| { | |
| "best_metric": 82.33239678451572, | |
| "best_model_checkpoint": "./whisper-small-dialect_all/checkpoint-5000", | |
| "epoch": 0.6034274680183442, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.003017137340091721, | |
| "grad_norm": 96.26067352294922, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 4.8022, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.006034274680183442, | |
| "grad_norm": 43.14243698120117, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 4.0872, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.009051412020275163, | |
| "grad_norm": 36.04750442504883, | |
| "learning_rate": 1.5e-06, | |
| "loss": 3.4292, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.012068549360366883, | |
| "grad_norm": 30.31016731262207, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 2.6464, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.015085686700458605, | |
| "grad_norm": 31.633859634399414, | |
| "learning_rate": 2.5e-06, | |
| "loss": 2.3122, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.018102824040550327, | |
| "grad_norm": 29.1795711517334, | |
| "learning_rate": 3e-06, | |
| "loss": 2.2551, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.021119961380642045, | |
| "grad_norm": 33.240631103515625, | |
| "learning_rate": 3.5e-06, | |
| "loss": 2.0591, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.024137098720733767, | |
| "grad_norm": 25.619850158691406, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 2.0698, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02715423606082549, | |
| "grad_norm": 28.270214080810547, | |
| "learning_rate": 4.5e-06, | |
| "loss": 1.986, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.03017137340091721, | |
| "grad_norm": 28.568641662597656, | |
| "learning_rate": 5e-06, | |
| "loss": 1.9292, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.03318851074100893, | |
| "grad_norm": 26.848628997802734, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 1.8674, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.036205648081100654, | |
| "grad_norm": 25.980438232421875, | |
| "learning_rate": 6e-06, | |
| "loss": 1.8075, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.039222785421192376, | |
| "grad_norm": 32.828521728515625, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 1.7452, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.04223992276128409, | |
| "grad_norm": 22.947566986083984, | |
| "learning_rate": 7e-06, | |
| "loss": 1.6252, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.04525706010137581, | |
| "grad_norm": 26.654155731201172, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 1.4486, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.048274197441467534, | |
| "grad_norm": 31.73291778564453, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 1.5101, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.051291334781559256, | |
| "grad_norm": 28.088008880615234, | |
| "learning_rate": 8.5e-06, | |
| "loss": 1.6028, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.05430847212165098, | |
| "grad_norm": 31.480905532836914, | |
| "learning_rate": 9e-06, | |
| "loss": 1.5145, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.0573256094617427, | |
| "grad_norm": 28.74748420715332, | |
| "learning_rate": 9.5e-06, | |
| "loss": 1.4598, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.06034274680183442, | |
| "grad_norm": 25.75771141052246, | |
| "learning_rate": 1e-05, | |
| "loss": 1.4026, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.06335988414192614, | |
| "grad_norm": 26.966527938842773, | |
| "learning_rate": 9.944444444444445e-06, | |
| "loss": 1.5448, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.06637702148201786, | |
| "grad_norm": 22.180273056030273, | |
| "learning_rate": 9.88888888888889e-06, | |
| "loss": 1.4776, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.06939415882210959, | |
| "grad_norm": 22.99480438232422, | |
| "learning_rate": 9.833333333333333e-06, | |
| "loss": 1.5805, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.07241129616220131, | |
| "grad_norm": 22.988994598388672, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 1.4311, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.07542843350229303, | |
| "grad_norm": 24.09758758544922, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 1.4485, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.07844557084238475, | |
| "grad_norm": 23.2039852142334, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 1.3868, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.08146270818247647, | |
| "grad_norm": 24.07797622680664, | |
| "learning_rate": 9.611111111111112e-06, | |
| "loss": 1.3593, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.08447984552256818, | |
| "grad_norm": 28.38634490966797, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 1.4899, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.0874969828626599, | |
| "grad_norm": 16.204975128173828, | |
| "learning_rate": 9.5e-06, | |
| "loss": 1.464, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.09051412020275162, | |
| "grad_norm": 25.585763931274414, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 1.4399, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.09353125754284335, | |
| "grad_norm": 26.01482582092285, | |
| "learning_rate": 9.38888888888889e-06, | |
| "loss": 1.3702, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.09654839488293507, | |
| "grad_norm": 28.53825569152832, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 1.4086, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.09956553222302679, | |
| "grad_norm": 22.171388626098633, | |
| "learning_rate": 9.277777777777778e-06, | |
| "loss": 1.4689, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.10258266956311851, | |
| "grad_norm": 24.513687133789062, | |
| "learning_rate": 9.222222222222224e-06, | |
| "loss": 1.4648, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.10559980690321023, | |
| "grad_norm": 21.335763931274414, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 1.3645, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.10861694424330195, | |
| "grad_norm": 23.623140335083008, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 1.3909, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.11163408158339368, | |
| "grad_norm": 22.183908462524414, | |
| "learning_rate": 9.055555555555556e-06, | |
| "loss": 1.459, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.1146512189234854, | |
| "grad_norm": 22.47441864013672, | |
| "learning_rate": 9e-06, | |
| "loss": 1.4225, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.11766835626357712, | |
| "grad_norm": 21.476436614990234, | |
| "learning_rate": 8.944444444444446e-06, | |
| "loss": 1.3905, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.12068549360366884, | |
| "grad_norm": 18.394203186035156, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 1.4523, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.12068549360366884, | |
| "eval_cer": 61.733491557731654, | |
| "eval_loss": 1.384151577949524, | |
| "eval_runtime": 4550.9714, | |
| "eval_samples_per_second": 3.642, | |
| "eval_steps_per_second": 0.455, | |
| "eval_wer": 91.60325450714119, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.12370263094376056, | |
| "grad_norm": 22.194650650024414, | |
| "learning_rate": 8.833333333333334e-06, | |
| "loss": 1.2796, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.12671976828385229, | |
| "grad_norm": 21.923349380493164, | |
| "learning_rate": 8.777777777777778e-06, | |
| "loss": 1.4362, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.129736905623944, | |
| "grad_norm": 18.93937873840332, | |
| "learning_rate": 8.722222222222224e-06, | |
| "loss": 1.4115, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.13275404296403573, | |
| "grad_norm": 23.666305541992188, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 1.2855, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.13577118030412744, | |
| "grad_norm": 25.00994873046875, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 1.3238, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.13878831764421917, | |
| "grad_norm": 27.25493812561035, | |
| "learning_rate": 8.555555555555556e-06, | |
| "loss": 1.3858, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.14180545498431088, | |
| "grad_norm": 25.662391662597656, | |
| "learning_rate": 8.5e-06, | |
| "loss": 1.3905, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.14482259232440262, | |
| "grad_norm": 18.837465286254883, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 1.396, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.14783972966449432, | |
| "grad_norm": 24.336353302001953, | |
| "learning_rate": 8.38888888888889e-06, | |
| "loss": 1.4919, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.15085686700458606, | |
| "grad_norm": 26.946441650390625, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 1.3943, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.15387400434467777, | |
| "grad_norm": 22.371461868286133, | |
| "learning_rate": 8.277777777777778e-06, | |
| "loss": 1.2847, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.1568911416847695, | |
| "grad_norm": 23.828168869018555, | |
| "learning_rate": 8.222222222222222e-06, | |
| "loss": 1.4082, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.1599082790248612, | |
| "grad_norm": 26.558212280273438, | |
| "learning_rate": 8.166666666666668e-06, | |
| "loss": 1.3358, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.16292541636495295, | |
| "grad_norm": 21.01841163635254, | |
| "learning_rate": 8.111111111111112e-06, | |
| "loss": 1.4203, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.16594255370504465, | |
| "grad_norm": 21.629016876220703, | |
| "learning_rate": 8.055555555555557e-06, | |
| "loss": 1.3997, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.16895969104513636, | |
| "grad_norm": 29.030927658081055, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 1.3639, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.1719768283852281, | |
| "grad_norm": 21.38187599182129, | |
| "learning_rate": 7.944444444444445e-06, | |
| "loss": 1.2888, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.1749939657253198, | |
| "grad_norm": 22.521839141845703, | |
| "learning_rate": 7.88888888888889e-06, | |
| "loss": 1.3174, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.17801110306541154, | |
| "grad_norm": 23.328771591186523, | |
| "learning_rate": 7.833333333333333e-06, | |
| "loss": 1.352, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.18102824040550325, | |
| "grad_norm": 30.692312240600586, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 1.3676, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.18404537774559498, | |
| "grad_norm": 15.439107894897461, | |
| "learning_rate": 7.722222222222223e-06, | |
| "loss": 1.294, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.1870625150856867, | |
| "grad_norm": 23.497426986694336, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 1.3519, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.19007965242577843, | |
| "grad_norm": 24.0831356048584, | |
| "learning_rate": 7.611111111111111e-06, | |
| "loss": 1.4069, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.19309678976587014, | |
| "grad_norm": 13.516092300415039, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 1.2309, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.19611392710596187, | |
| "grad_norm": 24.730207443237305, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 1.3205, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.19913106444605358, | |
| "grad_norm": 20.125049591064453, | |
| "learning_rate": 7.444444444444445e-06, | |
| "loss": 1.4066, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.20214820178614532, | |
| "grad_norm": 22.153867721557617, | |
| "learning_rate": 7.38888888888889e-06, | |
| "loss": 1.3224, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.20516533912623702, | |
| "grad_norm": 21.373519897460938, | |
| "learning_rate": 7.333333333333333e-06, | |
| "loss": 1.4275, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.20818247646632876, | |
| "grad_norm": 21.73772430419922, | |
| "learning_rate": 7.277777777777778e-06, | |
| "loss": 1.3157, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.21119961380642047, | |
| "grad_norm": 21.263460159301758, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 1.3194, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.2142167511465122, | |
| "grad_norm": 21.77660369873047, | |
| "learning_rate": 7.166666666666667e-06, | |
| "loss": 1.3158, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.2172338884866039, | |
| "grad_norm": 26.697940826416016, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 1.4215, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.22025102582669562, | |
| "grad_norm": 21.863849639892578, | |
| "learning_rate": 7.055555555555557e-06, | |
| "loss": 1.3072, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.22326816316678735, | |
| "grad_norm": 21.956884384155273, | |
| "learning_rate": 7e-06, | |
| "loss": 1.3314, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.22628530050687906, | |
| "grad_norm": 23.326414108276367, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 1.4175, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.2293024378469708, | |
| "grad_norm": 21.866300582885742, | |
| "learning_rate": 6.88888888888889e-06, | |
| "loss": 1.4136, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.2323195751870625, | |
| "grad_norm": 26.420286178588867, | |
| "learning_rate": 6.833333333333334e-06, | |
| "loss": 1.3006, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.23533671252715424, | |
| "grad_norm": 19.8134822845459, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 1.3161, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.23835384986724595, | |
| "grad_norm": 19.25493049621582, | |
| "learning_rate": 6.7222222222222235e-06, | |
| "loss": 1.2434, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.24137098720733768, | |
| "grad_norm": 25.278705596923828, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 1.289, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.24137098720733768, | |
| "eval_cer": 61.0700703600228, | |
| "eval_loss": 1.3041576147079468, | |
| "eval_runtime": 4601.9427, | |
| "eval_samples_per_second": 3.601, | |
| "eval_steps_per_second": 0.45, | |
| "eval_wer": 90.03355966596426, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2443881245474294, | |
| "grad_norm": 23.50843620300293, | |
| "learning_rate": 6.6111111111111115e-06, | |
| "loss": 1.2585, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.24740526188752113, | |
| "grad_norm": 22.55748748779297, | |
| "learning_rate": 6.555555555555556e-06, | |
| "loss": 1.3618, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.25042239922761284, | |
| "grad_norm": 27.278711318969727, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 1.2642, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.25343953656770457, | |
| "grad_norm": 19.05206298828125, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 1.1832, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.2564566739077963, | |
| "grad_norm": 21.810386657714844, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 1.3212, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.259473811247888, | |
| "grad_norm": 24.683704376220703, | |
| "learning_rate": 6.333333333333333e-06, | |
| "loss": 1.3897, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.2624909485879797, | |
| "grad_norm": 22.081329345703125, | |
| "learning_rate": 6.277777777777778e-06, | |
| "loss": 1.3165, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.26550808592807146, | |
| "grad_norm": 25.10523796081543, | |
| "learning_rate": 6.222222222222223e-06, | |
| "loss": 1.2884, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.2685252232681632, | |
| "grad_norm": 20.280773162841797, | |
| "learning_rate": 6.166666666666667e-06, | |
| "loss": 1.2945, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.2715423606082549, | |
| "grad_norm": 23.45345115661621, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 1.3067, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.2745594979483466, | |
| "grad_norm": 22.918188095092773, | |
| "learning_rate": 6.055555555555555e-06, | |
| "loss": 1.4076, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.27757663528843834, | |
| "grad_norm": 20.786130905151367, | |
| "learning_rate": 6e-06, | |
| "loss": 1.2956, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.28059377262853, | |
| "grad_norm": 20.77971076965332, | |
| "learning_rate": 5.944444444444445e-06, | |
| "loss": 1.3064, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.28361090996862176, | |
| "grad_norm": 23.5718994140625, | |
| "learning_rate": 5.88888888888889e-06, | |
| "loss": 1.2232, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.2866280473087135, | |
| "grad_norm": 20.429147720336914, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 1.2631, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.28964518464880523, | |
| "grad_norm": 18.190717697143555, | |
| "learning_rate": 5.777777777777778e-06, | |
| "loss": 1.3747, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.2926623219888969, | |
| "grad_norm": 19.36117172241211, | |
| "learning_rate": 5.722222222222222e-06, | |
| "loss": 1.2209, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.29567945932898865, | |
| "grad_norm": 21.805431365966797, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 1.2741, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.2986965966690804, | |
| "grad_norm": 23.5611572265625, | |
| "learning_rate": 5.611111111111112e-06, | |
| "loss": 1.2424, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.3017137340091721, | |
| "grad_norm": 17.004058837890625, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 1.2091, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.3047308713492638, | |
| "grad_norm": 15.842552185058594, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 1.288, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.30774800868935553, | |
| "grad_norm": 24.12178611755371, | |
| "learning_rate": 5.444444444444445e-06, | |
| "loss": 1.2938, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.31076514602944727, | |
| "grad_norm": 20.887136459350586, | |
| "learning_rate": 5.388888888888889e-06, | |
| "loss": 1.3591, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.313782283369539, | |
| "grad_norm": 20.898900985717773, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 1.2786, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.3167994207096307, | |
| "grad_norm": 26.384042739868164, | |
| "learning_rate": 5.2777777777777785e-06, | |
| "loss": 1.2195, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.3198165580497224, | |
| "grad_norm": 23.89251708984375, | |
| "learning_rate": 5.2222222222222226e-06, | |
| "loss": 1.2215, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.32283369538981416, | |
| "grad_norm": 23.179443359375, | |
| "learning_rate": 5.1666666666666675e-06, | |
| "loss": 1.2479, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.3258508327299059, | |
| "grad_norm": 21.939584732055664, | |
| "learning_rate": 5.1111111111111115e-06, | |
| "loss": 1.2047, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.3288679700699976, | |
| "grad_norm": 18.902935028076172, | |
| "learning_rate": 5.0555555555555555e-06, | |
| "loss": 1.2943, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.3318851074100893, | |
| "grad_norm": 25.96902084350586, | |
| "learning_rate": 5e-06, | |
| "loss": 1.2992, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.33490224475018104, | |
| "grad_norm": 19.23320960998535, | |
| "learning_rate": 4.944444444444445e-06, | |
| "loss": 1.1501, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.3379193820902727, | |
| "grad_norm": 19.24538803100586, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 1.2567, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.34093651943036446, | |
| "grad_norm": 21.92410659790039, | |
| "learning_rate": 4.833333333333333e-06, | |
| "loss": 1.3391, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.3439536567704562, | |
| "grad_norm": 21.012697219848633, | |
| "learning_rate": 4.777777777777778e-06, | |
| "loss": 1.4432, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.34697079411054793, | |
| "grad_norm": 23.614274978637695, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 1.3435, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.3499879314506396, | |
| "grad_norm": 21.29604148864746, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 1.2999, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.35300506879073135, | |
| "grad_norm": 28.61496925354004, | |
| "learning_rate": 4.611111111111112e-06, | |
| "loss": 1.311, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.3560222061308231, | |
| "grad_norm": 26.641523361206055, | |
| "learning_rate": 4.555555555555556e-06, | |
| "loss": 1.352, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.3590393434709148, | |
| "grad_norm": 25.239164352416992, | |
| "learning_rate": 4.5e-06, | |
| "loss": 1.2567, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.3620564808110065, | |
| "grad_norm": 15.89976692199707, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 1.2534, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.3620564808110065, | |
| "eval_cer": 58.7271656964521, | |
| "eval_loss": 1.2533084154129028, | |
| "eval_runtime": 4548.7652, | |
| "eval_samples_per_second": 3.643, | |
| "eval_steps_per_second": 0.456, | |
| "eval_wer": 87.17025677046749, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.36507361815109823, | |
| "grad_norm": 21.1923885345459, | |
| "learning_rate": 4.388888888888889e-06, | |
| "loss": 1.2428, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.36809075549118997, | |
| "grad_norm": 23.164690017700195, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 1.3261, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.3711078928312817, | |
| "grad_norm": 26.1407413482666, | |
| "learning_rate": 4.277777777777778e-06, | |
| "loss": 1.2174, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.3741250301713734, | |
| "grad_norm": 21.50725555419922, | |
| "learning_rate": 4.222222222222223e-06, | |
| "loss": 1.291, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.3771421675114651, | |
| "grad_norm": 28.819339752197266, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 1.2261, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.38015930485155686, | |
| "grad_norm": 17.817045211791992, | |
| "learning_rate": 4.111111111111111e-06, | |
| "loss": 1.1822, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.38317644219164854, | |
| "grad_norm": 22.199352264404297, | |
| "learning_rate": 4.055555555555556e-06, | |
| "loss": 1.3147, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.38619357953174027, | |
| "grad_norm": 16.548112869262695, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 1.2316, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.389210716871832, | |
| "grad_norm": 22.51311683654785, | |
| "learning_rate": 3.944444444444445e-06, | |
| "loss": 1.2971, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.39222785421192374, | |
| "grad_norm": 22.15917205810547, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 1.3681, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.3952449915520154, | |
| "grad_norm": 27.50135040283203, | |
| "learning_rate": 3.833333333333334e-06, | |
| "loss": 1.2541, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.39826212889210716, | |
| "grad_norm": 21.28223991394043, | |
| "learning_rate": 3.777777777777778e-06, | |
| "loss": 1.2734, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.4012792662321989, | |
| "grad_norm": 19.02390480041504, | |
| "learning_rate": 3.7222222222222225e-06, | |
| "loss": 1.2023, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.40429640357229063, | |
| "grad_norm": 15.156362533569336, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 1.2511, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.4073135409123823, | |
| "grad_norm": 23.127460479736328, | |
| "learning_rate": 3.6111111111111115e-06, | |
| "loss": 1.3393, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.41033067825247405, | |
| "grad_norm": 23.209978103637695, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 1.2491, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.4133478155925658, | |
| "grad_norm": 17.73348045349121, | |
| "learning_rate": 3.5e-06, | |
| "loss": 1.2695, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.4163649529326575, | |
| "grad_norm": 23.244314193725586, | |
| "learning_rate": 3.444444444444445e-06, | |
| "loss": 1.3284, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.4193820902727492, | |
| "grad_norm": 17.389102935791016, | |
| "learning_rate": 3.3888888888888893e-06, | |
| "loss": 1.2235, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.42239922761284093, | |
| "grad_norm": 18.45189666748047, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 1.2259, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.42541636495293267, | |
| "grad_norm": 18.34843635559082, | |
| "learning_rate": 3.277777777777778e-06, | |
| "loss": 1.3223, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.4284335022930244, | |
| "grad_norm": 17.53985023498535, | |
| "learning_rate": 3.2222222222222227e-06, | |
| "loss": 1.27, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.4314506396331161, | |
| "grad_norm": 20.862815856933594, | |
| "learning_rate": 3.1666666666666667e-06, | |
| "loss": 1.2712, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.4344677769732078, | |
| "grad_norm": 28.492572784423828, | |
| "learning_rate": 3.1111111111111116e-06, | |
| "loss": 1.2335, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.43748491431329956, | |
| "grad_norm": 18.365795135498047, | |
| "learning_rate": 3.055555555555556e-06, | |
| "loss": 1.2517, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.44050205165339124, | |
| "grad_norm": 22.894845962524414, | |
| "learning_rate": 3e-06, | |
| "loss": 1.3125, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.44351918899348297, | |
| "grad_norm": 17.793073654174805, | |
| "learning_rate": 2.944444444444445e-06, | |
| "loss": 1.2094, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.4465363263335747, | |
| "grad_norm": 21.477651596069336, | |
| "learning_rate": 2.888888888888889e-06, | |
| "loss": 1.2371, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.44955346367366644, | |
| "grad_norm": 23.065067291259766, | |
| "learning_rate": 2.8333333333333335e-06, | |
| "loss": 1.1843, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.4525706010137581, | |
| "grad_norm": 23.395771026611328, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 1.2631, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.45558773835384986, | |
| "grad_norm": 20.306285858154297, | |
| "learning_rate": 2.7222222222222224e-06, | |
| "loss": 1.2584, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.4586048756939416, | |
| "grad_norm": 19.325054168701172, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 1.2586, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.46162201303403333, | |
| "grad_norm": 22.971973419189453, | |
| "learning_rate": 2.6111111111111113e-06, | |
| "loss": 1.2183, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.464639150374125, | |
| "grad_norm": 28.512895584106445, | |
| "learning_rate": 2.5555555555555557e-06, | |
| "loss": 1.2019, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.46765628771421675, | |
| "grad_norm": 22.517488479614258, | |
| "learning_rate": 2.5e-06, | |
| "loss": 1.188, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.4706734250543085, | |
| "grad_norm": 21.99509620666504, | |
| "learning_rate": 2.4444444444444447e-06, | |
| "loss": 1.2423, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.4736905623944002, | |
| "grad_norm": 28.65657615661621, | |
| "learning_rate": 2.388888888888889e-06, | |
| "loss": 1.2421, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.4767076997344919, | |
| "grad_norm": 18.730836868286133, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 1.199, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.47972483707458363, | |
| "grad_norm": 15.609992980957031, | |
| "learning_rate": 2.277777777777778e-06, | |
| "loss": 1.2331, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.48274197441467537, | |
| "grad_norm": 20.7346134185791, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 1.2008, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.48274197441467537, | |
| "eval_cer": 61.418973846254474, | |
| "eval_loss": 1.2168192863464355, | |
| "eval_runtime": 4689.4507, | |
| "eval_samples_per_second": 3.534, | |
| "eval_steps_per_second": 0.442, | |
| "eval_wer": 88.51947241083275, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.4857591117547671, | |
| "grad_norm": 22.913969039916992, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 1.1668, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.4887762490948588, | |
| "grad_norm": 22.071813583374023, | |
| "learning_rate": 2.1111111111111114e-06, | |
| "loss": 1.2253, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.4917933864349505, | |
| "grad_norm": 26.147418975830078, | |
| "learning_rate": 2.0555555555555555e-06, | |
| "loss": 1.2845, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.49481052377504225, | |
| "grad_norm": 14.699922561645508, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 1.1683, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.49782766111513393, | |
| "grad_norm": 22.841310501098633, | |
| "learning_rate": 1.944444444444445e-06, | |
| "loss": 1.2777, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.5008447984552257, | |
| "grad_norm": 25.15740394592285, | |
| "learning_rate": 1.888888888888889e-06, | |
| "loss": 1.1747, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.5038619357953174, | |
| "grad_norm": 21.897415161132812, | |
| "learning_rate": 1.8333333333333333e-06, | |
| "loss": 1.1542, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.5068790731354091, | |
| "grad_norm": 21.65673065185547, | |
| "learning_rate": 1.777777777777778e-06, | |
| "loss": 1.2236, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.5098962104755008, | |
| "grad_norm": 24.261625289916992, | |
| "learning_rate": 1.7222222222222224e-06, | |
| "loss": 1.2967, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.5129133478155926, | |
| "grad_norm": 20.909135818481445, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 1.1574, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.5159304851556843, | |
| "grad_norm": 23.735315322875977, | |
| "learning_rate": 1.6111111111111113e-06, | |
| "loss": 1.3281, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.518947622495776, | |
| "grad_norm": 22.57413673400879, | |
| "learning_rate": 1.5555555555555558e-06, | |
| "loss": 1.216, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.5219647598358678, | |
| "grad_norm": 14.66985034942627, | |
| "learning_rate": 1.5e-06, | |
| "loss": 1.2625, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.5249818971759594, | |
| "grad_norm": 23.085771560668945, | |
| "learning_rate": 1.4444444444444445e-06, | |
| "loss": 1.1991, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.5279990345160511, | |
| "grad_norm": 23.1724853515625, | |
| "learning_rate": 1.3888888888888892e-06, | |
| "loss": 1.3677, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.5310161718561429, | |
| "grad_norm": 20.612730026245117, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 1.2346, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.5340333091962346, | |
| "grad_norm": 20.71513557434082, | |
| "learning_rate": 1.2777777777777779e-06, | |
| "loss": 1.2159, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.5370504465363264, | |
| "grad_norm": 21.999197006225586, | |
| "learning_rate": 1.2222222222222223e-06, | |
| "loss": 1.1703, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.5400675838764181, | |
| "grad_norm": 25.04129981994629, | |
| "learning_rate": 1.1666666666666668e-06, | |
| "loss": 1.1681, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.5430847212165097, | |
| "grad_norm": 27.193050384521484, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 1.1838, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.5461018585566015, | |
| "grad_norm": 25.8516788482666, | |
| "learning_rate": 1.0555555555555557e-06, | |
| "loss": 1.1382, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.5491189958966932, | |
| "grad_norm": 20.00945281982422, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 1.1503, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.5521361332367849, | |
| "grad_norm": 19.037160873413086, | |
| "learning_rate": 9.444444444444445e-07, | |
| "loss": 1.2207, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.5551532705768767, | |
| "grad_norm": 15.488086700439453, | |
| "learning_rate": 8.88888888888889e-07, | |
| "loss": 1.2648, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.5581704079169684, | |
| "grad_norm": 20.632287979125977, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 1.2193, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.56118754525706, | |
| "grad_norm": 17.698545455932617, | |
| "learning_rate": 7.777777777777779e-07, | |
| "loss": 1.108, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.5642046825971518, | |
| "grad_norm": 19.926164627075195, | |
| "learning_rate": 7.222222222222222e-07, | |
| "loss": 1.2295, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.5672218199372435, | |
| "grad_norm": 22.911317825317383, | |
| "learning_rate": 6.666666666666667e-07, | |
| "loss": 1.178, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.5702389572773353, | |
| "grad_norm": 22.65654754638672, | |
| "learning_rate": 6.111111111111112e-07, | |
| "loss": 1.2775, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.573256094617427, | |
| "grad_norm": 20.710227966308594, | |
| "learning_rate": 5.555555555555555e-07, | |
| "loss": 1.1441, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.5762732319575187, | |
| "grad_norm": 26.962419509887695, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 1.1739, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.5792903692976105, | |
| "grad_norm": 16.164112091064453, | |
| "learning_rate": 4.444444444444445e-07, | |
| "loss": 1.1939, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.5823075066377021, | |
| "grad_norm": 18.468446731567383, | |
| "learning_rate": 3.8888888888888895e-07, | |
| "loss": 1.1682, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.5853246439777938, | |
| "grad_norm": 26.14542007446289, | |
| "learning_rate": 3.3333333333333335e-07, | |
| "loss": 1.1946, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.5883417813178856, | |
| "grad_norm": 23.07315444946289, | |
| "learning_rate": 2.7777777777777776e-07, | |
| "loss": 1.2364, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.5913589186579773, | |
| "grad_norm": 22.729957580566406, | |
| "learning_rate": 2.2222222222222224e-07, | |
| "loss": 1.1943, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.5943760559980691, | |
| "grad_norm": 28.493045806884766, | |
| "learning_rate": 1.6666666666666668e-07, | |
| "loss": 1.2507, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.5973931933381608, | |
| "grad_norm": 18.747501373291016, | |
| "learning_rate": 1.1111111111111112e-07, | |
| "loss": 1.2096, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.6004103306782524, | |
| "grad_norm": 22.211803436279297, | |
| "learning_rate": 5.555555555555556e-08, | |
| "loss": 1.2027, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 0.6034274680183442, | |
| "grad_norm": 21.037738800048828, | |
| "learning_rate": 0.0, | |
| "loss": 1.3156, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.6034274680183442, | |
| "eval_cer": 55.08526363984638, | |
| "eval_loss": 1.199288010597229, | |
| "eval_runtime": 4552.6671, | |
| "eval_samples_per_second": 3.64, | |
| "eval_steps_per_second": 0.455, | |
| "eval_wer": 82.33239678451572, | |
| "step": 5000 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 1000, | |
| "total_flos": 1.15434160128e+19, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |