591836735e-06, | |
| "loss": 0.0584, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 2.1052631578947367, | |
| "grad_norm": 3.581019163131714, | |
| "learning_rate": 5.2051020408163264e-06, | |
| "loss": 0.035, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 2.116704805491991, | |
| "grad_norm": 1.7799072265625, | |
| "learning_rate": 5.198724489795918e-06, | |
| "loss": 0.0534, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 2.128146453089245, | |
| "grad_norm": 4.093613624572754, | |
| "learning_rate": 5.19234693877551e-06, | |
| "loss": 0.0299, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 2.139588100686499, | |
| "grad_norm": 4.328442573547363, | |
| "learning_rate": 5.185969387755103e-06, | |
| "loss": 0.0504, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 2.151029748283753, | |
| "grad_norm": 4.18818998336792, | |
| "learning_rate": 5.179591836734694e-06, | |
| "loss": 0.0371, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.1624713958810067, | |
| "grad_norm": 4.5619587898254395, | |
| "learning_rate": 5.173214285714286e-06, | |
| "loss": 0.0514, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 2.1739130434782608, | |
| "grad_norm": 2.6793291568756104, | |
| "learning_rate": 5.166836734693878e-06, | |
| "loss": 0.0415, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.185354691075515, | |
| "grad_norm": 3.8424887657165527, | |
| "learning_rate": 5.16045918367347e-06, | |
| "loss": 0.0584, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 2.196796338672769, | |
| "grad_norm": 1.9429773092269897, | |
| "learning_rate": 5.154081632653061e-06, | |
| "loss": 0.0361, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.208237986270023, | |
| "grad_norm": 2.7046594619750977, | |
| "learning_rate": 5.147704081632653e-06, | |
| "loss": 0.0501, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 2.219679633867277, | |
| "grad_norm": 0.6767886281013489, | |
| "learning_rate": 5.141326530612245e-06, | |
| "loss": 0.0284, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.2311212814645307, | |
| "grad_norm": 1.749976634979248, | |
| "learning_rate": 5.134948979591837e-06, | |
| "loss": 0.0641, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 2.242562929061785, | |
| "grad_norm": 5.310193061828613, | |
| "learning_rate": 5.128571428571429e-06, | |
| "loss": 0.0392, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.254004576659039, | |
| "grad_norm": 3.5149447917938232, | |
| "learning_rate": 5.122193877551021e-06, | |
| "loss": 0.064, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 2.265446224256293, | |
| "grad_norm": 4.282719135284424, | |
| "learning_rate": 5.115816326530612e-06, | |
| "loss": 0.0433, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.276887871853547, | |
| "grad_norm": 2.836113929748535, | |
| "learning_rate": 5.109438775510205e-06, | |
| "loss": 0.0547, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 2.288329519450801, | |
| "grad_norm": 3.2092835903167725, | |
| "learning_rate": 5.103061224489796e-06, | |
| "loss": 0.0339, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.288329519450801, | |
| "eval_loss": 0.16944891214370728, | |
| "eval_runtime": 5496.3202, | |
| "eval_samples_per_second": 1.732, | |
| "eval_steps_per_second": 0.217, | |
| "eval_wer": 0.10248288219107954, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.2997711670480547, | |
| "grad_norm": 2.923719882965088, | |
| "learning_rate": 5.096683673469388e-06, | |
| "loss": 0.063, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 2.311212814645309, | |
| "grad_norm": 3.116055727005005, | |
| "learning_rate": 5.090306122448979e-06, | |
| "loss": 0.0394, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 2.322654462242563, | |
| "grad_norm": 2.1266860961914062, | |
| "learning_rate": 5.083928571428572e-06, | |
| "loss": 0.0497, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 2.334096109839817, | |
| "grad_norm": 3.964972972869873, | |
| "learning_rate": 5.077551020408163e-06, | |
| "loss": 0.0355, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 2.345537757437071, | |
| "grad_norm": 3.5802409648895264, | |
| "learning_rate": 5.071173469387756e-06, | |
| "loss": 0.0483, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 2.356979405034325, | |
| "grad_norm": 4.559951305389404, | |
| "learning_rate": 5.064795918367347e-06, | |
| "loss": 0.0365, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 2.3684210526315788, | |
| "grad_norm": 2.1596357822418213, | |
| "learning_rate": 5.0584183673469396e-06, | |
| "loss": 0.0507, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 2.379862700228833, | |
| "grad_norm": 6.24343204498291, | |
| "learning_rate": 5.052040816326531e-06, | |
| "loss": 0.0394, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 2.391304347826087, | |
| "grad_norm": 3.8776490688323975, | |
| "learning_rate": 5.045663265306123e-06, | |
| "loss": 0.0519, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 2.402745995423341, | |
| "grad_norm": 5.080620288848877, | |
| "learning_rate": 5.039285714285714e-06, | |
| "loss": 0.0372, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 2.414187643020595, | |
| "grad_norm": 2.12448787689209, | |
| "learning_rate": 5.032908163265307e-06, | |
| "loss": 0.0578, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 2.425629290617849, | |
| "grad_norm": 2.113422393798828, | |
| "learning_rate": 5.026530612244898e-06, | |
| "loss": 0.0381, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 2.437070938215103, | |
| "grad_norm": 3.075899124145508, | |
| "learning_rate": 5.02015306122449e-06, | |
| "loss": 0.051, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 2.448512585812357, | |
| "grad_norm": 4.633288860321045, | |
| "learning_rate": 5.013775510204082e-06, | |
| "loss": 0.0477, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 2.459954233409611, | |
| "grad_norm": 3.1297614574432373, | |
| "learning_rate": 5.007397959183674e-06, | |
| "loss": 0.0518, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 2.471395881006865, | |
| "grad_norm": 1.0485554933547974, | |
| "learning_rate": 5.001020408163266e-06, | |
| "loss": 0.047, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 2.482837528604119, | |
| "grad_norm": 2.6167232990264893, | |
| "learning_rate": 4.9946428571428575e-06, | |
| "loss": 0.0515, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 2.494279176201373, | |
| "grad_norm": 3.787649393081665, | |
| "learning_rate": 4.988265306122449e-06, | |
| "loss": 0.0406, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 2.505720823798627, | |
| "grad_norm": 3.5073530673980713, | |
| "learning_rate": 4.981887755102041e-06, | |
| "loss": 0.0451, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 2.517162471395881, | |
| "grad_norm": 3.0327556133270264, | |
| "learning_rate": 4.975510204081633e-06, | |
| "loss": 0.0381, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.528604118993135, | |
| "grad_norm": 4.393261432647705, | |
| "learning_rate": 4.9691326530612246e-06, | |
| "loss": 0.0505, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 2.540045766590389, | |
| "grad_norm": 1.906299114227295, | |
| "learning_rate": 4.962755102040816e-06, | |
| "loss": 0.0406, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 2.551487414187643, | |
| "grad_norm": 3.095064401626587, | |
| "learning_rate": 4.9563775510204085e-06, | |
| "loss": 0.0545, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 2.5629290617848968, | |
| "grad_norm": 2.1432371139526367, | |
| "learning_rate": 4.950000000000001e-06, | |
| "loss": 0.038, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 2.5743707093821513, | |
| "grad_norm": 2.254021167755127, | |
| "learning_rate": 4.9436224489795924e-06, | |
| "loss": 0.05, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 2.585812356979405, | |
| "grad_norm": 4.11069917678833, | |
| "learning_rate": 4.937244897959184e-06, | |
| "loss": 0.0484, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 2.597254004576659, | |
| "grad_norm": 2.1790149211883545, | |
| "learning_rate": 4.9308673469387755e-06, | |
| "loss": 0.0584, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 2.608695652173913, | |
| "grad_norm": 2.150855541229248, | |
| "learning_rate": 4.924489795918368e-06, | |
| "loss": 0.0355, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 2.620137299771167, | |
| "grad_norm": 2.5849735736846924, | |
| "learning_rate": 4.9181122448979595e-06, | |
| "loss": 0.0486, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 2.6315789473684212, | |
| "grad_norm": 3.598619222640991, | |
| "learning_rate": 4.911734693877551e-06, | |
| "loss": 0.043, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 2.643020594965675, | |
| "grad_norm": 3.094461441040039, | |
| "learning_rate": 4.9053571428571425e-06, | |
| "loss": 0.0497, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 2.654462242562929, | |
| "grad_norm": 6.460962772369385, | |
| "learning_rate": 4.898979591836735e-06, | |
| "loss": 0.0369, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 2.665903890160183, | |
| "grad_norm": 2.5654125213623047, | |
| "learning_rate": 4.892602040816327e-06, | |
| "loss": 0.0517, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 2.677345537757437, | |
| "grad_norm": 1.984237790107727, | |
| "learning_rate": 4.886224489795919e-06, | |
| "loss": 0.0374, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 2.688787185354691, | |
| "grad_norm": 2.4424266815185547, | |
| "learning_rate": 4.87984693877551e-06, | |
| "loss": 0.0586, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 2.700228832951945, | |
| "grad_norm": 3.130814552307129, | |
| "learning_rate": 4.873469387755102e-06, | |
| "loss": 0.0379, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 2.7116704805491993, | |
| "grad_norm": 1.8555734157562256, | |
| "learning_rate": 4.867091836734694e-06, | |
| "loss": 0.0472, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 2.723112128146453, | |
| "grad_norm": 2.3630707263946533, | |
| "learning_rate": 4.860714285714286e-06, | |
| "loss": 0.0412, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 2.734553775743707, | |
| "grad_norm": 2.636423110961914, | |
| "learning_rate": 4.8543367346938774e-06, | |
| "loss": 0.0509, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 2.745995423340961, | |
| "grad_norm": 6.911414623260498, | |
| "learning_rate": 4.847959183673469e-06, | |
| "loss": 0.0399, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.757437070938215, | |
| "grad_norm": 3.120143413543701, | |
| "learning_rate": 4.841581632653061e-06, | |
| "loss": 0.0517, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 2.7688787185354693, | |
| "grad_norm": 2.109907627105713, | |
| "learning_rate": 4.835204081632654e-06, | |
| "loss": 0.0299, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 2.780320366132723, | |
| "grad_norm": 3.6579062938690186, | |
| "learning_rate": 4.828826530612245e-06, | |
| "loss": 0.0494, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 2.791762013729977, | |
| "grad_norm": 0.9419932961463928, | |
| "learning_rate": 4.822448979591837e-06, | |
| "loss": 0.0296, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 2.803203661327231, | |
| "grad_norm": 1.7163337469100952, | |
| "learning_rate": 4.816071428571429e-06, | |
| "loss": 0.0512, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 2.814645308924485, | |
| "grad_norm": 3.6414005756378174, | |
| "learning_rate": 4.809693877551021e-06, | |
| "loss": 0.0516, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 2.8260869565217392, | |
| "grad_norm": 2.7387633323669434, | |
| "learning_rate": 4.803571428571429e-06, | |
| "loss": 0.0491, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 2.837528604118993, | |
| "grad_norm": 6.319782733917236, | |
| "learning_rate": 4.7971938775510205e-06, | |
| "loss": 0.0418, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 2.8489702517162474, | |
| "grad_norm": 3.232647180557251, | |
| "learning_rate": 4.790816326530612e-06, | |
| "loss": 0.0492, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 2.860411899313501, | |
| "grad_norm": 2.536766767501831, | |
| "learning_rate": 4.784438775510204e-06, | |
| "loss": 0.0394, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 2.871853546910755, | |
| "grad_norm": 3.481765031814575, | |
| "learning_rate": 4.778061224489796e-06, | |
| "loss": 0.0563, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 2.883295194508009, | |
| "grad_norm": 2.599041700363159, | |
| "learning_rate": 4.771683673469388e-06, | |
| "loss": 0.041, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 2.8947368421052633, | |
| "grad_norm": 2.9180805683135986, | |
| "learning_rate": 4.76530612244898e-06, | |
| "loss": 0.0455, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 2.9061784897025174, | |
| "grad_norm": 5.283931255340576, | |
| "learning_rate": 4.758928571428571e-06, | |
| "loss": 0.045, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 2.917620137299771, | |
| "grad_norm": 3.000849485397339, | |
| "learning_rate": 4.752551020408164e-06, | |
| "loss": 0.0451, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 2.929061784897025, | |
| "grad_norm": 5.314061641693115, | |
| "learning_rate": 4.746173469387755e-06, | |
| "loss": 0.0401, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 2.940503432494279, | |
| "grad_norm": 4.263325214385986, | |
| "learning_rate": 4.739795918367347e-06, | |
| "loss": 0.0481, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 2.9519450800915332, | |
| "grad_norm": 3.5418529510498047, | |
| "learning_rate": 4.7334183673469384e-06, | |
| "loss": 0.0494, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 2.9633867276887873, | |
| "grad_norm": 3.0929176807403564, | |
| "learning_rate": 4.727040816326531e-06, | |
| "loss": 0.0583, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 2.974828375286041, | |
| "grad_norm": 3.1736650466918945, | |
| "learning_rate": 4.720663265306122e-06, | |
| "loss": 0.0377, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.9862700228832955, | |
| "grad_norm": 2.7548913955688477, | |
| "learning_rate": 4.714285714285715e-06, | |
| "loss": 0.0385, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 2.997711670480549, | |
| "grad_norm": 3.6645278930664062, | |
| "learning_rate": 4.707908163265306e-06, | |
| "loss": 0.0454, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 3.009153318077803, | |
| "grad_norm": 1.9902344942092896, | |
| "learning_rate": 4.701530612244899e-06, | |
| "loss": 0.0287, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 3.0205949656750573, | |
| "grad_norm": 7.351073265075684, | |
| "learning_rate": 4.69515306122449e-06, | |
| "loss": 0.0237, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 3.0320366132723113, | |
| "grad_norm": 3.1090986728668213, | |
| "learning_rate": 4.688775510204082e-06, | |
| "loss": 0.0232, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 3.0434782608695654, | |
| "grad_norm": 1.9417141675949097, | |
| "learning_rate": 4.682397959183673e-06, | |
| "loss": 0.0227, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 3.054919908466819, | |
| "grad_norm": 4.676495552062988, | |
| "learning_rate": 4.676020408163266e-06, | |
| "loss": 0.0369, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 3.066361556064073, | |
| "grad_norm": 1.6039929389953613, | |
| "learning_rate": 4.669642857142857e-06, | |
| "loss": 0.0184, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 3.077803203661327, | |
| "grad_norm": 1.7776933908462524, | |
| "learning_rate": 4.663265306122449e-06, | |
| "loss": 0.0244, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 3.0892448512585813, | |
| "grad_norm": 4.828967571258545, | |
| "learning_rate": 4.656887755102041e-06, | |
| "loss": 0.0312, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 3.1006864988558354, | |
| "grad_norm": 8.169689178466797, | |
| "learning_rate": 4.650510204081633e-06, | |
| "loss": 0.0328, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 3.1121281464530894, | |
| "grad_norm": 3.101473093032837, | |
| "learning_rate": 4.644132653061225e-06, | |
| "loss": 0.016, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 3.123569794050343, | |
| "grad_norm": 0.8072414398193359, | |
| "learning_rate": 4.637755102040817e-06, | |
| "loss": 0.0299, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 3.135011441647597, | |
| "grad_norm": 1.684613585472107, | |
| "learning_rate": 4.631377551020408e-06, | |
| "loss": 0.03, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 3.1464530892448512, | |
| "grad_norm": 4.120328426361084, | |
| "learning_rate": 4.625e-06, | |
| "loss": 0.0336, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 3.1578947368421053, | |
| "grad_norm": 3.2325098514556885, | |
| "learning_rate": 4.618622448979592e-06, | |
| "loss": 0.0197, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 3.1693363844393594, | |
| "grad_norm": 2.3901312351226807, | |
| "learning_rate": 4.612244897959184e-06, | |
| "loss": 0.0251, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 3.1807780320366135, | |
| "grad_norm": 3.9952516555786133, | |
| "learning_rate": 4.605867346938775e-06, | |
| "loss": 0.0322, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 3.192219679633867, | |
| "grad_norm": 3.828606605529785, | |
| "learning_rate": 4.599489795918368e-06, | |
| "loss": 0.0406, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 3.203661327231121, | |
| "grad_norm": 1.8031266927719116, | |
| "learning_rate": 4.59311224489796e-06, | |
| "loss": 0.0236, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 3.2151029748283753, | |
| "grad_norm": 3.989572286605835, | |
| "learning_rate": 4.5867346938775516e-06, | |
| "loss": 0.0289, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 3.2265446224256293, | |
| "grad_norm": 5.1198296546936035, | |
| "learning_rate": 4.580357142857143e-06, | |
| "loss": 0.0257, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 3.2379862700228834, | |
| "grad_norm": 3.9181671142578125, | |
| "learning_rate": 4.573979591836735e-06, | |
| "loss": 0.0344, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 3.2494279176201375, | |
| "grad_norm": 2.100790500640869, | |
| "learning_rate": 4.567602040816327e-06, | |
| "loss": 0.0182, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 3.260869565217391, | |
| "grad_norm": 1.8334282636642456, | |
| "learning_rate": 4.561224489795919e-06, | |
| "loss": 0.0287, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 3.272311212814645, | |
| "grad_norm": 4.7145771980285645, | |
| "learning_rate": 4.55484693877551e-06, | |
| "loss": 0.0339, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 3.2837528604118993, | |
| "grad_norm": 3.8231897354125977, | |
| "learning_rate": 4.548469387755102e-06, | |
| "loss": 0.0243, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 3.2951945080091534, | |
| "grad_norm": 5.662996768951416, | |
| "learning_rate": 4.542091836734694e-06, | |
| "loss": 0.0281, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 3.3066361556064074, | |
| "grad_norm": 1.8065590858459473, | |
| "learning_rate": 4.5357142857142865e-06, | |
| "loss": 0.0361, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 3.3180778032036615, | |
| "grad_norm": 4.327584266662598, | |
| "learning_rate": 4.529336734693878e-06, | |
| "loss": 0.0285, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 3.329519450800915, | |
| "grad_norm": 3.929110527038574, | |
| "learning_rate": 4.5229591836734695e-06, | |
| "loss": 0.029, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 3.3409610983981692, | |
| "grad_norm": 4.633998870849609, | |
| "learning_rate": 4.516581632653061e-06, | |
| "loss": 0.0277, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 3.3524027459954233, | |
| "grad_norm": 1.1148935556411743, | |
| "learning_rate": 4.5102040816326535e-06, | |
| "loss": 0.0319, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 3.3638443935926774, | |
| "grad_norm": 8.381105422973633, | |
| "learning_rate": 4.503826530612245e-06, | |
| "loss": 0.0214, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 3.3752860411899315, | |
| "grad_norm": 3.111279010772705, | |
| "learning_rate": 4.4974489795918366e-06, | |
| "loss": 0.0341, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 3.386727688787185, | |
| "grad_norm": 2.937441349029541, | |
| "learning_rate": 4.491071428571428e-06, | |
| "loss": 0.0228, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 3.398169336384439, | |
| "grad_norm": 2.1108193397521973, | |
| "learning_rate": 4.484693877551021e-06, | |
| "loss": 0.0276, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 3.4096109839816933, | |
| "grad_norm": 5.9128031730651855, | |
| "learning_rate": 4.478316326530613e-06, | |
| "loss": 0.0273, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 3.4210526315789473, | |
| "grad_norm": 0.7438346743583679, | |
| "learning_rate": 4.4719387755102044e-06, | |
| "loss": 0.0305, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 3.4324942791762014, | |
| "grad_norm": 7.93667459487915, | |
| "learning_rate": 4.465561224489796e-06, | |
| "loss": 0.0265, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 3.4439359267734555, | |
| "grad_norm": 4.7441277503967285, | |
| "learning_rate": 4.459183673469388e-06, | |
| "loss": 0.0285, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 3.4553775743707096, | |
| "grad_norm": 6.173753261566162, | |
| "learning_rate": 4.45280612244898e-06, | |
| "loss": 0.0341, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 3.466819221967963, | |
| "grad_norm": 1.397835373878479, | |
| "learning_rate": 4.4464285714285715e-06, | |
| "loss": 0.0304, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 3.4782608695652173, | |
| "grad_norm": 1.049957275390625, | |
| "learning_rate": 4.440051020408163e-06, | |
| "loss": 0.0315, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 3.4897025171624714, | |
| "grad_norm": 1.9728070497512817, | |
| "learning_rate": 4.433673469387755e-06, | |
| "loss": 0.0345, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 3.5011441647597255, | |
| "grad_norm": 3.764899492263794, | |
| "learning_rate": 4.427295918367347e-06, | |
| "loss": 0.0312, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 3.5125858123569795, | |
| "grad_norm": 3.4769232273101807, | |
| "learning_rate": 4.420918367346939e-06, | |
| "loss": 0.0314, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 3.524027459954233, | |
| "grad_norm": 2.4961886405944824, | |
| "learning_rate": 4.414540816326531e-06, | |
| "loss": 0.0269, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 3.5354691075514877, | |
| "grad_norm": 2.6600871086120605, | |
| "learning_rate": 4.408163265306122e-06, | |
| "loss": 0.0304, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 3.5469107551487413, | |
| "grad_norm": 3.568831205368042, | |
| "learning_rate": 4.401785714285715e-06, | |
| "loss": 0.026, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 3.5583524027459954, | |
| "grad_norm": 2.488873243331909, | |
| "learning_rate": 4.395408163265306e-06, | |
| "loss": 0.0319, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 3.5697940503432495, | |
| "grad_norm": 3.0144600868225098, | |
| "learning_rate": 4.389030612244898e-06, | |
| "loss": 0.0241, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 3.5812356979405036, | |
| "grad_norm": 5.78762149810791, | |
| "learning_rate": 4.3826530612244894e-06, | |
| "loss": 0.0335, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 3.5926773455377576, | |
| "grad_norm": 7.364430904388428, | |
| "learning_rate": 4.376275510204082e-06, | |
| "loss": 0.03, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 3.6041189931350113, | |
| "grad_norm": 1.7618129253387451, | |
| "learning_rate": 4.369897959183673e-06, | |
| "loss": 0.0269, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 3.6155606407322654, | |
| "grad_norm": 2.5900871753692627, | |
| "learning_rate": 4.363520408163266e-06, | |
| "loss": 0.021, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 3.6270022883295194, | |
| "grad_norm": 1.509843111038208, | |
| "learning_rate": 4.357142857142857e-06, | |
| "loss": 0.0355, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 3.6384439359267735, | |
| "grad_norm": 7.716503143310547, | |
| "learning_rate": 4.35076530612245e-06, | |
| "loss": 0.0257, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 3.6498855835240276, | |
| "grad_norm": 1.4650408029556274, | |
| "learning_rate": 4.344387755102041e-06, | |
| "loss": 0.0377, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 3.6613272311212812, | |
| "grad_norm": 2.744730234146118, | |
| "learning_rate": 4.338010204081633e-06, | |
| "loss": 0.0277, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 3.6727688787185357, | |
| "grad_norm": 3.6976919174194336, | |
| "learning_rate": 4.331632653061224e-06, | |
| "loss": 0.0282, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 3.6842105263157894, | |
| "grad_norm": 7.903183460235596, | |
| "learning_rate": 4.325255102040817e-06, | |
| "loss": 0.0324, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 3.6956521739130435, | |
| "grad_norm": 2.9782605171203613, | |
| "learning_rate": 4.318877551020408e-06, | |
| "loss": 0.0304, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 3.7070938215102975, | |
| "grad_norm": 3.6786653995513916, | |
| "learning_rate": 4.3125e-06, | |
| "loss": 0.0304, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 3.7185354691075516, | |
| "grad_norm": 2.000511884689331, | |
| "learning_rate": 4.306122448979592e-06, | |
| "loss": 0.0266, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 3.7299771167048057, | |
| "grad_norm": 3.0682075023651123, | |
| "learning_rate": 4.299744897959184e-06, | |
| "loss": 0.0314, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 3.7414187643020593, | |
| "grad_norm": 3.421633720397949, | |
| "learning_rate": 4.293367346938776e-06, | |
| "loss": 0.0296, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 3.7528604118993134, | |
| "grad_norm": 1.5098000764846802, | |
| "learning_rate": 4.286989795918368e-06, | |
| "loss": 0.0211, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 3.7643020594965675, | |
| "grad_norm": 1.048941731452942, | |
| "learning_rate": 4.280612244897959e-06, | |
| "loss": 0.032, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 3.7757437070938216, | |
| "grad_norm": 13.178288459777832, | |
| "learning_rate": 4.274234693877551e-06, | |
| "loss": 0.0266, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 3.7871853546910756, | |
| "grad_norm": 2.7297298908233643, | |
| "learning_rate": 4.267857142857143e-06, | |
| "loss": 0.0254, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 3.7986270022883293, | |
| "grad_norm": 5.409793376922607, | |
| "learning_rate": 4.261479591836735e-06, | |
| "loss": 0.0283, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 3.8100686498855834, | |
| "grad_norm": 1.6612489223480225, | |
| "learning_rate": 4.255102040816326e-06, | |
| "loss": 0.0263, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 3.8215102974828374, | |
| "grad_norm": 9.250696182250977, | |
| "learning_rate": 4.248724489795919e-06, | |
| "loss": 0.0265, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 3.8329519450800915, | |
| "grad_norm": 4.991941928863525, | |
| "learning_rate": 4.242346938775511e-06, | |
| "loss": 0.0314, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 3.8443935926773456, | |
| "grad_norm": 2.8890774250030518, | |
| "learning_rate": 4.2359693877551025e-06, | |
| "loss": 0.0222, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 3.8558352402745997, | |
| "grad_norm": 2.2660281658172607, | |
| "learning_rate": 4.229591836734694e-06, | |
| "loss": 0.0281, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 3.8672768878718538, | |
| "grad_norm": 2.7722203731536865, | |
| "learning_rate": 4.223214285714286e-06, | |
| "loss": 0.0315, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 3.8787185354691074, | |
| "grad_norm": 3.3059165477752686, | |
| "learning_rate": 4.216836734693878e-06, | |
| "loss": 0.0349, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 3.8901601830663615, | |
| "grad_norm": 4.720456600189209, | |
| "learning_rate": 4.2104591836734696e-06, | |
| "loss": 0.0171, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 3.9016018306636155, | |
| "grad_norm": 2.5770797729492188, | |
| "learning_rate": 4.204081632653061e-06, | |
| "loss": 0.0285, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 3.9130434782608696, | |
| "grad_norm": 7.210494041442871, | |
| "learning_rate": 4.197704081632653e-06, | |
| "loss": 0.0243, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 3.9244851258581237, | |
| "grad_norm": 0.5488982796669006, | |
| "learning_rate": 4.191326530612245e-06, | |
| "loss": 0.0293, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 3.9359267734553773, | |
| "grad_norm": 4.990447044372559, | |
| "learning_rate": 4.1849489795918374e-06, | |
| "loss": 0.0262, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 3.9473684210526314, | |
| "grad_norm": 4.837020397186279, | |
| "learning_rate": 4.178571428571429e-06, | |
| "loss": 0.0366, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 3.9588100686498855, | |
| "grad_norm": 8.747390747070312, | |
| "learning_rate": 4.1721938775510205e-06, | |
| "loss": 0.0228, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 3.9702517162471396, | |
| "grad_norm": 0.8226907849311829, | |
| "learning_rate": 4.166071428571429e-06, | |
| "loss": 0.0315, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 3.9816933638443937, | |
| "grad_norm": 4.007413864135742, | |
| "learning_rate": 4.159693877551021e-06, | |
| "loss": 0.0251, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 3.9931350114416477, | |
| "grad_norm": 2.9124577045440674, | |
| "learning_rate": 4.153316326530613e-06, | |
| "loss": 0.0287, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 4.004576659038902, | |
| "grad_norm": 2.2100229263305664, | |
| "learning_rate": 4.146938775510204e-06, | |
| "loss": 0.0337, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 4.016018306636155, | |
| "grad_norm": 1.504098653793335, | |
| "learning_rate": 4.140561224489796e-06, | |
| "loss": 0.0166, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 4.02745995423341, | |
| "grad_norm": 2.594133138656616, | |
| "learning_rate": 4.134183673469388e-06, | |
| "loss": 0.0193, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 4.038901601830664, | |
| "grad_norm": 3.2971105575561523, | |
| "learning_rate": 4.12780612244898e-06, | |
| "loss": 0.0116, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 4.050343249427917, | |
| "grad_norm": 2.0989925861358643, | |
| "learning_rate": 4.121428571428572e-06, | |
| "loss": 0.0199, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 4.061784897025172, | |
| "grad_norm": 2.077739715576172, | |
| "learning_rate": 4.115306122448979e-06, | |
| "loss": 0.0132, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 4.073226544622425, | |
| "grad_norm": 2.2226176261901855, | |
| "learning_rate": 4.108928571428572e-06, | |
| "loss": 0.0203, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 4.08466819221968, | |
| "grad_norm": 5.787581443786621, | |
| "learning_rate": 4.102551020408163e-06, | |
| "loss": 0.0194, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 4.0961098398169336, | |
| "grad_norm": 1.7698355913162231, | |
| "learning_rate": 4.096173469387756e-06, | |
| "loss": 0.0225, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 4.107551487414188, | |
| "grad_norm": 7.658657073974609, | |
| "learning_rate": 4.089795918367347e-06, | |
| "loss": 0.017, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 4.118993135011442, | |
| "grad_norm": 1.4201191663742065, | |
| "learning_rate": 4.083418367346939e-06, | |
| "loss": 0.025, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 4.130434782608695, | |
| "grad_norm": 1.703041434288025, | |
| "learning_rate": 4.07704081632653e-06, | |
| "loss": 0.0114, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 4.14187643020595, | |
| "grad_norm": 0.8880120515823364, | |
| "learning_rate": 4.070663265306123e-06, | |
| "loss": 0.0247, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 4.1533180778032035, | |
| "grad_norm": 2.5007543563842773, | |
| "learning_rate": 4.064285714285714e-06, | |
| "loss": 0.0221, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 4.164759725400458, | |
| "grad_norm": 9.973050117492676, | |
| "learning_rate": 4.057908163265306e-06, | |
| "loss": 0.0212, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 4.176201372997712, | |
| "grad_norm": 3.308204412460327, | |
| "learning_rate": 4.051530612244898e-06, | |
| "loss": 0.0117, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 4.187643020594965, | |
| "grad_norm": 1.699859380722046, | |
| "learning_rate": 4.0451530612244905e-06, | |
| "loss": 0.0227, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 4.19908466819222, | |
| "grad_norm": 1.368072271347046, | |
| "learning_rate": 4.038775510204082e-06, | |
| "loss": 0.0105, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 4.2105263157894735, | |
| "grad_norm": 0.6647465229034424, | |
| "learning_rate": 4.032397959183674e-06, | |
| "loss": 0.022, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 4.221967963386728, | |
| "grad_norm": 4.245569705963135, | |
| "learning_rate": 4.026020408163265e-06, | |
| "loss": 0.0151, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 4.233409610983982, | |
| "grad_norm": 0.7587606906890869, | |
| "learning_rate": 4.0196428571428576e-06, | |
| "loss": 0.0235, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 4.244851258581235, | |
| "grad_norm": 2.736632823944092, | |
| "learning_rate": 4.013265306122449e-06, | |
| "loss": 0.0157, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 4.25629290617849, | |
| "grad_norm": 2.0381782054901123, | |
| "learning_rate": 4.006887755102041e-06, | |
| "loss": 0.0188, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 4.267734553775743, | |
| "grad_norm": 1.4942119121551514, | |
| "learning_rate": 4.000510204081632e-06, | |
| "loss": 0.0101, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 4.279176201372998, | |
| "grad_norm": 3.967541217803955, | |
| "learning_rate": 3.994132653061225e-06, | |
| "loss": 0.021, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 4.290617848970252, | |
| "grad_norm": 0.7107511758804321, | |
| "learning_rate": 3.987755102040817e-06, | |
| "loss": 0.0106, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 4.302059496567506, | |
| "grad_norm": 1.9319146871566772, | |
| "learning_rate": 3.9813775510204085e-06, | |
| "loss": 0.0278, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 4.31350114416476, | |
| "grad_norm": 4.877601146697998, | |
| "learning_rate": 3.975e-06, | |
| "loss": 0.0157, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 4.324942791762013, | |
| "grad_norm": 3.034367799758911, | |
| "learning_rate": 3.968622448979592e-06, | |
| "loss": 0.0248, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 4.336384439359268, | |
| "grad_norm": 0.23764902353286743, | |
| "learning_rate": 3.962244897959184e-06, | |
| "loss": 0.0105, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 4.3478260869565215, | |
| "grad_norm": 3.390343189239502, | |
| "learning_rate": 3.9558673469387755e-06, | |
| "loss": 0.0252, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 4.359267734553776, | |
| "grad_norm": 0.8049426078796387, | |
| "learning_rate": 3.949489795918367e-06, | |
| "loss": 0.0127, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 4.37070938215103, | |
| "grad_norm": 2.284611463546753, | |
| "learning_rate": 3.9431122448979595e-06, | |
| "loss": 0.0295, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 4.382151029748284, | |
| "grad_norm": 1.1632159948349, | |
| "learning_rate": 3.936734693877552e-06, | |
| "loss": 0.014, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 4.393592677345538, | |
| "grad_norm": 3.538940906524658, | |
| "learning_rate": 3.930357142857143e-06, | |
| "loss": 0.0254, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 4.4050343249427915, | |
| "grad_norm": 3.8108420372009277, | |
| "learning_rate": 3.923979591836735e-06, | |
| "loss": 0.0131, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 4.416475972540046, | |
| "grad_norm": 0.45241427421569824, | |
| "learning_rate": 3.9176020408163265e-06, | |
| "loss": 0.0201, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 4.4279176201373, | |
| "grad_norm": 3.477997064590454, | |
| "learning_rate": 3.911224489795919e-06, | |
| "loss": 0.0174, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 4.439359267734554, | |
| "grad_norm": 1.1490719318389893, | |
| "learning_rate": 3.9048469387755104e-06, | |
| "loss": 0.0151, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 4.450800915331808, | |
| "grad_norm": 7.397353649139404, | |
| "learning_rate": 3.898469387755102e-06, | |
| "loss": 0.0169, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 4.462242562929061, | |
| "grad_norm": 0.5268431305885315, | |
| "learning_rate": 3.8920918367346935e-06, | |
| "loss": 0.0246, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 4.473684210526316, | |
| "grad_norm": 1.0522817373275757, | |
| "learning_rate": 3.885714285714286e-06, | |
| "loss": 0.0159, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 4.48512585812357, | |
| "grad_norm": 1.1104140281677246, | |
| "learning_rate": 3.879336734693878e-06, | |
| "loss": 0.0312, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 4.496567505720824, | |
| "grad_norm": 11.163028717041016, | |
| "learning_rate": 3.87295918367347e-06, | |
| "loss": 0.0169, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 4.508009153318078, | |
| "grad_norm": 3.165876865386963, | |
| "learning_rate": 3.866581632653061e-06, | |
| "loss": 0.0159, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 4.519450800915331, | |
| "grad_norm": 3.8829312324523926, | |
| "learning_rate": 3.860204081632653e-06, | |
| "loss": 0.0163, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 4.530892448512586, | |
| "grad_norm": 3.46303129196167, | |
| "learning_rate": 3.853826530612245e-06, | |
| "loss": 0.0279, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 4.5423340961098395, | |
| "grad_norm": 1.1138535737991333, | |
| "learning_rate": 3.847448979591837e-06, | |
| "loss": 0.0154, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 4.553775743707094, | |
| "grad_norm": 4.287326335906982, | |
| "learning_rate": 3.841071428571428e-06, | |
| "loss": 0.0193, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 4.565217391304348, | |
| "grad_norm": 0.4523741602897644, | |
| "learning_rate": 3.83469387755102e-06, | |
| "loss": 0.0126, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 4.576659038901602, | |
| "grad_norm": 0.8860822916030884, | |
| "learning_rate": 3.828316326530612e-06, | |
| "loss": 0.0281, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 4.576659038901602, | |
| "eval_loss": 0.1852317899465561, | |
| "eval_runtime": 5272.523, | |
| "eval_samples_per_second": 1.806, | |
| "eval_steps_per_second": 0.226, | |
| "eval_wer": 0.10049913611057784, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 4.588100686498856, | |
| "grad_norm": 2.3640730381011963, | |
| "learning_rate": 3.821938775510205e-06, | |
| "loss": 0.0134, | |
| "step": 10025 | |
| }, | |
| { | |
| "epoch": 4.5995423340961095, | |
| "grad_norm": 0.9958900809288025, | |
| "learning_rate": 3.815561224489796e-06, | |
| "loss": 0.017, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 4.610983981693364, | |
| "grad_norm": 4.021085739135742, | |
| "learning_rate": 3.8091836734693882e-06, | |
| "loss": 0.0169, | |
| "step": 10075 | |
| }, | |
| { | |
| "epoch": 4.622425629290618, | |
| "grad_norm": 1.908926010131836, | |
| "learning_rate": 3.8028061224489798e-06, | |
| "loss": 0.0208, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 4.633867276887872, | |
| "grad_norm": 4.975993633270264, | |
| "learning_rate": 3.7964285714285717e-06, | |
| "loss": 0.0133, | |
| "step": 10125 | |
| }, | |
| { | |
| "epoch": 4.645308924485126, | |
| "grad_norm": 0.652805745601654, | |
| "learning_rate": 3.7900510204081633e-06, | |
| "loss": 0.0208, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 4.65675057208238, | |
| "grad_norm": 1.805553913116455, | |
| "learning_rate": 3.7836734693877553e-06, | |
| "loss": 0.0132, | |
| "step": 10175 | |
| }, | |
| { | |
| "epoch": 4.668192219679634, | |
| "grad_norm": 0.581511378288269, | |
| "learning_rate": 3.777295918367347e-06, | |
| "loss": 0.0243, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 4.679633867276888, | |
| "grad_norm": 4.517154693603516, | |
| "learning_rate": 3.7709183673469388e-06, | |
| "loss": 0.0142, | |
| "step": 10225 | |
| }, | |
| { | |
| "epoch": 4.691075514874142, | |
| "grad_norm": 3.5025482177734375, | |
| "learning_rate": 3.764540816326531e-06, | |
| "loss": 0.0258, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 4.702517162471396, | |
| "grad_norm": 2.5942628383636475, | |
| "learning_rate": 3.7581632653061227e-06, | |
| "loss": 0.0121, | |
| "step": 10275 | |
| }, | |
| { | |
| "epoch": 4.71395881006865, | |
| "grad_norm": 0.7120451927185059, | |
| "learning_rate": 3.7517857142857147e-06, | |
| "loss": 0.0254, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 4.725400457665904, | |
| "grad_norm": 8.78650951385498, | |
| "learning_rate": 3.745408163265306e-06, | |
| "loss": 0.0141, | |
| "step": 10325 | |
| }, | |
| { | |
| "epoch": 4.7368421052631575, | |
| "grad_norm": 1.8899227380752563, | |
| "learning_rate": 3.739030612244898e-06, | |
| "loss": 0.0264, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 4.748283752860412, | |
| "grad_norm": 0.5672324895858765, | |
| "learning_rate": 3.7326530612244897e-06, | |
| "loss": 0.0095, | |
| "step": 10375 | |
| }, | |
| { | |
| "epoch": 4.759725400457666, | |
| "grad_norm": 0.8572821617126465, | |
| "learning_rate": 3.7262755102040817e-06, | |
| "loss": 0.0236, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 4.77116704805492, | |
| "grad_norm": 0.23919886350631714, | |
| "learning_rate": 3.7198979591836732e-06, | |
| "loss": 0.0117, | |
| "step": 10425 | |
| }, | |
| { | |
| "epoch": 4.782608695652174, | |
| "grad_norm": 0.6619164943695068, | |
| "learning_rate": 3.713520408163265e-06, | |
| "loss": 0.0256, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 4.7940503432494275, | |
| "grad_norm": 1.8749737739562988, | |
| "learning_rate": 3.7071428571428576e-06, | |
| "loss": 0.0132, | |
| "step": 10475 | |
| }, | |
| { | |
| "epoch": 4.805491990846682, | |
| "grad_norm": 3.1002092361450195, | |
| "learning_rate": 3.7007653061224496e-06, | |
| "loss": 0.0264, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 4.816933638443936, | |
| "grad_norm": 1.090328335762024, | |
| "learning_rate": 3.694387755102041e-06, | |
| "loss": 0.0166, | |
| "step": 10525 | |
| }, | |
| { | |
| "epoch": 4.82837528604119, | |
| "grad_norm": 0.8507818579673767, | |
| "learning_rate": 3.688010204081633e-06, | |
| "loss": 0.0206, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 4.839816933638444, | |
| "grad_norm": 2.137056589126587, | |
| "learning_rate": 3.6816326530612246e-06, | |
| "loss": 0.0114, | |
| "step": 10575 | |
| }, | |
| { | |
| "epoch": 4.851258581235698, | |
| "grad_norm": 3.8611929416656494, | |
| "learning_rate": 3.6752551020408166e-06, | |
| "loss": 0.0214, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 4.862700228832952, | |
| "grad_norm": 5.636245250701904, | |
| "learning_rate": 3.668877551020408e-06, | |
| "loss": 0.017, | |
| "step": 10625 | |
| }, | |
| { | |
| "epoch": 4.874141876430206, | |
| "grad_norm": 1.076035499572754, | |
| "learning_rate": 3.6625e-06, | |
| "loss": 0.0316, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 4.88558352402746, | |
| "grad_norm": 2.079132556915283, | |
| "learning_rate": 3.6561224489795916e-06, | |
| "loss": 0.0136, | |
| "step": 10675 | |
| }, | |
| { | |
| "epoch": 4.897025171624714, | |
| "grad_norm": 2.008894443511963, | |
| "learning_rate": 3.649744897959184e-06, | |
| "loss": 0.0239, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 4.908466819221968, | |
| "grad_norm": 0.4755445420742035, | |
| "learning_rate": 3.643367346938776e-06, | |
| "loss": 0.0131, | |
| "step": 10725 | |
| }, | |
| { | |
| "epoch": 4.919908466819222, | |
| "grad_norm": 2.093269109725952, | |
| "learning_rate": 3.6369897959183675e-06, | |
| "loss": 0.023, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 4.931350114416476, | |
| "grad_norm": 0.36701831221580505, | |
| "learning_rate": 3.6306122448979595e-06, | |
| "loss": 0.0137, | |
| "step": 10775 | |
| }, | |
| { | |
| "epoch": 4.94279176201373, | |
| "grad_norm": 2.4676878452301025, | |
| "learning_rate": 3.624234693877551e-06, | |
| "loss": 0.0324, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 4.954233409610984, | |
| "grad_norm": 2.1514317989349365, | |
| "learning_rate": 3.617857142857143e-06, | |
| "loss": 0.0163, | |
| "step": 10825 | |
| }, | |
| { | |
| "epoch": 4.965675057208238, | |
| "grad_norm": 0.4548572897911072, | |
| "learning_rate": 3.6114795918367346e-06, | |
| "loss": 0.0225, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 4.977116704805492, | |
| "grad_norm": 4.841091632843018, | |
| "learning_rate": 3.6051020408163265e-06, | |
| "loss": 0.015, | |
| "step": 10875 | |
| }, | |
| { | |
| "epoch": 4.988558352402746, | |
| "grad_norm": 1.0846531391143799, | |
| "learning_rate": 3.598724489795918e-06, | |
| "loss": 0.0241, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 3.79087495803833, | |
| "learning_rate": 3.592346938775511e-06, | |
| "loss": 0.0228, | |
| "step": 10925 | |
| }, | |
| { | |
| "epoch": 5.011441647597254, | |
| "grad_norm": 7.431537628173828, | |
| "learning_rate": 3.5859693877551024e-06, | |
| "loss": 0.0131, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 5.022883295194508, | |
| "grad_norm": 9.645064353942871, | |
| "learning_rate": 3.5795918367346944e-06, | |
| "loss": 0.0099, | |
| "step": 10975 | |
| }, | |
| { | |
| "epoch": 5.034324942791762, | |
| "grad_norm": 4.066143035888672, | |
| "learning_rate": 3.573214285714286e-06, | |
| "loss": 0.0122, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 5.045766590389016, | |
| "grad_norm": 15.450593948364258, | |
| "learning_rate": 3.566836734693878e-06, | |
| "loss": 0.0238, | |
| "step": 11025 | |
| }, | |
| { | |
| "epoch": 5.05720823798627, | |
| "grad_norm": 4.155067443847656, | |
| "learning_rate": 3.5604591836734694e-06, | |
| "loss": 0.0146, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 5.068649885583524, | |
| "grad_norm": 1.2085702419281006, | |
| "learning_rate": 3.5540816326530614e-06, | |
| "loss": 0.0205, | |
| "step": 11075 | |
| }, | |
| { | |
| "epoch": 5.080091533180778, | |
| "grad_norm": 2.0568387508392334, | |
| "learning_rate": 3.547704081632653e-06, | |
| "loss": 0.0104, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 5.091533180778032, | |
| "grad_norm": 2.4178097248077393, | |
| "learning_rate": 3.541326530612245e-06, | |
| "loss": 0.0152, | |
| "step": 11125 | |
| }, | |
| { | |
| "epoch": 5.102974828375286, | |
| "grad_norm": 6.744433879852295, | |
| "learning_rate": 3.5349489795918373e-06, | |
| "loss": 0.0118, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 5.11441647597254, | |
| "grad_norm": 2.1950619220733643, | |
| "learning_rate": 3.528571428571429e-06, | |
| "loss": 0.013, | |
| "step": 11175 | |
| }, | |
| { | |
| "epoch": 5.125858123569794, | |
| "grad_norm": 7.050522804260254, | |
| "learning_rate": 3.522193877551021e-06, | |
| "loss": 0.0117, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 5.137299771167048, | |
| "grad_norm": 18.47918128967285, | |
| "learning_rate": 3.5158163265306124e-06, | |
| "loss": 0.0153, | |
| "step": 11225 | |
| }, | |
| { | |
| "epoch": 5.148741418764302, | |
| "grad_norm": 0.19902612268924713, | |
| "learning_rate": 3.5094387755102043e-06, | |
| "loss": 0.0111, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 5.160183066361556, | |
| "grad_norm": 4.996516704559326, | |
| "learning_rate": 3.503061224489796e-06, | |
| "loss": 0.0147, | |
| "step": 11275 | |
| }, | |
| { | |
| "epoch": 5.17162471395881, | |
| "grad_norm": 0.06878898292779922, | |
| "learning_rate": 3.496683673469388e-06, | |
| "loss": 0.0084, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 5.183066361556064, | |
| "grad_norm": 3.7109427452087402, | |
| "learning_rate": 3.4903061224489794e-06, | |
| "loss": 0.0171, | |
| "step": 11325 | |
| }, | |
| { | |
| "epoch": 5.194508009153318, | |
| "grad_norm": 1.2318646907806396, | |
| "learning_rate": 3.484183673469388e-06, | |
| "loss": 0.0104, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 5.2059496567505725, | |
| "grad_norm": 5.8200364112854, | |
| "learning_rate": 3.4778061224489795e-06, | |
| "loss": 0.0178, | |
| "step": 11375 | |
| }, | |
| { | |
| "epoch": 5.217391304347826, | |
| "grad_norm": 2.4029898643493652, | |
| "learning_rate": 3.4714285714285715e-06, | |
| "loss": 0.0143, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 5.22883295194508, | |
| "grad_norm": 5.977914810180664, | |
| "learning_rate": 3.465051020408164e-06, | |
| "loss": 0.0137, | |
| "step": 11425 | |
| }, | |
| { | |
| "epoch": 5.240274599542334, | |
| "grad_norm": 1.1518189907073975, | |
| "learning_rate": 3.4586734693877554e-06, | |
| "loss": 0.0115, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 5.251716247139588, | |
| "grad_norm": 9.050216674804688, | |
| "learning_rate": 3.4522959183673474e-06, | |
| "loss": 0.019, | |
| "step": 11475 | |
| }, | |
| { | |
| "epoch": 5.2631578947368425, | |
| "grad_norm": 2.823286294937134, | |
| "learning_rate": 3.445918367346939e-06, | |
| "loss": 0.0105, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 5.274599542334096, | |
| "grad_norm": 2.077913284301758, | |
| "learning_rate": 3.439540816326531e-06, | |
| "loss": 0.012, | |
| "step": 11525 | |
| }, | |
| { | |
| "epoch": 5.28604118993135, | |
| "grad_norm": 1.8525944948196411, | |
| "learning_rate": 3.4331632653061224e-06, | |
| "loss": 0.0128, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 5.297482837528604, | |
| "grad_norm": 4.57568883895874, | |
| "learning_rate": 3.4267857142857144e-06, | |
| "loss": 0.0168, | |
| "step": 11575 | |
| }, | |
| { | |
| "epoch": 5.308924485125858, | |
| "grad_norm": 0.7349223494529724, | |
| "learning_rate": 3.420408163265306e-06, | |
| "loss": 0.0103, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 5.320366132723112, | |
| "grad_norm": 8.281305313110352, | |
| "learning_rate": 3.414030612244898e-06, | |
| "loss": 0.0142, | |
| "step": 11625 | |
| }, | |
| { | |
| "epoch": 5.331807780320366, | |
| "grad_norm": 2.828233480453491, | |
| "learning_rate": 3.4076530612244903e-06, | |
| "loss": 0.0074, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 5.34324942791762, | |
| "grad_norm": 3.6413841247558594, | |
| "learning_rate": 3.4012755102040823e-06, | |
| "loss": 0.0145, | |
| "step": 11675 | |
| }, | |
| { | |
| "epoch": 5.354691075514874, | |
| "grad_norm": 8.595261573791504, | |
| "learning_rate": 3.394897959183674e-06, | |
| "loss": 0.0104, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 5.366132723112128, | |
| "grad_norm": 8.777588844299316, | |
| "learning_rate": 3.3885204081632658e-06, | |
| "loss": 0.016, | |
| "step": 11725 | |
| }, | |
| { | |
| "epoch": 5.377574370709382, | |
| "grad_norm": 0.81954026222229, | |
| "learning_rate": 3.3821428571428573e-06, | |
| "loss": 0.0076, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 5.389016018306636, | |
| "grad_norm": 2.593818187713623, | |
| "learning_rate": 3.3757653061224493e-06, | |
| "loss": 0.0178, | |
| "step": 11775 | |
| }, | |
| { | |
| "epoch": 5.4004576659038905, | |
| "grad_norm": 1.6006801128387451, | |
| "learning_rate": 3.369387755102041e-06, | |
| "loss": 0.0099, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 5.411899313501144, | |
| "grad_norm": 3.7885050773620605, | |
| "learning_rate": 3.363010204081633e-06, | |
| "loss": 0.0165, | |
| "step": 11825 | |
| }, | |
| { | |
| "epoch": 5.423340961098398, | |
| "grad_norm": 1.194949984550476, | |
| "learning_rate": 3.3566326530612243e-06, | |
| "loss": 0.0108, | |
| "step": 11850 | |
| }, | |
| { | |
| "epoch": 5.434782608695652, | |
| "grad_norm": 11.738698959350586, | |
| "learning_rate": 3.3502551020408167e-06, | |
| "loss": 0.0215, | |
| "step": 11875 | |
| }, | |
| { | |
| "epoch": 5.446224256292906, | |
| "grad_norm": 2.309065103530884, | |
| "learning_rate": 3.3438775510204087e-06, | |
| "loss": 0.0121, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 5.4576659038901605, | |
| "grad_norm": 7.016556262969971, | |
| "learning_rate": 3.3375000000000002e-06, | |
| "loss": 0.0138, | |
| "step": 11925 | |
| }, | |
| { | |
| "epoch": 5.469107551487414, | |
| "grad_norm": 2.991405963897705, | |
| "learning_rate": 3.331122448979592e-06, | |
| "loss": 0.0146, | |
| "step": 11950 | |
| }, | |
| { | |
| "epoch": 5.480549199084669, | |
| "grad_norm": 7.811606407165527, | |
| "learning_rate": 3.3247448979591837e-06, | |
| "loss": 0.0194, | |
| "step": 11975 | |
| }, | |
| { | |
| "epoch": 5.491990846681922, | |
| "grad_norm": 3.6345748901367188, | |
| "learning_rate": 3.3183673469387757e-06, | |
| "loss": 0.0142, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 5.503432494279176, | |
| "grad_norm": 7.627600193023682, | |
| "learning_rate": 3.3119897959183673e-06, | |
| "loss": 0.0179, | |
| "step": 12025 | |
| }, | |
| { | |
| "epoch": 5.51487414187643, | |
| "grad_norm": 1.8679994344711304, | |
| "learning_rate": 3.3056122448979592e-06, | |
| "loss": 0.0105, | |
| "step": 12050 | |
| }, | |
| { | |
| "epoch": 5.526315789473684, | |
| "grad_norm": 0.5164907574653625, | |
| "learning_rate": 3.2992346938775508e-06, | |
| "loss": 0.0105, | |
| "step": 12075 | |
| }, | |
| { | |
| "epoch": 5.537757437070939, | |
| "grad_norm": 0.4254048764705658, | |
| "learning_rate": 3.2928571428571427e-06, | |
| "loss": 0.0081, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 5.549199084668192, | |
| "grad_norm": 5.356523513793945, | |
| "learning_rate": 3.286479591836735e-06, | |
| "loss": 0.0218, | |
| "step": 12125 | |
| }, | |
| { | |
| "epoch": 5.560640732265446, | |
| "grad_norm": 3.510284662246704, | |
| "learning_rate": 3.280102040816327e-06, | |
| "loss": 0.0085, | |
| "step": 12150 | |
| }, | |
| { | |
| "epoch": 5.5720823798627, | |
| "grad_norm": 6.946488380432129, | |
| "learning_rate": 3.2737244897959186e-06, | |
| "loss": 0.0184, | |
| "step": 12175 | |
| }, | |
| { | |
| "epoch": 5.583524027459954, | |
| "grad_norm": 0.3557557463645935, | |
| "learning_rate": 3.2673469387755106e-06, | |
| "loss": 0.0106, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 5.5949656750572085, | |
| "grad_norm": 3.596979856491089, | |
| "learning_rate": 3.260969387755102e-06, | |
| "loss": 0.0258, | |
| "step": 12225 | |
| }, | |
| { | |
| "epoch": 5.606407322654462, | |
| "grad_norm": 0.7150298953056335, | |
| "learning_rate": 3.254591836734694e-06, | |
| "loss": 0.0094, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 5.617848970251716, | |
| "grad_norm": 4.756287097930908, | |
| "learning_rate": 3.2482142857142857e-06, | |
| "loss": 0.0164, | |
| "step": 12275 | |
| }, | |
| { | |
| "epoch": 5.62929061784897, | |
| "grad_norm": 3.6292340755462646, | |
| "learning_rate": 3.2418367346938776e-06, | |
| "loss": 0.0107, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 5.640732265446224, | |
| "grad_norm": 9.82845687866211, | |
| "learning_rate": 3.235459183673469e-06, | |
| "loss": 0.019, | |
| "step": 12325 | |
| }, | |
| { | |
| "epoch": 5.6521739130434785, | |
| "grad_norm": 0.3698252737522125, | |
| "learning_rate": 3.2290816326530616e-06, | |
| "loss": 0.0079, | |
| "step": 12350 | |
| }, | |
| { | |
| "epoch": 5.663615560640732, | |
| "grad_norm": 1.8457624912261963, | |
| "learning_rate": 3.2227040816326535e-06, | |
| "loss": 0.023, | |
| "step": 12375 | |
| }, | |
| { | |
| "epoch": 5.675057208237987, | |
| "grad_norm": 0.7149168848991394, | |
| "learning_rate": 3.216326530612245e-06, | |
| "loss": 0.016, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 5.68649885583524, | |
| "grad_norm": 4.959986209869385, | |
| "learning_rate": 3.209948979591837e-06, | |
| "loss": 0.0184, | |
| "step": 12425 | |
| }, | |
| { | |
| "epoch": 5.697940503432494, | |
| "grad_norm": 1.3709676265716553, | |
| "learning_rate": 3.2035714285714286e-06, | |
| "loss": 0.0118, | |
| "step": 12450 | |
| }, | |
| { | |
| "epoch": 5.709382151029748, | |
| "grad_norm": 4.3518829345703125, | |
| "learning_rate": 3.1971938775510205e-06, | |
| "loss": 0.0172, | |
| "step": 12475 | |
| }, | |
| { | |
| "epoch": 5.720823798627002, | |
| "grad_norm": 4.1297502517700195, | |
| "learning_rate": 3.190816326530612e-06, | |
| "loss": 0.0126, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 5.732265446224257, | |
| "grad_norm": 4.261162281036377, | |
| "learning_rate": 3.184438775510204e-06, | |
| "loss": 0.0133, | |
| "step": 12525 | |
| }, | |
| { | |
| "epoch": 5.74370709382151, | |
| "grad_norm": 3.297361135482788, | |
| "learning_rate": 3.1780612244897956e-06, | |
| "loss": 0.0116, | |
| "step": 12550 | |
| }, | |
| { | |
| "epoch": 5.755148741418765, | |
| "grad_norm": 6.726646423339844, | |
| "learning_rate": 3.171683673469388e-06, | |
| "loss": 0.019, | |
| "step": 12575 | |
| }, | |
| { | |
| "epoch": 5.766590389016018, | |
| "grad_norm": 0.5207387208938599, | |
| "learning_rate": 3.16530612244898e-06, | |
| "loss": 0.0092, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 5.778032036613272, | |
| "grad_norm": 11.70545768737793, | |
| "learning_rate": 3.158928571428572e-06, | |
| "loss": 0.024, | |
| "step": 12625 | |
| }, | |
| { | |
| "epoch": 5.7894736842105265, | |
| "grad_norm": 0.3776039481163025, | |
| "learning_rate": 3.1525510204081635e-06, | |
| "loss": 0.01, | |
| "step": 12650 | |
| }, | |
| { | |
| "epoch": 5.80091533180778, | |
| "grad_norm": 16.962390899658203, | |
| "learning_rate": 3.1461734693877554e-06, | |
| "loss": 0.0147, | |
| "step": 12675 | |
| }, | |
| { | |
| "epoch": 5.812356979405035, | |
| "grad_norm": 3.64579701423645, | |
| "learning_rate": 3.139795918367347e-06, | |
| "loss": 0.011, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 5.823798627002288, | |
| "grad_norm": 6.796038627624512, | |
| "learning_rate": 3.133418367346939e-06, | |
| "loss": 0.0138, | |
| "step": 12725 | |
| }, | |
| { | |
| "epoch": 5.835240274599542, | |
| "grad_norm": 1.9765185117721558, | |
| "learning_rate": 3.1270408163265305e-06, | |
| "loss": 0.011, | |
| "step": 12750 | |
| }, | |
| { | |
| "epoch": 5.8466819221967965, | |
| "grad_norm": 4.846302509307861, | |
| "learning_rate": 3.1206632653061225e-06, | |
| "loss": 0.0107, | |
| "step": 12775 | |
| }, | |
| { | |
| "epoch": 5.85812356979405, | |
| "grad_norm": 3.6651906967163086, | |
| "learning_rate": 3.1142857142857144e-06, | |
| "loss": 0.0094, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 5.869565217391305, | |
| "grad_norm": 6.605918884277344, | |
| "learning_rate": 3.107908163265306e-06, | |
| "loss": 0.0158, | |
| "step": 12825 | |
| }, | |
| { | |
| "epoch": 5.881006864988558, | |
| "grad_norm": 3.4638562202453613, | |
| "learning_rate": 3.1015306122448984e-06, | |
| "loss": 0.0119, | |
| "step": 12850 | |
| }, | |
| { | |
| "epoch": 5.892448512585812, | |
| "grad_norm": 1.832309603691101, | |
| "learning_rate": 3.09515306122449e-06, | |
| "loss": 0.0164, | |
| "step": 12875 | |
| }, | |
| { | |
| "epoch": 5.9038901601830664, | |
| "grad_norm": 0.39115744829177856, | |
| "learning_rate": 3.088775510204082e-06, | |
| "loss": 0.0137, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 5.91533180778032, | |
| "grad_norm": 5.2722086906433105, | |
| "learning_rate": 3.0823979591836734e-06, | |
| "loss": 0.0173, | |
| "step": 12925 | |
| }, | |
| { | |
| "epoch": 5.926773455377575, | |
| "grad_norm": 0.5123651623725891, | |
| "learning_rate": 3.0760204081632654e-06, | |
| "loss": 0.0109, | |
| "step": 12950 | |
| }, | |
| { | |
| "epoch": 5.938215102974828, | |
| "grad_norm": 5.007115840911865, | |
| "learning_rate": 3.0696428571428573e-06, | |
| "loss": 0.0161, | |
| "step": 12975 | |
| }, | |
| { | |
| "epoch": 5.949656750572083, | |
| "grad_norm": 1.0899899005889893, | |
| "learning_rate": 3.0632653061224493e-06, | |
| "loss": 0.0098, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 5.961098398169336, | |
| "grad_norm": 6.381730079650879, | |
| "learning_rate": 3.056887755102041e-06, | |
| "loss": 0.0206, | |
| "step": 13025 | |
| }, | |
| { | |
| "epoch": 5.97254004576659, | |
| "grad_norm": 2.857933521270752, | |
| "learning_rate": 3.050510204081633e-06, | |
| "loss": 0.0124, | |
| "step": 13050 | |
| }, | |
| { | |
| "epoch": 5.983981693363845, | |
| "grad_norm": 9.316118240356445, | |
| "learning_rate": 3.0441326530612248e-06, | |
| "loss": 0.0134, | |
| "step": 13075 | |
| }, | |
| { | |
| "epoch": 5.995423340961098, | |
| "grad_norm": 0.3497377932071686, | |
| "learning_rate": 3.0377551020408163e-06, | |
| "loss": 0.0106, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 6.006864988558353, | |
| "grad_norm": 4.825676918029785, | |
| "learning_rate": 3.0313775510204083e-06, | |
| "loss": 0.0167, | |
| "step": 13125 | |
| }, | |
| { | |
| "epoch": 6.018306636155606, | |
| "grad_norm": 8.057607650756836, | |
| "learning_rate": 3.0250000000000003e-06, | |
| "loss": 0.0133, | |
| "step": 13150 | |
| }, | |
| { | |
| "epoch": 6.02974828375286, | |
| "grad_norm": 0.27666619420051575, | |
| "learning_rate": 3.018622448979592e-06, | |
| "loss": 0.0215, | |
| "step": 13175 | |
| }, | |
| { | |
| "epoch": 6.0411899313501145, | |
| "grad_norm": 0.5118196606636047, | |
| "learning_rate": 3.0122448979591838e-06, | |
| "loss": 0.0144, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 6.052631578947368, | |
| "grad_norm": 12.31119155883789, | |
| "learning_rate": 3.0058673469387757e-06, | |
| "loss": 0.0134, | |
| "step": 13225 | |
| }, | |
| { | |
| "epoch": 6.064073226544623, | |
| "grad_norm": 0.9528910517692566, | |
| "learning_rate": 2.9994897959183673e-06, | |
| "loss": 0.0096, | |
| "step": 13250 | |
| }, | |
| { | |
| "epoch": 6.075514874141876, | |
| "grad_norm": 1.3786743879318237, | |
| "learning_rate": 2.9931122448979592e-06, | |
| "loss": 0.0155, | |
| "step": 13275 | |
| }, | |
| { | |
| "epoch": 6.086956521739131, | |
| "grad_norm": 4.0814337730407715, | |
| "learning_rate": 2.9867346938775512e-06, | |
| "loss": 0.0113, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 6.0983981693363845, | |
| "grad_norm": 0.17644235491752625, | |
| "learning_rate": 2.980357142857143e-06, | |
| "loss": 0.0086, | |
| "step": 13325 | |
| }, | |
| { | |
| "epoch": 6.109839816933638, | |
| "grad_norm": 2.6412882804870605, | |
| "learning_rate": 2.9739795918367347e-06, | |
| "loss": 0.0087, | |
| "step": 13350 | |
| }, | |
| { | |
| "epoch": 6.121281464530893, | |
| "grad_norm": 4.250133514404297, | |
| "learning_rate": 2.9676020408163267e-06, | |
| "loss": 0.0123, | |
| "step": 13375 | |
| }, | |
| { | |
| "epoch": 6.132723112128146, | |
| "grad_norm": 1.6422185897827148, | |
| "learning_rate": 2.9612244897959182e-06, | |
| "loss": 0.0067, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 6.144164759725401, | |
| "grad_norm": 3.366658926010132, | |
| "learning_rate": 2.9548469387755106e-06, | |
| "loss": 0.0117, | |
| "step": 13425 | |
| }, | |
| { | |
| "epoch": 6.155606407322654, | |
| "grad_norm": 6.940051555633545, | |
| "learning_rate": 2.948469387755102e-06, | |
| "loss": 0.0072, | |
| "step": 13450 | |
| }, | |
| { | |
| "epoch": 6.167048054919908, | |
| "grad_norm": 0.4400598108768463, | |
| "learning_rate": 2.942091836734694e-06, | |
| "loss": 0.0159, | |
| "step": 13475 | |
| }, | |
| { | |
| "epoch": 6.178489702517163, | |
| "grad_norm": 3.105194091796875, | |
| "learning_rate": 2.9357142857142857e-06, | |
| "loss": 0.0091, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 6.189931350114416, | |
| "grad_norm": 1.1099435091018677, | |
| "learning_rate": 2.9293367346938776e-06, | |
| "loss": 0.019, | |
| "step": 13525 | |
| }, | |
| { | |
| "epoch": 6.201372997711671, | |
| "grad_norm": 1.7562593221664429, | |
| "learning_rate": 2.9229591836734696e-06, | |
| "loss": 0.0084, | |
| "step": 13550 | |
| }, | |
| { | |
| "epoch": 6.212814645308924, | |
| "grad_norm": 2.2326951026916504, | |
| "learning_rate": 2.916581632653061e-06, | |
| "loss": 0.0148, | |
| "step": 13575 | |
| }, | |
| { | |
| "epoch": 6.224256292906179, | |
| "grad_norm": 0.38852858543395996, | |
| "learning_rate": 2.910204081632653e-06, | |
| "loss": 0.0068, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 6.2356979405034325, | |
| "grad_norm": 3.4514925479888916, | |
| "learning_rate": 2.903826530612245e-06, | |
| "loss": 0.0111, | |
| "step": 13625 | |
| }, | |
| { | |
| "epoch": 6.247139588100686, | |
| "grad_norm": 2.002838134765625, | |
| "learning_rate": 2.897448979591837e-06, | |
| "loss": 0.0082, | |
| "step": 13650 | |
| }, | |
| { | |
| "epoch": 6.258581235697941, | |
| "grad_norm": 0.08688002824783325, | |
| "learning_rate": 2.8910714285714286e-06, | |
| "loss": 0.0267, | |
| "step": 13675 | |
| }, | |
| { | |
| "epoch": 6.270022883295194, | |
| "grad_norm": 2.080207347869873, | |
| "learning_rate": 2.8846938775510206e-06, | |
| "loss": 0.0096, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 6.281464530892449, | |
| "grad_norm": 0.9220284819602966, | |
| "learning_rate": 2.878316326530612e-06, | |
| "loss": 0.0086, | |
| "step": 13725 | |
| }, | |
| { | |
| "epoch": 6.2929061784897025, | |
| "grad_norm": 0.35067400336265564, | |
| "learning_rate": 2.8719387755102045e-06, | |
| "loss": 0.0052, | |
| "step": 13750 | |
| }, | |
| { | |
| "epoch": 6.304347826086957, | |
| "grad_norm": 3.5281693935394287, | |
| "learning_rate": 2.865561224489796e-06, | |
| "loss": 0.0191, | |
| "step": 13775 | |
| }, | |
| { | |
| "epoch": 6.315789473684211, | |
| "grad_norm": 3.336193323135376, | |
| "learning_rate": 2.859183673469388e-06, | |
| "loss": 0.0063, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 6.327231121281464, | |
| "grad_norm": 2.0798017978668213, | |
| "learning_rate": 2.8528061224489796e-06, | |
| "loss": 0.0132, | |
| "step": 13825 | |
| }, | |
| { | |
| "epoch": 6.338672768878719, | |
| "grad_norm": 0.20700007677078247, | |
| "learning_rate": 2.8464285714285715e-06, | |
| "loss": 0.0061, | |
| "step": 13850 | |
| }, | |
| { | |
| "epoch": 6.350114416475972, | |
| "grad_norm": 0.1666710078716278, | |
| "learning_rate": 2.8400510204081635e-06, | |
| "loss": 0.0073, | |
| "step": 13875 | |
| }, | |
| { | |
| "epoch": 6.361556064073227, | |
| "grad_norm": 4.459512233734131, | |
| "learning_rate": 2.8336734693877555e-06, | |
| "loss": 0.0107, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 6.372997711670481, | |
| "grad_norm": 2.873020887374878, | |
| "learning_rate": 2.827295918367347e-06, | |
| "loss": 0.0124, | |
| "step": 13925 | |
| }, | |
| { | |
| "epoch": 6.384439359267734, | |
| "grad_norm": 0.23313076794147491, | |
| "learning_rate": 2.820918367346939e-06, | |
| "loss": 0.012, | |
| "step": 13950 | |
| }, | |
| { | |
| "epoch": 6.395881006864989, | |
| "grad_norm": 0.10002599656581879, | |
| "learning_rate": 2.814540816326531e-06, | |
| "loss": 0.01, | |
| "step": 13975 | |
| }, | |
| { | |
| "epoch": 6.407322654462242, | |
| "grad_norm": 0.2222910374403, | |
| "learning_rate": 2.8081632653061225e-06, | |
| "loss": 0.0061, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 6.418764302059497, | |
| "grad_norm": 1.97636079788208, | |
| "learning_rate": 2.8017857142857144e-06, | |
| "loss": 0.0125, | |
| "step": 14025 | |
| }, | |
| { | |
| "epoch": 6.4302059496567505, | |
| "grad_norm": 0.23159927129745483, | |
| "learning_rate": 2.795408163265306e-06, | |
| "loss": 0.0121, | |
| "step": 14050 | |
| }, | |
| { | |
| "epoch": 6.441647597254004, | |
| "grad_norm": 2.222355365753174, | |
| "learning_rate": 2.789030612244898e-06, | |
| "loss": 0.0082, | |
| "step": 14075 | |
| }, | |
| { | |
| "epoch": 6.453089244851259, | |
| "grad_norm": 1.612273931503296, | |
| "learning_rate": 2.78265306122449e-06, | |
| "loss": 0.0129, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 6.464530892448512, | |
| "grad_norm": 3.7675607204437256, | |
| "learning_rate": 2.776275510204082e-06, | |
| "loss": 0.0126, | |
| "step": 14125 | |
| }, | |
| { | |
| "epoch": 6.475972540045767, | |
| "grad_norm": 2.6924796104431152, | |
| "learning_rate": 2.7698979591836734e-06, | |
| "loss": 0.0123, | |
| "step": 14150 | |
| }, | |
| { | |
| "epoch": 6.4874141876430205, | |
| "grad_norm": 0.5433311462402344, | |
| "learning_rate": 2.7635204081632654e-06, | |
| "loss": 0.0083, | |
| "step": 14175 | |
| }, | |
| { | |
| "epoch": 6.498855835240275, | |
| "grad_norm": 1.2560780048370361, | |
| "learning_rate": 2.7571428571428574e-06, | |
| "loss": 0.0119, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 6.510297482837529, | |
| "grad_norm": 0.1697181910276413, | |
| "learning_rate": 2.7507653061224493e-06, | |
| "loss": 0.0142, | |
| "step": 14225 | |
| }, | |
| { | |
| "epoch": 6.521739130434782, | |
| "grad_norm": 2.489327907562256, | |
| "learning_rate": 2.744387755102041e-06, | |
| "loss": 0.0098, | |
| "step": 14250 | |
| }, | |
| { | |
| "epoch": 6.533180778032037, | |
| "grad_norm": 0.10007134824991226, | |
| "learning_rate": 2.738010204081633e-06, | |
| "loss": 0.0118, | |
| "step": 14275 | |
| }, | |
| { | |
| "epoch": 6.54462242562929, | |
| "grad_norm": 1.808995008468628, | |
| "learning_rate": 2.7316326530612244e-06, | |
| "loss": 0.0091, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 6.556064073226545, | |
| "grad_norm": 0.3042183816432953, | |
| "learning_rate": 2.7252551020408168e-06, | |
| "loss": 0.0117, | |
| "step": 14325 | |
| }, | |
| { | |
| "epoch": 6.567505720823799, | |
| "grad_norm": 0.8734502196311951, | |
| "learning_rate": 2.7188775510204083e-06, | |
| "loss": 0.0099, | |
| "step": 14350 | |
| }, | |
| { | |
| "epoch": 6.578947368421053, | |
| "grad_norm": 0.13821353018283844, | |
| "learning_rate": 2.7125000000000003e-06, | |
| "loss": 0.0115, | |
| "step": 14375 | |
| }, | |
| { | |
| "epoch": 6.590389016018307, | |
| "grad_norm": 0.3783830404281616, | |
| "learning_rate": 2.706122448979592e-06, | |
| "loss": 0.0068, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 6.60183066361556, | |
| "grad_norm": 0.3584430515766144, | |
| "learning_rate": 2.699744897959184e-06, | |
| "loss": 0.0101, | |
| "step": 14425 | |
| }, | |
| { | |
| "epoch": 6.613272311212815, | |
| "grad_norm": 3.8657848834991455, | |
| "learning_rate": 2.6933673469387758e-06, | |
| "loss": 0.0098, | |
| "step": 14450 | |
| }, | |
| { | |
| "epoch": 6.6247139588100685, | |
| "grad_norm": 0.2408173382282257, | |
| "learning_rate": 2.6869897959183673e-06, | |
| "loss": 0.0112, | |
| "step": 14475 | |
| }, | |
| { | |
| "epoch": 6.636155606407323, | |
| "grad_norm": 0.08185845613479614, | |
| "learning_rate": 2.6806122448979593e-06, | |
| "loss": 0.0059, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 6.647597254004577, | |
| "grad_norm": 0.0738307312130928, | |
| "learning_rate": 2.674234693877551e-06, | |
| "loss": 0.0132, | |
| "step": 14525 | |
| }, | |
| { | |
| "epoch": 6.65903890160183, | |
| "grad_norm": 3.3000831604003906, | |
| "learning_rate": 2.667857142857143e-06, | |
| "loss": 0.01, | |
| "step": 14550 | |
| }, | |
| { | |
| "epoch": 6.670480549199085, | |
| "grad_norm": 1.2340450286865234, | |
| "learning_rate": 2.6614795918367348e-06, | |
| "loss": 0.0141, | |
| "step": 14575 | |
| }, | |
| { | |
| "epoch": 6.6819221967963385, | |
| "grad_norm": 2.966970682144165, | |
| "learning_rate": 2.6551020408163267e-06, | |
| "loss": 0.0052, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 6.693363844393593, | |
| "grad_norm": 1.181470274925232, | |
| "learning_rate": 2.6487244897959183e-06, | |
| "loss": 0.015, | |
| "step": 14625 | |
| }, | |
| { | |
| "epoch": 6.704805491990847, | |
| "grad_norm": 0.6192874908447266, | |
| "learning_rate": 2.6423469387755102e-06, | |
| "loss": 0.0063, | |
| "step": 14650 | |
| }, | |
| { | |
| "epoch": 6.7162471395881, | |
| "grad_norm": 2.0739264488220215, | |
| "learning_rate": 2.635969387755102e-06, | |
| "loss": 0.0145, | |
| "step": 14675 | |
| }, | |
| { | |
| "epoch": 6.727688787185355, | |
| "grad_norm": 1.339400291442871, | |
| "learning_rate": 2.629591836734694e-06, | |
| "loss": 0.0076, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 6.739130434782608, | |
| "grad_norm": 3.128563404083252, | |
| "learning_rate": 2.6232142857142857e-06, | |
| "loss": 0.0192, | |
| "step": 14725 | |
| }, | |
| { | |
| "epoch": 6.750572082379863, | |
| "grad_norm": 3.493795394897461, | |
| "learning_rate": 2.6168367346938777e-06, | |
| "loss": 0.005, | |
| "step": 14750 | |
| }, | |
| { | |
| "epoch": 6.762013729977117, | |
| "grad_norm": 1.3682916164398193, | |
| "learning_rate": 2.6104591836734696e-06, | |
| "loss": 0.0134, | |
| "step": 14775 | |
| }, | |
| { | |
| "epoch": 6.77345537757437, | |
| "grad_norm": 1.8985130786895752, | |
| "learning_rate": 2.6040816326530616e-06, | |
| "loss": 0.0082, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 6.784897025171625, | |
| "grad_norm": 0.5169524550437927, | |
| "learning_rate": 2.597704081632653e-06, | |
| "loss": 0.0202, | |
| "step": 14825 | |
| }, | |
| { | |
| "epoch": 6.796338672768878, | |
| "grad_norm": 0.1607659012079239, | |
| "learning_rate": 2.591326530612245e-06, | |
| "loss": 0.0084, | |
| "step": 14850 | |
| }, | |
| { | |
| "epoch": 6.807780320366133, | |
| "grad_norm": 0.6245440244674683, | |
| "learning_rate": 2.5849489795918367e-06, | |
| "loss": 0.0152, | |
| "step": 14875 | |
| }, | |
| { | |
| "epoch": 6.8192219679633865, | |
| "grad_norm": 0.2351321130990982, | |
| "learning_rate": 2.5785714285714286e-06, | |
| "loss": 0.0071, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 6.830663615560641, | |
| "grad_norm": 5.026946067810059, | |
| "learning_rate": 2.5721938775510206e-06, | |
| "loss": 0.0168, | |
| "step": 14925 | |
| }, | |
| { | |
| "epoch": 6.842105263157895, | |
| "grad_norm": 1.034845232963562, | |
| "learning_rate": 2.565816326530612e-06, | |
| "loss": 0.0084, | |
| "step": 14950 | |
| }, | |
| { | |
| "epoch": 6.853546910755149, | |
| "grad_norm": 0.16656899452209473, | |
| "learning_rate": 2.559438775510204e-06, | |
| "loss": 0.0148, | |
| "step": 14975 | |
| }, | |
| { | |
| "epoch": 6.864988558352403, | |
| "grad_norm": 7.876341819763184, | |
| "learning_rate": 2.553061224489796e-06, | |
| "loss": 0.0139, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 6.864988558352403, | |
| "eval_loss": 0.20922012627124786, | |
| "eval_runtime": 5296.0961, | |
| "eval_samples_per_second": 1.798, | |
| "eval_steps_per_second": 0.225, | |
| "eval_wer": 0.10019517501759775, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 6.8764302059496565, | |
| "grad_norm": 1.0131422281265259, | |
| "learning_rate": 2.546683673469388e-06, | |
| "loss": 0.0175, | |
| "step": 15025 | |
| }, | |
| { | |
| "epoch": 6.887871853546911, | |
| "grad_norm": 0.43789076805114746, | |
| "learning_rate": 2.5403061224489796e-06, | |
| "loss": 0.0071, | |
| "step": 15050 | |
| }, | |
| { | |
| "epoch": 6.899313501144165, | |
| "grad_norm": 0.10620521008968353, | |
| "learning_rate": 2.5339285714285716e-06, | |
| "loss": 0.0118, | |
| "step": 15075 | |
| }, | |
| { | |
| "epoch": 6.910755148741419, | |
| "grad_norm": 0.22217608988285065, | |
| "learning_rate": 2.527551020408163e-06, | |
| "loss": 0.0103, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 6.922196796338673, | |
| "grad_norm": 1.88958740234375, | |
| "learning_rate": 2.5211734693877555e-06, | |
| "loss": 0.0146, | |
| "step": 15125 | |
| }, | |
| { | |
| "epoch": 6.933638443935926, | |
| "grad_norm": 0.2760096788406372, | |
| "learning_rate": 2.514795918367347e-06, | |
| "loss": 0.0051, | |
| "step": 15150 | |
| }, | |
| { | |
| "epoch": 6.945080091533181, | |
| "grad_norm": 0.23248428106307983, | |
| "learning_rate": 2.508418367346939e-06, | |
| "loss": 0.0133, | |
| "step": 15175 | |
| }, | |
| { | |
| "epoch": 6.956521739130435, | |
| "grad_norm": 1.852455496788025, | |
| "learning_rate": 2.5020408163265305e-06, | |
| "loss": 0.0072, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 6.967963386727689, | |
| "grad_norm": 0.9175077676773071, | |
| "learning_rate": 2.495663265306123e-06, | |
| "loss": 0.0118, | |
| "step": 15225 | |
| }, | |
| { | |
| "epoch": 6.979405034324943, | |
| "grad_norm": 4.601647853851318, | |
| "learning_rate": 2.4892857142857145e-06, | |
| "loss": 0.0081, | |
| "step": 15250 | |
| }, | |
| { | |
| "epoch": 6.990846681922196, | |
| "grad_norm": 0.6206708550453186, | |
| "learning_rate": 2.4829081632653064e-06, | |
| "loss": 0.0128, | |
| "step": 15275 | |
| }, | |
| { | |
| "epoch": 7.002288329519451, | |
| "grad_norm": 1.0712556838989258, | |
| "learning_rate": 2.476530612244898e-06, | |
| "loss": 0.0097, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 7.0137299771167045, | |
| "grad_norm": 0.7450961470603943, | |
| "learning_rate": 2.47015306122449e-06, | |
| "loss": 0.0058, | |
| "step": 15325 | |
| }, | |
| { | |
| "epoch": 7.025171624713959, | |
| "grad_norm": 1.869686245918274, | |
| "learning_rate": 2.463775510204082e-06, | |
| "loss": 0.0188, | |
| "step": 15350 | |
| }, | |
| { | |
| "epoch": 7.036613272311213, | |
| "grad_norm": 0.30196335911750793, | |
| "learning_rate": 2.4573979591836735e-06, | |
| "loss": 0.0056, | |
| "step": 15375 | |
| }, | |
| { | |
| "epoch": 7.048054919908467, | |
| "grad_norm": 1.2585679292678833, | |
| "learning_rate": 2.4510204081632654e-06, | |
| "loss": 0.0092, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 7.059496567505721, | |
| "grad_norm": 0.07888475805521011, | |
| "learning_rate": 2.444642857142857e-06, | |
| "loss": 0.0066, | |
| "step": 15425 | |
| }, | |
| { | |
| "epoch": 7.0709382151029745, | |
| "grad_norm": 1.7455697059631348, | |
| "learning_rate": 2.4382653061224494e-06, | |
| "loss": 0.0107, | |
| "step": 15450 | |
| }, | |
| { | |
| "epoch": 7.082379862700229, | |
| "grad_norm": 0.1373489499092102, | |
| "learning_rate": 2.431887755102041e-06, | |
| "loss": 0.0055, | |
| "step": 15475 | |
| }, | |
| { | |
| "epoch": 7.093821510297483, | |
| "grad_norm": 2.089559555053711, | |
| "learning_rate": 2.425510204081633e-06, | |
| "loss": 0.0107, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 7.105263157894737, | |
| "grad_norm": 5.042351245880127, | |
| "learning_rate": 2.4191326530612244e-06, | |
| "loss": 0.0052, | |
| "step": 15525 | |
| }, | |
| { | |
| "epoch": 7.116704805491991, | |
| "grad_norm": 0.10954425483942032, | |
| "learning_rate": 2.4127551020408164e-06, | |
| "loss": 0.0091, | |
| "step": 15550 | |
| }, | |
| { | |
| "epoch": 7.128146453089244, | |
| "grad_norm": 0.06460034102201462, | |
| "learning_rate": 2.4063775510204083e-06, | |
| "loss": 0.0047, | |
| "step": 15575 | |
| }, | |
| { | |
| "epoch": 7.139588100686499, | |
| "grad_norm": 2.1633071899414062, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.0126, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 7.151029748283753, | |
| "grad_norm": 0.39932990074157715, | |
| "learning_rate": 2.393622448979592e-06, | |
| "loss": 0.0068, | |
| "step": 15625 | |
| }, | |
| { | |
| "epoch": 7.162471395881007, | |
| "grad_norm": 0.31883901357650757, | |
| "learning_rate": 2.387244897959184e-06, | |
| "loss": 0.0121, | |
| "step": 15650 | |
| }, | |
| { | |
| "epoch": 7.173913043478261, | |
| "grad_norm": 0.2664627134799957, | |
| "learning_rate": 2.380867346938776e-06, | |
| "loss": 0.0085, | |
| "step": 15675 | |
| }, | |
| { | |
| "epoch": 7.185354691075515, | |
| "grad_norm": 2.2720088958740234, | |
| "learning_rate": 2.3744897959183678e-06, | |
| "loss": 0.0195, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 7.196796338672769, | |
| "grad_norm": 0.13204744458198547, | |
| "learning_rate": 2.3681122448979593e-06, | |
| "loss": 0.0072, | |
| "step": 15725 | |
| }, | |
| { | |
| "epoch": 7.2082379862700225, | |
| "grad_norm": 1.5910881757736206, | |
| "learning_rate": 2.3617346938775513e-06, | |
| "loss": 0.0127, | |
| "step": 15750 | |
| }, | |
| { | |
| "epoch": 7.219679633867277, | |
| "grad_norm": 2.159229040145874, | |
| "learning_rate": 2.355357142857143e-06, | |
| "loss": 0.008, | |
| "step": 15775 | |
| }, | |
| { | |
| "epoch": 7.231121281464531, | |
| "grad_norm": 1.8656185865402222, | |
| "learning_rate": 2.3489795918367348e-06, | |
| "loss": 0.0058, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 7.242562929061785, | |
| "grad_norm": 2.223982334136963, | |
| "learning_rate": 2.3426020408163267e-06, | |
| "loss": 0.0088, | |
| "step": 15825 | |
| }, | |
| { | |
| "epoch": 7.254004576659039, | |
| "grad_norm": 0.8876869082450867, | |
| "learning_rate": 2.3362244897959183e-06, | |
| "loss": 0.0173, | |
| "step": 15850 | |
| }, | |
| { | |
| "epoch": 7.2654462242562925, | |
| "grad_norm": 0.246218740940094, | |
| "learning_rate": 2.3298469387755103e-06, | |
| "loss": 0.0043, | |
| "step": 15875 | |
| }, | |
| { | |
| "epoch": 7.276887871853547, | |
| "grad_norm": 1.6510305404663086, | |
| "learning_rate": 2.3234693877551022e-06, | |
| "loss": 0.0149, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 7.288329519450801, | |
| "grad_norm": 1.7581284046173096, | |
| "learning_rate": 2.317091836734694e-06, | |
| "loss": 0.0055, | |
| "step": 15925 | |
| }, | |
| { | |
| "epoch": 7.299771167048055, | |
| "grad_norm": 1.8961583375930786, | |
| "learning_rate": 2.3107142857142857e-06, | |
| "loss": 0.0087, | |
| "step": 15950 | |
| }, | |
| { | |
| "epoch": 7.311212814645309, | |
| "grad_norm": 0.2390100657939911, | |
| "learning_rate": 2.3043367346938777e-06, | |
| "loss": 0.0104, | |
| "step": 15975 | |
| }, | |
| { | |
| "epoch": 7.322654462242563, | |
| "grad_norm": 1.6409028768539429, | |
| "learning_rate": 2.2979591836734692e-06, | |
| "loss": 0.0062, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 7.334096109839817, | |
| "grad_norm": 0.10882013291120529, | |
| "learning_rate": 2.2915816326530616e-06, | |
| "loss": 0.0048, | |
| "step": 16025 | |
| }, | |
| { | |
| "epoch": 7.345537757437071, | |
| "grad_norm": 1.4074586629867554, | |
| "learning_rate": 2.285204081632653e-06, | |
| "loss": 0.0178, | |
| "step": 16050 | |
| }, | |
| { | |
| "epoch": 7.356979405034325, | |
| "grad_norm": 0.09742479026317596, | |
| "learning_rate": 2.278826530612245e-06, | |
| "loss": 0.0067, | |
| "step": 16075 | |
| }, | |
| { | |
| "epoch": 7.368421052631579, | |
| "grad_norm": 2.2490932941436768, | |
| "learning_rate": 2.2724489795918367e-06, | |
| "loss": 0.0142, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 7.379862700228833, | |
| "grad_norm": 0.0923023372888565, | |
| "learning_rate": 2.2660714285714287e-06, | |
| "loss": 0.0101, | |
| "step": 16125 | |
| }, | |
| { | |
| "epoch": 7.391304347826087, | |
| "grad_norm": 1.3611345291137695, | |
| "learning_rate": 2.2596938775510206e-06, | |
| "loss": 0.0118, | |
| "step": 16150 | |
| }, | |
| { | |
| "epoch": 7.4027459954233406, | |
| "grad_norm": 2.053800106048584, | |
| "learning_rate": 2.2533163265306126e-06, | |
| "loss": 0.0041, | |
| "step": 16175 | |
| }, | |
| { | |
| "epoch": 7.414187643020595, | |
| "grad_norm": 2.127472162246704, | |
| "learning_rate": 2.246938775510204e-06, | |
| "loss": 0.0177, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 7.425629290617849, | |
| "grad_norm": 4.369998931884766, | |
| "learning_rate": 2.240561224489796e-06, | |
| "loss": 0.0065, | |
| "step": 16225 | |
| }, | |
| { | |
| "epoch": 7.437070938215103, | |
| "grad_norm": 2.2975997924804688, | |
| "learning_rate": 2.234183673469388e-06, | |
| "loss": 0.0076, | |
| "step": 16250 | |
| }, | |
| { | |
| "epoch": 7.448512585812357, | |
| "grad_norm": 2.5363821983337402, | |
| "learning_rate": 2.2278061224489796e-06, | |
| "loss": 0.0111, | |
| "step": 16275 | |
| }, | |
| { | |
| "epoch": 7.459954233409611, | |
| "grad_norm": 0.09316561371088028, | |
| "learning_rate": 2.2214285714285716e-06, | |
| "loss": 0.0095, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 7.471395881006865, | |
| "grad_norm": 0.7145214676856995, | |
| "learning_rate": 2.215051020408163e-06, | |
| "loss": 0.0055, | |
| "step": 16325 | |
| }, | |
| { | |
| "epoch": 7.482837528604119, | |
| "grad_norm": 2.495028018951416, | |
| "learning_rate": 2.208673469387755e-06, | |
| "loss": 0.0112, | |
| "step": 16350 | |
| }, | |
| { | |
| "epoch": 7.494279176201373, | |
| "grad_norm": 0.3236389458179474, | |
| "learning_rate": 2.202295918367347e-06, | |
| "loss": 0.0044, | |
| "step": 16375 | |
| }, | |
| { | |
| "epoch": 7.505720823798627, | |
| "grad_norm": 1.0565695762634277, | |
| "learning_rate": 2.195918367346939e-06, | |
| "loss": 0.0129, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 7.517162471395881, | |
| "grad_norm": 1.8314106464385986, | |
| "learning_rate": 2.1895408163265306e-06, | |
| "loss": 0.0052, | |
| "step": 16425 | |
| }, | |
| { | |
| "epoch": 7.528604118993135, | |
| "grad_norm": 2.1195669174194336, | |
| "learning_rate": 2.1831632653061225e-06, | |
| "loss": 0.0212, | |
| "step": 16450 | |
| }, | |
| { | |
| "epoch": 7.540045766590389, | |
| "grad_norm": 1.8161754608154297, | |
| "learning_rate": 2.1767857142857145e-06, | |
| "loss": 0.0066, | |
| "step": 16475 | |
| }, | |
| { | |
| "epoch": 7.551487414187643, | |
| "grad_norm": 0.14074856042861938, | |
| "learning_rate": 2.1704081632653065e-06, | |
| "loss": 0.0068, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 7.562929061784897, | |
| "grad_norm": 0.07693963497877121, | |
| "learning_rate": 2.164030612244898e-06, | |
| "loss": 0.005, | |
| "step": 16525 | |
| }, | |
| { | |
| "epoch": 7.574370709382151, | |
| "grad_norm": 1.0592836141586304, | |
| "learning_rate": 2.15765306122449e-06, | |
| "loss": 0.0124, | |
| "step": 16550 | |
| }, | |
| { | |
| "epoch": 7.585812356979405, | |
| "grad_norm": 3.276205539703369, | |
| "learning_rate": 2.1512755102040815e-06, | |
| "loss": 0.0077, | |
| "step": 16575 | |
| }, | |
| { | |
| "epoch": 7.597254004576659, | |
| "grad_norm": 0.2712470293045044, | |
| "learning_rate": 2.1448979591836735e-06, | |
| "loss": 0.0137, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 7.608695652173913, | |
| "grad_norm": 2.913771629333496, | |
| "learning_rate": 2.1385204081632655e-06, | |
| "loss": 0.0084, | |
| "step": 16625 | |
| }, | |
| { | |
| "epoch": 7.620137299771167, | |
| "grad_norm": 0.8734631538391113, | |
| "learning_rate": 2.132142857142857e-06, | |
| "loss": 0.0162, | |
| "step": 16650 | |
| }, | |
| { | |
| "epoch": 7.631578947368421, | |
| "grad_norm": 2.229383707046509, | |
| "learning_rate": 2.125765306122449e-06, | |
| "loss": 0.0089, | |
| "step": 16675 | |
| }, | |
| { | |
| "epoch": 7.643020594965675, | |
| "grad_norm": 1.0993914604187012, | |
| "learning_rate": 2.119387755102041e-06, | |
| "loss": 0.0105, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 7.654462242562929, | |
| "grad_norm": 0.42427483201026917, | |
| "learning_rate": 2.113010204081633e-06, | |
| "loss": 0.0063, | |
| "step": 16725 | |
| }, | |
| { | |
| "epoch": 7.665903890160183, | |
| "grad_norm": 1.0200995206832886, | |
| "learning_rate": 2.1066326530612244e-06, | |
| "loss": 0.012, | |
| "step": 16750 | |
| }, | |
| { | |
| "epoch": 7.6773455377574376, | |
| "grad_norm": 0.11753875017166138, | |
| "learning_rate": 2.1002551020408164e-06, | |
| "loss": 0.0069, | |
| "step": 16775 | |
| }, | |
| { | |
| "epoch": 7.688787185354691, | |
| "grad_norm": 1.3853702545166016, | |
| "learning_rate": 2.0941326530612246e-06, | |
| "loss": 0.015, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 7.700228832951945, | |
| "grad_norm": 0.7108054757118225, | |
| "learning_rate": 2.0877551020408165e-06, | |
| "loss": 0.0094, | |
| "step": 16825 | |
| }, | |
| { | |
| "epoch": 7.711670480549199, | |
| "grad_norm": 0.9410815834999084, | |
| "learning_rate": 2.081377551020408e-06, | |
| "loss": 0.0091, | |
| "step": 16850 | |
| }, | |
| { | |
| "epoch": 7.723112128146453, | |
| "grad_norm": 1.315185308456421, | |
| "learning_rate": 2.075e-06, | |
| "loss": 0.0066, | |
| "step": 16875 | |
| }, | |
| { | |
| "epoch": 7.7345537757437075, | |
| "grad_norm": 2.3978397846221924, | |
| "learning_rate": 2.068622448979592e-06, | |
| "loss": 0.0082, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 7.745995423340961, | |
| "grad_norm": 0.3152960538864136, | |
| "learning_rate": 2.0622448979591835e-06, | |
| "loss": 0.0045, | |
| "step": 16925 | |
| }, | |
| { | |
| "epoch": 7.757437070938215, | |
| "grad_norm": 3.8207294940948486, | |
| "learning_rate": 2.0558673469387755e-06, | |
| "loss": 0.0139, | |
| "step": 16950 | |
| }, | |
| { | |
| "epoch": 7.768878718535469, | |
| "grad_norm": 0.7664352059364319, | |
| "learning_rate": 2.0494897959183675e-06, | |
| "loss": 0.0082, | |
| "step": 16975 | |
| }, | |
| { | |
| "epoch": 7.780320366132723, | |
| "grad_norm": 2.621455192565918, | |
| "learning_rate": 2.0431122448979594e-06, | |
| "loss": 0.0138, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 7.7917620137299775, | |
| "grad_norm": 2.7599635124206543, | |
| "learning_rate": 2.036734693877551e-06, | |
| "loss": 0.006, | |
| "step": 17025 | |
| }, | |
| { | |
| "epoch": 7.803203661327231, | |
| "grad_norm": 2.1494710445404053, | |
| "learning_rate": 2.030357142857143e-06, | |
| "loss": 0.0133, | |
| "step": 17050 | |
| }, | |
| { | |
| "epoch": 7.814645308924485, | |
| "grad_norm": 2.6911447048187256, | |
| "learning_rate": 2.0239795918367345e-06, | |
| "loss": 0.0079, | |
| "step": 17075 | |
| }, | |
| { | |
| "epoch": 7.826086956521739, | |
| "grad_norm": 1.8577873706817627, | |
| "learning_rate": 2.017602040816327e-06, | |
| "loss": 0.0098, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 7.837528604118993, | |
| "grad_norm": 0.08260419219732285, | |
| "learning_rate": 2.0112244897959184e-06, | |
| "loss": 0.0073, | |
| "step": 17125 | |
| }, | |
| { | |
| "epoch": 7.848970251716247, | |
| "grad_norm": 0.34523025155067444, | |
| "learning_rate": 2.0048469387755104e-06, | |
| "loss": 0.0097, | |
| "step": 17150 | |
| }, | |
| { | |
| "epoch": 7.860411899313501, | |
| "grad_norm": 2.8545644283294678, | |
| "learning_rate": 1.998469387755102e-06, | |
| "loss": 0.0093, | |
| "step": 17175 | |
| }, | |
| { | |
| "epoch": 7.871853546910755, | |
| "grad_norm": 0.9209850430488586, | |
| "learning_rate": 1.9920918367346943e-06, | |
| "loss": 0.0193, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 7.883295194508009, | |
| "grad_norm": 0.15244260430335999, | |
| "learning_rate": 1.985714285714286e-06, | |
| "loss": 0.0074, | |
| "step": 17225 | |
| }, | |
| { | |
| "epoch": 7.894736842105263, | |
| "grad_norm": 0.9773492813110352, | |
| "learning_rate": 1.979336734693878e-06, | |
| "loss": 0.0102, | |
| "step": 17250 | |
| }, | |
| { | |
| "epoch": 7.906178489702517, | |
| "grad_norm": 3.8897087574005127, | |
| "learning_rate": 1.9729591836734694e-06, | |
| "loss": 0.0092, | |
| "step": 17275 | |
| }, | |
| { | |
| "epoch": 7.917620137299771, | |
| "grad_norm": 1.736915946006775, | |
| "learning_rate": 1.9665816326530614e-06, | |
| "loss": 0.0087, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 7.9290617848970255, | |
| "grad_norm": 0.2637374699115753, | |
| "learning_rate": 1.9602040816326533e-06, | |
| "loss": 0.0056, | |
| "step": 17325 | |
| }, | |
| { | |
| "epoch": 7.940503432494279, | |
| "grad_norm": 0.23918192088603973, | |
| "learning_rate": 1.953826530612245e-06, | |
| "loss": 0.0101, | |
| "step": 17350 | |
| }, | |
| { | |
| "epoch": 7.951945080091534, | |
| "grad_norm": 2.3854148387908936, | |
| "learning_rate": 1.947448979591837e-06, | |
| "loss": 0.0082, | |
| "step": 17375 | |
| }, | |
| { | |
| "epoch": 7.963386727688787, | |
| "grad_norm": 1.505005955696106, | |
| "learning_rate": 1.9410714285714284e-06, | |
| "loss": 0.0092, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 7.974828375286041, | |
| "grad_norm": 5.391610145568848, | |
| "learning_rate": 1.9346938775510208e-06, | |
| "loss": 0.0088, | |
| "step": 17425 | |
| }, | |
| { | |
| "epoch": 7.9862700228832955, | |
| "grad_norm": 0.1221591904759407, | |
| "learning_rate": 1.9283163265306123e-06, | |
| "loss": 0.0139, | |
| "step": 17450 | |
| }, | |
| { | |
| "epoch": 7.997711670480549, | |
| "grad_norm": 3.126054286956787, | |
| "learning_rate": 1.9219387755102043e-06, | |
| "loss": 0.006, | |
| "step": 17475 | |
| }, | |
| { | |
| "epoch": 8.009153318077804, | |
| "grad_norm": 0.13957707583904266, | |
| "learning_rate": 1.915561224489796e-06, | |
| "loss": 0.0126, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 8.020594965675057, | |
| "grad_norm": 0.3743535876274109, | |
| "learning_rate": 1.909183673469388e-06, | |
| "loss": 0.0066, | |
| "step": 17525 | |
| }, | |
| { | |
| "epoch": 8.03203661327231, | |
| "grad_norm": 2.4367594718933105, | |
| "learning_rate": 1.9028061224489798e-06, | |
| "loss": 0.0095, | |
| "step": 17550 | |
| }, | |
| { | |
| "epoch": 8.043478260869565, | |
| "grad_norm": 0.27416449785232544, | |
| "learning_rate": 1.8964285714285715e-06, | |
| "loss": 0.0061, | |
| "step": 17575 | |
| }, | |
| { | |
| "epoch": 8.05491990846682, | |
| "grad_norm": 2.7768595218658447, | |
| "learning_rate": 1.8900510204081633e-06, | |
| "loss": 0.0079, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 8.066361556064074, | |
| "grad_norm": 1.4709582328796387, | |
| "learning_rate": 1.883673469387755e-06, | |
| "loss": 0.0077, | |
| "step": 17625 | |
| }, | |
| { | |
| "epoch": 8.077803203661327, | |
| "grad_norm": 2.9024922847747803, | |
| "learning_rate": 1.8772959183673472e-06, | |
| "loss": 0.0044, | |
| "step": 17650 | |
| }, | |
| { | |
| "epoch": 8.08924485125858, | |
| "grad_norm": 1.4688998460769653, | |
| "learning_rate": 1.870918367346939e-06, | |
| "loss": 0.0051, | |
| "step": 17675 | |
| }, | |
| { | |
| "epoch": 8.100686498855834, | |
| "grad_norm": 0.9765781760215759, | |
| "learning_rate": 1.8645408163265307e-06, | |
| "loss": 0.0117, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 8.11212814645309, | |
| "grad_norm": 0.23804131150245667, | |
| "learning_rate": 1.8581632653061225e-06, | |
| "loss": 0.0079, | |
| "step": 17725 | |
| }, | |
| { | |
| "epoch": 8.123569794050344, | |
| "grad_norm": 0.10259401798248291, | |
| "learning_rate": 1.8517857142857142e-06, | |
| "loss": 0.0063, | |
| "step": 17750 | |
| }, | |
| { | |
| "epoch": 8.135011441647597, | |
| "grad_norm": 3.3533835411071777, | |
| "learning_rate": 1.8454081632653064e-06, | |
| "loss": 0.0072, | |
| "step": 17775 | |
| }, | |
| { | |
| "epoch": 8.14645308924485, | |
| "grad_norm": 0.26105380058288574, | |
| "learning_rate": 1.8390306122448982e-06, | |
| "loss": 0.0123, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 8.157894736842104, | |
| "grad_norm": 3.717785358428955, | |
| "learning_rate": 1.83265306122449e-06, | |
| "loss": 0.0057, | |
| "step": 17825 | |
| }, | |
| { | |
| "epoch": 8.16933638443936, | |
| "grad_norm": 0.14977721869945526, | |
| "learning_rate": 1.8262755102040817e-06, | |
| "loss": 0.0099, | |
| "step": 17850 | |
| }, | |
| { | |
| "epoch": 8.180778032036613, | |
| "grad_norm": 0.706764817237854, | |
| "learning_rate": 1.8198979591836736e-06, | |
| "loss": 0.0088, | |
| "step": 17875 | |
| }, | |
| { | |
| "epoch": 8.192219679633867, | |
| "grad_norm": 0.2173665463924408, | |
| "learning_rate": 1.8135204081632654e-06, | |
| "loss": 0.0047, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 8.20366132723112, | |
| "grad_norm": 1.6431961059570312, | |
| "learning_rate": 1.8071428571428571e-06, | |
| "loss": 0.0079, | |
| "step": 17925 | |
| }, | |
| { | |
| "epoch": 8.215102974828376, | |
| "grad_norm": 1.231646180152893, | |
| "learning_rate": 1.8007653061224491e-06, | |
| "loss": 0.0086, | |
| "step": 17950 | |
| }, | |
| { | |
| "epoch": 8.22654462242563, | |
| "grad_norm": 0.3211919069290161, | |
| "learning_rate": 1.7943877551020409e-06, | |
| "loss": 0.0093, | |
| "step": 17975 | |
| }, | |
| { | |
| "epoch": 8.237986270022883, | |
| "grad_norm": 0.0584256611764431, | |
| "learning_rate": 1.7880102040816328e-06, | |
| "loss": 0.0072, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 8.249427917620137, | |
| "grad_norm": 1.3097387552261353, | |
| "learning_rate": 1.7816326530612246e-06, | |
| "loss": 0.0064, | |
| "step": 18025 | |
| }, | |
| { | |
| "epoch": 8.26086956521739, | |
| "grad_norm": 2.2777936458587646, | |
| "learning_rate": 1.7752551020408163e-06, | |
| "loss": 0.0062, | |
| "step": 18050 | |
| }, | |
| { | |
| "epoch": 8.272311212814646, | |
| "grad_norm": 2.830906867980957, | |
| "learning_rate": 1.768877551020408e-06, | |
| "loss": 0.0071, | |
| "step": 18075 | |
| }, | |
| { | |
| "epoch": 8.2837528604119, | |
| "grad_norm": 1.2469624280929565, | |
| "learning_rate": 1.7624999999999999e-06, | |
| "loss": 0.0086, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 8.295194508009153, | |
| "grad_norm": 1.5804486274719238, | |
| "learning_rate": 1.756122448979592e-06, | |
| "loss": 0.0052, | |
| "step": 18125 | |
| }, | |
| { | |
| "epoch": 8.306636155606407, | |
| "grad_norm": 2.0198347568511963, | |
| "learning_rate": 1.7497448979591838e-06, | |
| "loss": 0.009, | |
| "step": 18150 | |
| }, | |
| { | |
| "epoch": 8.31807780320366, | |
| "grad_norm": 1.086702585220337, | |
| "learning_rate": 1.7433673469387755e-06, | |
| "loss": 0.0034, | |
| "step": 18175 | |
| }, | |
| { | |
| "epoch": 8.329519450800916, | |
| "grad_norm": 0.40313854813575745, | |
| "learning_rate": 1.7369897959183673e-06, | |
| "loss": 0.011, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 8.34096109839817, | |
| "grad_norm": 1.1296424865722656, | |
| "learning_rate": 1.7306122448979595e-06, | |
| "loss": 0.0051, | |
| "step": 18225 | |
| }, | |
| { | |
| "epoch": 8.352402745995423, | |
| "grad_norm": 0.12974560260772705, | |
| "learning_rate": 1.7242346938775512e-06, | |
| "loss": 0.0103, | |
| "step": 18250 | |
| }, | |
| { | |
| "epoch": 8.363844393592677, | |
| "grad_norm": 6.7116265296936035, | |
| "learning_rate": 1.717857142857143e-06, | |
| "loss": 0.0076, | |
| "step": 18275 | |
| }, | |
| { | |
| "epoch": 8.37528604118993, | |
| "grad_norm": 3.5247104167938232, | |
| "learning_rate": 1.7114795918367347e-06, | |
| "loss": 0.0096, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 8.386727688787186, | |
| "grad_norm": 1.0144484043121338, | |
| "learning_rate": 1.7051020408163265e-06, | |
| "loss": 0.0073, | |
| "step": 18325 | |
| }, | |
| { | |
| "epoch": 8.39816933638444, | |
| "grad_norm": 0.3811889886856079, | |
| "learning_rate": 1.6987244897959185e-06, | |
| "loss": 0.0056, | |
| "step": 18350 | |
| }, | |
| { | |
| "epoch": 8.409610983981693, | |
| "grad_norm": 0.449440598487854, | |
| "learning_rate": 1.6923469387755102e-06, | |
| "loss": 0.0056, | |
| "step": 18375 | |
| }, | |
| { | |
| "epoch": 8.421052631578947, | |
| "grad_norm": 1.2429654598236084, | |
| "learning_rate": 1.685969387755102e-06, | |
| "loss": 0.0089, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 8.4324942791762, | |
| "grad_norm": 0.530453085899353, | |
| "learning_rate": 1.6795918367346937e-06, | |
| "loss": 0.0054, | |
| "step": 18425 | |
| }, | |
| { | |
| "epoch": 8.443935926773456, | |
| "grad_norm": 0.31273970007896423, | |
| "learning_rate": 1.673214285714286e-06, | |
| "loss": 0.0072, | |
| "step": 18450 | |
| }, | |
| { | |
| "epoch": 8.45537757437071, | |
| "grad_norm": 5.236429214477539, | |
| "learning_rate": 1.6668367346938777e-06, | |
| "loss": 0.005, | |
| "step": 18475 | |
| }, | |
| { | |
| "epoch": 8.466819221967963, | |
| "grad_norm": 0.06535414606332779, | |
| "learning_rate": 1.6604591836734694e-06, | |
| "loss": 0.0065, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 8.478260869565217, | |
| "grad_norm": 1.5802868604660034, | |
| "learning_rate": 1.6540816326530612e-06, | |
| "loss": 0.0063, | |
| "step": 18525 | |
| }, | |
| { | |
| "epoch": 8.48970251716247, | |
| "grad_norm": 1.6625992059707642, | |
| "learning_rate": 1.647704081632653e-06, | |
| "loss": 0.0106, | |
| "step": 18550 | |
| }, | |
| { | |
| "epoch": 8.501144164759726, | |
| "grad_norm": 0.21594946086406708, | |
| "learning_rate": 1.6413265306122451e-06, | |
| "loss": 0.0036, | |
| "step": 18575 | |
| }, | |
| { | |
| "epoch": 8.51258581235698, | |
| "grad_norm": 0.22096717357635498, | |
| "learning_rate": 1.6349489795918369e-06, | |
| "loss": 0.009, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 8.524027459954233, | |
| "grad_norm": 0.08814064413309097, | |
| "learning_rate": 1.6285714285714286e-06, | |
| "loss": 0.0021, | |
| "step": 18625 | |
| }, | |
| { | |
| "epoch": 8.535469107551487, | |
| "grad_norm": 0.07081279903650284, | |
| "learning_rate": 1.6221938775510204e-06, | |
| "loss": 0.0077, | |
| "step": 18650 | |
| }, | |
| { | |
| "epoch": 8.546910755148742, | |
| "grad_norm": 0.4350265562534332, | |
| "learning_rate": 1.6158163265306126e-06, | |
| "loss": 0.0069, | |
| "step": 18675 | |
| }, | |
| { | |
| "epoch": 8.558352402745996, | |
| "grad_norm": 1.913025975227356, | |
| "learning_rate": 1.6094387755102043e-06, | |
| "loss": 0.0094, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 8.56979405034325, | |
| "grad_norm": 0.0708305835723877, | |
| "learning_rate": 1.603061224489796e-06, | |
| "loss": 0.0066, | |
| "step": 18725 | |
| }, | |
| { | |
| "epoch": 8.581235697940503, | |
| "grad_norm": 0.2957993149757385, | |
| "learning_rate": 1.5966836734693878e-06, | |
| "loss": 0.0103, | |
| "step": 18750 | |
| }, | |
| { | |
| "epoch": 8.592677345537757, | |
| "grad_norm": 0.3358778953552246, | |
| "learning_rate": 1.5903061224489796e-06, | |
| "loss": 0.0064, | |
| "step": 18775 | |
| }, | |
| { | |
| "epoch": 8.604118993135012, | |
| "grad_norm": 6.7304887771606445, | |
| "learning_rate": 1.5839285714285715e-06, | |
| "loss": 0.0093, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 8.615560640732266, | |
| "grad_norm": 3.724275588989258, | |
| "learning_rate": 1.5775510204081633e-06, | |
| "loss": 0.0084, | |
| "step": 18825 | |
| }, | |
| { | |
| "epoch": 8.62700228832952, | |
| "grad_norm": 0.04007994756102562, | |
| "learning_rate": 1.571173469387755e-06, | |
| "loss": 0.0136, | |
| "step": 18850 | |
| }, | |
| { | |
| "epoch": 8.638443935926773, | |
| "grad_norm": 0.05183762311935425, | |
| "learning_rate": 1.5647959183673468e-06, | |
| "loss": 0.0061, | |
| "step": 18875 | |
| }, | |
| { | |
| "epoch": 8.649885583524027, | |
| "grad_norm": 3.223602771759033, | |
| "learning_rate": 1.5584183673469388e-06, | |
| "loss": 0.0098, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 8.661327231121282, | |
| "grad_norm": 3.744605541229248, | |
| "learning_rate": 1.5520408163265307e-06, | |
| "loss": 0.0106, | |
| "step": 18925 | |
| }, | |
| { | |
| "epoch": 8.672768878718536, | |
| "grad_norm": 0.07725165784358978, | |
| "learning_rate": 1.5456632653061225e-06, | |
| "loss": 0.0076, | |
| "step": 18950 | |
| }, | |
| { | |
| "epoch": 8.68421052631579, | |
| "grad_norm": 14.24994945526123, | |
| "learning_rate": 1.5392857142857143e-06, | |
| "loss": 0.0086, | |
| "step": 18975 | |
| }, | |
| { | |
| "epoch": 8.695652173913043, | |
| "grad_norm": 0.2724073529243469, | |
| "learning_rate": 1.5329081632653062e-06, | |
| "loss": 0.0141, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 8.707093821510298, | |
| "grad_norm": 0.03878064826130867, | |
| "learning_rate": 1.526530612244898e-06, | |
| "loss": 0.0071, | |
| "step": 19025 | |
| }, | |
| { | |
| "epoch": 8.718535469107552, | |
| "grad_norm": 0.631288468837738, | |
| "learning_rate": 1.5204081632653061e-06, | |
| "loss": 0.011, | |
| "step": 19050 | |
| }, | |
| { | |
| "epoch": 8.729977116704806, | |
| "grad_norm": 9.200640678405762, | |
| "learning_rate": 1.5140306122448979e-06, | |
| "loss": 0.009, | |
| "step": 19075 | |
| }, | |
| { | |
| "epoch": 8.74141876430206, | |
| "grad_norm": 0.18505507707595825, | |
| "learning_rate": 1.5076530612244898e-06, | |
| "loss": 0.0084, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 8.752860411899313, | |
| "grad_norm": 5.33227014541626, | |
| "learning_rate": 1.5012755102040816e-06, | |
| "loss": 0.0057, | |
| "step": 19125 | |
| }, | |
| { | |
| "epoch": 8.764302059496568, | |
| "grad_norm": 0.31844860315322876, | |
| "learning_rate": 1.4948979591836736e-06, | |
| "loss": 0.0081, | |
| "step": 19150 | |
| }, | |
| { | |
| "epoch": 8.775743707093822, | |
| "grad_norm": 0.5912889242172241, | |
| "learning_rate": 1.4885204081632653e-06, | |
| "loss": 0.0083, | |
| "step": 19175 | |
| }, | |
| { | |
| "epoch": 8.787185354691076, | |
| "grad_norm": 0.05793588235974312, | |
| "learning_rate": 1.4821428571428573e-06, | |
| "loss": 0.0061, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 8.79862700228833, | |
| "grad_norm": 3.5258617401123047, | |
| "learning_rate": 1.475765306122449e-06, | |
| "loss": 0.0091, | |
| "step": 19225 | |
| }, | |
| { | |
| "epoch": 8.810068649885583, | |
| "grad_norm": 0.5657800436019897, | |
| "learning_rate": 1.4693877551020408e-06, | |
| "loss": 0.0082, | |
| "step": 19250 | |
| }, | |
| { | |
| "epoch": 8.821510297482838, | |
| "grad_norm": 0.11896498501300812, | |
| "learning_rate": 1.4630102040816328e-06, | |
| "loss": 0.0061, | |
| "step": 19275 | |
| }, | |
| { | |
| "epoch": 8.832951945080092, | |
| "grad_norm": 2.1698849201202393, | |
| "learning_rate": 1.4566326530612245e-06, | |
| "loss": 0.0071, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 8.844393592677346, | |
| "grad_norm": 2.6796391010284424, | |
| "learning_rate": 1.4502551020408165e-06, | |
| "loss": 0.0064, | |
| "step": 19325 | |
| }, | |
| { | |
| "epoch": 8.8558352402746, | |
| "grad_norm": 3.8971478939056396, | |
| "learning_rate": 1.4438775510204082e-06, | |
| "loss": 0.0071, | |
| "step": 19350 | |
| }, | |
| { | |
| "epoch": 8.867276887871853, | |
| "grad_norm": 3.8938040733337402, | |
| "learning_rate": 1.4375000000000002e-06, | |
| "loss": 0.0034, | |
| "step": 19375 | |
| }, | |
| { | |
| "epoch": 8.878718535469108, | |
| "grad_norm": 0.1528436839580536, | |
| "learning_rate": 1.431122448979592e-06, | |
| "loss": 0.0083, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 8.890160183066362, | |
| "grad_norm": 0.024963609874248505, | |
| "learning_rate": 1.4247448979591837e-06, | |
| "loss": 0.0048, | |
| "step": 19425 | |
| }, | |
| { | |
| "epoch": 8.901601830663616, | |
| "grad_norm": 0.10767273604869843, | |
| "learning_rate": 1.4183673469387757e-06, | |
| "loss": 0.0048, | |
| "step": 19450 | |
| }, | |
| { | |
| "epoch": 8.91304347826087, | |
| "grad_norm": 0.09147515892982483, | |
| "learning_rate": 1.4119897959183674e-06, | |
| "loss": 0.0054, | |
| "step": 19475 | |
| }, | |
| { | |
| "epoch": 8.924485125858123, | |
| "grad_norm": 0.6032863855361938, | |
| "learning_rate": 1.4056122448979592e-06, | |
| "loss": 0.0112, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 8.935926773455378, | |
| "grad_norm": 0.663342297077179, | |
| "learning_rate": 1.399234693877551e-06, | |
| "loss": 0.0089, | |
| "step": 19525 | |
| }, | |
| { | |
| "epoch": 8.947368421052632, | |
| "grad_norm": 0.2562825381755829, | |
| "learning_rate": 1.392857142857143e-06, | |
| "loss": 0.0052, | |
| "step": 19550 | |
| }, | |
| { | |
| "epoch": 8.958810068649885, | |
| "grad_norm": 1.2865808010101318, | |
| "learning_rate": 1.3864795918367347e-06, | |
| "loss": 0.0105, | |
| "step": 19575 | |
| }, | |
| { | |
| "epoch": 8.97025171624714, | |
| "grad_norm": 6.299253940582275, | |
| "learning_rate": 1.3801020408163266e-06, | |
| "loss": 0.0086, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 8.981693363844393, | |
| "grad_norm": 0.08225379884243011, | |
| "learning_rate": 1.3737244897959184e-06, | |
| "loss": 0.0036, | |
| "step": 19625 | |
| }, | |
| { | |
| "epoch": 8.993135011441648, | |
| "grad_norm": 0.18679049611091614, | |
| "learning_rate": 1.3673469387755102e-06, | |
| "loss": 0.0039, | |
| "step": 19650 | |
| }, | |
| { | |
| "epoch": 9.004576659038902, | |
| "grad_norm": 0.39510053396224976, | |
| "learning_rate": 1.3609693877551021e-06, | |
| "loss": 0.0088, | |
| "step": 19675 | |
| }, | |
| { | |
| "epoch": 9.016018306636155, | |
| "grad_norm": 0.06154235452413559, | |
| "learning_rate": 1.3545918367346939e-06, | |
| "loss": 0.0053, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 9.027459954233409, | |
| "grad_norm": 0.6689733862876892, | |
| "learning_rate": 1.3482142857142858e-06, | |
| "loss": 0.0163, | |
| "step": 19725 | |
| }, | |
| { | |
| "epoch": 9.038901601830664, | |
| "grad_norm": 0.4997970759868622, | |
| "learning_rate": 1.3418367346938776e-06, | |
| "loss": 0.0049, | |
| "step": 19750 | |
| }, | |
| { | |
| "epoch": 9.050343249427918, | |
| "grad_norm": 0.23966868221759796, | |
| "learning_rate": 1.3354591836734696e-06, | |
| "loss": 0.0061, | |
| "step": 19775 | |
| }, | |
| { | |
| "epoch": 9.061784897025172, | |
| "grad_norm": 3.6731507778167725, | |
| "learning_rate": 1.3290816326530613e-06, | |
| "loss": 0.0085, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 9.073226544622425, | |
| "grad_norm": 1.2267433404922485, | |
| "learning_rate": 1.3227040816326533e-06, | |
| "loss": 0.0082, | |
| "step": 19825 | |
| }, | |
| { | |
| "epoch": 9.084668192219679, | |
| "grad_norm": 3.816725492477417, | |
| "learning_rate": 1.316326530612245e-06, | |
| "loss": 0.0031, | |
| "step": 19850 | |
| }, | |
| { | |
| "epoch": 9.096109839816934, | |
| "grad_norm": 0.5195302963256836, | |
| "learning_rate": 1.3099489795918368e-06, | |
| "loss": 0.0081, | |
| "step": 19875 | |
| }, | |
| { | |
| "epoch": 9.107551487414188, | |
| "grad_norm": 0.04216615855693817, | |
| "learning_rate": 1.3035714285714286e-06, | |
| "loss": 0.006, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 9.118993135011442, | |
| "grad_norm": 0.15574586391448975, | |
| "learning_rate": 1.2971938775510203e-06, | |
| "loss": 0.0089, | |
| "step": 19925 | |
| }, | |
| { | |
| "epoch": 9.130434782608695, | |
| "grad_norm": 0.7792266011238098, | |
| "learning_rate": 1.2908163265306123e-06, | |
| "loss": 0.0054, | |
| "step": 19950 | |
| }, | |
| { | |
| "epoch": 9.141876430205949, | |
| "grad_norm": 0.07065439969301224, | |
| "learning_rate": 1.284438775510204e-06, | |
| "loss": 0.0075, | |
| "step": 19975 | |
| }, | |
| { | |
| "epoch": 9.153318077803204, | |
| "grad_norm": 1.0782091617584229, | |
| "learning_rate": 1.278061224489796e-06, | |
| "loss": 0.0044, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 9.153318077803204, | |
| "eval_loss": 0.20870625972747803, | |
| "eval_runtime": 5198.5381, | |
| "eval_samples_per_second": 1.832, | |
| "eval_steps_per_second": 0.229, | |
| "eval_wer": 0.0960037115249248, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 9.164759725400458, | |
| "grad_norm": 2.809683322906494, | |
| "learning_rate": 1.2716836734693878e-06, | |
| "loss": 0.0099, | |
| "step": 20025 | |
| }, | |
| { | |
| "epoch": 9.176201372997712, | |
| "grad_norm": 0.10887058824300766, | |
| "learning_rate": 1.2653061224489797e-06, | |
| "loss": 0.0042, | |
| "step": 20050 | |
| }, | |
| { | |
| "epoch": 9.187643020594965, | |
| "grad_norm": 1.1026456356048584, | |
| "learning_rate": 1.2589285714285715e-06, | |
| "loss": 0.0086, | |
| "step": 20075 | |
| }, | |
| { | |
| "epoch": 9.199084668192219, | |
| "grad_norm": 2.4570958614349365, | |
| "learning_rate": 1.2525510204081632e-06, | |
| "loss": 0.0041, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 9.210526315789474, | |
| "grad_norm": 0.12701238691806793, | |
| "learning_rate": 1.2461734693877552e-06, | |
| "loss": 0.0069, | |
| "step": 20125 | |
| }, | |
| { | |
| "epoch": 9.221967963386728, | |
| "grad_norm": 0.08014000952243805, | |
| "learning_rate": 1.239795918367347e-06, | |
| "loss": 0.0038, | |
| "step": 20150 | |
| }, | |
| { | |
| "epoch": 9.233409610983982, | |
| "grad_norm": 0.03979151323437691, | |
| "learning_rate": 1.233418367346939e-06, | |
| "loss": 0.0055, | |
| "step": 20175 | |
| }, | |
| { | |
| "epoch": 9.244851258581235, | |
| "grad_norm": 0.17478398978710175, | |
| "learning_rate": 1.2270408163265307e-06, | |
| "loss": 0.0042, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 9.256292906178489, | |
| "grad_norm": 1.452242136001587, | |
| "learning_rate": 1.2206632653061226e-06, | |
| "loss": 0.0062, | |
| "step": 20225 | |
| }, | |
| { | |
| "epoch": 9.267734553775744, | |
| "grad_norm": 1.9169418811798096, | |
| "learning_rate": 1.2142857142857144e-06, | |
| "loss": 0.005, | |
| "step": 20250 | |
| }, | |
| { | |
| "epoch": 9.279176201372998, | |
| "grad_norm": 0.07548157870769501, | |
| "learning_rate": 1.2079081632653064e-06, | |
| "loss": 0.0065, | |
| "step": 20275 | |
| }, | |
| { | |
| "epoch": 9.290617848970252, | |
| "grad_norm": 0.0355396643280983, | |
| "learning_rate": 1.2015306122448981e-06, | |
| "loss": 0.0059, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 9.302059496567505, | |
| "grad_norm": 0.0616096630692482, | |
| "learning_rate": 1.1951530612244899e-06, | |
| "loss": 0.0067, | |
| "step": 20325 | |
| }, | |
| { | |
| "epoch": 9.31350114416476, | |
| "grad_norm": 4.230183124542236, | |
| "learning_rate": 1.1887755102040816e-06, | |
| "loss": 0.0054, | |
| "step": 20350 | |
| }, | |
| { | |
| "epoch": 9.324942791762014, | |
| "grad_norm": 4.540670871734619, | |
| "learning_rate": 1.1823979591836734e-06, | |
| "loss": 0.0102, | |
| "step": 20375 | |
| }, | |
| { | |
| "epoch": 9.336384439359268, | |
| "grad_norm": 0.9268777370452881, | |
| "learning_rate": 1.1760204081632654e-06, | |
| "loss": 0.0075, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 9.347826086956522, | |
| "grad_norm": 1.7255213260650635, | |
| "learning_rate": 1.1696428571428571e-06, | |
| "loss": 0.0065, | |
| "step": 20425 | |
| }, | |
| { | |
| "epoch": 9.359267734553775, | |
| "grad_norm": 2.455357074737549, | |
| "learning_rate": 1.163265306122449e-06, | |
| "loss": 0.0031, | |
| "step": 20450 | |
| }, | |
| { | |
| "epoch": 9.37070938215103, | |
| "grad_norm": 0.6424604058265686, | |
| "learning_rate": 1.1568877551020408e-06, | |
| "loss": 0.0111, | |
| "step": 20475 | |
| }, | |
| { | |
| "epoch": 9.382151029748284, | |
| "grad_norm": 5.830212116241455, | |
| "learning_rate": 1.1505102040816326e-06, | |
| "loss": 0.0076, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 9.393592677345538, | |
| "grad_norm": 0.1393570601940155, | |
| "learning_rate": 1.1441326530612246e-06, | |
| "loss": 0.0067, | |
| "step": 20525 | |
| }, | |
| { | |
| "epoch": 9.405034324942791, | |
| "grad_norm": 0.03577861189842224, | |
| "learning_rate": 1.1377551020408163e-06, | |
| "loss": 0.003, | |
| "step": 20550 | |
| }, | |
| { | |
| "epoch": 9.416475972540045, | |
| "grad_norm": 0.26087144017219543, | |
| "learning_rate": 1.1313775510204083e-06, | |
| "loss": 0.0056, | |
| "step": 20575 | |
| }, | |
| { | |
| "epoch": 9.4279176201373, | |
| "grad_norm": 0.6032670140266418, | |
| "learning_rate": 1.125e-06, | |
| "loss": 0.0038, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 9.439359267734554, | |
| "grad_norm": 0.40521496534347534, | |
| "learning_rate": 1.118622448979592e-06, | |
| "loss": 0.0042, | |
| "step": 20625 | |
| }, | |
| { | |
| "epoch": 9.450800915331808, | |
| "grad_norm": 0.04099424555897713, | |
| "learning_rate": 1.1122448979591838e-06, | |
| "loss": 0.0049, | |
| "step": 20650 | |
| }, | |
| { | |
| "epoch": 9.462242562929061, | |
| "grad_norm": 0.08814136683940887, | |
| "learning_rate": 1.1058673469387757e-06, | |
| "loss": 0.0113, | |
| "step": 20675 | |
| }, | |
| { | |
| "epoch": 9.473684210526315, | |
| "grad_norm": 0.07661563158035278, | |
| "learning_rate": 1.0994897959183675e-06, | |
| "loss": 0.0043, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 9.48512585812357, | |
| "grad_norm": 0.3395727872848511, | |
| "learning_rate": 1.0931122448979592e-06, | |
| "loss": 0.0107, | |
| "step": 20725 | |
| }, | |
| { | |
| "epoch": 9.496567505720824, | |
| "grad_norm": 0.7869210243225098, | |
| "learning_rate": 1.0867346938775512e-06, | |
| "loss": 0.0051, | |
| "step": 20750 | |
| }, | |
| { | |
| "epoch": 9.508009153318078, | |
| "grad_norm": 0.7539875507354736, | |
| "learning_rate": 1.080357142857143e-06, | |
| "loss": 0.0056, | |
| "step": 20775 | |
| }, | |
| { | |
| "epoch": 9.519450800915331, | |
| "grad_norm": 16.64691162109375, | |
| "learning_rate": 1.0739795918367347e-06, | |
| "loss": 0.0066, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 9.530892448512585, | |
| "grad_norm": 0.7613354325294495, | |
| "learning_rate": 1.0676020408163265e-06, | |
| "loss": 0.0139, | |
| "step": 20825 | |
| }, | |
| { | |
| "epoch": 9.54233409610984, | |
| "grad_norm": 3.5544533729553223, | |
| "learning_rate": 1.0612244897959184e-06, | |
| "loss": 0.0058, | |
| "step": 20850 | |
| }, | |
| { | |
| "epoch": 9.553775743707094, | |
| "grad_norm": 0.14786852896213531, | |
| "learning_rate": 1.0548469387755102e-06, | |
| "loss": 0.0072, | |
| "step": 20875 | |
| }, | |
| { | |
| "epoch": 9.565217391304348, | |
| "grad_norm": 0.15534640848636627, | |
| "learning_rate": 1.0484693877551021e-06, | |
| "loss": 0.0045, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 9.576659038901601, | |
| "grad_norm": 0.23513126373291016, | |
| "learning_rate": 1.042091836734694e-06, | |
| "loss": 0.0042, | |
| "step": 20925 | |
| }, | |
| { | |
| "epoch": 9.588100686498855, | |
| "grad_norm": 0.39259815216064453, | |
| "learning_rate": 1.0357142857142857e-06, | |
| "loss": 0.0052, | |
| "step": 20950 | |
| }, | |
| { | |
| "epoch": 9.59954233409611, | |
| "grad_norm": 3.45914626121521, | |
| "learning_rate": 1.0293367346938776e-06, | |
| "loss": 0.0083, | |
| "step": 20975 | |
| }, | |
| { | |
| "epoch": 9.610983981693364, | |
| "grad_norm": 0.8460655212402344, | |
| "learning_rate": 1.0229591836734694e-06, | |
| "loss": 0.006, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 9.622425629290618, | |
| "grad_norm": 0.3035661280155182, | |
| "learning_rate": 1.0165816326530613e-06, | |
| "loss": 0.0066, | |
| "step": 21025 | |
| }, | |
| { | |
| "epoch": 9.633867276887871, | |
| "grad_norm": 0.07911881059408188, | |
| "learning_rate": 1.010204081632653e-06, | |
| "loss": 0.0044, | |
| "step": 21050 | |
| }, | |
| { | |
| "epoch": 9.645308924485127, | |
| "grad_norm": 1.1485878229141235, | |
| "learning_rate": 1.003826530612245e-06, | |
| "loss": 0.0061, | |
| "step": 21075 | |
| }, | |
| { | |
| "epoch": 9.65675057208238, | |
| "grad_norm": 3.1659584045410156, | |
| "learning_rate": 9.974489795918368e-07, | |
| "loss": 0.0062, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 9.668192219679634, | |
| "grad_norm": 0.08537878096103668, | |
| "learning_rate": 9.910714285714286e-07, | |
| "loss": 0.0065, | |
| "step": 21125 | |
| }, | |
| { | |
| "epoch": 9.679633867276888, | |
| "grad_norm": 0.27491432428359985, | |
| "learning_rate": 9.846938775510205e-07, | |
| "loss": 0.0054, | |
| "step": 21150 | |
| }, | |
| { | |
| "epoch": 9.691075514874141, | |
| "grad_norm": 1.1356456279754639, | |
| "learning_rate": 9.783163265306123e-07, | |
| "loss": 0.0106, | |
| "step": 21175 | |
| }, | |
| { | |
| "epoch": 9.702517162471397, | |
| "grad_norm": 0.0873272642493248, | |
| "learning_rate": 9.71938775510204e-07, | |
| "loss": 0.0028, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 9.71395881006865, | |
| "grad_norm": 0.33347004652023315, | |
| "learning_rate": 9.655612244897958e-07, | |
| "loss": 0.0134, | |
| "step": 21225 | |
| }, | |
| { | |
| "epoch": 9.725400457665904, | |
| "grad_norm": 0.14462488889694214, | |
| "learning_rate": 9.591836734693878e-07, | |
| "loss": 0.0046, | |
| "step": 21250 | |
| }, | |
| { | |
| "epoch": 9.736842105263158, | |
| "grad_norm": 1.358343482017517, | |
| "learning_rate": 9.528061224489796e-07, | |
| "loss": 0.0103, | |
| "step": 21275 | |
| }, | |
| { | |
| "epoch": 9.748283752860411, | |
| "grad_norm": 0.039507463574409485, | |
| "learning_rate": 9.464285714285715e-07, | |
| "loss": 0.0041, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 9.759725400457667, | |
| "grad_norm": 0.7528636455535889, | |
| "learning_rate": 9.400510204081633e-07, | |
| "loss": 0.0028, | |
| "step": 21325 | |
| }, | |
| { | |
| "epoch": 9.77116704805492, | |
| "grad_norm": 1.8518726825714111, | |
| "learning_rate": 9.33673469387755e-07, | |
| "loss": 0.0054, | |
| "step": 21350 | |
| }, | |
| { | |
| "epoch": 9.782608695652174, | |
| "grad_norm": 0.5261640548706055, | |
| "learning_rate": 9.27295918367347e-07, | |
| "loss": 0.008, | |
| "step": 21375 | |
| }, | |
| { | |
| "epoch": 9.794050343249427, | |
| "grad_norm": 0.48610520362854004, | |
| "learning_rate": 9.209183673469387e-07, | |
| "loss": 0.0045, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 9.805491990846681, | |
| "grad_norm": 0.35371649265289307, | |
| "learning_rate": 9.145408163265307e-07, | |
| "loss": 0.0067, | |
| "step": 21425 | |
| }, | |
| { | |
| "epoch": 9.816933638443937, | |
| "grad_norm": 4.049976825714111, | |
| "learning_rate": 9.081632653061225e-07, | |
| "loss": 0.0065, | |
| "step": 21450 | |
| }, | |
| { | |
| "epoch": 9.82837528604119, | |
| "grad_norm": 4.127640724182129, | |
| "learning_rate": 9.017857142857144e-07, | |
| "loss": 0.009, | |
| "step": 21475 | |
| }, | |
| { | |
| "epoch": 9.839816933638444, | |
| "grad_norm": 0.15992511808872223, | |
| "learning_rate": 8.954081632653062e-07, | |
| "loss": 0.0047, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 9.851258581235697, | |
| "grad_norm": 0.5393081307411194, | |
| "learning_rate": 8.89030612244898e-07, | |
| "loss": 0.0054, | |
| "step": 21525 | |
| }, | |
| { | |
| "epoch": 9.862700228832953, | |
| "grad_norm": 4.514695167541504, | |
| "learning_rate": 8.826530612244898e-07, | |
| "loss": 0.0054, | |
| "step": 21550 | |
| }, | |
| { | |
| "epoch": 9.874141876430206, | |
| "grad_norm": 1.2299213409423828, | |
| "learning_rate": 8.762755102040816e-07, | |
| "loss": 0.0042, | |
| "step": 21575 | |
| }, | |
| { | |
| "epoch": 9.88558352402746, | |
| "grad_norm": 0.12707316875457764, | |
| "learning_rate": 8.698979591836735e-07, | |
| "loss": 0.0062, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 9.897025171624714, | |
| "grad_norm": 0.8991334438323975, | |
| "learning_rate": 8.635204081632653e-07, | |
| "loss": 0.0078, | |
| "step": 21625 | |
| }, | |
| { | |
| "epoch": 9.908466819221967, | |
| "grad_norm": 0.9138942360877991, | |
| "learning_rate": 8.571428571428572e-07, | |
| "loss": 0.0034, | |
| "step": 21650 | |
| }, | |
| { | |
| "epoch": 9.919908466819223, | |
| "grad_norm": 0.07458051294088364, | |
| "learning_rate": 8.50765306122449e-07, | |
| "loss": 0.0061, | |
| "step": 21675 | |
| }, | |
| { | |
| "epoch": 9.931350114416476, | |
| "grad_norm": 0.15481892228126526, | |
| "learning_rate": 8.44387755102041e-07, | |
| "loss": 0.0062, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 9.94279176201373, | |
| "grad_norm": 0.4163806736469269, | |
| "learning_rate": 8.380102040816327e-07, | |
| "loss": 0.0037, | |
| "step": 21725 | |
| }, | |
| { | |
| "epoch": 9.954233409610984, | |
| "grad_norm": 0.1567489504814148, | |
| "learning_rate": 8.316326530612246e-07, | |
| "loss": 0.0041, | |
| "step": 21750 | |
| }, | |
| { | |
| "epoch": 9.965675057208237, | |
| "grad_norm": 2.551042079925537, | |
| "learning_rate": 8.252551020408163e-07, | |
| "loss": 0.0093, | |
| "step": 21775 | |
| }, | |
| { | |
| "epoch": 9.977116704805493, | |
| "grad_norm": 0.11054814606904984, | |
| "learning_rate": 8.188775510204081e-07, | |
| "loss": 0.0058, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 9.988558352402746, | |
| "grad_norm": 0.0360272191464901, | |
| "learning_rate": 8.125000000000001e-07, | |
| "loss": 0.0072, | |
| "step": 21825 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "grad_norm": 5.81298303604126, | |
| "learning_rate": 8.061224489795918e-07, | |
| "loss": 0.0068, | |
| "step": 21850 | |
| }, | |
| { | |
| "epoch": 10.011441647597254, | |
| "grad_norm": 0.0239973533898592, | |
| "learning_rate": 7.997448979591838e-07, | |
| "loss": 0.0039, | |
| "step": 21875 | |
| }, | |
| { | |
| "epoch": 10.022883295194507, | |
| "grad_norm": 0.14009559154510498, | |
| "learning_rate": 7.933673469387755e-07, | |
| "loss": 0.0083, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 10.034324942791763, | |
| "grad_norm": 0.05973900109529495, | |
| "learning_rate": 7.869897959183675e-07, | |
| "loss": 0.0067, | |
| "step": 21925 | |
| }, | |
| { | |
| "epoch": 10.045766590389016, | |
| "grad_norm": 2.9854466915130615, | |
| "learning_rate": 7.806122448979593e-07, | |
| "loss": 0.005, | |
| "step": 21950 | |
| }, | |
| { | |
| "epoch": 10.05720823798627, | |
| "grad_norm": 2.130392074584961, | |
| "learning_rate": 7.74234693877551e-07, | |
| "loss": 0.0026, | |
| "step": 21975 | |
| }, | |
| { | |
| "epoch": 10.068649885583524, | |
| "grad_norm": 13.454663276672363, | |
| "learning_rate": 7.678571428571429e-07, | |
| "loss": 0.0094, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 10.080091533180777, | |
| "grad_norm": 2.081770420074463, | |
| "learning_rate": 7.614795918367347e-07, | |
| "loss": 0.0069, | |
| "step": 22025 | |
| }, | |
| { | |
| "epoch": 10.091533180778033, | |
| "grad_norm": 1.7810285091400146, | |
| "learning_rate": 7.551020408163266e-07, | |
| "loss": 0.0113, | |
| "step": 22050 | |
| }, | |
| { | |
| "epoch": 10.102974828375286, | |
| "grad_norm": 3.7953577041625977, | |
| "learning_rate": 7.487244897959183e-07, | |
| "loss": 0.005, | |
| "step": 22075 | |
| }, | |
| { | |
| "epoch": 10.11441647597254, | |
| "grad_norm": 7.428182601928711, | |
| "learning_rate": 7.423469387755102e-07, | |
| "loss": 0.0077, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 10.125858123569794, | |
| "grad_norm": 4.6249165534973145, | |
| "learning_rate": 7.359693877551021e-07, | |
| "loss": 0.0032, | |
| "step": 22125 | |
| }, | |
| { | |
| "epoch": 10.137299771167047, | |
| "grad_norm": 7.158110618591309, | |
| "learning_rate": 7.295918367346939e-07, | |
| "loss": 0.0071, | |
| "step": 22150 | |
| }, | |
| { | |
| "epoch": 10.148741418764303, | |
| "grad_norm": 0.05203917250037193, | |
| "learning_rate": 7.232142857142858e-07, | |
| "loss": 0.0025, | |
| "step": 22175 | |
| }, | |
| { | |
| "epoch": 10.160183066361556, | |
| "grad_norm": 0.4155661165714264, | |
| "learning_rate": 7.168367346938775e-07, | |
| "loss": 0.0057, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 10.17162471395881, | |
| "grad_norm": 0.3524133563041687, | |
| "learning_rate": 7.104591836734694e-07, | |
| "loss": 0.0042, | |
| "step": 22225 | |
| }, | |
| { | |
| "epoch": 10.183066361556063, | |
| "grad_norm": 7.646109580993652, | |
| "learning_rate": 7.040816326530613e-07, | |
| "loss": 0.0074, | |
| "step": 22250 | |
| }, | |
| { | |
| "epoch": 10.194508009153319, | |
| "grad_norm": 1.1890356540679932, | |
| "learning_rate": 6.977040816326531e-07, | |
| "loss": 0.0024, | |
| "step": 22275 | |
| }, | |
| { | |
| "epoch": 10.205949656750573, | |
| "grad_norm": 2.9643213748931885, | |
| "learning_rate": 6.913265306122449e-07, | |
| "loss": 0.0113, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 10.217391304347826, | |
| "grad_norm": 0.13656121492385864, | |
| "learning_rate": 6.849489795918367e-07, | |
| "loss": 0.0027, | |
| "step": 22325 | |
| }, | |
| { | |
| "epoch": 10.22883295194508, | |
| "grad_norm": 3.043562173843384, | |
| "learning_rate": 6.785714285714286e-07, | |
| "loss": 0.015, | |
| "step": 22350 | |
| }, | |
| { | |
| "epoch": 10.240274599542333, | |
| "grad_norm": 0.23451216518878937, | |
| "learning_rate": 6.721938775510205e-07, | |
| "loss": 0.004, | |
| "step": 22375 | |
| }, | |
| { | |
| "epoch": 10.251716247139589, | |
| "grad_norm": 6.707335948944092, | |
| "learning_rate": 6.658163265306123e-07, | |
| "loss": 0.0029, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 10.263157894736842, | |
| "grad_norm": 0.055560849606990814, | |
| "learning_rate": 6.594387755102041e-07, | |
| "loss": 0.0088, | |
| "step": 22425 | |
| }, | |
| { | |
| "epoch": 10.274599542334096, | |
| "grad_norm": 8.039435386657715, | |
| "learning_rate": 6.53061224489796e-07, | |
| "loss": 0.0068, | |
| "step": 22450 | |
| }, | |
| { | |
| "epoch": 10.28604118993135, | |
| "grad_norm": 0.11787508428096771, | |
| "learning_rate": 6.466836734693878e-07, | |
| "loss": 0.0035, | |
| "step": 22475 | |
| }, | |
| { | |
| "epoch": 10.297482837528603, | |
| "grad_norm": 1.5206620693206787, | |
| "learning_rate": 6.403061224489796e-07, | |
| "loss": 0.0094, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 10.308924485125859, | |
| "grad_norm": 0.029644370079040527, | |
| "learning_rate": 6.339285714285714e-07, | |
| "loss": 0.0054, | |
| "step": 22525 | |
| }, | |
| { | |
| "epoch": 10.320366132723112, | |
| "grad_norm": 0.48167529702186584, | |
| "learning_rate": 6.275510204081633e-07, | |
| "loss": 0.0117, | |
| "step": 22550 | |
| }, | |
| { | |
| "epoch": 10.331807780320366, | |
| "grad_norm": 0.14305919408798218, | |
| "learning_rate": 6.211734693877551e-07, | |
| "loss": 0.0048, | |
| "step": 22575 | |
| }, | |
| { | |
| "epoch": 10.34324942791762, | |
| "grad_norm": 1.8980730772018433, | |
| "learning_rate": 6.14795918367347e-07, | |
| "loss": 0.0026, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 10.354691075514873, | |
| "grad_norm": 1.3038541078567505, | |
| "learning_rate": 6.084183673469388e-07, | |
| "loss": 0.0039, | |
| "step": 22625 | |
| }, | |
| { | |
| "epoch": 10.366132723112129, | |
| "grad_norm": 5.523509502410889, | |
| "learning_rate": 6.020408163265306e-07, | |
| "loss": 0.0066, | |
| "step": 22650 | |
| }, | |
| { | |
| "epoch": 10.377574370709382, | |
| "grad_norm": 0.04104507714509964, | |
| "learning_rate": 5.956632653061225e-07, | |
| "loss": 0.0042, | |
| "step": 22675 | |
| }, | |
| { | |
| "epoch": 10.389016018306636, | |
| "grad_norm": 1.2055227756500244, | |
| "learning_rate": 5.892857142857143e-07, | |
| "loss": 0.0038, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 10.40045766590389, | |
| "grad_norm": 0.02921094372868538, | |
| "learning_rate": 5.829081632653061e-07, | |
| "loss": 0.0036, | |
| "step": 22725 | |
| }, | |
| { | |
| "epoch": 10.411899313501145, | |
| "grad_norm": 16.179323196411133, | |
| "learning_rate": 5.76530612244898e-07, | |
| "loss": 0.0049, | |
| "step": 22750 | |
| }, | |
| { | |
| "epoch": 10.423340961098399, | |
| "grad_norm": 0.6530442833900452, | |
| "learning_rate": 5.701530612244898e-07, | |
| "loss": 0.0022, | |
| "step": 22775 | |
| }, | |
| { | |
| "epoch": 10.434782608695652, | |
| "grad_norm": 8.9596529006958, | |
| "learning_rate": 5.637755102040817e-07, | |
| "loss": 0.0077, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 10.446224256292906, | |
| "grad_norm": 0.03742032125592232, | |
| "learning_rate": 5.573979591836735e-07, | |
| "loss": 0.004, | |
| "step": 22825 | |
| }, | |
| { | |
| "epoch": 10.45766590389016, | |
| "grad_norm": 4.956221103668213, | |
| "learning_rate": 5.510204081632653e-07, | |
| "loss": 0.0062, | |
| "step": 22850 | |
| }, | |
| { | |
| "epoch": 10.469107551487415, | |
| "grad_norm": 0.04372161999344826, | |
| "learning_rate": 5.446428571428572e-07, | |
| "loss": 0.0029, | |
| "step": 22875 | |
| }, | |
| { | |
| "epoch": 10.480549199084669, | |
| "grad_norm": 2.4792611598968506, | |
| "learning_rate": 5.38265306122449e-07, | |
| "loss": 0.0047, | |
| "step": 22900 | |
| }, | |
| { | |
| "epoch": 10.491990846681922, | |
| "grad_norm": 1.2810579538345337, | |
| "learning_rate": 5.318877551020408e-07, | |
| "loss": 0.0054, | |
| "step": 22925 | |
| }, | |
| { | |
| "epoch": 10.503432494279176, | |
| "grad_norm": 16.370330810546875, | |
| "learning_rate": 5.255102040816326e-07, | |
| "loss": 0.0114, | |
| "step": 22950 | |
| }, | |
| { | |
| "epoch": 10.51487414187643, | |
| "grad_norm": 0.6860277652740479, | |
| "learning_rate": 5.191326530612245e-07, | |
| "loss": 0.004, | |
| "step": 22975 | |
| }, | |
| { | |
| "epoch": 10.526315789473685, | |
| "grad_norm": 1.6368955373764038, | |
| "learning_rate": 5.127551020408164e-07, | |
| "loss": 0.0074, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 10.537757437070939, | |
| "grad_norm": 1.6935875415802002, | |
| "learning_rate": 5.063775510204082e-07, | |
| "loss": 0.0043, | |
| "step": 23025 | |
| }, | |
| { | |
| "epoch": 10.549199084668192, | |
| "grad_norm": 0.27796322107315063, | |
| "learning_rate": 5.002551020408164e-07, | |
| "loss": 0.0075, | |
| "step": 23050 | |
| }, | |
| { | |
| "epoch": 10.560640732265446, | |
| "grad_norm": 0.05163831263780594, | |
| "learning_rate": 4.938775510204081e-07, | |
| "loss": 0.0033, | |
| "step": 23075 | |
| }, | |
| { | |
| "epoch": 10.5720823798627, | |
| "grad_norm": 3.6930410861968994, | |
| "learning_rate": 4.875e-07, | |
| "loss": 0.0067, | |
| "step": 23100 | |
| }, | |
| { | |
| "epoch": 10.583524027459955, | |
| "grad_norm": 2.0602188110351562, | |
| "learning_rate": 4.811224489795919e-07, | |
| "loss": 0.0046, | |
| "step": 23125 | |
| }, | |
| { | |
| "epoch": 10.594965675057209, | |
| "grad_norm": 6.743757724761963, | |
| "learning_rate": 4.747448979591837e-07, | |
| "loss": 0.0119, | |
| "step": 23150 | |
| }, | |
| { | |
| "epoch": 10.606407322654462, | |
| "grad_norm": 0.08728916943073273, | |
| "learning_rate": 4.683673469387756e-07, | |
| "loss": 0.0023, | |
| "step": 23175 | |
| }, | |
| { | |
| "epoch": 10.617848970251716, | |
| "grad_norm": 0.6951027512550354, | |
| "learning_rate": 4.6198979591836744e-07, | |
| "loss": 0.0067, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 10.62929061784897, | |
| "grad_norm": 0.9762528538703918, | |
| "learning_rate": 4.556122448979592e-07, | |
| "loss": 0.0033, | |
| "step": 23225 | |
| }, | |
| { | |
| "epoch": 10.640732265446225, | |
| "grad_norm": 4.385350704193115, | |
| "learning_rate": 4.49234693877551e-07, | |
| "loss": 0.006, | |
| "step": 23250 | |
| }, | |
| { | |
| "epoch": 10.652173913043478, | |
| "grad_norm": 0.12361292541027069, | |
| "learning_rate": 4.4285714285714286e-07, | |
| "loss": 0.002, | |
| "step": 23275 | |
| }, | |
| { | |
| "epoch": 10.663615560640732, | |
| "grad_norm": 4.548694610595703, | |
| "learning_rate": 4.364795918367347e-07, | |
| "loss": 0.0063, | |
| "step": 23300 | |
| }, | |
| { | |
| "epoch": 10.675057208237986, | |
| "grad_norm": 3.6494503021240234, | |
| "learning_rate": 4.301020408163266e-07, | |
| "loss": 0.005, | |
| "step": 23325 | |
| }, | |
| { | |
| "epoch": 10.68649885583524, | |
| "grad_norm": 5.889870643615723, | |
| "learning_rate": 4.237244897959184e-07, | |
| "loss": 0.0057, | |
| "step": 23350 | |
| }, | |
| { | |
| "epoch": 10.697940503432495, | |
| "grad_norm": 3.1003196239471436, | |
| "learning_rate": 4.1734693877551025e-07, | |
| "loss": 0.0073, | |
| "step": 23375 | |
| }, | |
| { | |
| "epoch": 10.709382151029748, | |
| "grad_norm": 3.9601292610168457, | |
| "learning_rate": 4.109693877551021e-07, | |
| "loss": 0.0089, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 10.720823798627002, | |
| "grad_norm": 0.19110934436321259, | |
| "learning_rate": 4.0459183673469387e-07, | |
| "loss": 0.0035, | |
| "step": 23425 | |
| }, | |
| { | |
| "epoch": 10.732265446224256, | |
| "grad_norm": 6.233257293701172, | |
| "learning_rate": 3.9821428571428573e-07, | |
| "loss": 0.0094, | |
| "step": 23450 | |
| }, | |
| { | |
| "epoch": 10.743707093821511, | |
| "grad_norm": 0.034573789685964584, | |
| "learning_rate": 3.9183673469387754e-07, | |
| "loss": 0.0049, | |
| "step": 23475 | |
| }, | |
| { | |
| "epoch": 10.755148741418765, | |
| "grad_norm": 4.074402332305908, | |
| "learning_rate": 3.854591836734694e-07, | |
| "loss": 0.0098, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 10.766590389016018, | |
| "grad_norm": 0.19702082872390747, | |
| "learning_rate": 3.7908163265306126e-07, | |
| "loss": 0.0034, | |
| "step": 23525 | |
| }, | |
| { | |
| "epoch": 10.778032036613272, | |
| "grad_norm": 6.4033522605896, | |
| "learning_rate": 3.7270408163265307e-07, | |
| "loss": 0.0061, | |
| "step": 23550 | |
| }, | |
| { | |
| "epoch": 10.789473684210526, | |
| "grad_norm": 0.6741847395896912, | |
| "learning_rate": 3.6632653061224493e-07, | |
| "loss": 0.0042, | |
| "step": 23575 | |
| }, | |
| { | |
| "epoch": 10.800915331807781, | |
| "grad_norm": 0.5898193717002869, | |
| "learning_rate": 3.5994897959183674e-07, | |
| "loss": 0.0067, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 10.812356979405035, | |
| "grad_norm": 0.05853106081485748, | |
| "learning_rate": 3.535714285714286e-07, | |
| "loss": 0.0068, | |
| "step": 23625 | |
| }, | |
| { | |
| "epoch": 10.823798627002288, | |
| "grad_norm": 0.1718439757823944, | |
| "learning_rate": 3.4719387755102046e-07, | |
| "loss": 0.0067, | |
| "step": 23650 | |
| }, | |
| { | |
| "epoch": 10.835240274599542, | |
| "grad_norm": 0.39581310749053955, | |
| "learning_rate": 3.4081632653061227e-07, | |
| "loss": 0.0035, | |
| "step": 23675 | |
| }, | |
| { | |
| "epoch": 10.846681922196796, | |
| "grad_norm": 3.5774476528167725, | |
| "learning_rate": 3.344387755102041e-07, | |
| "loss": 0.0081, | |
| "step": 23700 | |
| }, | |
| { | |
| "epoch": 10.858123569794051, | |
| "grad_norm": 0.2080162763595581, | |
| "learning_rate": 3.2806122448979594e-07, | |
| "loss": 0.0038, | |
| "step": 23725 | |
| }, | |
| { | |
| "epoch": 10.869565217391305, | |
| "grad_norm": 0.9511691331863403, | |
| "learning_rate": 3.216836734693878e-07, | |
| "loss": 0.0075, | |
| "step": 23750 | |
| }, | |
| { | |
| "epoch": 10.881006864988558, | |
| "grad_norm": 2.3875999450683594, | |
| "learning_rate": 3.153061224489796e-07, | |
| "loss": 0.0054, | |
| "step": 23775 | |
| }, | |
| { | |
| "epoch": 10.892448512585812, | |
| "grad_norm": 0.2756766080856323, | |
| "learning_rate": 3.089285714285714e-07, | |
| "loss": 0.0081, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 10.903890160183066, | |
| "grad_norm": 0.8285855054855347, | |
| "learning_rate": 3.025510204081633e-07, | |
| "loss": 0.0039, | |
| "step": 23825 | |
| }, | |
| { | |
| "epoch": 10.915331807780321, | |
| "grad_norm": 7.757495403289795, | |
| "learning_rate": 2.9617346938775513e-07, | |
| "loss": 0.0048, | |
| "step": 23850 | |
| }, | |
| { | |
| "epoch": 10.926773455377575, | |
| "grad_norm": 3.3156180381774902, | |
| "learning_rate": 2.8979591836734694e-07, | |
| "loss": 0.0044, | |
| "step": 23875 | |
| }, | |
| { | |
| "epoch": 10.938215102974828, | |
| "grad_norm": 1.0153356790542603, | |
| "learning_rate": 2.8341836734693875e-07, | |
| "loss": 0.0043, | |
| "step": 23900 | |
| }, | |
| { | |
| "epoch": 10.949656750572082, | |
| "grad_norm": 14.048026084899902, | |
| "learning_rate": 2.770408163265306e-07, | |
| "loss": 0.0062, | |
| "step": 23925 | |
| }, | |
| { | |
| "epoch": 10.961098398169337, | |
| "grad_norm": 3.859445810317993, | |
| "learning_rate": 2.7066326530612247e-07, | |
| "loss": 0.0067, | |
| "step": 23950 | |
| }, | |
| { | |
| "epoch": 10.972540045766591, | |
| "grad_norm": 2.134991407394409, | |
| "learning_rate": 2.6428571428571433e-07, | |
| "loss": 0.0033, | |
| "step": 23975 | |
| }, | |
| { | |
| "epoch": 10.983981693363845, | |
| "grad_norm": 9.151762008666992, | |
| "learning_rate": 2.579081632653061e-07, | |
| "loss": 0.0121, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 10.995423340961098, | |
| "grad_norm": 0.3495076298713684, | |
| "learning_rate": 2.5153061224489795e-07, | |
| "loss": 0.0048, | |
| "step": 24025 | |
| }, | |
| { | |
| "epoch": 11.006864988558352, | |
| "grad_norm": 0.10012809187173843, | |
| "learning_rate": 2.451530612244898e-07, | |
| "loss": 0.0055, | |
| "step": 24050 | |
| }, | |
| { | |
| "epoch": 11.018306636155607, | |
| "grad_norm": 0.06472992897033691, | |
| "learning_rate": 2.3877551020408167e-07, | |
| "loss": 0.0019, | |
| "step": 24075 | |
| }, | |
| { | |
| "epoch": 11.02974828375286, | |
| "grad_norm": 0.16011592745780945, | |
| "learning_rate": 2.323979591836735e-07, | |
| "loss": 0.0075, | |
| "step": 24100 | |
| }, | |
| { | |
| "epoch": 11.041189931350115, | |
| "grad_norm": 2.186831474304199, | |
| "learning_rate": 2.2602040816326532e-07, | |
| "loss": 0.0037, | |
| "step": 24125 | |
| }, | |
| { | |
| "epoch": 11.052631578947368, | |
| "grad_norm": 0.32710880041122437, | |
| "learning_rate": 2.1964285714285715e-07, | |
| "loss": 0.0073, | |
| "step": 24150 | |
| }, | |
| { | |
| "epoch": 11.064073226544622, | |
| "grad_norm": 0.054156772792339325, | |
| "learning_rate": 2.1326530612244898e-07, | |
| "loss": 0.0052, | |
| "step": 24175 | |
| }, | |
| { | |
| "epoch": 11.075514874141877, | |
| "grad_norm": 0.07884764671325684, | |
| "learning_rate": 2.0688775510204085e-07, | |
| "loss": 0.003, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 11.08695652173913, | |
| "grad_norm": 0.04855496436357498, | |
| "learning_rate": 2.0051020408163265e-07, | |
| "loss": 0.0039, | |
| "step": 24225 | |
| }, | |
| { | |
| "epoch": 11.098398169336384, | |
| "grad_norm": 3.0451197624206543, | |
| "learning_rate": 1.941326530612245e-07, | |
| "loss": 0.0037, | |
| "step": 24250 | |
| }, | |
| { | |
| "epoch": 11.109839816933638, | |
| "grad_norm": 2.733774185180664, | |
| "learning_rate": 1.8775510204081635e-07, | |
| "loss": 0.0049, | |
| "step": 24275 | |
| }, | |
| { | |
| "epoch": 11.121281464530892, | |
| "grad_norm": 0.01790427230298519, | |
| "learning_rate": 1.816326530612245e-07, | |
| "loss": 0.0082, | |
| "step": 24300 | |
| }, | |
| { | |
| "epoch": 11.132723112128147, | |
| "grad_norm": 24.361324310302734, | |
| "learning_rate": 1.7525510204081634e-07, | |
| "loss": 0.0067, | |
| "step": 24325 | |
| }, | |
| { | |
| "epoch": 11.1441647597254, | |
| "grad_norm": 0.0411471389234066, | |
| "learning_rate": 1.6887755102040817e-07, | |
| "loss": 0.0056, | |
| "step": 24350 | |
| }, | |
| { | |
| "epoch": 11.155606407322654, | |
| "grad_norm": 0.5140661597251892, | |
| "learning_rate": 1.625e-07, | |
| "loss": 0.0036, | |
| "step": 24375 | |
| }, | |
| { | |
| "epoch": 11.167048054919908, | |
| "grad_norm": 0.36049628257751465, | |
| "learning_rate": 1.5612244897959184e-07, | |
| "loss": 0.008, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 11.178489702517162, | |
| "grad_norm": 0.6188111305236816, | |
| "learning_rate": 1.4974489795918368e-07, | |
| "loss": 0.0021, | |
| "step": 24425 | |
| }, | |
| { | |
| "epoch": 11.189931350114417, | |
| "grad_norm": 0.12496360391378403, | |
| "learning_rate": 1.433673469387755e-07, | |
| "loss": 0.0052, | |
| "step": 24450 | |
| }, | |
| { | |
| "epoch": 11.20137299771167, | |
| "grad_norm": 2.986814498901367, | |
| "learning_rate": 1.3698979591836734e-07, | |
| "loss": 0.0062, | |
| "step": 24475 | |
| }, | |
| { | |
| "epoch": 11.212814645308924, | |
| "grad_norm": 1.2586897611618042, | |
| "learning_rate": 1.3061224489795918e-07, | |
| "loss": 0.0067, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 11.224256292906178, | |
| "grad_norm": 2.0878169536590576, | |
| "learning_rate": 1.2423469387755104e-07, | |
| "loss": 0.0052, | |
| "step": 24525 | |
| }, | |
| { | |
| "epoch": 11.235697940503432, | |
| "grad_norm": 0.07096085697412491, | |
| "learning_rate": 1.1785714285714286e-07, | |
| "loss": 0.0073, | |
| "step": 24550 | |
| }, | |
| { | |
| "epoch": 11.247139588100687, | |
| "grad_norm": 0.042739126831293106, | |
| "learning_rate": 1.1147959183673471e-07, | |
| "loss": 0.0043, | |
| "step": 24575 | |
| }, | |
| { | |
| "epoch": 11.25858123569794, | |
| "grad_norm": 1.214879035949707, | |
| "learning_rate": 1.0510204081632653e-07, | |
| "loss": 0.0076, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 11.270022883295194, | |
| "grad_norm": 0.14526697993278503, | |
| "learning_rate": 9.872448979591838e-08, | |
| "loss": 0.0041, | |
| "step": 24625 | |
| }, | |
| { | |
| "epoch": 11.281464530892448, | |
| "grad_norm": 2.3068532943725586, | |
| "learning_rate": 9.23469387755102e-08, | |
| "loss": 0.011, | |
| "step": 24650 | |
| }, | |
| { | |
| "epoch": 11.292906178489703, | |
| "grad_norm": 0.07109973579645157, | |
| "learning_rate": 8.596938775510205e-08, | |
| "loss": 0.002, | |
| "step": 24675 | |
| }, | |
| { | |
| "epoch": 11.304347826086957, | |
| "grad_norm": 0.05076288804411888, | |
| "learning_rate": 7.959183673469388e-08, | |
| "loss": 0.0091, | |
| "step": 24700 | |
| }, | |
| { | |
| "epoch": 11.31578947368421, | |
| "grad_norm": 0.026290345937013626, | |
| "learning_rate": 7.321428571428572e-08, | |
| "loss": 0.0057, | |
| "step": 24725 | |
| }, | |
| { | |
| "epoch": 11.327231121281464, | |
| "grad_norm": 0.04591425135731697, | |
| "learning_rate": 6.683673469387755e-08, | |
| "loss": 0.0046, | |
| "step": 24750 | |
| }, | |
| { | |
| "epoch": 11.338672768878718, | |
| "grad_norm": 0.057588379830121994, | |
| "learning_rate": 6.04591836734694e-08, | |
| "loss": 0.0051, | |
| "step": 24775 | |
| }, | |
| { | |
| "epoch": 11.350114416475973, | |
| "grad_norm": 0.19686830043792725, | |
| "learning_rate": 5.4081632653061234e-08, | |
| "loss": 0.0064, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 11.361556064073227, | |
| "grad_norm": 0.039152998477220535, | |
| "learning_rate": 4.770408163265306e-08, | |
| "loss": 0.004, | |
| "step": 24825 | |
| }, | |
| { | |
| "epoch": 11.37299771167048, | |
| "grad_norm": 0.1060929223895073, | |
| "learning_rate": 4.1326530612244896e-08, | |
| "loss": 0.0053, | |
| "step": 24850 | |
| }, | |
| { | |
| "epoch": 11.384439359267734, | |
| "grad_norm": 0.057997722178697586, | |
| "learning_rate": 3.494897959183673e-08, | |
| "loss": 0.0036, | |
| "step": 24875 | |
| }, | |
| { | |
| "epoch": 11.395881006864988, | |
| "grad_norm": 0.3169647753238678, | |
| "learning_rate": 2.8571428571428575e-08, | |
| "loss": 0.0059, | |
| "step": 24900 | |
| }, | |
| { | |
| "epoch": 11.407322654462243, | |
| "grad_norm": 0.020627867430448532, | |
| "learning_rate": 2.2193877551020407e-08, | |
| "loss": 0.0043, | |
| "step": 24925 | |
| }, | |
| { | |
| "epoch": 11.418764302059497, | |
| "grad_norm": 0.04437786713242531, | |
| "learning_rate": 1.5816326530612248e-08, | |
| "loss": 0.0075, | |
| "step": 24950 | |
| }, | |
| { | |
| "epoch": 11.43020594965675, | |
| "grad_norm": 0.03256097808480263, | |
| "learning_rate": 9.438775510204082e-09, | |
| "loss": 0.0022, | |
| "step": 24975 | |
| }, | |
| { | |
| "epoch": 11.441647597254004, | |
| "grad_norm": 3.5547521114349365, | |
| "learning_rate": 3.0612244897959183e-09, | |
| "loss": 0.0055, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 11.441647597254004, | |
| "eval_loss": 0.21081247925758362, | |
| "eval_runtime": 5131.1271, | |
| "eval_samples_per_second": 1.856, | |
| "eval_steps_per_second": 0.232, | |
| "eval_wer": 0.09488385486657708, | |
| "step": 25000 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 25000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 12, | |
| "save_steps": 5000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 4.081970563920691e+20, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |