| { | |
| "best_metric": 32.56401305446938, | |
| "best_model_checkpoint": "./whisper-tiny-hi/checkpoint-15000", | |
| "epoch": 2.5942580421999306, | |
| "eval_steps": 1000, | |
| "global_step": 15000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004323763403666551, | |
| "grad_norm": 16.286670684814453, | |
| "learning_rate": 4.6000000000000004e-07, | |
| "loss": 2.1388, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.008647526807333102, | |
| "grad_norm": 12.285518646240234, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 2.0213, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.012971290210999653, | |
| "grad_norm": 12.98936653137207, | |
| "learning_rate": 1.46e-06, | |
| "loss": 1.852, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.017295053614666205, | |
| "grad_norm": 11.324748992919922, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 1.6847, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.021618817018332757, | |
| "grad_norm": 10.082378387451172, | |
| "learning_rate": 2.46e-06, | |
| "loss": 1.4744, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.025942580421999307, | |
| "grad_norm": 10.503169059753418, | |
| "learning_rate": 2.96e-06, | |
| "loss": 1.3716, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.03026634382566586, | |
| "grad_norm": 10.596274375915527, | |
| "learning_rate": 3.46e-06, | |
| "loss": 1.2115, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.03459010722933241, | |
| "grad_norm": 8.612921714782715, | |
| "learning_rate": 3.96e-06, | |
| "loss": 1.0883, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.03891387063299896, | |
| "grad_norm": 9.459671020507812, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.9957, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.043237634036665515, | |
| "grad_norm": 9.334480285644531, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.8869, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.04756139744033207, | |
| "grad_norm": 8.94874095916748, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.8336, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.051885160843998614, | |
| "grad_norm": 7.839649677276611, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.8094, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.05620892424766517, | |
| "grad_norm": 11.369429588317871, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.76, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.06053268765133172, | |
| "grad_norm": 7.519564628601074, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.7319, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.06485645105499827, | |
| "grad_norm": 10.504302024841309, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.6995, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.06918021445866482, | |
| "grad_norm": 8.440424919128418, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 0.6782, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.07350397786233137, | |
| "grad_norm": 9.647106170654297, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.6475, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.07782774126599792, | |
| "grad_norm": 9.011175155639648, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.6515, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.08215150466966448, | |
| "grad_norm": 7.202056884765625, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.6173, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.08647526807333103, | |
| "grad_norm": 8.170417785644531, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 0.5954, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.09079903147699758, | |
| "grad_norm": 7.681850910186768, | |
| "learning_rate": 9.984137931034483e-06, | |
| "loss": 0.5821, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.09512279488066414, | |
| "grad_norm": 6.318211078643799, | |
| "learning_rate": 9.96689655172414e-06, | |
| "loss": 0.5575, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.09944655828433069, | |
| "grad_norm": 7.533044815063477, | |
| "learning_rate": 9.949655172413793e-06, | |
| "loss": 0.5778, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.10377032168799723, | |
| "grad_norm": 9.38764476776123, | |
| "learning_rate": 9.93241379310345e-06, | |
| "loss": 0.5735, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.10809408509166378, | |
| "grad_norm": 6.581768035888672, | |
| "learning_rate": 9.915172413793104e-06, | |
| "loss": 0.5358, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.11241784849533033, | |
| "grad_norm": 8.789082527160645, | |
| "learning_rate": 9.897931034482759e-06, | |
| "loss": 0.5619, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.11674161189899689, | |
| "grad_norm": 8.615039825439453, | |
| "learning_rate": 9.880689655172414e-06, | |
| "loss": 0.5211, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.12106537530266344, | |
| "grad_norm": 8.99618911743164, | |
| "learning_rate": 9.86344827586207e-06, | |
| "loss": 0.5039, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.12538913870632998, | |
| "grad_norm": 7.325024127960205, | |
| "learning_rate": 9.846206896551725e-06, | |
| "loss": 0.4967, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.12971290210999653, | |
| "grad_norm": 7.267101287841797, | |
| "learning_rate": 9.82896551724138e-06, | |
| "loss": 0.4913, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.13403666551366308, | |
| "grad_norm": 8.017245292663574, | |
| "learning_rate": 9.811724137931035e-06, | |
| "loss": 0.4841, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.13836042891732964, | |
| "grad_norm": 6.565851211547852, | |
| "learning_rate": 9.79448275862069e-06, | |
| "loss": 0.4826, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.1426841923209962, | |
| "grad_norm": 9.583740234375, | |
| "learning_rate": 9.777241379310347e-06, | |
| "loss": 0.4966, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.14700795572466274, | |
| "grad_norm": 6.968817234039307, | |
| "learning_rate": 9.760000000000001e-06, | |
| "loss": 0.4736, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.1513317191283293, | |
| "grad_norm": 6.600325107574463, | |
| "learning_rate": 9.742758620689656e-06, | |
| "loss": 0.5066, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.15565548253199585, | |
| "grad_norm": 7.23114538192749, | |
| "learning_rate": 9.725517241379311e-06, | |
| "loss": 0.4672, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.1599792459356624, | |
| "grad_norm": 9.536112785339355, | |
| "learning_rate": 9.708275862068966e-06, | |
| "loss": 0.4724, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.16430300933932895, | |
| "grad_norm": 8.753514289855957, | |
| "learning_rate": 9.691034482758621e-06, | |
| "loss": 0.458, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.1686267727429955, | |
| "grad_norm": 9.407349586486816, | |
| "learning_rate": 9.673793103448277e-06, | |
| "loss": 0.4598, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.17295053614666206, | |
| "grad_norm": 6.315821647644043, | |
| "learning_rate": 9.65655172413793e-06, | |
| "loss": 0.4332, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.17295053614666206, | |
| "eval_loss": 0.4459797739982605, | |
| "eval_runtime": 5597.1729, | |
| "eval_samples_per_second": 4.132, | |
| "eval_steps_per_second": 0.517, | |
| "eval_wer": 49.49288587647923, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.1772742995503286, | |
| "grad_norm": 6.963139533996582, | |
| "learning_rate": 9.639310344827587e-06, | |
| "loss": 0.4519, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.18159806295399517, | |
| "grad_norm": 6.76003360748291, | |
| "learning_rate": 9.622068965517242e-06, | |
| "loss": 0.4627, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.18592182635766172, | |
| "grad_norm": 6.6258416175842285, | |
| "learning_rate": 9.604827586206897e-06, | |
| "loss": 0.4318, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.19024558976132827, | |
| "grad_norm": 6.607818603515625, | |
| "learning_rate": 9.587586206896554e-06, | |
| "loss": 0.4362, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.19456935316499482, | |
| "grad_norm": 6.816850662231445, | |
| "learning_rate": 9.570344827586208e-06, | |
| "loss": 0.4379, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.19889311656866138, | |
| "grad_norm": 7.295541763305664, | |
| "learning_rate": 9.553103448275863e-06, | |
| "loss": 0.4327, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.2032168799723279, | |
| "grad_norm": 5.751529693603516, | |
| "learning_rate": 9.535862068965518e-06, | |
| "loss": 0.434, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.20754064337599445, | |
| "grad_norm": 7.614989280700684, | |
| "learning_rate": 9.518620689655173e-06, | |
| "loss": 0.4214, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.211864406779661, | |
| "grad_norm": 7.232028961181641, | |
| "learning_rate": 9.501379310344828e-06, | |
| "loss": 0.4246, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.21618817018332756, | |
| "grad_norm": 5.579376697540283, | |
| "learning_rate": 9.484137931034484e-06, | |
| "loss": 0.4118, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.2205119335869941, | |
| "grad_norm": 6.2615580558776855, | |
| "learning_rate": 9.46689655172414e-06, | |
| "loss": 0.4286, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.22483569699066067, | |
| "grad_norm": 6.8341240882873535, | |
| "learning_rate": 9.449655172413794e-06, | |
| "loss": 0.411, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.22915946039432722, | |
| "grad_norm": 5.939593315124512, | |
| "learning_rate": 9.432413793103449e-06, | |
| "loss": 0.4254, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.23348322379799377, | |
| "grad_norm": 6.054545879364014, | |
| "learning_rate": 9.415172413793104e-06, | |
| "loss": 0.4094, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.23780698720166032, | |
| "grad_norm": 6.636862277984619, | |
| "learning_rate": 9.397931034482759e-06, | |
| "loss": 0.4256, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.24213075060532688, | |
| "grad_norm": 6.010420322418213, | |
| "learning_rate": 9.380689655172415e-06, | |
| "loss": 0.4267, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.24645451400899343, | |
| "grad_norm": 8.598302841186523, | |
| "learning_rate": 9.363448275862069e-06, | |
| "loss": 0.4223, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.25077827741265996, | |
| "grad_norm": 6.458983421325684, | |
| "learning_rate": 9.346206896551725e-06, | |
| "loss": 0.4123, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.25510204081632654, | |
| "grad_norm": 6.562103748321533, | |
| "learning_rate": 9.32896551724138e-06, | |
| "loss": 0.4, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.25942580421999306, | |
| "grad_norm": 7.095609188079834, | |
| "learning_rate": 9.311724137931035e-06, | |
| "loss": 0.4121, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.26374956762365964, | |
| "grad_norm": 6.2590789794921875, | |
| "learning_rate": 9.294482758620691e-06, | |
| "loss": 0.395, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.26807333102732617, | |
| "grad_norm": 7.436612606048584, | |
| "learning_rate": 9.277241379310346e-06, | |
| "loss": 0.4271, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.27239709443099275, | |
| "grad_norm": 6.528996467590332, | |
| "learning_rate": 9.260000000000001e-06, | |
| "loss": 0.3954, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.2767208578346593, | |
| "grad_norm": 7.369086265563965, | |
| "learning_rate": 9.242758620689656e-06, | |
| "loss": 0.388, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.28104462123832585, | |
| "grad_norm": 5.275778770446777, | |
| "learning_rate": 9.225517241379311e-06, | |
| "loss": 0.3789, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.2853683846419924, | |
| "grad_norm": 5.707178115844727, | |
| "learning_rate": 9.208275862068966e-06, | |
| "loss": 0.3748, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.28969214804565896, | |
| "grad_norm": 6.311554908752441, | |
| "learning_rate": 9.191034482758622e-06, | |
| "loss": 0.399, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.2940159114493255, | |
| "grad_norm": 5.5442633628845215, | |
| "learning_rate": 9.173793103448277e-06, | |
| "loss": 0.3857, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.29833967485299207, | |
| "grad_norm": 5.790552616119385, | |
| "learning_rate": 9.156551724137932e-06, | |
| "loss": 0.3914, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.3026634382566586, | |
| "grad_norm": 6.97217321395874, | |
| "learning_rate": 9.139310344827587e-06, | |
| "loss": 0.3832, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.30698720166032517, | |
| "grad_norm": 7.6458353996276855, | |
| "learning_rate": 9.122068965517242e-06, | |
| "loss": 0.3863, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.3113109650639917, | |
| "grad_norm": 6.375527381896973, | |
| "learning_rate": 9.104827586206897e-06, | |
| "loss": 0.3717, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.3156347284676583, | |
| "grad_norm": 7.621360778808594, | |
| "learning_rate": 9.087586206896553e-06, | |
| "loss": 0.4053, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.3199584918713248, | |
| "grad_norm": 5.717822551727295, | |
| "learning_rate": 9.070344827586206e-06, | |
| "loss": 0.375, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.3242822552749913, | |
| "grad_norm": 6.005116939544678, | |
| "learning_rate": 9.053103448275863e-06, | |
| "loss": 0.407, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.3286060186786579, | |
| "grad_norm": 6.100674629211426, | |
| "learning_rate": 9.035862068965518e-06, | |
| "loss": 0.383, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.33292978208232443, | |
| "grad_norm": 5.4287495613098145, | |
| "learning_rate": 9.018620689655173e-06, | |
| "loss": 0.3685, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.337253545485991, | |
| "grad_norm": 6.621429920196533, | |
| "learning_rate": 9.00137931034483e-06, | |
| "loss": 0.3858, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.34157730888965754, | |
| "grad_norm": 7.438424110412598, | |
| "learning_rate": 8.984137931034484e-06, | |
| "loss": 0.3874, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.3459010722933241, | |
| "grad_norm": 7.171720027923584, | |
| "learning_rate": 8.966896551724139e-06, | |
| "loss": 0.3796, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.3459010722933241, | |
| "eval_loss": 0.3686262369155884, | |
| "eval_runtime": 5442.7156, | |
| "eval_samples_per_second": 4.249, | |
| "eval_steps_per_second": 0.531, | |
| "eval_wer": 43.32270451494399, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.35022483569699064, | |
| "grad_norm": 5.1768059730529785, | |
| "learning_rate": 8.949655172413794e-06, | |
| "loss": 0.3639, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.3545485991006572, | |
| "grad_norm": 6.131287574768066, | |
| "learning_rate": 8.932413793103449e-06, | |
| "loss": 0.3684, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.35887236250432375, | |
| "grad_norm": 6.503818988800049, | |
| "learning_rate": 8.915172413793104e-06, | |
| "loss": 0.3804, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.36319612590799033, | |
| "grad_norm": 5.307014465332031, | |
| "learning_rate": 8.89793103448276e-06, | |
| "loss": 0.3778, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.36751988931165686, | |
| "grad_norm": 4.741936683654785, | |
| "learning_rate": 8.880689655172415e-06, | |
| "loss": 0.3642, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.37184365271532344, | |
| "grad_norm": 5.130821228027344, | |
| "learning_rate": 8.864137931034484e-06, | |
| "loss": 0.3771, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.37616741611898996, | |
| "grad_norm": 6.177253723144531, | |
| "learning_rate": 8.846896551724138e-06, | |
| "loss": 0.3645, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.38049117952265654, | |
| "grad_norm": 5.380115509033203, | |
| "learning_rate": 8.829655172413793e-06, | |
| "loss": 0.3637, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.38481494292632307, | |
| "grad_norm": 7.195498943328857, | |
| "learning_rate": 8.81241379310345e-06, | |
| "loss": 0.3628, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.38913870632998965, | |
| "grad_norm": 6.12623929977417, | |
| "learning_rate": 8.795172413793103e-06, | |
| "loss": 0.3847, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.3934624697336562, | |
| "grad_norm": 6.666962623596191, | |
| "learning_rate": 8.77793103448276e-06, | |
| "loss": 0.3776, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.39778623313732275, | |
| "grad_norm": 6.3228325843811035, | |
| "learning_rate": 8.760689655172415e-06, | |
| "loss": 0.3603, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.4021099965409893, | |
| "grad_norm": 5.662914276123047, | |
| "learning_rate": 8.74344827586207e-06, | |
| "loss": 0.341, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.4064337599446558, | |
| "grad_norm": 6.619643211364746, | |
| "learning_rate": 8.726206896551724e-06, | |
| "loss": 0.3513, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.4107575233483224, | |
| "grad_norm": 5.331205368041992, | |
| "learning_rate": 8.70896551724138e-06, | |
| "loss": 0.3531, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.4150812867519889, | |
| "grad_norm": 6.8525824546813965, | |
| "learning_rate": 8.691724137931034e-06, | |
| "loss": 0.358, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.4194050501556555, | |
| "grad_norm": 6.635438442230225, | |
| "learning_rate": 8.67448275862069e-06, | |
| "loss": 0.3756, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.423728813559322, | |
| "grad_norm": 5.737441062927246, | |
| "learning_rate": 8.657241379310345e-06, | |
| "loss": 0.3502, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.4280525769629886, | |
| "grad_norm": 6.25765323638916, | |
| "learning_rate": 8.64e-06, | |
| "loss": 0.3632, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.4323763403666551, | |
| "grad_norm": 6.037199020385742, | |
| "learning_rate": 8.622758620689657e-06, | |
| "loss": 0.3622, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.4367001037703217, | |
| "grad_norm": 6.072882175445557, | |
| "learning_rate": 8.605517241379312e-06, | |
| "loss": 0.3559, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.4410238671739882, | |
| "grad_norm": 6.014054298400879, | |
| "learning_rate": 8.588275862068967e-06, | |
| "loss": 0.376, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.4453476305776548, | |
| "grad_norm": 6.040060043334961, | |
| "learning_rate": 8.571034482758621e-06, | |
| "loss": 0.3749, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.44967139398132133, | |
| "grad_norm": 5.293423652648926, | |
| "learning_rate": 8.553793103448276e-06, | |
| "loss": 0.3617, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.4539951573849879, | |
| "grad_norm": 6.393920421600342, | |
| "learning_rate": 8.536551724137931e-06, | |
| "loss": 0.3506, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.45831892078865444, | |
| "grad_norm": 5.98794412612915, | |
| "learning_rate": 8.519310344827588e-06, | |
| "loss": 0.3523, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.462642684192321, | |
| "grad_norm": 5.70527458190918, | |
| "learning_rate": 8.502068965517241e-06, | |
| "loss": 0.3622, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.46696644759598754, | |
| "grad_norm": 6.078751087188721, | |
| "learning_rate": 8.484827586206898e-06, | |
| "loss": 0.3403, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.4712902109996541, | |
| "grad_norm": 6.641058444976807, | |
| "learning_rate": 8.467586206896552e-06, | |
| "loss": 0.3506, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.47561397440332065, | |
| "grad_norm": 6.066605091094971, | |
| "learning_rate": 8.450344827586207e-06, | |
| "loss": 0.3773, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.4799377378069872, | |
| "grad_norm": 6.622874736785889, | |
| "learning_rate": 8.433103448275862e-06, | |
| "loss": 0.3435, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.48426150121065376, | |
| "grad_norm": 5.157226085662842, | |
| "learning_rate": 8.415862068965519e-06, | |
| "loss": 0.3586, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.4885852646143203, | |
| "grad_norm": 5.92050838470459, | |
| "learning_rate": 8.398620689655174e-06, | |
| "loss": 0.352, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.49290902801798686, | |
| "grad_norm": 5.522551536560059, | |
| "learning_rate": 8.381379310344828e-06, | |
| "loss": 0.3447, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.4972327914216534, | |
| "grad_norm": 6.284322738647461, | |
| "learning_rate": 8.364137931034483e-06, | |
| "loss": 0.3477, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.5015565548253199, | |
| "grad_norm": 6.7316179275512695, | |
| "learning_rate": 8.346896551724138e-06, | |
| "loss": 0.3402, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.5058803182289865, | |
| "grad_norm": 5.808598518371582, | |
| "learning_rate": 8.329655172413795e-06, | |
| "loss": 0.3332, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.5102040816326531, | |
| "grad_norm": 5.109462261199951, | |
| "learning_rate": 8.31241379310345e-06, | |
| "loss": 0.3465, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.5145278450363197, | |
| "grad_norm": 6.471071720123291, | |
| "learning_rate": 8.295172413793104e-06, | |
| "loss": 0.3501, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.5188516084399861, | |
| "grad_norm": 7.2887139320373535, | |
| "learning_rate": 8.27793103448276e-06, | |
| "loss": 0.3414, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.5188516084399861, | |
| "eval_loss": 0.33114978671073914, | |
| "eval_runtime": 5410.7097, | |
| "eval_samples_per_second": 4.274, | |
| "eval_steps_per_second": 0.534, | |
| "eval_wer": 39.80441724633165, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.5231753718436527, | |
| "grad_norm": 5.811481475830078, | |
| "learning_rate": 8.260689655172414e-06, | |
| "loss": 0.3286, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.5274991352473193, | |
| "grad_norm": 5.4608869552612305, | |
| "learning_rate": 8.243448275862069e-06, | |
| "loss": 0.3255, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.5318228986509859, | |
| "grad_norm": 6.3298797607421875, | |
| "learning_rate": 8.226206896551726e-06, | |
| "loss": 0.3299, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.5361466620546523, | |
| "grad_norm": 5.198216915130615, | |
| "learning_rate": 8.208965517241379e-06, | |
| "loss": 0.3463, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.5404704254583189, | |
| "grad_norm": 4.894079685211182, | |
| "learning_rate": 8.191724137931035e-06, | |
| "loss": 0.3362, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.5447941888619855, | |
| "grad_norm": 5.518843650817871, | |
| "learning_rate": 8.17448275862069e-06, | |
| "loss": 0.3592, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.5491179522656521, | |
| "grad_norm": 5.926825046539307, | |
| "learning_rate": 8.157241379310345e-06, | |
| "loss": 0.3402, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.5534417156693185, | |
| "grad_norm": 5.385416507720947, | |
| "learning_rate": 8.14e-06, | |
| "loss": 0.3308, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.5577654790729851, | |
| "grad_norm": 4.563920497894287, | |
| "learning_rate": 8.122758620689657e-06, | |
| "loss": 0.3512, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.5620892424766517, | |
| "grad_norm": 5.665447235107422, | |
| "learning_rate": 8.105517241379311e-06, | |
| "loss": 0.3256, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.5664130058803182, | |
| "grad_norm": 6.340756416320801, | |
| "learning_rate": 8.088275862068966e-06, | |
| "loss": 0.3264, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.5707367692839848, | |
| "grad_norm": 5.052750587463379, | |
| "learning_rate": 8.071034482758621e-06, | |
| "loss": 0.3144, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.5750605326876513, | |
| "grad_norm": 6.486083984375, | |
| "learning_rate": 8.053793103448276e-06, | |
| "loss": 0.3306, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.5793842960913179, | |
| "grad_norm": 6.098331451416016, | |
| "learning_rate": 8.036551724137933e-06, | |
| "loss": 0.3363, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.5837080594949844, | |
| "grad_norm": 5.329631328582764, | |
| "learning_rate": 8.019310344827587e-06, | |
| "loss": 0.3245, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.588031822898651, | |
| "grad_norm": 5.355101108551025, | |
| "learning_rate": 8.002068965517242e-06, | |
| "loss": 0.3213, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.5923555863023175, | |
| "grad_norm": 6.330533504486084, | |
| "learning_rate": 7.984827586206897e-06, | |
| "loss": 0.338, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.5966793497059841, | |
| "grad_norm": 5.799704551696777, | |
| "learning_rate": 7.967586206896552e-06, | |
| "loss": 0.3348, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.6010031131096506, | |
| "grad_norm": 6.250985145568848, | |
| "learning_rate": 7.950344827586207e-06, | |
| "loss": 0.3275, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.6053268765133172, | |
| "grad_norm": 5.608683109283447, | |
| "learning_rate": 7.933103448275864e-06, | |
| "loss": 0.3264, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.6096506399169838, | |
| "grad_norm": 4.862841606140137, | |
| "learning_rate": 7.915862068965517e-06, | |
| "loss": 0.3352, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.6139744033206503, | |
| "grad_norm": 6.654427528381348, | |
| "learning_rate": 7.898620689655173e-06, | |
| "loss": 0.337, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.6182981667243168, | |
| "grad_norm": 4.726236820220947, | |
| "learning_rate": 7.881379310344828e-06, | |
| "loss": 0.34, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.6226219301279834, | |
| "grad_norm": 5.243809223175049, | |
| "learning_rate": 7.864137931034483e-06, | |
| "loss": 0.3181, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.62694569353165, | |
| "grad_norm": 5.479759216308594, | |
| "learning_rate": 7.84689655172414e-06, | |
| "loss": 0.3297, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.6312694569353166, | |
| "grad_norm": 5.421146392822266, | |
| "learning_rate": 7.829655172413794e-06, | |
| "loss": 0.3221, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.635593220338983, | |
| "grad_norm": 5.118608474731445, | |
| "learning_rate": 7.81241379310345e-06, | |
| "loss": 0.3081, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.6399169837426496, | |
| "grad_norm": 5.862517356872559, | |
| "learning_rate": 7.795172413793104e-06, | |
| "loss": 0.3128, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.6442407471463162, | |
| "grad_norm": 5.084294319152832, | |
| "learning_rate": 7.777931034482759e-06, | |
| "loss": 0.3245, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.6485645105499827, | |
| "grad_norm": 5.693409442901611, | |
| "learning_rate": 7.760689655172414e-06, | |
| "loss": 0.3167, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.6528882739536492, | |
| "grad_norm": 4.5151286125183105, | |
| "learning_rate": 7.74344827586207e-06, | |
| "loss": 0.3278, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.6572120373573158, | |
| "grad_norm": 6.0459442138671875, | |
| "learning_rate": 7.726206896551725e-06, | |
| "loss": 0.3403, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.6615358007609824, | |
| "grad_norm": 5.351093292236328, | |
| "learning_rate": 7.70896551724138e-06, | |
| "loss": 0.311, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.6658595641646489, | |
| "grad_norm": 5.750779151916504, | |
| "learning_rate": 7.691724137931035e-06, | |
| "loss": 0.3112, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.6701833275683154, | |
| "grad_norm": 5.5272216796875, | |
| "learning_rate": 7.67448275862069e-06, | |
| "loss": 0.32, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.674507090971982, | |
| "grad_norm": 5.890946865081787, | |
| "learning_rate": 7.657241379310345e-06, | |
| "loss": 0.3434, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.6788308543756486, | |
| "grad_norm": 6.403041839599609, | |
| "learning_rate": 7.640000000000001e-06, | |
| "loss": 0.3194, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.6831546177793151, | |
| "grad_norm": 4.705821990966797, | |
| "learning_rate": 7.622758620689655e-06, | |
| "loss": 0.3038, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.6874783811829817, | |
| "grad_norm": 4.945014953613281, | |
| "learning_rate": 7.605517241379311e-06, | |
| "loss": 0.3414, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.6918021445866482, | |
| "grad_norm": 6.610292434692383, | |
| "learning_rate": 7.588275862068966e-06, | |
| "loss": 0.3374, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.6918021445866482, | |
| "eval_loss": 0.312212198972702, | |
| "eval_runtime": 5649.9248, | |
| "eval_samples_per_second": 4.093, | |
| "eval_steps_per_second": 0.512, | |
| "eval_wer": 38.40712982074705, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.6961259079903148, | |
| "grad_norm": 5.513612270355225, | |
| "learning_rate": 7.571034482758622e-06, | |
| "loss": 0.3368, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.7004496713939813, | |
| "grad_norm": 4.718435764312744, | |
| "learning_rate": 7.553793103448277e-06, | |
| "loss": 0.291, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.7047734347976479, | |
| "grad_norm": 5.8622965812683105, | |
| "learning_rate": 7.5365517241379315e-06, | |
| "loss": 0.2975, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.7090971982013144, | |
| "grad_norm": 4.535427570343018, | |
| "learning_rate": 7.519310344827587e-06, | |
| "loss": 0.2925, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.713420961604981, | |
| "grad_norm": 5.352747440338135, | |
| "learning_rate": 7.502068965517242e-06, | |
| "loss": 0.3202, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.7177447250086475, | |
| "grad_norm": 4.6951446533203125, | |
| "learning_rate": 7.484827586206898e-06, | |
| "loss": 0.3239, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.7220684884123141, | |
| "grad_norm": 5.972590923309326, | |
| "learning_rate": 7.467586206896552e-06, | |
| "loss": 0.3203, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.7263922518159807, | |
| "grad_norm": 5.971949577331543, | |
| "learning_rate": 7.4503448275862075e-06, | |
| "loss": 0.3207, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.7307160152196471, | |
| "grad_norm": 5.5973334312438965, | |
| "learning_rate": 7.433103448275862e-06, | |
| "loss": 0.3248, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.7350397786233137, | |
| "grad_norm": 5.686942100524902, | |
| "learning_rate": 7.415862068965518e-06, | |
| "loss": 0.3349, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.7393635420269803, | |
| "grad_norm": 4.615256309509277, | |
| "learning_rate": 7.398620689655173e-06, | |
| "loss": 0.3259, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.7436873054306469, | |
| "grad_norm": 7.120574951171875, | |
| "learning_rate": 7.381379310344829e-06, | |
| "loss": 0.3444, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.7480110688343133, | |
| "grad_norm": 5.114111423492432, | |
| "learning_rate": 7.364137931034483e-06, | |
| "loss": 0.315, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.7523348322379799, | |
| "grad_norm": 5.820385456085205, | |
| "learning_rate": 7.346896551724138e-06, | |
| "loss": 0.3149, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.7566585956416465, | |
| "grad_norm": 5.231871128082275, | |
| "learning_rate": 7.330344827586208e-06, | |
| "loss": 0.3251, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.7609823590453131, | |
| "grad_norm": 5.228817462921143, | |
| "learning_rate": 7.313103448275863e-06, | |
| "loss": 0.3162, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.7653061224489796, | |
| "grad_norm": 4.916074752807617, | |
| "learning_rate": 7.2958620689655175e-06, | |
| "loss": 0.304, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.7696298858526461, | |
| "grad_norm": 5.931058883666992, | |
| "learning_rate": 7.278620689655172e-06, | |
| "loss": 0.3304, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.7739536492563127, | |
| "grad_norm": 5.873823165893555, | |
| "learning_rate": 7.261379310344828e-06, | |
| "loss": 0.3398, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.7782774126599793, | |
| "grad_norm": 5.279876708984375, | |
| "learning_rate": 7.244137931034483e-06, | |
| "loss": 0.3064, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.7826011760636458, | |
| "grad_norm": 4.812004089355469, | |
| "learning_rate": 7.226896551724139e-06, | |
| "loss": 0.3063, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.7869249394673123, | |
| "grad_norm": 6.433504581451416, | |
| "learning_rate": 7.2096551724137944e-06, | |
| "loss": 0.3224, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.7912487028709789, | |
| "grad_norm": 5.206650257110596, | |
| "learning_rate": 7.1924137931034485e-06, | |
| "loss": 0.3247, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.7955724662746455, | |
| "grad_norm": 5.398522853851318, | |
| "learning_rate": 7.175172413793104e-06, | |
| "loss": 0.3017, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.799896229678312, | |
| "grad_norm": 4.88048791885376, | |
| "learning_rate": 7.157931034482759e-06, | |
| "loss": 0.3218, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.8042199930819786, | |
| "grad_norm": 4.6434407234191895, | |
| "learning_rate": 7.140689655172415e-06, | |
| "loss": 0.3134, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.8085437564856451, | |
| "grad_norm": 5.189647674560547, | |
| "learning_rate": 7.12344827586207e-06, | |
| "loss": 0.3068, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.8128675198893116, | |
| "grad_norm": 5.023260116577148, | |
| "learning_rate": 7.106206896551725e-06, | |
| "loss": 0.2977, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.8171912832929782, | |
| "grad_norm": 5.623253345489502, | |
| "learning_rate": 7.088965517241379e-06, | |
| "loss": 0.3083, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.8215150466966448, | |
| "grad_norm": 5.7590789794921875, | |
| "learning_rate": 7.071724137931035e-06, | |
| "loss": 0.3113, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.8258388101003113, | |
| "grad_norm": 5.757692813873291, | |
| "learning_rate": 7.05448275862069e-06, | |
| "loss": 0.3204, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.8301625735039778, | |
| "grad_norm": 4.696317195892334, | |
| "learning_rate": 7.037241379310346e-06, | |
| "loss": 0.2966, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.8344863369076444, | |
| "grad_norm": 5.004659175872803, | |
| "learning_rate": 7.0200000000000006e-06, | |
| "loss": 0.3017, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.838810100311311, | |
| "grad_norm": 5.880341529846191, | |
| "learning_rate": 7.002758620689655e-06, | |
| "loss": 0.3177, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.8431338637149776, | |
| "grad_norm": 5.527527809143066, | |
| "learning_rate": 6.98551724137931e-06, | |
| "loss": 0.3056, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.847457627118644, | |
| "grad_norm": 5.48380708694458, | |
| "learning_rate": 6.968275862068966e-06, | |
| "loss": 0.2952, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.8517813905223106, | |
| "grad_norm": 3.9720652103424072, | |
| "learning_rate": 6.951034482758622e-06, | |
| "loss": 0.2961, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.8561051539259772, | |
| "grad_norm": 5.183377265930176, | |
| "learning_rate": 6.933793103448277e-06, | |
| "loss": 0.3051, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.8604289173296438, | |
| "grad_norm": 5.809779644012451, | |
| "learning_rate": 6.916551724137932e-06, | |
| "loss": 0.3191, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 0.8647526807333102, | |
| "grad_norm": 5.579798698425293, | |
| "learning_rate": 6.899310344827586e-06, | |
| "loss": 0.3185, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.8647526807333102, | |
| "eval_loss": 0.29861804842948914, | |
| "eval_runtime": 5403.1228, | |
| "eval_samples_per_second": 4.28, | |
| "eval_steps_per_second": 0.535, | |
| "eval_wer": 37.68833742945034, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.8690764441369768, | |
| "grad_norm": 3.995452880859375, | |
| "learning_rate": 6.882068965517242e-06, | |
| "loss": 0.3095, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 0.8734002075406434, | |
| "grad_norm": 5.617555618286133, | |
| "learning_rate": 6.864827586206897e-06, | |
| "loss": 0.3139, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.8777239709443099, | |
| "grad_norm": 5.809163570404053, | |
| "learning_rate": 6.847586206896553e-06, | |
| "loss": 0.3209, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 0.8820477343479765, | |
| "grad_norm": 6.078767776489258, | |
| "learning_rate": 6.8303448275862075e-06, | |
| "loss": 0.3019, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.886371497751643, | |
| "grad_norm": 5.000216007232666, | |
| "learning_rate": 6.813103448275863e-06, | |
| "loss": 0.3195, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 0.8906952611553096, | |
| "grad_norm": 4.086789131164551, | |
| "learning_rate": 6.795862068965517e-06, | |
| "loss": 0.2903, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.8950190245589761, | |
| "grad_norm": 6.727631568908691, | |
| "learning_rate": 6.778620689655173e-06, | |
| "loss": 0.3091, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 0.8993427879626427, | |
| "grad_norm": 5.265336513519287, | |
| "learning_rate": 6.761379310344828e-06, | |
| "loss": 0.309, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.9036665513663092, | |
| "grad_norm": 4.668389320373535, | |
| "learning_rate": 6.7441379310344836e-06, | |
| "loss": 0.2931, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 0.9079903147699758, | |
| "grad_norm": 5.374257564544678, | |
| "learning_rate": 6.7268965517241384e-06, | |
| "loss": 0.2978, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.9123140781736423, | |
| "grad_norm": 4.311838150024414, | |
| "learning_rate": 6.709655172413793e-06, | |
| "loss": 0.2889, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 0.9166378415773089, | |
| "grad_norm": 4.521867275238037, | |
| "learning_rate": 6.692413793103448e-06, | |
| "loss": 0.3059, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.9209616049809755, | |
| "grad_norm": 5.439430236816406, | |
| "learning_rate": 6.675172413793104e-06, | |
| "loss": 0.3191, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 0.925285368384642, | |
| "grad_norm": 6.107820987701416, | |
| "learning_rate": 6.65793103448276e-06, | |
| "loss": 0.3078, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.9296091317883085, | |
| "grad_norm": 5.370652675628662, | |
| "learning_rate": 6.6406896551724145e-06, | |
| "loss": 0.3103, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 0.9339328951919751, | |
| "grad_norm": 6.220107555389404, | |
| "learning_rate": 6.62344827586207e-06, | |
| "loss": 0.3074, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.9382566585956417, | |
| "grad_norm": 6.766858100891113, | |
| "learning_rate": 6.606206896551724e-06, | |
| "loss": 0.2855, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 0.9425804219993082, | |
| "grad_norm": 5.859302043914795, | |
| "learning_rate": 6.58896551724138e-06, | |
| "loss": 0.311, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 0.9469041854029747, | |
| "grad_norm": 5.242767810821533, | |
| "learning_rate": 6.571724137931035e-06, | |
| "loss": 0.3057, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 0.9512279488066413, | |
| "grad_norm": 5.731853008270264, | |
| "learning_rate": 6.5544827586206905e-06, | |
| "loss": 0.3083, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.9555517122103079, | |
| "grad_norm": 5.8285040855407715, | |
| "learning_rate": 6.537241379310345e-06, | |
| "loss": 0.2989, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 0.9598754756139743, | |
| "grad_norm": 6.33150053024292, | |
| "learning_rate": 6.520000000000001e-06, | |
| "loss": 0.2985, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 0.9641992390176409, | |
| "grad_norm": 4.892938613891602, | |
| "learning_rate": 6.502758620689655e-06, | |
| "loss": 0.3094, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 0.9685230024213075, | |
| "grad_norm": 4.8297648429870605, | |
| "learning_rate": 6.485517241379311e-06, | |
| "loss": 0.3145, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.9728467658249741, | |
| "grad_norm": 5.7645111083984375, | |
| "learning_rate": 6.468275862068966e-06, | |
| "loss": 0.3013, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 0.9771705292286406, | |
| "grad_norm": 4.90895414352417, | |
| "learning_rate": 6.4510344827586214e-06, | |
| "loss": 0.3054, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 0.9814942926323071, | |
| "grad_norm": 5.324610710144043, | |
| "learning_rate": 6.433793103448276e-06, | |
| "loss": 0.3134, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 0.9858180560359737, | |
| "grad_norm": 4.568467140197754, | |
| "learning_rate": 6.416551724137931e-06, | |
| "loss": 0.2974, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.9901418194396403, | |
| "grad_norm": 5.177453994750977, | |
| "learning_rate": 6.399310344827587e-06, | |
| "loss": 0.3156, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 0.9944655828433068, | |
| "grad_norm": 6.221437454223633, | |
| "learning_rate": 6.382068965517242e-06, | |
| "loss": 0.3109, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.9987893462469734, | |
| "grad_norm": 5.171840667724609, | |
| "learning_rate": 6.3648275862068975e-06, | |
| "loss": 0.315, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 1.0031131096506398, | |
| "grad_norm": 5.1609907150268555, | |
| "learning_rate": 6.347586206896552e-06, | |
| "loss": 0.2943, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.0074368730543064, | |
| "grad_norm": 5.415836811065674, | |
| "learning_rate": 6.330344827586208e-06, | |
| "loss": 0.2833, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 1.011760636457973, | |
| "grad_norm": 5.87399435043335, | |
| "learning_rate": 6.313103448275862e-06, | |
| "loss": 0.2806, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 1.0160843998616396, | |
| "grad_norm": 6.694667816162109, | |
| "learning_rate": 6.295862068965518e-06, | |
| "loss": 0.2844, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 1.0204081632653061, | |
| "grad_norm": 6.607420444488525, | |
| "learning_rate": 6.278620689655173e-06, | |
| "loss": 0.3028, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.0247319266689727, | |
| "grad_norm": 5.000158309936523, | |
| "learning_rate": 6.261379310344828e-06, | |
| "loss": 0.287, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 1.0290556900726393, | |
| "grad_norm": 4.624439239501953, | |
| "learning_rate": 6.244137931034483e-06, | |
| "loss": 0.2848, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 1.0333794534763059, | |
| "grad_norm": 4.626707553863525, | |
| "learning_rate": 6.226896551724139e-06, | |
| "loss": 0.2767, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 1.0377032168799722, | |
| "grad_norm": 6.180722236633301, | |
| "learning_rate": 6.209655172413793e-06, | |
| "loss": 0.2898, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.0377032168799722, | |
| "eval_loss": 0.28753867745399475, | |
| "eval_runtime": 5032.2441, | |
| "eval_samples_per_second": 4.596, | |
| "eval_steps_per_second": 0.574, | |
| "eval_wer": 35.51532588144612, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.0420269802836388, | |
| "grad_norm": 5.8476176261901855, | |
| "learning_rate": 6.192413793103449e-06, | |
| "loss": 0.2917, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 1.0463507436873054, | |
| "grad_norm": 4.507814407348633, | |
| "learning_rate": 6.175172413793104e-06, | |
| "loss": 0.2886, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 1.050674507090972, | |
| "grad_norm": 5.021352767944336, | |
| "learning_rate": 6.157931034482759e-06, | |
| "loss": 0.2953, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 1.0549982704946386, | |
| "grad_norm": 4.907651901245117, | |
| "learning_rate": 6.140689655172414e-06, | |
| "loss": 0.28, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 1.0593220338983051, | |
| "grad_norm": 4.90705680847168, | |
| "learning_rate": 6.123448275862069e-06, | |
| "loss": 0.2906, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 1.0636457973019717, | |
| "grad_norm": 5.755337715148926, | |
| "learning_rate": 6.106206896551725e-06, | |
| "loss": 0.2886, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 1.067969560705638, | |
| "grad_norm": 3.9755401611328125, | |
| "learning_rate": 6.08896551724138e-06, | |
| "loss": 0.283, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 1.0722933241093047, | |
| "grad_norm": 4.78421688079834, | |
| "learning_rate": 6.071724137931035e-06, | |
| "loss": 0.3046, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 1.0766170875129712, | |
| "grad_norm": 5.140844345092773, | |
| "learning_rate": 6.05448275862069e-06, | |
| "loss": 0.2558, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 1.0809408509166378, | |
| "grad_norm": 4.85251522064209, | |
| "learning_rate": 6.037241379310346e-06, | |
| "loss": 0.2634, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 1.0852646143203044, | |
| "grad_norm": 5.412033557891846, | |
| "learning_rate": 6.02e-06, | |
| "loss": 0.267, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 1.089588377723971, | |
| "grad_norm": 4.728926658630371, | |
| "learning_rate": 6.002758620689656e-06, | |
| "loss": 0.2644, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.0939121411276376, | |
| "grad_norm": 4.522409915924072, | |
| "learning_rate": 5.9855172413793105e-06, | |
| "loss": 0.2853, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 1.098235904531304, | |
| "grad_norm": 4.903046607971191, | |
| "learning_rate": 5.968275862068966e-06, | |
| "loss": 0.2836, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 1.1025596679349705, | |
| "grad_norm": 5.107143878936768, | |
| "learning_rate": 5.951724137931036e-06, | |
| "loss": 0.2775, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 1.106883431338637, | |
| "grad_norm": 4.579159259796143, | |
| "learning_rate": 5.93448275862069e-06, | |
| "loss": 0.2855, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 1.1112071947423037, | |
| "grad_norm": 5.995217323303223, | |
| "learning_rate": 5.917241379310345e-06, | |
| "loss": 0.2759, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 1.1155309581459703, | |
| "grad_norm": 4.0890350341796875, | |
| "learning_rate": 5.9e-06, | |
| "loss": 0.2806, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 1.1198547215496368, | |
| "grad_norm": 4.915022850036621, | |
| "learning_rate": 5.882758620689656e-06, | |
| "loss": 0.2787, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 1.1241784849533034, | |
| "grad_norm": 5.1101579666137695, | |
| "learning_rate": 5.865517241379311e-06, | |
| "loss": 0.2725, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.12850224835697, | |
| "grad_norm": 5.38516902923584, | |
| "learning_rate": 5.848275862068966e-06, | |
| "loss": 0.2921, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 1.1328260117606366, | |
| "grad_norm": 5.741593360900879, | |
| "learning_rate": 5.831034482758621e-06, | |
| "loss": 0.2747, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 1.137149775164303, | |
| "grad_norm": 4.1872053146362305, | |
| "learning_rate": 5.813793103448276e-06, | |
| "loss": 0.2992, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 1.1414735385679695, | |
| "grad_norm": 5.309129238128662, | |
| "learning_rate": 5.796551724137931e-06, | |
| "loss": 0.2776, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.145797301971636, | |
| "grad_norm": 4.281647682189941, | |
| "learning_rate": 5.779310344827587e-06, | |
| "loss": 0.2961, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 1.1501210653753027, | |
| "grad_norm": 5.709052085876465, | |
| "learning_rate": 5.762068965517243e-06, | |
| "loss": 0.2934, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 1.1544448287789693, | |
| "grad_norm": 5.323013782501221, | |
| "learning_rate": 5.744827586206897e-06, | |
| "loss": 0.2814, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 1.1587685921826358, | |
| "grad_norm": 4.633469104766846, | |
| "learning_rate": 5.727586206896552e-06, | |
| "loss": 0.2805, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 1.1630923555863024, | |
| "grad_norm": 5.769145488739014, | |
| "learning_rate": 5.710344827586207e-06, | |
| "loss": 0.2909, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 1.1674161189899688, | |
| "grad_norm": 4.406848907470703, | |
| "learning_rate": 5.693103448275863e-06, | |
| "loss": 0.2717, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 1.1717398823936354, | |
| "grad_norm": 5.0518269538879395, | |
| "learning_rate": 5.675862068965518e-06, | |
| "loss": 0.2822, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 1.176063645797302, | |
| "grad_norm": 5.016385078430176, | |
| "learning_rate": 5.6586206896551735e-06, | |
| "loss": 0.2574, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 1.1803874092009685, | |
| "grad_norm": 6.169381618499756, | |
| "learning_rate": 5.6413793103448275e-06, | |
| "loss": 0.2934, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 1.184711172604635, | |
| "grad_norm": 4.3018879890441895, | |
| "learning_rate": 5.624137931034483e-06, | |
| "loss": 0.2707, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 1.1890349360083017, | |
| "grad_norm": 5.160860538482666, | |
| "learning_rate": 5.606896551724138e-06, | |
| "loss": 0.2674, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 1.1933586994119683, | |
| "grad_norm": 4.624250888824463, | |
| "learning_rate": 5.589655172413794e-06, | |
| "loss": 0.2896, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 1.1976824628156346, | |
| "grad_norm": 4.681013584136963, | |
| "learning_rate": 5.572413793103449e-06, | |
| "loss": 0.2827, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 1.2020062262193012, | |
| "grad_norm": 5.956193447113037, | |
| "learning_rate": 5.555172413793104e-06, | |
| "loss": 0.2655, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 1.2063299896229678, | |
| "grad_norm": 5.1831278800964355, | |
| "learning_rate": 5.5379310344827585e-06, | |
| "loss": 0.2847, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 1.2106537530266344, | |
| "grad_norm": 5.0179595947265625, | |
| "learning_rate": 5.520689655172414e-06, | |
| "loss": 0.2787, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.2106537530266344, | |
| "eval_loss": 0.28157490491867065, | |
| "eval_runtime": 5064.7694, | |
| "eval_samples_per_second": 4.566, | |
| "eval_steps_per_second": 0.571, | |
| "eval_wer": 35.47549709835601, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.214977516430301, | |
| "grad_norm": 3.484829902648926, | |
| "learning_rate": 5.503448275862069e-06, | |
| "loss": 0.2671, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 1.2193012798339675, | |
| "grad_norm": 5.8819169998168945, | |
| "learning_rate": 5.486206896551725e-06, | |
| "loss": 0.2879, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 1.223625043237634, | |
| "grad_norm": 4.751934051513672, | |
| "learning_rate": 5.4689655172413805e-06, | |
| "loss": 0.2732, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 1.2279488066413007, | |
| "grad_norm": 6.112524032592773, | |
| "learning_rate": 5.4517241379310345e-06, | |
| "loss": 0.281, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 1.2322725700449673, | |
| "grad_norm": 5.031464099884033, | |
| "learning_rate": 5.43448275862069e-06, | |
| "loss": 0.2796, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 1.2365963334486336, | |
| "grad_norm": 4.925386905670166, | |
| "learning_rate": 5.417241379310345e-06, | |
| "loss": 0.2772, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 1.2409200968523002, | |
| "grad_norm": 4.831676483154297, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.2754, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 1.2452438602559668, | |
| "grad_norm": 5.259826183319092, | |
| "learning_rate": 5.382758620689656e-06, | |
| "loss": 0.2807, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.2495676236596334, | |
| "grad_norm": 5.998917102813721, | |
| "learning_rate": 5.365517241379311e-06, | |
| "loss": 0.264, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 1.2538913870633, | |
| "grad_norm": 4.769485950469971, | |
| "learning_rate": 5.3482758620689654e-06, | |
| "loss": 0.2742, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 1.2582151504669665, | |
| "grad_norm": 4.706541538238525, | |
| "learning_rate": 5.331034482758621e-06, | |
| "loss": 0.2644, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 1.262538913870633, | |
| "grad_norm": 4.766952991485596, | |
| "learning_rate": 5.313793103448276e-06, | |
| "loss": 0.2863, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 1.2668626772742995, | |
| "grad_norm": 4.333870887756348, | |
| "learning_rate": 5.296551724137932e-06, | |
| "loss": 0.275, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 1.271186440677966, | |
| "grad_norm": 5.625321388244629, | |
| "learning_rate": 5.279310344827587e-06, | |
| "loss": 0.2775, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 1.2755102040816326, | |
| "grad_norm": 4.249492645263672, | |
| "learning_rate": 5.2620689655172415e-06, | |
| "loss": 0.259, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 1.2798339674852992, | |
| "grad_norm": 6.244980812072754, | |
| "learning_rate": 5.244827586206896e-06, | |
| "loss": 0.2656, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 1.2841577308889658, | |
| "grad_norm": 4.6464056968688965, | |
| "learning_rate": 5.227586206896552e-06, | |
| "loss": 0.2644, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 1.2884814942926324, | |
| "grad_norm": 4.944653511047363, | |
| "learning_rate": 5.210344827586208e-06, | |
| "loss": 0.282, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 1.292805257696299, | |
| "grad_norm": 6.058406352996826, | |
| "learning_rate": 5.193103448275863e-06, | |
| "loss": 0.2717, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 1.2971290210999653, | |
| "grad_norm": 5.668537139892578, | |
| "learning_rate": 5.175862068965518e-06, | |
| "loss": 0.2731, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.3014527845036319, | |
| "grad_norm": 4.982054710388184, | |
| "learning_rate": 5.158620689655172e-06, | |
| "loss": 0.2579, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 1.3057765479072985, | |
| "grad_norm": 4.244757175445557, | |
| "learning_rate": 5.141379310344828e-06, | |
| "loss": 0.2684, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 1.310100311310965, | |
| "grad_norm": 4.842339515686035, | |
| "learning_rate": 5.124137931034483e-06, | |
| "loss": 0.2795, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 1.3144240747146316, | |
| "grad_norm": 4.516208171844482, | |
| "learning_rate": 5.106896551724139e-06, | |
| "loss": 0.2724, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 1.3187478381182982, | |
| "grad_norm": 4.063875675201416, | |
| "learning_rate": 5.0896551724137936e-06, | |
| "loss": 0.2816, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 1.3230716015219648, | |
| "grad_norm": 4.514278888702393, | |
| "learning_rate": 5.072413793103449e-06, | |
| "loss": 0.2664, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 1.3273953649256311, | |
| "grad_norm": 5.441287994384766, | |
| "learning_rate": 5.055172413793103e-06, | |
| "loss": 0.2656, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 1.331719128329298, | |
| "grad_norm": 4.973991870880127, | |
| "learning_rate": 5.037931034482759e-06, | |
| "loss": 0.2751, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 1.3360428917329643, | |
| "grad_norm": 5.129386901855469, | |
| "learning_rate": 5.020689655172414e-06, | |
| "loss": 0.2644, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 1.340366655136631, | |
| "grad_norm": 5.174379348754883, | |
| "learning_rate": 5.00344827586207e-06, | |
| "loss": 0.2638, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 1.3446904185402975, | |
| "grad_norm": 3.6351191997528076, | |
| "learning_rate": 4.9862068965517245e-06, | |
| "loss": 0.2712, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 1.349014181943964, | |
| "grad_norm": 5.315875053405762, | |
| "learning_rate": 4.968965517241379e-06, | |
| "loss": 0.2663, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 1.3533379453476306, | |
| "grad_norm": 5.3713202476501465, | |
| "learning_rate": 4.951724137931035e-06, | |
| "loss": 0.2867, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 1.357661708751297, | |
| "grad_norm": 4.494086265563965, | |
| "learning_rate": 4.93448275862069e-06, | |
| "loss": 0.2633, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 1.3619854721549638, | |
| "grad_norm": 4.476729869842529, | |
| "learning_rate": 4.917241379310345e-06, | |
| "loss": 0.2646, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 1.3663092355586302, | |
| "grad_norm": 4.904832363128662, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 0.2543, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 1.3706329989622967, | |
| "grad_norm": 6.566470623016357, | |
| "learning_rate": 4.882758620689655e-06, | |
| "loss": 0.2778, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 1.3749567623659633, | |
| "grad_norm": 7.662662982940674, | |
| "learning_rate": 4.86551724137931e-06, | |
| "loss": 0.2749, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 1.37928052576963, | |
| "grad_norm": 4.465984344482422, | |
| "learning_rate": 4.848275862068966e-06, | |
| "loss": 0.2497, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 1.3836042891732965, | |
| "grad_norm": 4.482520580291748, | |
| "learning_rate": 4.831034482758621e-06, | |
| "loss": 0.2856, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.3836042891732965, | |
| "eval_loss": 0.27571046352386475, | |
| "eval_runtime": 5061.2899, | |
| "eval_samples_per_second": 4.569, | |
| "eval_steps_per_second": 0.571, | |
| "eval_wer": 35.15030679877327, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.387928052576963, | |
| "grad_norm": 4.781071662902832, | |
| "learning_rate": 4.813793103448276e-06, | |
| "loss": 0.2509, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 1.3922518159806296, | |
| "grad_norm": 5.495136260986328, | |
| "learning_rate": 4.7965517241379314e-06, | |
| "loss": 0.2804, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 1.396575579384296, | |
| "grad_norm": 5.323382377624512, | |
| "learning_rate": 4.779310344827587e-06, | |
| "loss": 0.2662, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 1.4008993427879626, | |
| "grad_norm": 4.909964084625244, | |
| "learning_rate": 4.762068965517242e-06, | |
| "loss": 0.2815, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 1.4052231061916292, | |
| "grad_norm": 4.8298420906066895, | |
| "learning_rate": 4.744827586206897e-06, | |
| "loss": 0.2605, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 1.4095468695952957, | |
| "grad_norm": 5.043326377868652, | |
| "learning_rate": 4.727586206896553e-06, | |
| "loss": 0.2805, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 1.4138706329989623, | |
| "grad_norm": 4.897740364074707, | |
| "learning_rate": 4.7103448275862075e-06, | |
| "loss": 0.2724, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 1.418194396402629, | |
| "grad_norm": 4.258267402648926, | |
| "learning_rate": 4.693103448275862e-06, | |
| "loss": 0.2559, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 1.4225181598062955, | |
| "grad_norm": 4.238452911376953, | |
| "learning_rate": 4.675862068965517e-06, | |
| "loss": 0.2633, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 1.4268419232099618, | |
| "grad_norm": 5.237506866455078, | |
| "learning_rate": 4.658620689655173e-06, | |
| "loss": 0.2695, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 1.4311656866136286, | |
| "grad_norm": 7.0735182762146, | |
| "learning_rate": 4.641379310344828e-06, | |
| "loss": 0.2792, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 1.435489450017295, | |
| "grad_norm": 5.525311470031738, | |
| "learning_rate": 4.624137931034483e-06, | |
| "loss": 0.2861, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 1.4398132134209616, | |
| "grad_norm": 5.706710338592529, | |
| "learning_rate": 4.606896551724138e-06, | |
| "loss": 0.2616, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 1.4441369768246282, | |
| "grad_norm": 4.5631794929504395, | |
| "learning_rate": 4.589655172413793e-06, | |
| "loss": 0.2528, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 1.4484607402282947, | |
| "grad_norm": 4.35201358795166, | |
| "learning_rate": 4.572413793103448e-06, | |
| "loss": 0.2669, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 1.4527845036319613, | |
| "grad_norm": 5.7140793800354, | |
| "learning_rate": 4.555172413793104e-06, | |
| "loss": 0.275, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 1.4571082670356277, | |
| "grad_norm": 6.433730602264404, | |
| "learning_rate": 4.537931034482759e-06, | |
| "loss": 0.2718, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 1.4614320304392945, | |
| "grad_norm": 4.385723114013672, | |
| "learning_rate": 4.521379310344828e-06, | |
| "loss": 0.2751, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 1.4657557938429608, | |
| "grad_norm": 4.835318565368652, | |
| "learning_rate": 4.504137931034483e-06, | |
| "loss": 0.273, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 1.4700795572466274, | |
| "grad_norm": 5.070703506469727, | |
| "learning_rate": 4.486896551724138e-06, | |
| "loss": 0.2659, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.474403320650294, | |
| "grad_norm": 4.764357089996338, | |
| "learning_rate": 4.4696551724137936e-06, | |
| "loss": 0.2614, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 1.4787270840539606, | |
| "grad_norm": 4.433307647705078, | |
| "learning_rate": 4.452413793103449e-06, | |
| "loss": 0.2773, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 1.4830508474576272, | |
| "grad_norm": 5.220912456512451, | |
| "learning_rate": 4.435172413793104e-06, | |
| "loss": 0.2739, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 1.4873746108612937, | |
| "grad_norm": 4.9389753341674805, | |
| "learning_rate": 4.417931034482759e-06, | |
| "loss": 0.2801, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 1.4916983742649603, | |
| "grad_norm": 5.52213191986084, | |
| "learning_rate": 4.400689655172414e-06, | |
| "loss": 0.2713, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 1.4960221376686267, | |
| "grad_norm": 4.186790943145752, | |
| "learning_rate": 4.38344827586207e-06, | |
| "loss": 0.2702, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 1.5003459010722935, | |
| "grad_norm": 5.3168535232543945, | |
| "learning_rate": 4.3662068965517245e-06, | |
| "loss": 0.2744, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 1.5046696644759598, | |
| "grad_norm": 5.070104122161865, | |
| "learning_rate": 4.348965517241379e-06, | |
| "loss": 0.282, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 1.5089934278796264, | |
| "grad_norm": 4.920200347900391, | |
| "learning_rate": 4.331724137931035e-06, | |
| "loss": 0.262, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 1.513317191283293, | |
| "grad_norm": 4.704413414001465, | |
| "learning_rate": 4.31448275862069e-06, | |
| "loss": 0.2539, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 1.5176409546869594, | |
| "grad_norm": 5.695456027984619, | |
| "learning_rate": 4.297241379310345e-06, | |
| "loss": 0.2843, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 1.5219647180906262, | |
| "grad_norm": 5.497166633605957, | |
| "learning_rate": 4.2800000000000005e-06, | |
| "loss": 0.2813, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 1.5262884814942925, | |
| "grad_norm": 6.1521100997924805, | |
| "learning_rate": 4.262758620689655e-06, | |
| "loss": 0.2835, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 1.5306122448979593, | |
| "grad_norm": 4.628076076507568, | |
| "learning_rate": 4.24551724137931e-06, | |
| "loss": 0.2685, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 1.5349360083016257, | |
| "grad_norm": 6.371509552001953, | |
| "learning_rate": 4.228275862068966e-06, | |
| "loss": 0.2596, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 1.5392597717052923, | |
| "grad_norm": 6.203851699829102, | |
| "learning_rate": 4.211034482758621e-06, | |
| "loss": 0.2801, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 1.5435835351089588, | |
| "grad_norm": 5.120136260986328, | |
| "learning_rate": 4.193793103448276e-06, | |
| "loss": 0.2749, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 1.5479072985126254, | |
| "grad_norm": 4.296701908111572, | |
| "learning_rate": 4.1765517241379314e-06, | |
| "loss": 0.268, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 1.552231061916292, | |
| "grad_norm": 4.204896926879883, | |
| "learning_rate": 4.159310344827587e-06, | |
| "loss": 0.2731, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 1.5565548253199584, | |
| "grad_norm": 4.787944793701172, | |
| "learning_rate": 4.142068965517242e-06, | |
| "loss": 0.2672, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.5565548253199584, | |
| "eval_loss": 0.27039873600006104, | |
| "eval_runtime": 5069.0399, | |
| "eval_samples_per_second": 4.562, | |
| "eval_steps_per_second": 0.57, | |
| "eval_wer": 34.11897560369892, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.5608785887236252, | |
| "grad_norm": 5.445296764373779, | |
| "learning_rate": 4.124827586206897e-06, | |
| "loss": 0.256, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 1.5652023521272915, | |
| "grad_norm": 6.453240871429443, | |
| "learning_rate": 4.107586206896552e-06, | |
| "loss": 0.2559, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 1.569526115530958, | |
| "grad_norm": 6.632162570953369, | |
| "learning_rate": 4.0903448275862075e-06, | |
| "loss": 0.287, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 1.5738498789346247, | |
| "grad_norm": 5.117345809936523, | |
| "learning_rate": 4.073103448275862e-06, | |
| "loss": 0.2669, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 1.5781736423382913, | |
| "grad_norm": 4.2106804847717285, | |
| "learning_rate": 4.055862068965517e-06, | |
| "loss": 0.2566, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 1.5824974057419579, | |
| "grad_norm": 4.997186660766602, | |
| "learning_rate": 4.038620689655173e-06, | |
| "loss": 0.2718, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 1.5868211691456242, | |
| "grad_norm": 6.009954452514648, | |
| "learning_rate": 4.021379310344828e-06, | |
| "loss": 0.2759, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 1.591144932549291, | |
| "grad_norm": 4.32164192199707, | |
| "learning_rate": 4.004137931034483e-06, | |
| "loss": 0.2437, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 1.5954686959529574, | |
| "grad_norm": 4.823526382446289, | |
| "learning_rate": 3.986896551724138e-06, | |
| "loss": 0.2632, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 1.599792459356624, | |
| "grad_norm": 4.981450080871582, | |
| "learning_rate": 3.969655172413793e-06, | |
| "loss": 0.263, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 1.6041162227602905, | |
| "grad_norm": 4.784339427947998, | |
| "learning_rate": 3.952413793103448e-06, | |
| "loss": 0.2763, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 1.6084399861639571, | |
| "grad_norm": 4.33436393737793, | |
| "learning_rate": 3.935172413793104e-06, | |
| "loss": 0.2595, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 1.6127637495676237, | |
| "grad_norm": 5.4285688400268555, | |
| "learning_rate": 3.917931034482759e-06, | |
| "loss": 0.2783, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 1.61708751297129, | |
| "grad_norm": 5.2749786376953125, | |
| "learning_rate": 3.9006896551724144e-06, | |
| "loss": 0.2497, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 1.6214112763749569, | |
| "grad_norm": 4.453002452850342, | |
| "learning_rate": 3.883448275862069e-06, | |
| "loss": 0.2749, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 1.6257350397786232, | |
| "grad_norm": 4.48176383972168, | |
| "learning_rate": 3.866206896551725e-06, | |
| "loss": 0.2867, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 1.63005880318229, | |
| "grad_norm": 4.524690628051758, | |
| "learning_rate": 3.84896551724138e-06, | |
| "loss": 0.2569, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 1.6343825665859564, | |
| "grad_norm": 5.824161529541016, | |
| "learning_rate": 3.831724137931035e-06, | |
| "loss": 0.2774, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 1.638706329989623, | |
| "grad_norm": 4.404734134674072, | |
| "learning_rate": 3.81448275862069e-06, | |
| "loss": 0.2761, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 1.6430300933932895, | |
| "grad_norm": 4.873973369598389, | |
| "learning_rate": 3.7972413793103454e-06, | |
| "loss": 0.2595, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.6473538567969561, | |
| "grad_norm": 5.397193908691406, | |
| "learning_rate": 3.7800000000000002e-06, | |
| "loss": 0.2654, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 1.6516776202006227, | |
| "grad_norm": 5.546463966369629, | |
| "learning_rate": 3.7627586206896555e-06, | |
| "loss": 0.2579, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 1.656001383604289, | |
| "grad_norm": 4.908322334289551, | |
| "learning_rate": 3.745517241379311e-06, | |
| "loss": 0.2644, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 1.6603251470079559, | |
| "grad_norm": 4.3278985023498535, | |
| "learning_rate": 3.7282758620689657e-06, | |
| "loss": 0.2522, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 1.6646489104116222, | |
| "grad_norm": 3.946744918823242, | |
| "learning_rate": 3.711034482758621e-06, | |
| "loss": 0.2612, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 1.6689726738152888, | |
| "grad_norm": 4.440819263458252, | |
| "learning_rate": 3.693793103448276e-06, | |
| "loss": 0.2743, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 1.6732964372189554, | |
| "grad_norm": 4.908570766448975, | |
| "learning_rate": 3.676551724137931e-06, | |
| "loss": 0.2556, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 1.677620200622622, | |
| "grad_norm": 6.872124195098877, | |
| "learning_rate": 3.6593103448275864e-06, | |
| "loss": 0.2755, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 1.6819439640262885, | |
| "grad_norm": 5.950398921966553, | |
| "learning_rate": 3.6420689655172413e-06, | |
| "loss": 0.258, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 1.686267727429955, | |
| "grad_norm": 5.322935581207275, | |
| "learning_rate": 3.624827586206897e-06, | |
| "loss": 0.2603, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 1.6905914908336217, | |
| "grad_norm": 4.969674587249756, | |
| "learning_rate": 3.6075862068965523e-06, | |
| "loss": 0.2719, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 1.694915254237288, | |
| "grad_norm": 4.605515003204346, | |
| "learning_rate": 3.590344827586207e-06, | |
| "loss": 0.2731, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 1.6992390176409546, | |
| "grad_norm": 4.371634006500244, | |
| "learning_rate": 3.5731034482758625e-06, | |
| "loss": 0.2581, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 1.7035627810446212, | |
| "grad_norm": 5.099359512329102, | |
| "learning_rate": 3.5558620689655178e-06, | |
| "loss": 0.2702, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 1.7078865444482878, | |
| "grad_norm": 5.02194881439209, | |
| "learning_rate": 3.5386206896551726e-06, | |
| "loss": 0.2575, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 1.7122103078519544, | |
| "grad_norm": 5.109201431274414, | |
| "learning_rate": 3.521379310344828e-06, | |
| "loss": 0.2452, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 1.7165340712556207, | |
| "grad_norm": 4.52341890335083, | |
| "learning_rate": 3.5041379310344832e-06, | |
| "loss": 0.2708, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 1.7208578346592875, | |
| "grad_norm": 4.951146602630615, | |
| "learning_rate": 3.486896551724138e-06, | |
| "loss": 0.2612, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 1.725181598062954, | |
| "grad_norm": 5.593024730682373, | |
| "learning_rate": 3.4696551724137934e-06, | |
| "loss": 0.2716, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 1.7295053614666207, | |
| "grad_norm": 5.451072692871094, | |
| "learning_rate": 3.4524137931034487e-06, | |
| "loss": 0.2623, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.7295053614666207, | |
| "eval_loss": 0.26585614681243896, | |
| "eval_runtime": 5142.3561, | |
| "eval_samples_per_second": 4.497, | |
| "eval_steps_per_second": 0.562, | |
| "eval_wer": 33.59979570177145, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.733829124870287, | |
| "grad_norm": 4.803134441375732, | |
| "learning_rate": 3.4351724137931036e-06, | |
| "loss": 0.2573, | |
| "step": 10025 | |
| }, | |
| { | |
| "epoch": 1.7381528882739536, | |
| "grad_norm": 4.306927680969238, | |
| "learning_rate": 3.417931034482759e-06, | |
| "loss": 0.2617, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 1.7424766516776202, | |
| "grad_norm": 5.380494594573975, | |
| "learning_rate": 3.4006896551724137e-06, | |
| "loss": 0.2653, | |
| "step": 10075 | |
| }, | |
| { | |
| "epoch": 1.7468004150812868, | |
| "grad_norm": 4.847281455993652, | |
| "learning_rate": 3.383448275862069e-06, | |
| "loss": 0.267, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 1.7511241784849534, | |
| "grad_norm": 4.344338893890381, | |
| "learning_rate": 3.3662068965517243e-06, | |
| "loss": 0.2612, | |
| "step": 10125 | |
| }, | |
| { | |
| "epoch": 1.7554479418886197, | |
| "grad_norm": 5.100605487823486, | |
| "learning_rate": 3.34896551724138e-06, | |
| "loss": 0.2624, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 1.7597717052922865, | |
| "grad_norm": 6.067707061767578, | |
| "learning_rate": 3.331724137931035e-06, | |
| "loss": 0.2797, | |
| "step": 10175 | |
| }, | |
| { | |
| "epoch": 1.764095468695953, | |
| "grad_norm": 4.3519392013549805, | |
| "learning_rate": 3.31448275862069e-06, | |
| "loss": 0.2607, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 1.7684192320996195, | |
| "grad_norm": 4.383536338806152, | |
| "learning_rate": 3.297241379310345e-06, | |
| "loss": 0.2516, | |
| "step": 10225 | |
| }, | |
| { | |
| "epoch": 1.772742995503286, | |
| "grad_norm": 5.012591361999512, | |
| "learning_rate": 3.2800000000000004e-06, | |
| "loss": 0.2724, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 1.7770667589069526, | |
| "grad_norm": 4.1687774658203125, | |
| "learning_rate": 3.2627586206896557e-06, | |
| "loss": 0.2698, | |
| "step": 10275 | |
| }, | |
| { | |
| "epoch": 1.7813905223106192, | |
| "grad_norm": 5.5503034591674805, | |
| "learning_rate": 3.2455172413793105e-06, | |
| "loss": 0.2652, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 1.7857142857142856, | |
| "grad_norm": 5.016941547393799, | |
| "learning_rate": 3.228275862068966e-06, | |
| "loss": 0.248, | |
| "step": 10325 | |
| }, | |
| { | |
| "epoch": 1.7900380491179524, | |
| "grad_norm": 4.316959381103516, | |
| "learning_rate": 3.211034482758621e-06, | |
| "loss": 0.2598, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 1.7943618125216187, | |
| "grad_norm": 6.067855358123779, | |
| "learning_rate": 3.193793103448276e-06, | |
| "loss": 0.2677, | |
| "step": 10375 | |
| }, | |
| { | |
| "epoch": 1.7986855759252853, | |
| "grad_norm": 5.613093852996826, | |
| "learning_rate": 3.1765517241379313e-06, | |
| "loss": 0.2679, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 1.803009339328952, | |
| "grad_norm": 4.065937519073486, | |
| "learning_rate": 3.1593103448275866e-06, | |
| "loss": 0.2634, | |
| "step": 10425 | |
| }, | |
| { | |
| "epoch": 1.8073331027326185, | |
| "grad_norm": 5.0161638259887695, | |
| "learning_rate": 3.1427586206896555e-06, | |
| "loss": 0.2706, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 1.811656866136285, | |
| "grad_norm": 6.777268886566162, | |
| "learning_rate": 3.1255172413793104e-06, | |
| "loss": 0.2756, | |
| "step": 10475 | |
| }, | |
| { | |
| "epoch": 1.8159806295399514, | |
| "grad_norm": 6.550757884979248, | |
| "learning_rate": 3.1082758620689657e-06, | |
| "loss": 0.2704, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.8203043929436182, | |
| "grad_norm": 5.1796393394470215, | |
| "learning_rate": 3.091034482758621e-06, | |
| "loss": 0.2707, | |
| "step": 10525 | |
| }, | |
| { | |
| "epoch": 1.8246281563472846, | |
| "grad_norm": 5.0789570808410645, | |
| "learning_rate": 3.073793103448276e-06, | |
| "loss": 0.2592, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 1.8289519197509514, | |
| "grad_norm": 4.78272819519043, | |
| "learning_rate": 3.056551724137931e-06, | |
| "loss": 0.2616, | |
| "step": 10575 | |
| }, | |
| { | |
| "epoch": 1.8332756831546178, | |
| "grad_norm": 4.711808681488037, | |
| "learning_rate": 3.0393103448275864e-06, | |
| "loss": 0.2729, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 1.8375994465582843, | |
| "grad_norm": 4.821509838104248, | |
| "learning_rate": 3.0220689655172413e-06, | |
| "loss": 0.2522, | |
| "step": 10625 | |
| }, | |
| { | |
| "epoch": 1.841923209961951, | |
| "grad_norm": 4.584725379943848, | |
| "learning_rate": 3.004827586206897e-06, | |
| "loss": 0.2639, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 1.8462469733656173, | |
| "grad_norm": 6.850635051727295, | |
| "learning_rate": 2.9875862068965523e-06, | |
| "loss": 0.2658, | |
| "step": 10675 | |
| }, | |
| { | |
| "epoch": 1.850570736769284, | |
| "grad_norm": 6.138273239135742, | |
| "learning_rate": 2.970344827586207e-06, | |
| "loss": 0.2735, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 1.8548945001729504, | |
| "grad_norm": 5.5177507400512695, | |
| "learning_rate": 2.9531034482758625e-06, | |
| "loss": 0.2497, | |
| "step": 10725 | |
| }, | |
| { | |
| "epoch": 1.8592182635766172, | |
| "grad_norm": 5.395568370819092, | |
| "learning_rate": 2.9358620689655178e-06, | |
| "loss": 0.2556, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 1.8635420269802836, | |
| "grad_norm": 3.9228525161743164, | |
| "learning_rate": 2.9186206896551727e-06, | |
| "loss": 0.2697, | |
| "step": 10775 | |
| }, | |
| { | |
| "epoch": 1.8678657903839502, | |
| "grad_norm": 3.7661633491516113, | |
| "learning_rate": 2.901379310344828e-06, | |
| "loss": 0.24, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 1.8721895537876168, | |
| "grad_norm": 3.9159317016601562, | |
| "learning_rate": 2.884137931034483e-06, | |
| "loss": 0.2663, | |
| "step": 10825 | |
| }, | |
| { | |
| "epoch": 1.8765133171912833, | |
| "grad_norm": 5.218613147735596, | |
| "learning_rate": 2.866896551724138e-06, | |
| "loss": 0.2539, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 1.88083708059495, | |
| "grad_norm": 4.241786479949951, | |
| "learning_rate": 2.8496551724137934e-06, | |
| "loss": 0.2672, | |
| "step": 10875 | |
| }, | |
| { | |
| "epoch": 1.8851608439986163, | |
| "grad_norm": 5.012778282165527, | |
| "learning_rate": 2.8324137931034483e-06, | |
| "loss": 0.2624, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 1.889484607402283, | |
| "grad_norm": 5.514505863189697, | |
| "learning_rate": 2.8151724137931036e-06, | |
| "loss": 0.2635, | |
| "step": 10925 | |
| }, | |
| { | |
| "epoch": 1.8938083708059494, | |
| "grad_norm": 4.5488667488098145, | |
| "learning_rate": 2.797931034482759e-06, | |
| "loss": 0.2456, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 1.898132134209616, | |
| "grad_norm": 4.1149516105651855, | |
| "learning_rate": 2.7806896551724137e-06, | |
| "loss": 0.2644, | |
| "step": 10975 | |
| }, | |
| { | |
| "epoch": 1.9024558976132826, | |
| "grad_norm": 5.599881649017334, | |
| "learning_rate": 2.763448275862069e-06, | |
| "loss": 0.2772, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.9024558976132826, | |
| "eval_loss": 0.263325035572052, | |
| "eval_runtime": 5127.6582, | |
| "eval_samples_per_second": 4.51, | |
| "eval_steps_per_second": 0.564, | |
| "eval_wer": 33.32895997675873, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.9067796610169492, | |
| "grad_norm": 5.128805637359619, | |
| "learning_rate": 2.7462068965517243e-06, | |
| "loss": 0.2599, | |
| "step": 11025 | |
| }, | |
| { | |
| "epoch": 1.9111034244206158, | |
| "grad_norm": 6.218988418579102, | |
| "learning_rate": 2.728965517241379e-06, | |
| "loss": 0.271, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 1.9154271878242821, | |
| "grad_norm": 4.934520721435547, | |
| "learning_rate": 2.711724137931035e-06, | |
| "loss": 0.2633, | |
| "step": 11075 | |
| }, | |
| { | |
| "epoch": 1.919750951227949, | |
| "grad_norm": 5.033214092254639, | |
| "learning_rate": 2.69448275862069e-06, | |
| "loss": 0.2563, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 1.9240747146316153, | |
| "grad_norm": 5.1451416015625, | |
| "learning_rate": 2.677241379310345e-06, | |
| "loss": 0.2525, | |
| "step": 11125 | |
| }, | |
| { | |
| "epoch": 1.9283984780352819, | |
| "grad_norm": 5.570995330810547, | |
| "learning_rate": 2.6600000000000004e-06, | |
| "loss": 0.2688, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 1.9327222414389484, | |
| "grad_norm": 3.9599545001983643, | |
| "learning_rate": 2.6427586206896557e-06, | |
| "loss": 0.2533, | |
| "step": 11175 | |
| }, | |
| { | |
| "epoch": 1.937046004842615, | |
| "grad_norm": 4.358778476715088, | |
| "learning_rate": 2.6255172413793105e-06, | |
| "loss": 0.2596, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 1.9413697682462816, | |
| "grad_norm": 4.855038166046143, | |
| "learning_rate": 2.608275862068966e-06, | |
| "loss": 0.2651, | |
| "step": 11225 | |
| }, | |
| { | |
| "epoch": 1.945693531649948, | |
| "grad_norm": 6.604015350341797, | |
| "learning_rate": 2.5910344827586207e-06, | |
| "loss": 0.2601, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 1.9500172950536148, | |
| "grad_norm": 5.335809230804443, | |
| "learning_rate": 2.573793103448276e-06, | |
| "loss": 0.2574, | |
| "step": 11275 | |
| }, | |
| { | |
| "epoch": 1.9543410584572811, | |
| "grad_norm": 4.764793395996094, | |
| "learning_rate": 2.5565517241379313e-06, | |
| "loss": 0.2634, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 1.958664821860948, | |
| "grad_norm": 5.247544765472412, | |
| "learning_rate": 2.539310344827586e-06, | |
| "loss": 0.2664, | |
| "step": 11325 | |
| }, | |
| { | |
| "epoch": 1.9629885852646143, | |
| "grad_norm": 4.707311630249023, | |
| "learning_rate": 2.5220689655172414e-06, | |
| "loss": 0.2609, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 1.9673123486682809, | |
| "grad_norm": 4.89914083480835, | |
| "learning_rate": 2.5048275862068967e-06, | |
| "loss": 0.2578, | |
| "step": 11375 | |
| }, | |
| { | |
| "epoch": 1.9716361120719474, | |
| "grad_norm": 4.219529151916504, | |
| "learning_rate": 2.487586206896552e-06, | |
| "loss": 0.2609, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 1.975959875475614, | |
| "grad_norm": 5.0991435050964355, | |
| "learning_rate": 2.4703448275862073e-06, | |
| "loss": 0.2511, | |
| "step": 11425 | |
| }, | |
| { | |
| "epoch": 1.9802836388792806, | |
| "grad_norm": 4.904223918914795, | |
| "learning_rate": 2.453103448275862e-06, | |
| "loss": 0.257, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 1.984607402282947, | |
| "grad_norm": 4.135773658752441, | |
| "learning_rate": 2.4358620689655175e-06, | |
| "loss": 0.2673, | |
| "step": 11475 | |
| }, | |
| { | |
| "epoch": 1.9889311656866138, | |
| "grad_norm": 4.5873332023620605, | |
| "learning_rate": 2.4186206896551724e-06, | |
| "loss": 0.2598, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.9932549290902801, | |
| "grad_norm": 5.754453659057617, | |
| "learning_rate": 2.4013793103448277e-06, | |
| "loss": 0.2602, | |
| "step": 11525 | |
| }, | |
| { | |
| "epoch": 1.9975786924939467, | |
| "grad_norm": 5.299346923828125, | |
| "learning_rate": 2.384137931034483e-06, | |
| "loss": 0.2629, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 2.0019024558976133, | |
| "grad_norm": 4.596057415008545, | |
| "learning_rate": 2.3668965517241382e-06, | |
| "loss": 0.2577, | |
| "step": 11575 | |
| }, | |
| { | |
| "epoch": 2.0062262193012796, | |
| "grad_norm": 4.527500629425049, | |
| "learning_rate": 2.3496551724137935e-06, | |
| "loss": 0.2519, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 2.0105499827049464, | |
| "grad_norm": 4.149202823638916, | |
| "learning_rate": 2.3324137931034484e-06, | |
| "loss": 0.2485, | |
| "step": 11625 | |
| }, | |
| { | |
| "epoch": 2.014873746108613, | |
| "grad_norm": 4.757724761962891, | |
| "learning_rate": 2.3151724137931037e-06, | |
| "loss": 0.2445, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 2.0191975095122796, | |
| "grad_norm": 4.565730571746826, | |
| "learning_rate": 2.297931034482759e-06, | |
| "loss": 0.2376, | |
| "step": 11675 | |
| }, | |
| { | |
| "epoch": 2.023521272915946, | |
| "grad_norm": 4.340237140655518, | |
| "learning_rate": 2.280689655172414e-06, | |
| "loss": 0.2617, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 2.0278450363196128, | |
| "grad_norm": 5.368824005126953, | |
| "learning_rate": 2.263448275862069e-06, | |
| "loss": 0.2384, | |
| "step": 11725 | |
| }, | |
| { | |
| "epoch": 2.032168799723279, | |
| "grad_norm": 4.97426176071167, | |
| "learning_rate": 2.246206896551724e-06, | |
| "loss": 0.2501, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 2.0364925631269455, | |
| "grad_norm": 6.131598949432373, | |
| "learning_rate": 2.2289655172413797e-06, | |
| "loss": 0.2518, | |
| "step": 11775 | |
| }, | |
| { | |
| "epoch": 2.0408163265306123, | |
| "grad_norm": 4.28442907333374, | |
| "learning_rate": 2.2117241379310346e-06, | |
| "loss": 0.2653, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 2.0451400899342786, | |
| "grad_norm": 4.612010478973389, | |
| "learning_rate": 2.19448275862069e-06, | |
| "loss": 0.2447, | |
| "step": 11825 | |
| }, | |
| { | |
| "epoch": 2.0494638533379455, | |
| "grad_norm": 4.856114387512207, | |
| "learning_rate": 2.177241379310345e-06, | |
| "loss": 0.2456, | |
| "step": 11850 | |
| }, | |
| { | |
| "epoch": 2.053787616741612, | |
| "grad_norm": 5.475619792938232, | |
| "learning_rate": 2.16e-06, | |
| "loss": 0.2516, | |
| "step": 11875 | |
| }, | |
| { | |
| "epoch": 2.0581113801452786, | |
| "grad_norm": 4.844070911407471, | |
| "learning_rate": 2.1427586206896554e-06, | |
| "loss": 0.2354, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 2.062435143548945, | |
| "grad_norm": 4.885989189147949, | |
| "learning_rate": 2.1255172413793102e-06, | |
| "loss": 0.2427, | |
| "step": 11925 | |
| }, | |
| { | |
| "epoch": 2.0667589069526118, | |
| "grad_norm": 5.934988975524902, | |
| "learning_rate": 2.1082758620689655e-06, | |
| "loss": 0.2599, | |
| "step": 11950 | |
| }, | |
| { | |
| "epoch": 2.071082670356278, | |
| "grad_norm": 5.128411769866943, | |
| "learning_rate": 2.091034482758621e-06, | |
| "loss": 0.2457, | |
| "step": 11975 | |
| }, | |
| { | |
| "epoch": 2.0754064337599445, | |
| "grad_norm": 4.521295547485352, | |
| "learning_rate": 2.073793103448276e-06, | |
| "loss": 0.2481, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.0754064337599445, | |
| "eval_loss": 0.2604476511478424, | |
| "eval_runtime": 5099.3278, | |
| "eval_samples_per_second": 4.535, | |
| "eval_steps_per_second": 0.567, | |
| "eval_wer": 32.88803191925535, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.0797301971636113, | |
| "grad_norm": 4.452366828918457, | |
| "learning_rate": 2.0565517241379314e-06, | |
| "loss": 0.2485, | |
| "step": 12025 | |
| }, | |
| { | |
| "epoch": 2.0840539605672777, | |
| "grad_norm": 4.331884384155273, | |
| "learning_rate": 2.0393103448275863e-06, | |
| "loss": 0.2452, | |
| "step": 12050 | |
| }, | |
| { | |
| "epoch": 2.0883777239709445, | |
| "grad_norm": 4.478641510009766, | |
| "learning_rate": 2.0220689655172416e-06, | |
| "loss": 0.2485, | |
| "step": 12075 | |
| }, | |
| { | |
| "epoch": 2.092701487374611, | |
| "grad_norm": 4.945328235626221, | |
| "learning_rate": 2.004827586206897e-06, | |
| "loss": 0.2334, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 2.0970252507782776, | |
| "grad_norm": 5.300661087036133, | |
| "learning_rate": 1.9875862068965517e-06, | |
| "loss": 0.2457, | |
| "step": 12125 | |
| }, | |
| { | |
| "epoch": 2.101349014181944, | |
| "grad_norm": 5.748834133148193, | |
| "learning_rate": 1.970344827586207e-06, | |
| "loss": 0.2562, | |
| "step": 12150 | |
| }, | |
| { | |
| "epoch": 2.1056727775856103, | |
| "grad_norm": 4.078002452850342, | |
| "learning_rate": 1.9531034482758623e-06, | |
| "loss": 0.2496, | |
| "step": 12175 | |
| }, | |
| { | |
| "epoch": 2.109996540989277, | |
| "grad_norm": 4.0208258628845215, | |
| "learning_rate": 1.9358620689655176e-06, | |
| "loss": 0.2528, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 2.1143203043929435, | |
| "grad_norm": 5.57705545425415, | |
| "learning_rate": 1.9186206896551725e-06, | |
| "loss": 0.239, | |
| "step": 12225 | |
| }, | |
| { | |
| "epoch": 2.1186440677966103, | |
| "grad_norm": 4.870556831359863, | |
| "learning_rate": 1.9013793103448278e-06, | |
| "loss": 0.2438, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 2.1229678312002767, | |
| "grad_norm": 5.422897815704346, | |
| "learning_rate": 1.8841379310344829e-06, | |
| "loss": 0.2332, | |
| "step": 12275 | |
| }, | |
| { | |
| "epoch": 2.1272915946039435, | |
| "grad_norm": 5.2437052726745605, | |
| "learning_rate": 1.866896551724138e-06, | |
| "loss": 0.2505, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 2.13161535800761, | |
| "grad_norm": 5.451413631439209, | |
| "learning_rate": 1.8496551724137932e-06, | |
| "loss": 0.249, | |
| "step": 12325 | |
| }, | |
| { | |
| "epoch": 2.135939121411276, | |
| "grad_norm": 4.326362609863281, | |
| "learning_rate": 1.8324137931034483e-06, | |
| "loss": 0.2428, | |
| "step": 12350 | |
| }, | |
| { | |
| "epoch": 2.140262884814943, | |
| "grad_norm": 5.135347366333008, | |
| "learning_rate": 1.8151724137931036e-06, | |
| "loss": 0.2514, | |
| "step": 12375 | |
| }, | |
| { | |
| "epoch": 2.1445866482186093, | |
| "grad_norm": 4.476009368896484, | |
| "learning_rate": 1.797931034482759e-06, | |
| "loss": 0.2474, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 2.148910411622276, | |
| "grad_norm": 4.706514358520508, | |
| "learning_rate": 1.780689655172414e-06, | |
| "loss": 0.2376, | |
| "step": 12425 | |
| }, | |
| { | |
| "epoch": 2.1532341750259425, | |
| "grad_norm": 5.396040916442871, | |
| "learning_rate": 1.7641379310344827e-06, | |
| "loss": 0.2595, | |
| "step": 12450 | |
| }, | |
| { | |
| "epoch": 2.1575579384296093, | |
| "grad_norm": 4.83308744430542, | |
| "learning_rate": 1.7468965517241382e-06, | |
| "loss": 0.2447, | |
| "step": 12475 | |
| }, | |
| { | |
| "epoch": 2.1618817018332757, | |
| "grad_norm": 4.416563034057617, | |
| "learning_rate": 1.7296551724137933e-06, | |
| "loss": 0.2408, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 2.166205465236942, | |
| "grad_norm": 4.504754543304443, | |
| "learning_rate": 1.7124137931034484e-06, | |
| "loss": 0.2476, | |
| "step": 12525 | |
| }, | |
| { | |
| "epoch": 2.170529228640609, | |
| "grad_norm": 5.582792282104492, | |
| "learning_rate": 1.6951724137931035e-06, | |
| "loss": 0.2334, | |
| "step": 12550 | |
| }, | |
| { | |
| "epoch": 2.174852992044275, | |
| "grad_norm": 5.201076030731201, | |
| "learning_rate": 1.6779310344827588e-06, | |
| "loss": 0.2459, | |
| "step": 12575 | |
| }, | |
| { | |
| "epoch": 2.179176755447942, | |
| "grad_norm": 4.038005828857422, | |
| "learning_rate": 1.6606896551724139e-06, | |
| "loss": 0.2518, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 2.1835005188516083, | |
| "grad_norm": 4.951401233673096, | |
| "learning_rate": 1.643448275862069e-06, | |
| "loss": 0.2433, | |
| "step": 12625 | |
| }, | |
| { | |
| "epoch": 2.187824282255275, | |
| "grad_norm": 4.419375896453857, | |
| "learning_rate": 1.6262068965517242e-06, | |
| "loss": 0.2427, | |
| "step": 12650 | |
| }, | |
| { | |
| "epoch": 2.1921480456589415, | |
| "grad_norm": 4.835983753204346, | |
| "learning_rate": 1.6089655172413795e-06, | |
| "loss": 0.2348, | |
| "step": 12675 | |
| }, | |
| { | |
| "epoch": 2.196471809062608, | |
| "grad_norm": 4.035714149475098, | |
| "learning_rate": 1.5917241379310346e-06, | |
| "loss": 0.2481, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 2.2007955724662747, | |
| "grad_norm": 6.195335388183594, | |
| "learning_rate": 1.57448275862069e-06, | |
| "loss": 0.2559, | |
| "step": 12725 | |
| }, | |
| { | |
| "epoch": 2.205119335869941, | |
| "grad_norm": 4.89235258102417, | |
| "learning_rate": 1.557241379310345e-06, | |
| "loss": 0.2288, | |
| "step": 12750 | |
| }, | |
| { | |
| "epoch": 2.209443099273608, | |
| "grad_norm": 4.976417541503906, | |
| "learning_rate": 1.54e-06, | |
| "loss": 0.2412, | |
| "step": 12775 | |
| }, | |
| { | |
| "epoch": 2.213766862677274, | |
| "grad_norm": 4.204568862915039, | |
| "learning_rate": 1.5227586206896552e-06, | |
| "loss": 0.2607, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 2.218090626080941, | |
| "grad_norm": 4.752575397491455, | |
| "learning_rate": 1.5055172413793105e-06, | |
| "loss": 0.2328, | |
| "step": 12825 | |
| }, | |
| { | |
| "epoch": 2.2224143894846073, | |
| "grad_norm": 5.109005451202393, | |
| "learning_rate": 1.4882758620689655e-06, | |
| "loss": 0.2472, | |
| "step": 12850 | |
| }, | |
| { | |
| "epoch": 2.226738152888274, | |
| "grad_norm": 4.870945453643799, | |
| "learning_rate": 1.4710344827586208e-06, | |
| "loss": 0.2411, | |
| "step": 12875 | |
| }, | |
| { | |
| "epoch": 2.2310619162919405, | |
| "grad_norm": 4.599150657653809, | |
| "learning_rate": 1.4537931034482761e-06, | |
| "loss": 0.2475, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 2.235385679695607, | |
| "grad_norm": 5.0102057456970215, | |
| "learning_rate": 1.4365517241379312e-06, | |
| "loss": 0.2466, | |
| "step": 12925 | |
| }, | |
| { | |
| "epoch": 2.2397094430992737, | |
| "grad_norm": 4.145412921905518, | |
| "learning_rate": 1.4193103448275863e-06, | |
| "loss": 0.2484, | |
| "step": 12950 | |
| }, | |
| { | |
| "epoch": 2.24403320650294, | |
| "grad_norm": 3.842782735824585, | |
| "learning_rate": 1.4020689655172416e-06, | |
| "loss": 0.2357, | |
| "step": 12975 | |
| }, | |
| { | |
| "epoch": 2.248356969906607, | |
| "grad_norm": 4.585821151733398, | |
| "learning_rate": 1.3848275862068967e-06, | |
| "loss": 0.2436, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 2.248356969906607, | |
| "eval_loss": 0.25910165905952454, | |
| "eval_runtime": 5129.3259, | |
| "eval_samples_per_second": 4.509, | |
| "eval_steps_per_second": 0.564, | |
| "eval_wer": 32.73105965648846, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 2.252680733310273, | |
| "grad_norm": 4.6802568435668945, | |
| "learning_rate": 1.3675862068965517e-06, | |
| "loss": 0.2348, | |
| "step": 13025 | |
| }, | |
| { | |
| "epoch": 2.25700449671394, | |
| "grad_norm": 5.42781925201416, | |
| "learning_rate": 1.3503448275862068e-06, | |
| "loss": 0.2349, | |
| "step": 13050 | |
| }, | |
| { | |
| "epoch": 2.2613282601176063, | |
| "grad_norm": 5.2979536056518555, | |
| "learning_rate": 1.3331034482758623e-06, | |
| "loss": 0.2421, | |
| "step": 13075 | |
| }, | |
| { | |
| "epoch": 2.265652023521273, | |
| "grad_norm": 5.278492450714111, | |
| "learning_rate": 1.3158620689655174e-06, | |
| "loss": 0.2438, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 2.2699757869249395, | |
| "grad_norm": 4.465871810913086, | |
| "learning_rate": 1.2986206896551725e-06, | |
| "loss": 0.2624, | |
| "step": 13125 | |
| }, | |
| { | |
| "epoch": 2.274299550328606, | |
| "grad_norm": 4.552392959594727, | |
| "learning_rate": 1.2813793103448278e-06, | |
| "loss": 0.2708, | |
| "step": 13150 | |
| }, | |
| { | |
| "epoch": 2.2786233137322727, | |
| "grad_norm": 4.50985050201416, | |
| "learning_rate": 1.2641379310344829e-06, | |
| "loss": 0.2437, | |
| "step": 13175 | |
| }, | |
| { | |
| "epoch": 2.282947077135939, | |
| "grad_norm": 5.000662326812744, | |
| "learning_rate": 1.246896551724138e-06, | |
| "loss": 0.2499, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 2.287270840539606, | |
| "grad_norm": 4.699297904968262, | |
| "learning_rate": 1.2296551724137932e-06, | |
| "loss": 0.2546, | |
| "step": 13225 | |
| }, | |
| { | |
| "epoch": 2.291594603943272, | |
| "grad_norm": 4.824137210845947, | |
| "learning_rate": 1.2124137931034483e-06, | |
| "loss": 0.2566, | |
| "step": 13250 | |
| }, | |
| { | |
| "epoch": 2.295918367346939, | |
| "grad_norm": 5.796444416046143, | |
| "learning_rate": 1.1951724137931036e-06, | |
| "loss": 0.254, | |
| "step": 13275 | |
| }, | |
| { | |
| "epoch": 2.3002421307506054, | |
| "grad_norm": 4.286495208740234, | |
| "learning_rate": 1.1779310344827587e-06, | |
| "loss": 0.2258, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 2.3045658941542717, | |
| "grad_norm": 4.737240314483643, | |
| "learning_rate": 1.160689655172414e-06, | |
| "loss": 0.2295, | |
| "step": 13325 | |
| }, | |
| { | |
| "epoch": 2.3088896575579385, | |
| "grad_norm": 5.5657148361206055, | |
| "learning_rate": 1.143448275862069e-06, | |
| "loss": 0.2516, | |
| "step": 13350 | |
| }, | |
| { | |
| "epoch": 2.313213420961605, | |
| "grad_norm": 6.122943878173828, | |
| "learning_rate": 1.1262068965517242e-06, | |
| "loss": 0.257, | |
| "step": 13375 | |
| }, | |
| { | |
| "epoch": 2.3175371843652717, | |
| "grad_norm": 5.030111789703369, | |
| "learning_rate": 1.1089655172413795e-06, | |
| "loss": 0.2409, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 2.321860947768938, | |
| "grad_norm": 4.784854412078857, | |
| "learning_rate": 1.0917241379310345e-06, | |
| "loss": 0.2371, | |
| "step": 13425 | |
| }, | |
| { | |
| "epoch": 2.326184711172605, | |
| "grad_norm": 4.606825351715088, | |
| "learning_rate": 1.0744827586206898e-06, | |
| "loss": 0.2496, | |
| "step": 13450 | |
| }, | |
| { | |
| "epoch": 2.330508474576271, | |
| "grad_norm": 5.06074333190918, | |
| "learning_rate": 1.057241379310345e-06, | |
| "loss": 0.2328, | |
| "step": 13475 | |
| }, | |
| { | |
| "epoch": 2.3348322379799376, | |
| "grad_norm": 4.836973190307617, | |
| "learning_rate": 1.04e-06, | |
| "loss": 0.2485, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.3391560013836044, | |
| "grad_norm": 4.6793365478515625, | |
| "learning_rate": 1.0227586206896553e-06, | |
| "loss": 0.257, | |
| "step": 13525 | |
| }, | |
| { | |
| "epoch": 2.3434797647872707, | |
| "grad_norm": 4.693976879119873, | |
| "learning_rate": 1.0055172413793104e-06, | |
| "loss": 0.2445, | |
| "step": 13550 | |
| }, | |
| { | |
| "epoch": 2.3478035281909375, | |
| "grad_norm": 4.091955184936523, | |
| "learning_rate": 9.882758620689657e-07, | |
| "loss": 0.2425, | |
| "step": 13575 | |
| }, | |
| { | |
| "epoch": 2.352127291594604, | |
| "grad_norm": 4.664127349853516, | |
| "learning_rate": 9.710344827586207e-07, | |
| "loss": 0.2414, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 2.3564510549982707, | |
| "grad_norm": 4.556197166442871, | |
| "learning_rate": 9.53793103448276e-07, | |
| "loss": 0.2503, | |
| "step": 13625 | |
| }, | |
| { | |
| "epoch": 2.360774818401937, | |
| "grad_norm": 4.94804573059082, | |
| "learning_rate": 9.365517241379311e-07, | |
| "loss": 0.2394, | |
| "step": 13650 | |
| }, | |
| { | |
| "epoch": 2.3650985818056034, | |
| "grad_norm": 5.318421840667725, | |
| "learning_rate": 9.193103448275863e-07, | |
| "loss": 0.2515, | |
| "step": 13675 | |
| }, | |
| { | |
| "epoch": 2.36942234520927, | |
| "grad_norm": 4.128868579864502, | |
| "learning_rate": 9.020689655172414e-07, | |
| "loss": 0.2494, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 2.3737461086129366, | |
| "grad_norm": 4.886101245880127, | |
| "learning_rate": 8.848275862068967e-07, | |
| "loss": 0.2567, | |
| "step": 13725 | |
| }, | |
| { | |
| "epoch": 2.3780698720166034, | |
| "grad_norm": 5.137601852416992, | |
| "learning_rate": 8.675862068965518e-07, | |
| "loss": 0.2529, | |
| "step": 13750 | |
| }, | |
| { | |
| "epoch": 2.3823936354202697, | |
| "grad_norm": 5.788776397705078, | |
| "learning_rate": 8.50344827586207e-07, | |
| "loss": 0.2369, | |
| "step": 13775 | |
| }, | |
| { | |
| "epoch": 2.3867173988239365, | |
| "grad_norm": 5.918336868286133, | |
| "learning_rate": 8.331034482758621e-07, | |
| "loss": 0.2505, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 2.391041162227603, | |
| "grad_norm": 4.928556442260742, | |
| "learning_rate": 8.158620689655173e-07, | |
| "loss": 0.2406, | |
| "step": 13825 | |
| }, | |
| { | |
| "epoch": 2.3953649256312692, | |
| "grad_norm": 5.406356334686279, | |
| "learning_rate": 7.986206896551725e-07, | |
| "loss": 0.2393, | |
| "step": 13850 | |
| }, | |
| { | |
| "epoch": 2.399688689034936, | |
| "grad_norm": 4.0312018394470215, | |
| "learning_rate": 7.813793103448276e-07, | |
| "loss": 0.2369, | |
| "step": 13875 | |
| }, | |
| { | |
| "epoch": 2.4040124524386024, | |
| "grad_norm": 5.446890830993652, | |
| "learning_rate": 7.641379310344828e-07, | |
| "loss": 0.2313, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 2.408336215842269, | |
| "grad_norm": 5.433596611022949, | |
| "learning_rate": 7.468965517241381e-07, | |
| "loss": 0.2434, | |
| "step": 13925 | |
| }, | |
| { | |
| "epoch": 2.4126599792459356, | |
| "grad_norm": 3.490582227706909, | |
| "learning_rate": 7.296551724137932e-07, | |
| "loss": 0.2268, | |
| "step": 13950 | |
| }, | |
| { | |
| "epoch": 2.4169837426496024, | |
| "grad_norm": 4.078389644622803, | |
| "learning_rate": 7.124137931034484e-07, | |
| "loss": 0.2397, | |
| "step": 13975 | |
| }, | |
| { | |
| "epoch": 2.4213075060532687, | |
| "grad_norm": 5.175432205200195, | |
| "learning_rate": 6.951724137931034e-07, | |
| "loss": 0.243, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.4213075060532687, | |
| "eval_loss": 0.2580419182777405, | |
| "eval_runtime": 5125.701, | |
| "eval_samples_per_second": 4.512, | |
| "eval_steps_per_second": 0.564, | |
| "eval_wer": 32.74277400445614, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.425631269456935, | |
| "grad_norm": 5.08500337600708, | |
| "learning_rate": 6.779310344827587e-07, | |
| "loss": 0.2431, | |
| "step": 14025 | |
| }, | |
| { | |
| "epoch": 2.429955032860602, | |
| "grad_norm": 4.165167331695557, | |
| "learning_rate": 6.606896551724139e-07, | |
| "loss": 0.2549, | |
| "step": 14050 | |
| }, | |
| { | |
| "epoch": 2.4342787962642682, | |
| "grad_norm": 4.766578197479248, | |
| "learning_rate": 6.43448275862069e-07, | |
| "loss": 0.2325, | |
| "step": 14075 | |
| }, | |
| { | |
| "epoch": 2.438602559667935, | |
| "grad_norm": 4.852494716644287, | |
| "learning_rate": 6.262068965517242e-07, | |
| "loss": 0.2471, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 2.4429263230716014, | |
| "grad_norm": 5.093419551849365, | |
| "learning_rate": 6.089655172413794e-07, | |
| "loss": 0.2543, | |
| "step": 14125 | |
| }, | |
| { | |
| "epoch": 2.447250086475268, | |
| "grad_norm": 6.8064045906066895, | |
| "learning_rate": 5.917241379310346e-07, | |
| "loss": 0.2446, | |
| "step": 14150 | |
| }, | |
| { | |
| "epoch": 2.4515738498789346, | |
| "grad_norm": 6.039092063903809, | |
| "learning_rate": 5.744827586206897e-07, | |
| "loss": 0.2437, | |
| "step": 14175 | |
| }, | |
| { | |
| "epoch": 2.4558976132826014, | |
| "grad_norm": 4.570720672607422, | |
| "learning_rate": 5.572413793103449e-07, | |
| "loss": 0.2357, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 2.4602213766862677, | |
| "grad_norm": 6.016234397888184, | |
| "learning_rate": 5.4e-07, | |
| "loss": 0.2389, | |
| "step": 14225 | |
| }, | |
| { | |
| "epoch": 2.4645451400899345, | |
| "grad_norm": 5.1399407386779785, | |
| "learning_rate": 5.227586206896552e-07, | |
| "loss": 0.2551, | |
| "step": 14250 | |
| }, | |
| { | |
| "epoch": 2.468868903493601, | |
| "grad_norm": 4.784331798553467, | |
| "learning_rate": 5.055172413793104e-07, | |
| "loss": 0.2292, | |
| "step": 14275 | |
| }, | |
| { | |
| "epoch": 2.4731926668972672, | |
| "grad_norm": 4.694582462310791, | |
| "learning_rate": 4.882758620689656e-07, | |
| "loss": 0.238, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 2.477516430300934, | |
| "grad_norm": 4.42811393737793, | |
| "learning_rate": 4.710344827586207e-07, | |
| "loss": 0.2471, | |
| "step": 14325 | |
| }, | |
| { | |
| "epoch": 2.4818401937046004, | |
| "grad_norm": 5.3042731285095215, | |
| "learning_rate": 4.537931034482759e-07, | |
| "loss": 0.2469, | |
| "step": 14350 | |
| }, | |
| { | |
| "epoch": 2.486163957108267, | |
| "grad_norm": 5.875729560852051, | |
| "learning_rate": 4.365517241379311e-07, | |
| "loss": 0.2438, | |
| "step": 14375 | |
| }, | |
| { | |
| "epoch": 2.4904877205119336, | |
| "grad_norm": 4.108542442321777, | |
| "learning_rate": 4.193103448275863e-07, | |
| "loss": 0.2482, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 2.4948114839156004, | |
| "grad_norm": 4.775691986083984, | |
| "learning_rate": 4.020689655172414e-07, | |
| "loss": 0.2374, | |
| "step": 14425 | |
| }, | |
| { | |
| "epoch": 2.4991352473192667, | |
| "grad_norm": 4.536764621734619, | |
| "learning_rate": 3.848275862068966e-07, | |
| "loss": 0.2399, | |
| "step": 14450 | |
| }, | |
| { | |
| "epoch": 2.503459010722933, | |
| "grad_norm": 3.924809694290161, | |
| "learning_rate": 3.6758620689655174e-07, | |
| "loss": 0.2518, | |
| "step": 14475 | |
| }, | |
| { | |
| "epoch": 2.5077827741266, | |
| "grad_norm": 4.728981971740723, | |
| "learning_rate": 3.50344827586207e-07, | |
| "loss": 0.2688, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 2.5121065375302662, | |
| "grad_norm": 4.350174903869629, | |
| "learning_rate": 3.331034482758621e-07, | |
| "loss": 0.2495, | |
| "step": 14525 | |
| }, | |
| { | |
| "epoch": 2.516430300933933, | |
| "grad_norm": 5.987773418426514, | |
| "learning_rate": 3.158620689655173e-07, | |
| "loss": 0.2436, | |
| "step": 14550 | |
| }, | |
| { | |
| "epoch": 2.5207540643375994, | |
| "grad_norm": 3.9468040466308594, | |
| "learning_rate": 2.9862068965517244e-07, | |
| "loss": 0.2241, | |
| "step": 14575 | |
| }, | |
| { | |
| "epoch": 2.525077827741266, | |
| "grad_norm": 4.972886562347412, | |
| "learning_rate": 2.813793103448276e-07, | |
| "loss": 0.2383, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 2.5294015911449326, | |
| "grad_norm": 5.958413600921631, | |
| "learning_rate": 2.6413793103448276e-07, | |
| "loss": 0.2435, | |
| "step": 14625 | |
| }, | |
| { | |
| "epoch": 2.533725354548599, | |
| "grad_norm": 4.814051151275635, | |
| "learning_rate": 2.4689655172413795e-07, | |
| "loss": 0.2485, | |
| "step": 14650 | |
| }, | |
| { | |
| "epoch": 2.5380491179522657, | |
| "grad_norm": 5.348843574523926, | |
| "learning_rate": 2.2965517241379313e-07, | |
| "loss": 0.2542, | |
| "step": 14675 | |
| }, | |
| { | |
| "epoch": 2.542372881355932, | |
| "grad_norm": 4.570481300354004, | |
| "learning_rate": 2.124137931034483e-07, | |
| "loss": 0.2415, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 2.546696644759599, | |
| "grad_norm": 4.454525470733643, | |
| "learning_rate": 1.9517241379310346e-07, | |
| "loss": 0.2356, | |
| "step": 14725 | |
| }, | |
| { | |
| "epoch": 2.5510204081632653, | |
| "grad_norm": 6.421968460083008, | |
| "learning_rate": 1.7793103448275864e-07, | |
| "loss": 0.2421, | |
| "step": 14750 | |
| }, | |
| { | |
| "epoch": 2.555344171566932, | |
| "grad_norm": 4.840851306915283, | |
| "learning_rate": 1.606896551724138e-07, | |
| "loss": 0.2446, | |
| "step": 14775 | |
| }, | |
| { | |
| "epoch": 2.5596679349705984, | |
| "grad_norm": 5.429544925689697, | |
| "learning_rate": 1.4344827586206897e-07, | |
| "loss": 0.2477, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 2.5639916983742648, | |
| "grad_norm": 5.538628101348877, | |
| "learning_rate": 1.2620689655172415e-07, | |
| "loss": 0.2475, | |
| "step": 14825 | |
| }, | |
| { | |
| "epoch": 2.5683154617779316, | |
| "grad_norm": 4.517398357391357, | |
| "learning_rate": 1.0896551724137932e-07, | |
| "loss": 0.2448, | |
| "step": 14850 | |
| }, | |
| { | |
| "epoch": 2.572639225181598, | |
| "grad_norm": 4.563333034515381, | |
| "learning_rate": 9.172413793103449e-08, | |
| "loss": 0.2316, | |
| "step": 14875 | |
| }, | |
| { | |
| "epoch": 2.5769629885852647, | |
| "grad_norm": 4.876601696014404, | |
| "learning_rate": 7.448275862068966e-08, | |
| "loss": 0.2378, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 2.581286751988931, | |
| "grad_norm": 4.932113170623779, | |
| "learning_rate": 5.724137931034483e-08, | |
| "loss": 0.2549, | |
| "step": 14925 | |
| }, | |
| { | |
| "epoch": 2.585610515392598, | |
| "grad_norm": 5.311142921447754, | |
| "learning_rate": 4e-08, | |
| "loss": 0.2477, | |
| "step": 14950 | |
| }, | |
| { | |
| "epoch": 2.5899342787962643, | |
| "grad_norm": 5.145079135894775, | |
| "learning_rate": 2.2758620689655175e-08, | |
| "loss": 0.2361, | |
| "step": 14975 | |
| }, | |
| { | |
| "epoch": 2.5942580421999306, | |
| "grad_norm": 4.524856090545654, | |
| "learning_rate": 5.517241379310346e-09, | |
| "loss": 0.2344, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.5942580421999306, | |
| "eval_loss": 0.2568697929382324, | |
| "eval_runtime": 5134.1958, | |
| "eval_samples_per_second": 4.504, | |
| "eval_steps_per_second": 0.563, | |
| "eval_wer": 32.56401305446938, | |
| "step": 15000 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 15000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.90799081332736e+18, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |