| { | |
| "best_global_step": 2000, | |
| "best_metric": 38.607193210094984, | |
| "best_model_checkpoint": "./SALAMA_C7/checkpoint-2000", | |
| "epoch": 1.6294561010389081, | |
| "eval_steps": 2000, | |
| "global_step": 4000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.008148299042574863, | |
| "grad_norm": 15.915007591247559, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 1.0767, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.016296598085149726, | |
| "grad_norm": 10.589001655578613, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 0.9639, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.024444897127724588, | |
| "grad_norm": 11.770127296447754, | |
| "learning_rate": 5.9e-06, | |
| "loss": 0.9679, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.03259319617029945, | |
| "grad_norm": 11.00736141204834, | |
| "learning_rate": 7.9e-06, | |
| "loss": 0.9264, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.04074149521287431, | |
| "grad_norm": 8.219990730285645, | |
| "learning_rate": 9.9e-06, | |
| "loss": 0.7982, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.048889794255449176, | |
| "grad_norm": 7.805874824523926, | |
| "learning_rate": 9.973847212663455e-06, | |
| "loss": 0.7226, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.057038093298024034, | |
| "grad_norm": 6.7659430503845215, | |
| "learning_rate": 9.946317962835514e-06, | |
| "loss": 0.7378, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.0651863923405989, | |
| "grad_norm": 6.5482282638549805, | |
| "learning_rate": 9.91878871300757e-06, | |
| "loss": 0.716, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.07333469138317376, | |
| "grad_norm": 6.729607105255127, | |
| "learning_rate": 9.891259463179629e-06, | |
| "loss": 0.7048, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.08148299042574862, | |
| "grad_norm": 8.517411231994629, | |
| "learning_rate": 9.863730213351686e-06, | |
| "loss": 0.7016, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.0896312894683235, | |
| "grad_norm": 6.785459041595459, | |
| "learning_rate": 9.836200963523744e-06, | |
| "loss": 0.7395, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.09777958851089835, | |
| "grad_norm": 6.437356472015381, | |
| "learning_rate": 9.808671713695803e-06, | |
| "loss": 0.6579, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.10592788755347321, | |
| "grad_norm": 6.3125762939453125, | |
| "learning_rate": 9.781142463867861e-06, | |
| "loss": 0.6969, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.11407618659604807, | |
| "grad_norm": 6.493209362030029, | |
| "learning_rate": 9.753613214039918e-06, | |
| "loss": 0.6338, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.12222448563862294, | |
| "grad_norm": 8.855860710144043, | |
| "learning_rate": 9.726083964211977e-06, | |
| "loss": 0.6657, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.1303727846811978, | |
| "grad_norm": 5.335328578948975, | |
| "learning_rate": 9.698554714384033e-06, | |
| "loss": 0.6906, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.13852108372377267, | |
| "grad_norm": 6.506489276885986, | |
| "learning_rate": 9.671025464556092e-06, | |
| "loss": 0.6485, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.14666938276634753, | |
| "grad_norm": 5.023345947265625, | |
| "learning_rate": 9.64349621472815e-06, | |
| "loss": 0.6291, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.15481768180892239, | |
| "grad_norm": 8.132065773010254, | |
| "learning_rate": 9.615966964900207e-06, | |
| "loss": 0.6553, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.16296598085149724, | |
| "grad_norm": 7.615749359130859, | |
| "learning_rate": 9.588437715072266e-06, | |
| "loss": 0.6635, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.1711142798940721, | |
| "grad_norm": 6.487279415130615, | |
| "learning_rate": 9.560908465244322e-06, | |
| "loss": 0.7554, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.179262578936647, | |
| "grad_norm": 6.409642219543457, | |
| "learning_rate": 9.53337921541638e-06, | |
| "loss": 0.5839, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.18741087797922185, | |
| "grad_norm": 5.239986419677734, | |
| "learning_rate": 9.505849965588438e-06, | |
| "loss": 0.7271, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.1955591770217967, | |
| "grad_norm": 5.452706336975098, | |
| "learning_rate": 9.478320715760496e-06, | |
| "loss": 0.6758, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.20370747606437156, | |
| "grad_norm": 4.827511787414551, | |
| "learning_rate": 9.450791465932555e-06, | |
| "loss": 0.6178, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.21185577510694642, | |
| "grad_norm": 6.320280075073242, | |
| "learning_rate": 9.423262216104611e-06, | |
| "loss": 0.6781, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.22000407414952128, | |
| "grad_norm": 5.648658275604248, | |
| "learning_rate": 9.39573296627667e-06, | |
| "loss": 0.6448, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.22815237319209614, | |
| "grad_norm": 6.806399822235107, | |
| "learning_rate": 9.368203716448727e-06, | |
| "loss": 0.6206, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.23630067223467102, | |
| "grad_norm": 6.556358337402344, | |
| "learning_rate": 9.340674466620785e-06, | |
| "loss": 0.6105, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.24444897127724588, | |
| "grad_norm": 5.21808385848999, | |
| "learning_rate": 9.313145216792842e-06, | |
| "loss": 0.651, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.2525972703198207, | |
| "grad_norm": 6.145657539367676, | |
| "learning_rate": 9.2856159669649e-06, | |
| "loss": 0.6504, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.2607455693623956, | |
| "grad_norm": 4.201127529144287, | |
| "learning_rate": 9.258086717136959e-06, | |
| "loss": 0.5885, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.2688938684049705, | |
| "grad_norm": 5.38640022277832, | |
| "learning_rate": 9.230557467309017e-06, | |
| "loss": 0.5888, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.27704216744754534, | |
| "grad_norm": 5.217910289764404, | |
| "learning_rate": 9.203028217481074e-06, | |
| "loss": 0.5901, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.2851904664901202, | |
| "grad_norm": 5.227973461151123, | |
| "learning_rate": 9.175498967653133e-06, | |
| "loss": 0.7178, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.29333876553269506, | |
| "grad_norm": 6.235045909881592, | |
| "learning_rate": 9.147969717825191e-06, | |
| "loss": 0.6711, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.3014870645752699, | |
| "grad_norm": 5.588578224182129, | |
| "learning_rate": 9.120440467997248e-06, | |
| "loss": 0.522, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.30963536361784477, | |
| "grad_norm": 5.49254035949707, | |
| "learning_rate": 9.092911218169306e-06, | |
| "loss": 0.5962, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.31778366266041963, | |
| "grad_norm": 5.4190239906311035, | |
| "learning_rate": 9.065381968341363e-06, | |
| "loss": 0.5981, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.3259319617029945, | |
| "grad_norm": 6.350463390350342, | |
| "learning_rate": 9.037852718513422e-06, | |
| "loss": 0.6114, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.33408026074556935, | |
| "grad_norm": 6.231777191162109, | |
| "learning_rate": 9.010323468685478e-06, | |
| "loss": 0.6729, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.3422285597881442, | |
| "grad_norm": 6.590610027313232, | |
| "learning_rate": 8.982794218857537e-06, | |
| "loss": 0.6124, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.35037685883071906, | |
| "grad_norm": 5.504190444946289, | |
| "learning_rate": 8.955264969029595e-06, | |
| "loss": 0.6555, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.358525157873294, | |
| "grad_norm": 4.821358680725098, | |
| "learning_rate": 8.927735719201652e-06, | |
| "loss": 0.5743, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.36667345691586883, | |
| "grad_norm": 6.006564140319824, | |
| "learning_rate": 8.90020646937371e-06, | |
| "loss": 0.6157, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.3748217559584437, | |
| "grad_norm": 5.9578471183776855, | |
| "learning_rate": 8.872677219545767e-06, | |
| "loss": 0.6485, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.38297005500101855, | |
| "grad_norm": 5.149762153625488, | |
| "learning_rate": 8.845147969717826e-06, | |
| "loss": 0.5951, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.3911183540435934, | |
| "grad_norm": 6.143222332000732, | |
| "learning_rate": 8.817618719889883e-06, | |
| "loss": 0.5805, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.39926665308616827, | |
| "grad_norm": 5.496342658996582, | |
| "learning_rate": 8.790089470061941e-06, | |
| "loss": 0.6158, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.4074149521287431, | |
| "grad_norm": 5.14535665512085, | |
| "learning_rate": 8.762560220234e-06, | |
| "loss": 0.6036, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.415563251171318, | |
| "grad_norm": 5.73581600189209, | |
| "learning_rate": 8.735030970406058e-06, | |
| "loss": 0.5925, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.42371155021389284, | |
| "grad_norm": 4.455753326416016, | |
| "learning_rate": 8.707501720578115e-06, | |
| "loss": 0.5894, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.4318598492564677, | |
| "grad_norm": 6.366560935974121, | |
| "learning_rate": 8.679972470750173e-06, | |
| "loss": 0.6424, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.44000814829904256, | |
| "grad_norm": 5.11430025100708, | |
| "learning_rate": 8.652443220922232e-06, | |
| "loss": 0.5731, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.4481564473416174, | |
| "grad_norm": 6.0588812828063965, | |
| "learning_rate": 8.624913971094288e-06, | |
| "loss": 0.6257, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.4563047463841923, | |
| "grad_norm": 6.879603385925293, | |
| "learning_rate": 8.597384721266347e-06, | |
| "loss": 0.6427, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.4644530454267672, | |
| "grad_norm": 4.6673502922058105, | |
| "learning_rate": 8.569855471438404e-06, | |
| "loss": 0.5763, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.47260134446934204, | |
| "grad_norm": 5.316527843475342, | |
| "learning_rate": 8.542326221610462e-06, | |
| "loss": 0.6207, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.4807496435119169, | |
| "grad_norm": 5.4205780029296875, | |
| "learning_rate": 8.514796971782519e-06, | |
| "loss": 0.5435, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.48889794255449176, | |
| "grad_norm": 4.579510688781738, | |
| "learning_rate": 8.487267721954577e-06, | |
| "loss": 0.5937, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.4970462415970666, | |
| "grad_norm": 6.327101707458496, | |
| "learning_rate": 8.459738472126636e-06, | |
| "loss": 0.6186, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.5051945406396414, | |
| "grad_norm": 5.347973346710205, | |
| "learning_rate": 8.432209222298693e-06, | |
| "loss": 0.564, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.5133428396822164, | |
| "grad_norm": 5.44881010055542, | |
| "learning_rate": 8.404679972470751e-06, | |
| "loss": 0.5221, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.5214911387247912, | |
| "grad_norm": 5.802955150604248, | |
| "learning_rate": 8.377150722642808e-06, | |
| "loss": 0.5893, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.5296394377673661, | |
| "grad_norm": 5.931309223175049, | |
| "learning_rate": 8.349621472814866e-06, | |
| "loss": 0.6306, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.537787736809941, | |
| "grad_norm": 5.793541431427002, | |
| "learning_rate": 8.322092222986923e-06, | |
| "loss": 0.6199, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.5459360358525158, | |
| "grad_norm": 4.972778797149658, | |
| "learning_rate": 8.294562973158982e-06, | |
| "loss": 0.4771, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.5540843348950907, | |
| "grad_norm": 5.573855876922607, | |
| "learning_rate": 8.26703372333104e-06, | |
| "loss": 0.5054, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.5622326339376655, | |
| "grad_norm": 4.7720947265625, | |
| "learning_rate": 8.239504473503099e-06, | |
| "loss": 0.5599, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.5703809329802404, | |
| "grad_norm": 6.338388442993164, | |
| "learning_rate": 8.211975223675155e-06, | |
| "loss": 0.5612, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.5785292320228153, | |
| "grad_norm": 6.1034464836120605, | |
| "learning_rate": 8.184445973847214e-06, | |
| "loss": 0.5662, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.5866775310653901, | |
| "grad_norm": 5.423385143280029, | |
| "learning_rate": 8.15691672401927e-06, | |
| "loss": 0.6246, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.594825830107965, | |
| "grad_norm": 6.507946014404297, | |
| "learning_rate": 8.12938747419133e-06, | |
| "loss": 0.5879, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.6029741291505398, | |
| "grad_norm": 5.927179336547852, | |
| "learning_rate": 8.101858224363388e-06, | |
| "loss": 0.5479, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.6111224281931147, | |
| "grad_norm": 6.775645732879639, | |
| "learning_rate": 8.074328974535444e-06, | |
| "loss": 0.5809, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.6192707272356895, | |
| "grad_norm": 4.263314247131348, | |
| "learning_rate": 8.046799724707503e-06, | |
| "loss": 0.55, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.6274190262782644, | |
| "grad_norm": 4.671892166137695, | |
| "learning_rate": 8.01927047487956e-06, | |
| "loss": 0.529, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.6355673253208393, | |
| "grad_norm": 5.7395501136779785, | |
| "learning_rate": 7.991741225051618e-06, | |
| "loss": 0.5349, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.6437156243634141, | |
| "grad_norm": 5.88397216796875, | |
| "learning_rate": 7.964211975223675e-06, | |
| "loss": 0.5675, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.651863923405989, | |
| "grad_norm": 5.037132263183594, | |
| "learning_rate": 7.936682725395733e-06, | |
| "loss": 0.5838, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.6600122224485638, | |
| "grad_norm": 6.442205429077148, | |
| "learning_rate": 7.909153475567792e-06, | |
| "loss": 0.5275, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.6681605214911387, | |
| "grad_norm": 4.626034259796143, | |
| "learning_rate": 7.881624225739849e-06, | |
| "loss": 0.5251, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.6763088205337136, | |
| "grad_norm": 6.96089506149292, | |
| "learning_rate": 7.854094975911907e-06, | |
| "loss": 0.589, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.6844571195762884, | |
| "grad_norm": 5.303503036499023, | |
| "learning_rate": 7.826565726083964e-06, | |
| "loss": 0.6233, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.6926054186188633, | |
| "grad_norm": 6.015293121337891, | |
| "learning_rate": 7.799036476256022e-06, | |
| "loss": 0.5517, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.7007537176614381, | |
| "grad_norm": 5.348663806915283, | |
| "learning_rate": 7.77150722642808e-06, | |
| "loss": 0.5403, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.708902016704013, | |
| "grad_norm": 4.189513206481934, | |
| "learning_rate": 7.743977976600138e-06, | |
| "loss": 0.5817, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.717050315746588, | |
| "grad_norm": 7.446457386016846, | |
| "learning_rate": 7.716448726772196e-06, | |
| "loss": 0.5437, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.7251986147891628, | |
| "grad_norm": 4.759880542755127, | |
| "learning_rate": 7.688919476944255e-06, | |
| "loss": 0.5648, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.7333469138317377, | |
| "grad_norm": 5.6719584465026855, | |
| "learning_rate": 7.661390227116311e-06, | |
| "loss": 0.6118, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.7414952128743125, | |
| "grad_norm": 4.943815231323242, | |
| "learning_rate": 7.63386097728837e-06, | |
| "loss": 0.5326, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.7496435119168874, | |
| "grad_norm": 3.6254968643188477, | |
| "learning_rate": 7.6063317274604275e-06, | |
| "loss": 0.5087, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.7577918109594622, | |
| "grad_norm": 5.9594526290893555, | |
| "learning_rate": 7.578802477632485e-06, | |
| "loss": 0.532, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.7659401100020371, | |
| "grad_norm": 5.103460311889648, | |
| "learning_rate": 7.551273227804543e-06, | |
| "loss": 0.553, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.774088409044612, | |
| "grad_norm": 6.327749729156494, | |
| "learning_rate": 7.5237439779766004e-06, | |
| "loss": 0.5767, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.7822367080871868, | |
| "grad_norm": 5.8910088539123535, | |
| "learning_rate": 7.496214728148659e-06, | |
| "loss": 0.5948, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.7903850071297617, | |
| "grad_norm": 5.571481227874756, | |
| "learning_rate": 7.4686854783207165e-06, | |
| "loss": 0.5199, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.7985333061723365, | |
| "grad_norm": 4.894184589385986, | |
| "learning_rate": 7.441156228492774e-06, | |
| "loss": 0.5577, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.8066816052149114, | |
| "grad_norm": 5.402436256408691, | |
| "learning_rate": 7.413626978664832e-06, | |
| "loss": 0.5395, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.8148299042574862, | |
| "grad_norm": 5.212170600891113, | |
| "learning_rate": 7.3860977288368894e-06, | |
| "loss": 0.5319, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.8148299042574862, | |
| "eval_loss": 0.4680393934249878, | |
| "eval_runtime": 10851.5751, | |
| "eval_samples_per_second": 1.81, | |
| "eval_steps_per_second": 0.226, | |
| "eval_wer": 38.607193210094984, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.8229782033000611, | |
| "grad_norm": 5.673357963562012, | |
| "learning_rate": 7.358568479008947e-06, | |
| "loss": 0.5641, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.831126502342636, | |
| "grad_norm": 4.529426097869873, | |
| "learning_rate": 7.331039229181005e-06, | |
| "loss": 0.489, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.8392748013852108, | |
| "grad_norm": 4.871108055114746, | |
| "learning_rate": 7.303509979353063e-06, | |
| "loss": 0.5413, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.8474231004277857, | |
| "grad_norm": 4.407359600067139, | |
| "learning_rate": 7.275980729525121e-06, | |
| "loss": 0.4993, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.8555713994703605, | |
| "grad_norm": 4.965864181518555, | |
| "learning_rate": 7.248451479697178e-06, | |
| "loss": 0.5825, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.8637196985129354, | |
| "grad_norm": 6.825056552886963, | |
| "learning_rate": 7.220922229869237e-06, | |
| "loss": 0.5672, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.8718679975555103, | |
| "grad_norm": 5.5648040771484375, | |
| "learning_rate": 7.1933929800412945e-06, | |
| "loss": 0.5188, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.8800162965980851, | |
| "grad_norm": 6.5213799476623535, | |
| "learning_rate": 7.165863730213353e-06, | |
| "loss": 0.5806, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.88816459564066, | |
| "grad_norm": 5.0149993896484375, | |
| "learning_rate": 7.138334480385411e-06, | |
| "loss": 0.5729, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.8963128946832348, | |
| "grad_norm": 4.783459186553955, | |
| "learning_rate": 7.110805230557468e-06, | |
| "loss": 0.4991, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.9044611937258097, | |
| "grad_norm": 5.186647891998291, | |
| "learning_rate": 7.083275980729526e-06, | |
| "loss": 0.6553, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.9126094927683845, | |
| "grad_norm": 4.135760307312012, | |
| "learning_rate": 7.0557467309015835e-06, | |
| "loss": 0.5947, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.9207577918109595, | |
| "grad_norm": 5.20528507232666, | |
| "learning_rate": 7.028217481073641e-06, | |
| "loss": 0.5454, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.9289060908535344, | |
| "grad_norm": 5.948816299438477, | |
| "learning_rate": 7.000688231245699e-06, | |
| "loss": 0.5475, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.9370543898961092, | |
| "grad_norm": 5.592283248901367, | |
| "learning_rate": 6.973158981417757e-06, | |
| "loss": 0.5153, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.9452026889386841, | |
| "grad_norm": 5.162081241607666, | |
| "learning_rate": 6.945629731589815e-06, | |
| "loss": 0.4637, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.953350987981259, | |
| "grad_norm": 5.888420104980469, | |
| "learning_rate": 6.9181004817618725e-06, | |
| "loss": 0.4956, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.9614992870238338, | |
| "grad_norm": 6.354236602783203, | |
| "learning_rate": 6.89057123193393e-06, | |
| "loss": 0.4912, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.9696475860664087, | |
| "grad_norm": 4.269996166229248, | |
| "learning_rate": 6.863041982105988e-06, | |
| "loss": 0.489, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.9777958851089835, | |
| "grad_norm": 5.822803974151611, | |
| "learning_rate": 6.835512732278045e-06, | |
| "loss": 0.55, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.9859441841515584, | |
| "grad_norm": 4.092251777648926, | |
| "learning_rate": 6.807983482450103e-06, | |
| "loss": 0.5625, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.9940924831941332, | |
| "grad_norm": 6.759193420410156, | |
| "learning_rate": 6.7804542326221615e-06, | |
| "loss": 0.5363, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 1.0020370747606437, | |
| "grad_norm": 5.126402378082275, | |
| "learning_rate": 6.752924982794219e-06, | |
| "loss": 0.535, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 1.0101853738032185, | |
| "grad_norm": 5.261082172393799, | |
| "learning_rate": 6.725395732966277e-06, | |
| "loss": 0.4262, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 1.0183336728457935, | |
| "grad_norm": 4.407663345336914, | |
| "learning_rate": 6.697866483138335e-06, | |
| "loss": 0.4076, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.0264819718883682, | |
| "grad_norm": 5.785089492797852, | |
| "learning_rate": 6.670337233310394e-06, | |
| "loss": 0.4255, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 1.0346302709309432, | |
| "grad_norm": 4.659370422363281, | |
| "learning_rate": 6.642807983482451e-06, | |
| "loss": 0.4116, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 1.042778569973518, | |
| "grad_norm": 5.880095958709717, | |
| "learning_rate": 6.615278733654509e-06, | |
| "loss": 0.4943, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 1.050926869016093, | |
| "grad_norm": 4.94119930267334, | |
| "learning_rate": 6.587749483826567e-06, | |
| "loss": 0.3904, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 1.0590751680586679, | |
| "grad_norm": 4.008668422698975, | |
| "learning_rate": 6.560220233998624e-06, | |
| "loss": 0.403, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.0672234671012426, | |
| "grad_norm": 6.079473495483398, | |
| "learning_rate": 6.532690984170682e-06, | |
| "loss": 0.4833, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 1.0753717661438176, | |
| "grad_norm": 5.263147830963135, | |
| "learning_rate": 6.5051617343427395e-06, | |
| "loss": 0.4423, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 1.0835200651863923, | |
| "grad_norm": 4.1324543952941895, | |
| "learning_rate": 6.477632484514798e-06, | |
| "loss": 0.4631, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 1.0916683642289673, | |
| "grad_norm": 3.648576259613037, | |
| "learning_rate": 6.450103234686856e-06, | |
| "loss": 0.4002, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 1.099816663271542, | |
| "grad_norm": 5.458010673522949, | |
| "learning_rate": 6.422573984858913e-06, | |
| "loss": 0.4572, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.107964962314117, | |
| "grad_norm": 5.379840850830078, | |
| "learning_rate": 6.395044735030971e-06, | |
| "loss": 0.4283, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 1.1161132613566918, | |
| "grad_norm": 5.2092108726501465, | |
| "learning_rate": 6.3675154852030285e-06, | |
| "loss": 0.4384, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 1.1242615603992667, | |
| "grad_norm": 5.582503795623779, | |
| "learning_rate": 6.339986235375086e-06, | |
| "loss": 0.4202, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 1.1324098594418415, | |
| "grad_norm": 4.186822891235352, | |
| "learning_rate": 6.312456985547144e-06, | |
| "loss": 0.3972, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 1.1405581584844164, | |
| "grad_norm": 5.029235363006592, | |
| "learning_rate": 6.284927735719201e-06, | |
| "loss": 0.4368, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.1487064575269912, | |
| "grad_norm": 4.080201148986816, | |
| "learning_rate": 6.25739848589126e-06, | |
| "loss": 0.4561, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 1.1568547565695662, | |
| "grad_norm": 5.171266078948975, | |
| "learning_rate": 6.2298692360633175e-06, | |
| "loss": 0.4594, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 1.165003055612141, | |
| "grad_norm": 6.01788854598999, | |
| "learning_rate": 6.202339986235376e-06, | |
| "loss": 0.363, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 1.1731513546547159, | |
| "grad_norm": 5.374858379364014, | |
| "learning_rate": 6.1748107364074336e-06, | |
| "loss": 0.4782, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 1.1812996536972906, | |
| "grad_norm": 4.696728706359863, | |
| "learning_rate": 6.147281486579492e-06, | |
| "loss": 0.4598, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.1894479527398656, | |
| "grad_norm": 4.785463333129883, | |
| "learning_rate": 6.11975223675155e-06, | |
| "loss": 0.4502, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 1.1975962517824403, | |
| "grad_norm": 4.824402809143066, | |
| "learning_rate": 6.092222986923607e-06, | |
| "loss": 0.4216, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 1.2057445508250153, | |
| "grad_norm": 4.554792404174805, | |
| "learning_rate": 6.064693737095665e-06, | |
| "loss": 0.39, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 1.21389284986759, | |
| "grad_norm": 5.259772300720215, | |
| "learning_rate": 6.0371644872677226e-06, | |
| "loss": 0.4199, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 1.222041148910165, | |
| "grad_norm": 4.8776140213012695, | |
| "learning_rate": 6.00963523743978e-06, | |
| "loss": 0.4742, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.23018944795274, | |
| "grad_norm": 5.5702033042907715, | |
| "learning_rate": 5.982105987611838e-06, | |
| "loss": 0.3934, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 1.2383377469953147, | |
| "grad_norm": 5.325311660766602, | |
| "learning_rate": 5.954576737783896e-06, | |
| "loss": 0.3655, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 1.2464860460378895, | |
| "grad_norm": 3.9275801181793213, | |
| "learning_rate": 5.927047487955954e-06, | |
| "loss": 0.3965, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 1.2546343450804645, | |
| "grad_norm": 4.71364688873291, | |
| "learning_rate": 5.8995182381280116e-06, | |
| "loss": 0.4026, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 1.2627826441230394, | |
| "grad_norm": 4.507187843322754, | |
| "learning_rate": 5.871988988300069e-06, | |
| "loss": 0.4004, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.2709309431656142, | |
| "grad_norm": 4.939427375793457, | |
| "learning_rate": 5.844459738472127e-06, | |
| "loss": 0.407, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 1.279079242208189, | |
| "grad_norm": 5.642360210418701, | |
| "learning_rate": 5.8169304886441844e-06, | |
| "loss": 0.4273, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 1.2872275412507639, | |
| "grad_norm": 4.531342506408691, | |
| "learning_rate": 5.789401238816242e-06, | |
| "loss": 0.4455, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 1.2953758402933389, | |
| "grad_norm": 5.926376819610596, | |
| "learning_rate": 5.7618719889883005e-06, | |
| "loss": 0.4627, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 1.3035241393359136, | |
| "grad_norm": 3.818305730819702, | |
| "learning_rate": 5.734342739160358e-06, | |
| "loss": 0.3957, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.3116724383784886, | |
| "grad_norm": 4.200014114379883, | |
| "learning_rate": 5.706813489332416e-06, | |
| "loss": 0.4516, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 1.3198207374210633, | |
| "grad_norm": 5.380039691925049, | |
| "learning_rate": 5.679284239504474e-06, | |
| "loss": 0.4775, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 1.3279690364636383, | |
| "grad_norm": 6.382922172546387, | |
| "learning_rate": 5.651754989676532e-06, | |
| "loss": 0.3792, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 1.336117335506213, | |
| "grad_norm": 5.670558929443359, | |
| "learning_rate": 5.62422573984859e-06, | |
| "loss": 0.4301, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 1.344265634548788, | |
| "grad_norm": 4.901285171508789, | |
| "learning_rate": 5.596696490020648e-06, | |
| "loss": 0.4095, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.3524139335913627, | |
| "grad_norm": 5.272710800170898, | |
| "learning_rate": 5.569167240192706e-06, | |
| "loss": 0.3585, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 1.3605622326339377, | |
| "grad_norm": 5.442037582397461, | |
| "learning_rate": 5.541637990364763e-06, | |
| "loss": 0.4019, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 1.3687105316765125, | |
| "grad_norm": 4.817733287811279, | |
| "learning_rate": 5.514108740536821e-06, | |
| "loss": 0.368, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 1.3768588307190874, | |
| "grad_norm": 4.459897994995117, | |
| "learning_rate": 5.4865794907088785e-06, | |
| "loss": 0.3646, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 1.3850071297616622, | |
| "grad_norm": 5.927896976470947, | |
| "learning_rate": 5.459050240880936e-06, | |
| "loss": 0.4448, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.3931554288042372, | |
| "grad_norm": 5.564925193786621, | |
| "learning_rate": 5.431520991052995e-06, | |
| "loss": 0.3794, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 1.401303727846812, | |
| "grad_norm": 4.68287467956543, | |
| "learning_rate": 5.403991741225052e-06, | |
| "loss": 0.3931, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 1.4094520268893869, | |
| "grad_norm": 5.004433631896973, | |
| "learning_rate": 5.37646249139711e-06, | |
| "loss": 0.432, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 1.4176003259319616, | |
| "grad_norm": 5.418098449707031, | |
| "learning_rate": 5.3489332415691675e-06, | |
| "loss": 0.4533, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 1.4257486249745366, | |
| "grad_norm": 4.825730800628662, | |
| "learning_rate": 5.321403991741225e-06, | |
| "loss": 0.4159, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.4338969240171116, | |
| "grad_norm": 5.892017364501953, | |
| "learning_rate": 5.293874741913283e-06, | |
| "loss": 0.394, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 1.4420452230596863, | |
| "grad_norm": 5.600419521331787, | |
| "learning_rate": 5.26634549208534e-06, | |
| "loss": 0.4357, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 1.450193522102261, | |
| "grad_norm": 4.615939617156982, | |
| "learning_rate": 5.238816242257399e-06, | |
| "loss": 0.4029, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 1.458341821144836, | |
| "grad_norm": 4.384542465209961, | |
| "learning_rate": 5.2112869924294565e-06, | |
| "loss": 0.4042, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 1.466490120187411, | |
| "grad_norm": 5.981232643127441, | |
| "learning_rate": 5.183757742601514e-06, | |
| "loss": 0.3923, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.4746384192299857, | |
| "grad_norm": 5.077232837677002, | |
| "learning_rate": 5.156228492773573e-06, | |
| "loss": 0.4995, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 1.4827867182725605, | |
| "grad_norm": 5.000918865203857, | |
| "learning_rate": 5.12869924294563e-06, | |
| "loss": 0.4167, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 1.4909350173151354, | |
| "grad_norm": 2.953489303588867, | |
| "learning_rate": 5.101169993117689e-06, | |
| "loss": 0.3798, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 1.4990833163577104, | |
| "grad_norm": 5.880030632019043, | |
| "learning_rate": 5.073640743289746e-06, | |
| "loss": 0.4349, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 1.5072316154002852, | |
| "grad_norm": 5.726639270782471, | |
| "learning_rate": 5.046111493461804e-06, | |
| "loss": 0.3809, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.51537991444286, | |
| "grad_norm": 4.755875587463379, | |
| "learning_rate": 5.018582243633862e-06, | |
| "loss": 0.4345, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 1.5235282134854349, | |
| "grad_norm": 3.825904130935669, | |
| "learning_rate": 4.991052993805919e-06, | |
| "loss": 0.4304, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 1.5316765125280098, | |
| "grad_norm": 6.1705217361450195, | |
| "learning_rate": 4.963523743977977e-06, | |
| "loss": 0.4224, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 1.5398248115705846, | |
| "grad_norm": 5.098936557769775, | |
| "learning_rate": 4.9359944941500345e-06, | |
| "loss": 0.4254, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 1.5479731106131593, | |
| "grad_norm": 4.633610248565674, | |
| "learning_rate": 4.908465244322093e-06, | |
| "loss": 0.3542, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.5561214096557343, | |
| "grad_norm": 5.469082832336426, | |
| "learning_rate": 4.880935994494151e-06, | |
| "loss": 0.3577, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 1.5642697086983093, | |
| "grad_norm": 4.476110458374023, | |
| "learning_rate": 4.853406744666208e-06, | |
| "loss": 0.3931, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 1.5724180077408842, | |
| "grad_norm": 4.693641185760498, | |
| "learning_rate": 4.825877494838266e-06, | |
| "loss": 0.3793, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 1.580566306783459, | |
| "grad_norm": 5.025706768035889, | |
| "learning_rate": 4.7983482450103235e-06, | |
| "loss": 0.4152, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 1.5887146058260337, | |
| "grad_norm": 5.075003147125244, | |
| "learning_rate": 4.770818995182382e-06, | |
| "loss": 0.3813, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.5968629048686087, | |
| "grad_norm": 4.8126115798950195, | |
| "learning_rate": 4.74328974535444e-06, | |
| "loss": 0.408, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 1.6050112039111837, | |
| "grad_norm": 4.3969340324401855, | |
| "learning_rate": 4.715760495526497e-06, | |
| "loss": 0.3667, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 1.6131595029537584, | |
| "grad_norm": 4.514830589294434, | |
| "learning_rate": 4.688231245698555e-06, | |
| "loss": 0.4022, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 1.6213078019963332, | |
| "grad_norm": 5.9185309410095215, | |
| "learning_rate": 4.6607019958706125e-06, | |
| "loss": 0.4277, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 1.6294561010389081, | |
| "grad_norm": 4.446202754974365, | |
| "learning_rate": 4.633172746042671e-06, | |
| "loss": 0.4179, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.6294561010389081, | |
| "eval_loss": 0.3749663531780243, | |
| "eval_runtime": 13864.4556, | |
| "eval_samples_per_second": 1.416, | |
| "eval_steps_per_second": 0.177, | |
| "eval_wer": 60.07223542231883, | |
| "step": 4000 | |
| } | |
| ], | |
| "logging_steps": 20, | |
| "max_steps": 7365, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 2000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.69331595329536e+19, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |