| { | |
| "best_metric": 0.1500573754310608, | |
| "best_model_checkpoint": "d:\\\\whisper-large-v3-pt-3000h-3\\checkpoint-691", | |
| "epoch": 9.99638336347197, | |
| "eval_steps": 500, | |
| "global_step": 6910, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03616636528028933, | |
| "grad_norm": 5.811695098876953, | |
| "learning_rate": 2.5000000000000004e-07, | |
| "loss": 0.7755, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07233273056057866, | |
| "grad_norm": Infinity, | |
| "learning_rate": 4.900000000000001e-07, | |
| "loss": 1.2265, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.10849909584086799, | |
| "grad_norm": 3.2242519855499268, | |
| "learning_rate": 7.4e-07, | |
| "loss": 0.3546, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.14466546112115733, | |
| "grad_norm": 11.40543270111084, | |
| "learning_rate": 9.9e-07, | |
| "loss": 0.2988, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.18083182640144665, | |
| "grad_norm": 2.457425117492676, | |
| "learning_rate": 1.2400000000000002e-06, | |
| "loss": 0.174, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.21699819168173598, | |
| "grad_norm": 10.275086402893066, | |
| "learning_rate": 1.4900000000000001e-06, | |
| "loss": 0.2317, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.25316455696202533, | |
| "grad_norm": 2.2328078746795654, | |
| "learning_rate": 1.74e-06, | |
| "loss": 0.1487, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.28933092224231466, | |
| "grad_norm": 8.097187995910645, | |
| "learning_rate": 1.9900000000000004e-06, | |
| "loss": 0.205, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.325497287522604, | |
| "grad_norm": 2.458644390106201, | |
| "learning_rate": 2.24e-06, | |
| "loss": 0.1509, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.3616636528028933, | |
| "grad_norm": 8.181588172912598, | |
| "learning_rate": 2.4900000000000003e-06, | |
| "loss": 0.2057, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.39783001808318263, | |
| "grad_norm": 2.0923259258270264, | |
| "learning_rate": 2.7400000000000004e-06, | |
| "loss": 0.1579, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.43399638336347196, | |
| "grad_norm": 5.871222496032715, | |
| "learning_rate": 2.99e-06, | |
| "loss": 0.188, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.4701627486437613, | |
| "grad_norm": 2.132258653640747, | |
| "learning_rate": 3.2400000000000003e-06, | |
| "loss": 0.1513, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.5063291139240507, | |
| "grad_norm": 9.259086608886719, | |
| "learning_rate": 3.49e-06, | |
| "loss": 0.1773, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.5424954792043399, | |
| "grad_norm": 2.059004783630371, | |
| "learning_rate": 3.74e-06, | |
| "loss": 0.1469, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.5786618444846293, | |
| "grad_norm": 6.2740302085876465, | |
| "learning_rate": 3.990000000000001e-06, | |
| "loss": 0.1765, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.6148282097649186, | |
| "grad_norm": 2.315974473953247, | |
| "learning_rate": 4.24e-06, | |
| "loss": 0.1364, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.650994575045208, | |
| "grad_norm": 6.675321102142334, | |
| "learning_rate": 4.49e-06, | |
| "loss": 0.1921, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.6871609403254972, | |
| "grad_norm": 2.4726405143737793, | |
| "learning_rate": 4.74e-06, | |
| "loss": 0.1365, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.7233273056057866, | |
| "grad_norm": 7.037698745727539, | |
| "learning_rate": 4.9900000000000005e-06, | |
| "loss": 0.1792, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.759493670886076, | |
| "grad_norm": 2.1146976947784424, | |
| "learning_rate": 5.240000000000001e-06, | |
| "loss": 0.1347, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.7956600361663653, | |
| "grad_norm": 4.760638236999512, | |
| "learning_rate": 5.490000000000001e-06, | |
| "loss": 0.1702, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.8318264014466547, | |
| "grad_norm": 3.498173952102661, | |
| "learning_rate": 5.74e-06, | |
| "loss": 0.132, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.8679927667269439, | |
| "grad_norm": 6.879450798034668, | |
| "learning_rate": 5.99e-06, | |
| "loss": 0.1589, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.9041591320072333, | |
| "grad_norm": 2.3761587142944336, | |
| "learning_rate": 6.24e-06, | |
| "loss": 0.15, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.9403254972875226, | |
| "grad_norm": 5.599714756011963, | |
| "learning_rate": 6.4900000000000005e-06, | |
| "loss": 0.1801, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.976491862567812, | |
| "grad_norm": 3.146120071411133, | |
| "learning_rate": 6.740000000000001e-06, | |
| "loss": 0.1388, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.9996383363471971, | |
| "eval_loss": 0.1500573754310608, | |
| "eval_runtime": 40633.1057, | |
| "eval_samples_per_second": 0.234, | |
| "eval_steps_per_second": 0.029, | |
| "eval_wer": 0.10736707238949392, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 1.0126582278481013, | |
| "grad_norm": 1.3991518020629883, | |
| "learning_rate": 6.99e-06, | |
| "loss": 0.1468, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.0488245931283906, | |
| "grad_norm": 2.3540124893188477, | |
| "learning_rate": 7.24e-06, | |
| "loss": 0.0971, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.0849909584086799, | |
| "grad_norm": 1.6329786777496338, | |
| "learning_rate": 7.49e-06, | |
| "loss": 0.1155, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.1211573236889691, | |
| "grad_norm": 2.5000386238098145, | |
| "learning_rate": 7.74e-06, | |
| "loss": 0.0903, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.1573236889692586, | |
| "grad_norm": 1.3056849241256714, | |
| "learning_rate": 7.990000000000001e-06, | |
| "loss": 0.1255, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.193490054249548, | |
| "grad_norm": 2.1964187622070312, | |
| "learning_rate": 8.24e-06, | |
| "loss": 0.0987, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.2296564195298372, | |
| "grad_norm": 1.5802552700042725, | |
| "learning_rate": 8.49e-06, | |
| "loss": 0.1393, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.2658227848101267, | |
| "grad_norm": 1.7570548057556152, | |
| "learning_rate": 8.740000000000001e-06, | |
| "loss": 0.1094, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.301989150090416, | |
| "grad_norm": 1.5590579509735107, | |
| "learning_rate": 8.99e-06, | |
| "loss": 0.1276, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.3381555153707052, | |
| "grad_norm": 2.429781913757324, | |
| "learning_rate": 9.240000000000001e-06, | |
| "loss": 0.1023, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.3743218806509945, | |
| "grad_norm": 2.3684589862823486, | |
| "learning_rate": 9.49e-06, | |
| "loss": 0.1372, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.410488245931284, | |
| "grad_norm": 5.570235252380371, | |
| "learning_rate": 9.74e-06, | |
| "loss": 0.1122, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.4466546112115732, | |
| "grad_norm": 2.2822372913360596, | |
| "learning_rate": 9.990000000000001e-06, | |
| "loss": 0.1463, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.4828209764918625, | |
| "grad_norm": 2.185589075088501, | |
| "learning_rate": 9.959390862944164e-06, | |
| "loss": 0.1055, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.518987341772152, | |
| "grad_norm": 2.054370164871216, | |
| "learning_rate": 9.917089678511e-06, | |
| "loss": 0.1426, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.5551537070524413, | |
| "grad_norm": 2.4509952068328857, | |
| "learning_rate": 9.874788494077835e-06, | |
| "loss": 0.0985, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.5913200723327305, | |
| "grad_norm": 1.7492316961288452, | |
| "learning_rate": 9.83248730964467e-06, | |
| "loss": 0.1553, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.6274864376130198, | |
| "grad_norm": 2.5610837936401367, | |
| "learning_rate": 9.790186125211507e-06, | |
| "loss": 0.1235, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.663652802893309, | |
| "grad_norm": 1.8614284992218018, | |
| "learning_rate": 9.747884940778342e-06, | |
| "loss": 0.1524, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.6998191681735986, | |
| "grad_norm": 3.679600954055786, | |
| "learning_rate": 9.705583756345178e-06, | |
| "loss": 0.1237, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.7359855334538878, | |
| "grad_norm": 1.617661952972412, | |
| "learning_rate": 9.663282571912015e-06, | |
| "loss": 0.1341, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.7721518987341773, | |
| "grad_norm": 2.3510446548461914, | |
| "learning_rate": 9.62098138747885e-06, | |
| "loss": 0.1232, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.8083182640144666, | |
| "grad_norm": 1.9561399221420288, | |
| "learning_rate": 9.578680203045685e-06, | |
| "loss": 0.1294, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.8444846292947559, | |
| "grad_norm": 3.439204216003418, | |
| "learning_rate": 9.536379018612522e-06, | |
| "loss": 0.1119, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.8806509945750451, | |
| "grad_norm": 2.9221997261047363, | |
| "learning_rate": 9.494077834179358e-06, | |
| "loss": 0.1494, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.9168173598553344, | |
| "grad_norm": 2.623828411102295, | |
| "learning_rate": 9.451776649746193e-06, | |
| "loss": 0.1109, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.952983725135624, | |
| "grad_norm": 1.5950006246566772, | |
| "learning_rate": 9.40947546531303e-06, | |
| "loss": 0.134, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.9891500904159132, | |
| "grad_norm": 2.305098533630371, | |
| "learning_rate": 9.367174280879865e-06, | |
| "loss": 0.108, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.9992766726943942, | |
| "eval_loss": 0.16191039979457855, | |
| "eval_runtime": 48873.7373, | |
| "eval_samples_per_second": 0.194, | |
| "eval_steps_per_second": 0.024, | |
| "eval_wer": 0.11531069827033953, | |
| "step": 1382 | |
| }, | |
| { | |
| "epoch": 2.0253164556962027, | |
| "grad_norm": 1.2765154838562012, | |
| "learning_rate": 9.324873096446702e-06, | |
| "loss": 0.1064, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.061482820976492, | |
| "grad_norm": 2.579493999481201, | |
| "learning_rate": 9.282571912013538e-06, | |
| "loss": 0.0676, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.097649186256781, | |
| "grad_norm": 1.4855302572250366, | |
| "learning_rate": 9.240270727580373e-06, | |
| "loss": 0.0733, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.1338155515370705, | |
| "grad_norm": 1.7606139183044434, | |
| "learning_rate": 9.19796954314721e-06, | |
| "loss": 0.069, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.1699819168173597, | |
| "grad_norm": 1.383319616317749, | |
| "learning_rate": 9.155668358714045e-06, | |
| "loss": 0.069, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.206148282097649, | |
| "grad_norm": 3.2015483379364014, | |
| "learning_rate": 9.11336717428088e-06, | |
| "loss": 0.0725, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.2423146473779383, | |
| "grad_norm": 2.2851688861846924, | |
| "learning_rate": 9.071065989847716e-06, | |
| "loss": 0.0743, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.278481012658228, | |
| "grad_norm": 3.721728801727295, | |
| "learning_rate": 9.028764805414553e-06, | |
| "loss": 0.0704, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.3146473779385173, | |
| "grad_norm": 1.077505111694336, | |
| "learning_rate": 8.986463620981388e-06, | |
| "loss": 0.0743, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.3508137432188065, | |
| "grad_norm": 2.2788355350494385, | |
| "learning_rate": 8.944162436548224e-06, | |
| "loss": 0.071, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.386980108499096, | |
| "grad_norm": 3.0096864700317383, | |
| "learning_rate": 8.901861252115061e-06, | |
| "loss": 0.0808, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.423146473779385, | |
| "grad_norm": 3.3069839477539062, | |
| "learning_rate": 8.859560067681896e-06, | |
| "loss": 0.0731, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.4593128390596743, | |
| "grad_norm": 1.791975498199463, | |
| "learning_rate": 8.817258883248732e-06, | |
| "loss": 0.0788, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.495479204339964, | |
| "grad_norm": 1.9154149293899536, | |
| "learning_rate": 8.774957698815567e-06, | |
| "loss": 0.0737, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.5316455696202533, | |
| "grad_norm": 1.7026249170303345, | |
| "learning_rate": 8.732656514382404e-06, | |
| "loss": 0.0821, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.5678119349005426, | |
| "grad_norm": 1.5672781467437744, | |
| "learning_rate": 8.69035532994924e-06, | |
| "loss": 0.0708, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.603978300180832, | |
| "grad_norm": 1.1428169012069702, | |
| "learning_rate": 8.648054145516075e-06, | |
| "loss": 0.0825, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.640144665461121, | |
| "grad_norm": 4.432084560394287, | |
| "learning_rate": 8.60575296108291e-06, | |
| "loss": 0.0749, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.6763110307414104, | |
| "grad_norm": 2.4532995223999023, | |
| "learning_rate": 8.563451776649747e-06, | |
| "loss": 0.0856, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.7124773960216997, | |
| "grad_norm": 2.286349296569824, | |
| "learning_rate": 8.521150592216582e-06, | |
| "loss": 0.075, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 2.748643761301989, | |
| "grad_norm": 1.524868130683899, | |
| "learning_rate": 8.478849407783418e-06, | |
| "loss": 0.0829, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.7848101265822782, | |
| "grad_norm": 3.2034952640533447, | |
| "learning_rate": 8.436548223350255e-06, | |
| "loss": 0.0749, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 2.820976491862568, | |
| "grad_norm": 2.6012070178985596, | |
| "learning_rate": 8.39424703891709e-06, | |
| "loss": 0.089, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.857142857142857, | |
| "grad_norm": 2.747398614883423, | |
| "learning_rate": 8.351945854483925e-06, | |
| "loss": 0.0774, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 2.8933092224231465, | |
| "grad_norm": 1.3775649070739746, | |
| "learning_rate": 8.309644670050762e-06, | |
| "loss": 0.0795, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.9294755877034357, | |
| "grad_norm": 6.074094295501709, | |
| "learning_rate": 8.267343485617598e-06, | |
| "loss": 0.0658, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 2.965641952983725, | |
| "grad_norm": 2.2065982818603516, | |
| "learning_rate": 8.225042301184433e-06, | |
| "loss": 0.091, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.9989150090415913, | |
| "eval_loss": 0.1697043776512146, | |
| "eval_runtime": 49669.5944, | |
| "eval_samples_per_second": 0.191, | |
| "eval_steps_per_second": 0.024, | |
| "eval_wer": 0.11237988468930173, | |
| "step": 2073 | |
| }, | |
| { | |
| "epoch": 3.0018083182640143, | |
| "grad_norm": 1.5902258157730103, | |
| "learning_rate": 8.18274111675127e-06, | |
| "loss": 0.0754, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 3.037974683544304, | |
| "grad_norm": 1.6310546398162842, | |
| "learning_rate": 8.140439932318105e-06, | |
| "loss": 0.0409, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.0741410488245933, | |
| "grad_norm": 1.9858086109161377, | |
| "learning_rate": 8.098138747884942e-06, | |
| "loss": 0.0529, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 3.1103074141048825, | |
| "grad_norm": 0.8269912004470825, | |
| "learning_rate": 8.055837563451778e-06, | |
| "loss": 0.0376, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.146473779385172, | |
| "grad_norm": 0.8812470436096191, | |
| "learning_rate": 8.013536379018613e-06, | |
| "loss": 0.0522, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 3.182640144665461, | |
| "grad_norm": 1.5196110010147095, | |
| "learning_rate": 7.97123519458545e-06, | |
| "loss": 0.043, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.2188065099457503, | |
| "grad_norm": 1.6323350667953491, | |
| "learning_rate": 7.928934010152285e-06, | |
| "loss": 0.0577, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 3.2549728752260396, | |
| "grad_norm": 1.475954532623291, | |
| "learning_rate": 7.88663282571912e-06, | |
| "loss": 0.0396, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.291139240506329, | |
| "grad_norm": 1.9558379650115967, | |
| "learning_rate": 7.844331641285958e-06, | |
| "loss": 0.0433, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 3.3273056057866186, | |
| "grad_norm": 1.1998658180236816, | |
| "learning_rate": 7.802030456852793e-06, | |
| "loss": 0.0383, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.363471971066908, | |
| "grad_norm": 1.4807426929473877, | |
| "learning_rate": 7.759729272419628e-06, | |
| "loss": 0.06, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 3.399638336347197, | |
| "grad_norm": 0.8246263265609741, | |
| "learning_rate": 7.717428087986464e-06, | |
| "loss": 0.0443, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.4358047016274864, | |
| "grad_norm": 0.9691349267959595, | |
| "learning_rate": 7.675126903553301e-06, | |
| "loss": 0.0535, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 3.4719710669077757, | |
| "grad_norm": 0.7182671427726746, | |
| "learning_rate": 7.632825719120136e-06, | |
| "loss": 0.0417, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.508137432188065, | |
| "grad_norm": 1.2362239360809326, | |
| "learning_rate": 7.5905245346869715e-06, | |
| "loss": 0.0529, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 3.5443037974683547, | |
| "grad_norm": 1.6033586263656616, | |
| "learning_rate": 7.5482233502538086e-06, | |
| "loss": 0.0377, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.580470162748644, | |
| "grad_norm": 1.3885324001312256, | |
| "learning_rate": 7.505922165820644e-06, | |
| "loss": 0.056, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 3.616636528028933, | |
| "grad_norm": 0.925649881362915, | |
| "learning_rate": 7.463620981387479e-06, | |
| "loss": 0.0399, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.6528028933092225, | |
| "grad_norm": 2.046786308288574, | |
| "learning_rate": 7.4213197969543146e-06, | |
| "loss": 0.064, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 3.6889692585895117, | |
| "grad_norm": 2.5204615592956543, | |
| "learning_rate": 7.379018612521152e-06, | |
| "loss": 0.0392, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.725135623869801, | |
| "grad_norm": 1.6770093441009521, | |
| "learning_rate": 7.336717428087987e-06, | |
| "loss": 0.0508, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 3.7613019891500903, | |
| "grad_norm": 1.5132670402526855, | |
| "learning_rate": 7.294416243654823e-06, | |
| "loss": 0.0442, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.7974683544303796, | |
| "grad_norm": 1.3414489030838013, | |
| "learning_rate": 7.2521150592216584e-06, | |
| "loss": 0.0551, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 3.833634719710669, | |
| "grad_norm": 1.41124427318573, | |
| "learning_rate": 7.209813874788495e-06, | |
| "loss": 0.0382, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 3.8698010849909585, | |
| "grad_norm": 1.133798360824585, | |
| "learning_rate": 7.167512690355331e-06, | |
| "loss": 0.0585, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 3.905967450271248, | |
| "grad_norm": 2.0255684852600098, | |
| "learning_rate": 7.125211505922166e-06, | |
| "loss": 0.0417, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 3.942133815551537, | |
| "grad_norm": 1.0678452253341675, | |
| "learning_rate": 7.082910321489002e-06, | |
| "loss": 0.0578, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 3.9783001808318263, | |
| "grad_norm": 1.9355206489562988, | |
| "learning_rate": 7.0406091370558385e-06, | |
| "loss": 0.0461, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.17637395858764648, | |
| "eval_runtime": 49043.1261, | |
| "eval_samples_per_second": 0.194, | |
| "eval_steps_per_second": 0.024, | |
| "eval_wer": 0.1119795003203075, | |
| "step": 2765 | |
| }, | |
| { | |
| "epoch": 4.014466546112116, | |
| "grad_norm": 0.5305593013763428, | |
| "learning_rate": 6.998307952622674e-06, | |
| "loss": 0.0582, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 4.050632911392405, | |
| "grad_norm": 1.1978957653045654, | |
| "learning_rate": 6.956006768189509e-06, | |
| "loss": 0.0205, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 4.086799276672695, | |
| "grad_norm": 1.0628869533538818, | |
| "learning_rate": 6.913705583756346e-06, | |
| "loss": 0.024, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 4.122965641952984, | |
| "grad_norm": 3.1139473915100098, | |
| "learning_rate": 6.8714043993231816e-06, | |
| "loss": 0.0242, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 4.159132007233273, | |
| "grad_norm": 0.7646371126174927, | |
| "learning_rate": 6.829103214890017e-06, | |
| "loss": 0.0363, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 4.195298372513562, | |
| "grad_norm": 1.3536927700042725, | |
| "learning_rate": 6.786802030456854e-06, | |
| "loss": 0.0254, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 4.231464737793852, | |
| "grad_norm": 1.744105339050293, | |
| "learning_rate": 6.744500846023689e-06, | |
| "loss": 0.0362, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 4.267631103074141, | |
| "grad_norm": 0.8838434815406799, | |
| "learning_rate": 6.702199661590525e-06, | |
| "loss": 0.0236, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 4.30379746835443, | |
| "grad_norm": 1.056674838066101, | |
| "learning_rate": 6.659898477157361e-06, | |
| "loss": 0.0311, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 4.3399638336347195, | |
| "grad_norm": 1.42670476436615, | |
| "learning_rate": 6.617597292724197e-06, | |
| "loss": 0.0277, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.376130198915009, | |
| "grad_norm": 1.0685067176818848, | |
| "learning_rate": 6.575296108291032e-06, | |
| "loss": 0.0346, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 4.412296564195298, | |
| "grad_norm": 0.6285632252693176, | |
| "learning_rate": 6.5329949238578685e-06, | |
| "loss": 0.0271, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 4.448462929475587, | |
| "grad_norm": 1.1495152711868286, | |
| "learning_rate": 6.490693739424705e-06, | |
| "loss": 0.0339, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 4.484629294755877, | |
| "grad_norm": 1.4339568614959717, | |
| "learning_rate": 6.448392554991541e-06, | |
| "loss": 0.0218, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 4.520795660036167, | |
| "grad_norm": 0.9594871997833252, | |
| "learning_rate": 6.406091370558376e-06, | |
| "loss": 0.029, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 4.556962025316456, | |
| "grad_norm": 1.1738754510879517, | |
| "learning_rate": 6.3637901861252115e-06, | |
| "loss": 0.0312, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 4.593128390596745, | |
| "grad_norm": 1.202968716621399, | |
| "learning_rate": 6.3214890016920485e-06, | |
| "loss": 0.032, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 4.6292947558770345, | |
| "grad_norm": 1.722782850265503, | |
| "learning_rate": 6.279187817258884e-06, | |
| "loss": 0.0327, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 4.665461121157324, | |
| "grad_norm": 0.8393648862838745, | |
| "learning_rate": 6.236886632825719e-06, | |
| "loss": 0.0356, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 4.701627486437613, | |
| "grad_norm": 2.255413770675659, | |
| "learning_rate": 6.194585448392556e-06, | |
| "loss": 0.0298, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 4.737793851717902, | |
| "grad_norm": 1.082092523574829, | |
| "learning_rate": 6.152284263959392e-06, | |
| "loss": 0.035, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 4.773960216998192, | |
| "grad_norm": 1.9351752996444702, | |
| "learning_rate": 6.109983079526227e-06, | |
| "loss": 0.025, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 4.810126582278481, | |
| "grad_norm": 1.7391856908798218, | |
| "learning_rate": 6.067681895093063e-06, | |
| "loss": 0.0398, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 4.84629294755877, | |
| "grad_norm": 1.332651972770691, | |
| "learning_rate": 6.025380710659899e-06, | |
| "loss": 0.0269, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 4.882459312839059, | |
| "grad_norm": 1.1322509050369263, | |
| "learning_rate": 5.983079526226735e-06, | |
| "loss": 0.0296, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 4.918625678119349, | |
| "grad_norm": 1.640367865562439, | |
| "learning_rate": 5.940778341793571e-06, | |
| "loss": 0.0245, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 4.954792043399638, | |
| "grad_norm": 1.5555723905563354, | |
| "learning_rate": 5.898477157360406e-06, | |
| "loss": 0.0455, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 4.990958408679928, | |
| "grad_norm": 2.142411947250366, | |
| "learning_rate": 5.856175972927242e-06, | |
| "loss": 0.0264, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 4.999638336347197, | |
| "eval_loss": 0.20242072641849518, | |
| "eval_runtime": 49328.761, | |
| "eval_samples_per_second": 0.192, | |
| "eval_steps_per_second": 0.024, | |
| "eval_wer": 0.11330877642536835, | |
| "step": 3456 | |
| }, | |
| { | |
| "epoch": 5.027124773960217, | |
| "grad_norm": 2.6065480709075928, | |
| "learning_rate": 5.8138747884940785e-06, | |
| "loss": 0.0223, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 5.063291139240507, | |
| "grad_norm": 6.660514831542969, | |
| "learning_rate": 5.771573604060914e-06, | |
| "loss": 0.0196, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.099457504520796, | |
| "grad_norm": 0.8771660923957825, | |
| "learning_rate": 5.729272419627751e-06, | |
| "loss": 0.0211, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 5.135623869801085, | |
| "grad_norm": 5.249636650085449, | |
| "learning_rate": 5.686971235194586e-06, | |
| "loss": 0.0212, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 5.1717902350813745, | |
| "grad_norm": 0.7272325754165649, | |
| "learning_rate": 5.6446700507614216e-06, | |
| "loss": 0.018, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 5.207956600361664, | |
| "grad_norm": 1.247758388519287, | |
| "learning_rate": 5.602368866328257e-06, | |
| "loss": 0.0174, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 5.244122965641953, | |
| "grad_norm": 1.6683036088943481, | |
| "learning_rate": 5.560067681895094e-06, | |
| "loss": 0.0208, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 5.280289330922242, | |
| "grad_norm": 0.9653752446174622, | |
| "learning_rate": 5.517766497461929e-06, | |
| "loss": 0.0158, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 5.3164556962025316, | |
| "grad_norm": 2.857499361038208, | |
| "learning_rate": 5.475465313028765e-06, | |
| "loss": 0.0215, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 5.352622061482821, | |
| "grad_norm": 2.953227996826172, | |
| "learning_rate": 5.433164128595602e-06, | |
| "loss": 0.017, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 5.38878842676311, | |
| "grad_norm": 2.159450054168701, | |
| "learning_rate": 5.390862944162437e-06, | |
| "loss": 0.0199, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 5.424954792043399, | |
| "grad_norm": 2.423708200454712, | |
| "learning_rate": 5.348561759729273e-06, | |
| "loss": 0.0185, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 5.461121157323689, | |
| "grad_norm": 1.3024959564208984, | |
| "learning_rate": 5.3062605752961085e-06, | |
| "loss": 0.0212, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 5.497287522603978, | |
| "grad_norm": 1.6463954448699951, | |
| "learning_rate": 5.263959390862945e-06, | |
| "loss": 0.018, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 5.533453887884267, | |
| "grad_norm": 0.8385624885559082, | |
| "learning_rate": 5.221658206429781e-06, | |
| "loss": 0.0191, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 5.569620253164557, | |
| "grad_norm": 1.0305800437927246, | |
| "learning_rate": 5.179357021996616e-06, | |
| "loss": 0.0179, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 5.605786618444847, | |
| "grad_norm": 1.5330723524093628, | |
| "learning_rate": 5.137055837563452e-06, | |
| "loss": 0.0203, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 5.641952983725136, | |
| "grad_norm": 1.5121113061904907, | |
| "learning_rate": 5.0947546531302885e-06, | |
| "loss": 0.0187, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 5.678119349005425, | |
| "grad_norm": 0.41953858733177185, | |
| "learning_rate": 5.052453468697124e-06, | |
| "loss": 0.0207, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 5.714285714285714, | |
| "grad_norm": 3.4192609786987305, | |
| "learning_rate": 5.010152284263959e-06, | |
| "loss": 0.0176, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 5.750452079566004, | |
| "grad_norm": 3.3937628269195557, | |
| "learning_rate": 4.967851099830795e-06, | |
| "loss": 0.0207, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 5.786618444846293, | |
| "grad_norm": 1.6074724197387695, | |
| "learning_rate": 4.925549915397632e-06, | |
| "loss": 0.0207, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.822784810126582, | |
| "grad_norm": 0.9468584656715393, | |
| "learning_rate": 4.883248730964467e-06, | |
| "loss": 0.018, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 5.8589511754068715, | |
| "grad_norm": 1.2002973556518555, | |
| "learning_rate": 4.840947546531303e-06, | |
| "loss": 0.0151, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 5.895117540687161, | |
| "grad_norm": 1.0356348752975464, | |
| "learning_rate": 4.798646362098139e-06, | |
| "loss": 0.0201, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 5.93128390596745, | |
| "grad_norm": 1.657277226448059, | |
| "learning_rate": 4.756345177664975e-06, | |
| "loss": 0.0202, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 5.967450271247739, | |
| "grad_norm": 1.1826387643814087, | |
| "learning_rate": 4.714043993231811e-06, | |
| "loss": 0.0203, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 5.9992766726943945, | |
| "eval_loss": 0.22004875540733337, | |
| "eval_runtime": 49062.4528, | |
| "eval_samples_per_second": 0.194, | |
| "eval_steps_per_second": 0.024, | |
| "eval_wer": 0.10991351697629724, | |
| "step": 4147 | |
| }, | |
| { | |
| "epoch": 6.003616636528029, | |
| "grad_norm": 0.2818962335586548, | |
| "learning_rate": 4.671742808798647e-06, | |
| "loss": 0.0216, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 6.039783001808318, | |
| "grad_norm": 0.8754785656929016, | |
| "learning_rate": 4.631133671742809e-06, | |
| "loss": 0.0117, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 6.075949367088608, | |
| "grad_norm": 0.6040045619010925, | |
| "learning_rate": 4.588832487309645e-06, | |
| "loss": 0.0185, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 6.112115732368897, | |
| "grad_norm": 0.5590214729309082, | |
| "learning_rate": 4.546531302876481e-06, | |
| "loss": 0.0095, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 6.1482820976491865, | |
| "grad_norm": 1.0092438459396362, | |
| "learning_rate": 4.504230118443317e-06, | |
| "loss": 0.0132, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 6.184448462929476, | |
| "grad_norm": 0.68355393409729, | |
| "learning_rate": 4.461928934010153e-06, | |
| "loss": 0.0073, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 6.220614828209765, | |
| "grad_norm": 0.854202926158905, | |
| "learning_rate": 4.4196277495769884e-06, | |
| "loss": 0.0208, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 6.256781193490054, | |
| "grad_norm": 0.3510783314704895, | |
| "learning_rate": 4.377326565143825e-06, | |
| "loss": 0.0112, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 6.292947558770344, | |
| "grad_norm": 0.36134082078933716, | |
| "learning_rate": 4.33502538071066e-06, | |
| "loss": 0.0136, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 6.329113924050633, | |
| "grad_norm": 1.4613643884658813, | |
| "learning_rate": 4.292724196277496e-06, | |
| "loss": 0.0109, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 6.365280289330922, | |
| "grad_norm": 0.7109003067016602, | |
| "learning_rate": 4.2504230118443315e-06, | |
| "loss": 0.0179, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 6.401446654611211, | |
| "grad_norm": 0.505111813545227, | |
| "learning_rate": 4.208121827411168e-06, | |
| "loss": 0.0094, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 6.437613019891501, | |
| "grad_norm": 1.2527493238449097, | |
| "learning_rate": 4.165820642978004e-06, | |
| "loss": 0.0195, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 6.47377938517179, | |
| "grad_norm": 1.6427637338638306, | |
| "learning_rate": 4.123519458544839e-06, | |
| "loss": 0.0102, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 6.509945750452079, | |
| "grad_norm": 0.24043454229831696, | |
| "learning_rate": 4.081218274111675e-06, | |
| "loss": 0.0181, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 6.5461121157323685, | |
| "grad_norm": 0.9318906664848328, | |
| "learning_rate": 4.0389170896785115e-06, | |
| "loss": 0.011, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 6.582278481012658, | |
| "grad_norm": 0.46116721630096436, | |
| "learning_rate": 3.996615905245348e-06, | |
| "loss": 0.0208, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 6.618444846292948, | |
| "grad_norm": 1.371081829071045, | |
| "learning_rate": 3.954314720812183e-06, | |
| "loss": 0.0092, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 6.654611211573237, | |
| "grad_norm": 1.1018704175949097, | |
| "learning_rate": 3.912013536379019e-06, | |
| "loss": 0.014, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 6.6907775768535265, | |
| "grad_norm": 1.1697708368301392, | |
| "learning_rate": 3.869712351945855e-06, | |
| "loss": 0.0126, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 6.726943942133816, | |
| "grad_norm": 0.5917466878890991, | |
| "learning_rate": 3.827411167512691e-06, | |
| "loss": 0.0168, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 6.763110307414105, | |
| "grad_norm": 0.7262076735496521, | |
| "learning_rate": 3.785109983079526e-06, | |
| "loss": 0.0077, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 6.799276672694394, | |
| "grad_norm": 0.16528570652008057, | |
| "learning_rate": 3.7428087986463623e-06, | |
| "loss": 0.0142, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 6.8354430379746836, | |
| "grad_norm": 2.227794885635376, | |
| "learning_rate": 3.7005076142131985e-06, | |
| "loss": 0.0077, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 6.871609403254973, | |
| "grad_norm": 0.8085231184959412, | |
| "learning_rate": 3.6582064297800342e-06, | |
| "loss": 0.0154, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 6.907775768535262, | |
| "grad_norm": 1.447237253189087, | |
| "learning_rate": 3.61590524534687e-06, | |
| "loss": 0.0087, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 6.943942133815551, | |
| "grad_norm": 0.4593859910964966, | |
| "learning_rate": 3.5736040609137058e-06, | |
| "loss": 0.0131, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 6.980108499095841, | |
| "grad_norm": 1.0502936840057373, | |
| "learning_rate": 3.531302876480542e-06, | |
| "loss": 0.0129, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 6.998915009041592, | |
| "eval_loss": 0.22768299281597137, | |
| "eval_runtime": 48990.2447, | |
| "eval_samples_per_second": 0.194, | |
| "eval_steps_per_second": 0.024, | |
| "eval_wer": 0.11143497757847534, | |
| "step": 4838 | |
| }, | |
| { | |
| "epoch": 7.01627486437613, | |
| "grad_norm": 0.29863396286964417, | |
| "learning_rate": 3.4890016920473773e-06, | |
| "loss": 0.0131, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 7.052441229656419, | |
| "grad_norm": 1.6121262311935425, | |
| "learning_rate": 3.4467005076142135e-06, | |
| "loss": 0.0067, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 7.0886075949367084, | |
| "grad_norm": 0.682396411895752, | |
| "learning_rate": 3.4043993231810496e-06, | |
| "loss": 0.0104, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 7.124773960216999, | |
| "grad_norm": 1.451640009880066, | |
| "learning_rate": 3.362098138747885e-06, | |
| "loss": 0.0061, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 7.160940325497288, | |
| "grad_norm": 1.0819129943847656, | |
| "learning_rate": 3.319796954314721e-06, | |
| "loss": 0.0098, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 7.197106690777577, | |
| "grad_norm": 0.8002307415008545, | |
| "learning_rate": 3.277495769881557e-06, | |
| "loss": 0.0076, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 7.233273056057866, | |
| "grad_norm": 0.34655046463012695, | |
| "learning_rate": 3.235194585448393e-06, | |
| "loss": 0.0119, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.269439421338156, | |
| "grad_norm": 0.7425087690353394, | |
| "learning_rate": 3.1928934010152284e-06, | |
| "loss": 0.0051, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 7.305605786618445, | |
| "grad_norm": 0.48938363790512085, | |
| "learning_rate": 3.1505922165820646e-06, | |
| "loss": 0.0103, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 7.341772151898734, | |
| "grad_norm": 1.4168592691421509, | |
| "learning_rate": 3.1082910321489e-06, | |
| "loss": 0.0053, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 7.3779385171790235, | |
| "grad_norm": 0.7530602812767029, | |
| "learning_rate": 3.065989847715736e-06, | |
| "loss": 0.0124, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 7.414104882459313, | |
| "grad_norm": 0.81525719165802, | |
| "learning_rate": 3.0236886632825723e-06, | |
| "loss": 0.0047, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 7.450271247739602, | |
| "grad_norm": 0.5934809446334839, | |
| "learning_rate": 2.981387478849408e-06, | |
| "loss": 0.0073, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 7.486437613019891, | |
| "grad_norm": 1.5411005020141602, | |
| "learning_rate": 2.939086294416244e-06, | |
| "loss": 0.0076, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 7.522603978300181, | |
| "grad_norm": 3.0440473556518555, | |
| "learning_rate": 2.8967851099830796e-06, | |
| "loss": 0.0119, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 7.55877034358047, | |
| "grad_norm": 0.15616849064826965, | |
| "learning_rate": 2.8544839255499158e-06, | |
| "loss": 0.0071, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 7.594936708860759, | |
| "grad_norm": 0.6418558955192566, | |
| "learning_rate": 2.812182741116751e-06, | |
| "loss": 0.0125, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 7.631103074141048, | |
| "grad_norm": 1.245208978652954, | |
| "learning_rate": 2.7698815566835873e-06, | |
| "loss": 0.009, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 7.6672694394213385, | |
| "grad_norm": 0.6857640743255615, | |
| "learning_rate": 2.7275803722504235e-06, | |
| "loss": 0.0143, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 7.703435804701628, | |
| "grad_norm": 0.6764671802520752, | |
| "learning_rate": 2.6852791878172592e-06, | |
| "loss": 0.0081, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 7.739602169981917, | |
| "grad_norm": 0.6338791251182556, | |
| "learning_rate": 2.642978003384095e-06, | |
| "loss": 0.0128, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 7.775768535262206, | |
| "grad_norm": 1.4596633911132812, | |
| "learning_rate": 2.6006768189509308e-06, | |
| "loss": 0.0075, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 7.811934900542496, | |
| "grad_norm": 0.14179235696792603, | |
| "learning_rate": 2.558375634517767e-06, | |
| "loss": 0.0142, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 7.848101265822785, | |
| "grad_norm": 0.7647730112075806, | |
| "learning_rate": 2.5160744500846023e-06, | |
| "loss": 0.0068, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 7.884267631103074, | |
| "grad_norm": 0.1921517550945282, | |
| "learning_rate": 2.4737732656514385e-06, | |
| "loss": 0.0078, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 7.920433996383363, | |
| "grad_norm": 1.385634183883667, | |
| "learning_rate": 2.4314720812182742e-06, | |
| "loss": 0.0067, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 7.956600361663653, | |
| "grad_norm": 0.08770379424095154, | |
| "learning_rate": 2.38917089678511e-06, | |
| "loss": 0.007, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 7.992766726943942, | |
| "grad_norm": 2.0286593437194824, | |
| "learning_rate": 2.346869712351946e-06, | |
| "loss": 0.0091, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_loss": 0.2551898956298828, | |
| "eval_runtime": 49008.1266, | |
| "eval_samples_per_second": 0.194, | |
| "eval_steps_per_second": 0.024, | |
| "eval_wer": 0.10667841127482383, | |
| "step": 5530 | |
| }, | |
| { | |
| "epoch": 8.028933092224232, | |
| "grad_norm": 0.40499478578567505, | |
| "learning_rate": 2.304568527918782e-06, | |
| "loss": 0.011, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 8.065099457504521, | |
| "grad_norm": 1.5098172426223755, | |
| "learning_rate": 2.2622673434856177e-06, | |
| "loss": 0.0074, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 8.10126582278481, | |
| "grad_norm": 2.1074013710021973, | |
| "learning_rate": 2.219966159052454e-06, | |
| "loss": 0.0079, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 8.1374321880651, | |
| "grad_norm": 1.3512495756149292, | |
| "learning_rate": 2.1776649746192896e-06, | |
| "loss": 0.0049, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 8.17359855334539, | |
| "grad_norm": 0.27601808309555054, | |
| "learning_rate": 2.1353637901861254e-06, | |
| "loss": 0.0037, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 8.209764918625678, | |
| "grad_norm": 0.38552117347717285, | |
| "learning_rate": 2.093062605752961e-06, | |
| "loss": 0.0054, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 8.245931283905968, | |
| "grad_norm": 0.6290847063064575, | |
| "learning_rate": 2.050761421319797e-06, | |
| "loss": 0.0075, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 8.282097649186257, | |
| "grad_norm": 0.34640949964523315, | |
| "learning_rate": 2.008460236886633e-06, | |
| "loss": 0.0038, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 8.318264014466546, | |
| "grad_norm": 0.47082650661468506, | |
| "learning_rate": 1.966159052453469e-06, | |
| "loss": 0.0072, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 8.354430379746836, | |
| "grad_norm": 0.7971699237823486, | |
| "learning_rate": 1.9238578680203046e-06, | |
| "loss": 0.0074, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 8.390596745027125, | |
| "grad_norm": 1.3511334657669067, | |
| "learning_rate": 1.8815566835871408e-06, | |
| "loss": 0.0065, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 8.426763110307414, | |
| "grad_norm": 0.04755210131406784, | |
| "learning_rate": 1.8392554991539766e-06, | |
| "loss": 0.0054, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 8.462929475587703, | |
| "grad_norm": 0.7288948893547058, | |
| "learning_rate": 1.7969543147208123e-06, | |
| "loss": 0.0075, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 8.499095840867993, | |
| "grad_norm": 2.372441291809082, | |
| "learning_rate": 1.7546531302876483e-06, | |
| "loss": 0.0071, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 8.535262206148282, | |
| "grad_norm": 0.029580192640423775, | |
| "learning_rate": 1.712351945854484e-06, | |
| "loss": 0.0057, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 8.571428571428571, | |
| "grad_norm": 3.4771149158477783, | |
| "learning_rate": 1.6700507614213198e-06, | |
| "loss": 0.0063, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 8.60759493670886, | |
| "grad_norm": 0.47946012020111084, | |
| "learning_rate": 1.6277495769881558e-06, | |
| "loss": 0.0095, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 8.64376130198915, | |
| "grad_norm": 1.6082369089126587, | |
| "learning_rate": 1.5854483925549915e-06, | |
| "loss": 0.0061, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 8.679927667269439, | |
| "grad_norm": 0.8396530747413635, | |
| "learning_rate": 1.5431472081218277e-06, | |
| "loss": 0.005, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 8.716094032549728, | |
| "grad_norm": 0.683506965637207, | |
| "learning_rate": 1.5008460236886635e-06, | |
| "loss": 0.0042, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 8.752260397830018, | |
| "grad_norm": 0.23053808510303497, | |
| "learning_rate": 1.4585448392554992e-06, | |
| "loss": 0.0085, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 8.788426763110307, | |
| "grad_norm": 1.4277093410491943, | |
| "learning_rate": 1.4162436548223352e-06, | |
| "loss": 0.0057, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 8.824593128390596, | |
| "grad_norm": 0.8418155908584595, | |
| "learning_rate": 1.373942470389171e-06, | |
| "loss": 0.0085, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 8.860759493670885, | |
| "grad_norm": 0.16715645790100098, | |
| "learning_rate": 1.3316412859560067e-06, | |
| "loss": 0.0039, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 8.896925858951175, | |
| "grad_norm": 1.1576683521270752, | |
| "learning_rate": 1.2893401015228427e-06, | |
| "loss": 0.0056, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 8.933092224231466, | |
| "grad_norm": 4.466509819030762, | |
| "learning_rate": 1.2470389170896787e-06, | |
| "loss": 0.008, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 8.969258589511753, | |
| "grad_norm": 0.09006431698799133, | |
| "learning_rate": 1.2047377326565144e-06, | |
| "loss": 0.0063, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 8.999638336347196, | |
| "eval_loss": 0.25654470920562744, | |
| "eval_runtime": 49994.6568, | |
| "eval_samples_per_second": 0.19, | |
| "eval_steps_per_second": 0.024, | |
| "eval_wer": 0.10539718129404228, | |
| "step": 6221 | |
| }, | |
| { | |
| "epoch": 9.005424954792044, | |
| "grad_norm": 0.04043040797114372, | |
| "learning_rate": 1.1624365482233504e-06, | |
| "loss": 0.0099, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 9.041591320072333, | |
| "grad_norm": 0.31738966703414917, | |
| "learning_rate": 1.1201353637901864e-06, | |
| "loss": 0.0031, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 9.077757685352623, | |
| "grad_norm": 0.8276641368865967, | |
| "learning_rate": 1.0778341793570221e-06, | |
| "loss": 0.0064, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 9.113924050632912, | |
| "grad_norm": 0.3469080626964569, | |
| "learning_rate": 1.035532994923858e-06, | |
| "loss": 0.0032, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 9.150090415913201, | |
| "grad_norm": 0.6140421628952026, | |
| "learning_rate": 9.932318104906939e-07, | |
| "loss": 0.0088, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 9.18625678119349, | |
| "grad_norm": 1.1443873643875122, | |
| "learning_rate": 9.509306260575297e-07, | |
| "loss": 0.0032, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 9.22242314647378, | |
| "grad_norm": 0.21027229726314545, | |
| "learning_rate": 9.086294416243656e-07, | |
| "loss": 0.0105, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 9.258589511754069, | |
| "grad_norm": 0.2232620120048523, | |
| "learning_rate": 8.663282571912015e-07, | |
| "loss": 0.0047, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 9.294755877034358, | |
| "grad_norm": 0.5159393548965454, | |
| "learning_rate": 8.240270727580372e-07, | |
| "loss": 0.0069, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 9.330922242314648, | |
| "grad_norm": 0.5648090839385986, | |
| "learning_rate": 7.817258883248732e-07, | |
| "loss": 0.0031, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 9.367088607594937, | |
| "grad_norm": 0.24160954356193542, | |
| "learning_rate": 7.394247038917091e-07, | |
| "loss": 0.0076, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 9.403254972875226, | |
| "grad_norm": 0.21812845766544342, | |
| "learning_rate": 6.971235194585449e-07, | |
| "loss": 0.0036, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 9.439421338155515, | |
| "grad_norm": 0.3133035898208618, | |
| "learning_rate": 6.548223350253807e-07, | |
| "loss": 0.0086, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 9.475587703435805, | |
| "grad_norm": 0.16532699763774872, | |
| "learning_rate": 6.125211505922167e-07, | |
| "loss": 0.0029, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 9.511754068716094, | |
| "grad_norm": 0.5517584681510925, | |
| "learning_rate": 5.702199661590524e-07, | |
| "loss": 0.0096, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 9.547920433996383, | |
| "grad_norm": 0.2974016070365906, | |
| "learning_rate": 5.279187817258884e-07, | |
| "loss": 0.0018, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 9.584086799276673, | |
| "grad_norm": 0.7380957007408142, | |
| "learning_rate": 4.856175972927242e-07, | |
| "loss": 0.0091, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 9.620253164556962, | |
| "grad_norm": 0.2390100508928299, | |
| "learning_rate": 4.433164128595601e-07, | |
| "loss": 0.0024, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 9.656419529837251, | |
| "grad_norm": 0.22240670025348663, | |
| "learning_rate": 4.0101522842639594e-07, | |
| "loss": 0.0054, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 9.69258589511754, | |
| "grad_norm": 0.03914798051118851, | |
| "learning_rate": 3.5871404399323186e-07, | |
| "loss": 0.0029, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 9.72875226039783, | |
| "grad_norm": 0.5454278588294983, | |
| "learning_rate": 3.1641285956006767e-07, | |
| "loss": 0.0053, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 9.764918625678119, | |
| "grad_norm": 0.21154353022575378, | |
| "learning_rate": 2.741116751269036e-07, | |
| "loss": 0.0024, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 9.801084990958408, | |
| "grad_norm": 0.47212332487106323, | |
| "learning_rate": 2.3181049069373945e-07, | |
| "loss": 0.0081, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 9.837251356238697, | |
| "grad_norm": 0.09104613214731216, | |
| "learning_rate": 1.8950930626057532e-07, | |
| "loss": 0.0048, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 9.873417721518987, | |
| "grad_norm": 0.35835790634155273, | |
| "learning_rate": 1.4720812182741118e-07, | |
| "loss": 0.0085, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 9.909584086799276, | |
| "grad_norm": 0.13305124640464783, | |
| "learning_rate": 1.0490693739424705e-07, | |
| "loss": 0.0028, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 9.945750452079565, | |
| "grad_norm": 0.4060196280479431, | |
| "learning_rate": 6.260575296108291e-08, | |
| "loss": 0.0096, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 9.981916817359856, | |
| "grad_norm": 1.0367088317871094, | |
| "learning_rate": 2.0304568527918784e-08, | |
| "loss": 0.0019, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 9.99638336347197, | |
| "eval_loss": 0.26707419753074646, | |
| "eval_runtime": 51487.4721, | |
| "eval_samples_per_second": 0.184, | |
| "eval_steps_per_second": 0.023, | |
| "eval_wer": 0.10419602818705957, | |
| "step": 6910 | |
| }, | |
| { | |
| "epoch": 9.99638336347197, | |
| "step": 6910, | |
| "total_flos": 7.51132515607511e+20, | |
| "train_loss": 0.05691683961323822, | |
| "train_runtime": 650193.1322, | |
| "train_samples_per_second": 0.34, | |
| "train_steps_per_second": 0.011 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 6910, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 7.51132515607511e+20, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |