| { | |
| "best_global_step": 1000, | |
| "best_metric": 35.75192541548439, | |
| "best_model_checkpoint": "./whisper-large-v2/In_house_data/checkpoint-1000", | |
| "epoch": 2.7482806052269604, | |
| "eval_steps": 500, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.013755158184319119, | |
| "grad_norm": 21.1955509185791, | |
| "learning_rate": 2.666666666666667e-07, | |
| "loss": 3.905, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.027510316368638238, | |
| "grad_norm": 11.807918548583984, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 3.6009, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04126547455295736, | |
| "grad_norm": NaN, | |
| "learning_rate": 9.333333333333334e-07, | |
| "loss": 3.5008, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.055020632737276476, | |
| "grad_norm": 12.797011375427246, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 3.2938, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0687757909215956, | |
| "grad_norm": 11.184866905212402, | |
| "learning_rate": 1.5333333333333334e-06, | |
| "loss": 2.9611, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.08253094910591471, | |
| "grad_norm": 12.54674243927002, | |
| "learning_rate": 1.8666666666666669e-06, | |
| "loss": 2.5185, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.09628610729023383, | |
| "grad_norm": 6.0443196296691895, | |
| "learning_rate": 2.2e-06, | |
| "loss": 2.433, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.11004126547455295, | |
| "grad_norm": 5.060392379760742, | |
| "learning_rate": 2.5333333333333338e-06, | |
| "loss": 2.1781, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.12379642365887207, | |
| "grad_norm": 4.572280406951904, | |
| "learning_rate": 2.8000000000000003e-06, | |
| "loss": 2.0818, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.1375515818431912, | |
| "grad_norm": 5.422489643096924, | |
| "learning_rate": 3.133333333333334e-06, | |
| "loss": 2.1037, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.15130674002751032, | |
| "grad_norm": 4.844382286071777, | |
| "learning_rate": 3.4666666666666672e-06, | |
| "loss": 1.9358, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.16506189821182943, | |
| "grad_norm": 9.222413063049316, | |
| "learning_rate": 3.8000000000000005e-06, | |
| "loss": 1.8682, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.17881705639614856, | |
| "grad_norm": 4.705779075622559, | |
| "learning_rate": 4.133333333333333e-06, | |
| "loss": 1.6894, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.19257221458046767, | |
| "grad_norm": 9.01275634765625, | |
| "learning_rate": 4.4666666666666665e-06, | |
| "loss": 1.4863, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.2063273727647868, | |
| "grad_norm": 21.184419631958008, | |
| "learning_rate": 4.7333333333333335e-06, | |
| "loss": 1.4149, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.2200825309491059, | |
| "grad_norm": 17.06338119506836, | |
| "learning_rate": 5.0666666666666676e-06, | |
| "loss": 1.3539, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.23383768913342504, | |
| "grad_norm": 4.059953689575195, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 1.3184, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.24759284731774414, | |
| "grad_norm": 3.739663600921631, | |
| "learning_rate": 5.733333333333334e-06, | |
| "loss": 1.2847, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.2613480055020633, | |
| "grad_norm": 6.212688446044922, | |
| "learning_rate": 6.066666666666667e-06, | |
| "loss": 1.2833, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.2751031636863824, | |
| "grad_norm": 7.45237922668457, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 1.2058, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.28885832187070154, | |
| "grad_norm": 3.6827175617218018, | |
| "learning_rate": 6.733333333333334e-06, | |
| "loss": 1.1704, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.30261348005502064, | |
| "grad_norm": 3.8309476375579834, | |
| "learning_rate": 7.066666666666667e-06, | |
| "loss": 1.1297, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.31636863823933975, | |
| "grad_norm": 4.041064739227295, | |
| "learning_rate": 7.4e-06, | |
| "loss": 1.1995, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.33012379642365886, | |
| "grad_norm": 4.068305969238281, | |
| "learning_rate": 7.733333333333334e-06, | |
| "loss": 1.176, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.343878954607978, | |
| "grad_norm": 11.103217124938965, | |
| "learning_rate": 8.066666666666667e-06, | |
| "loss": 1.1701, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.3576341127922971, | |
| "grad_norm": 3.5990092754364014, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 1.0819, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.3713892709766162, | |
| "grad_norm": 4.472628593444824, | |
| "learning_rate": 8.733333333333333e-06, | |
| "loss": 1.1597, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.38514442916093533, | |
| "grad_norm": 4.027596473693848, | |
| "learning_rate": 9.066666666666667e-06, | |
| "loss": 1.1287, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.3988995873452545, | |
| "grad_norm": 3.919308662414551, | |
| "learning_rate": 9.4e-06, | |
| "loss": 1.1378, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.4126547455295736, | |
| "grad_norm": 3.2769739627838135, | |
| "learning_rate": 9.733333333333334e-06, | |
| "loss": 1.0651, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4264099037138927, | |
| "grad_norm": 3.678276300430298, | |
| "learning_rate": 9.997701149425289e-06, | |
| "loss": 1.0233, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.4401650618982118, | |
| "grad_norm": 3.9921329021453857, | |
| "learning_rate": 9.986206896551724e-06, | |
| "loss": 0.9842, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.45392022008253097, | |
| "grad_norm": 4.81342077255249, | |
| "learning_rate": 9.974712643678162e-06, | |
| "loss": 0.9774, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.4676753782668501, | |
| "grad_norm": 3.1831002235412598, | |
| "learning_rate": 9.963218390804599e-06, | |
| "loss": 0.9691, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.4814305364511692, | |
| "grad_norm": 3.0080406665802, | |
| "learning_rate": 9.951724137931035e-06, | |
| "loss": 0.8912, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.4951856946354883, | |
| "grad_norm": 4.207084655761719, | |
| "learning_rate": 9.940229885057472e-06, | |
| "loss": 0.9708, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.5089408528198074, | |
| "grad_norm": 3.468383550643921, | |
| "learning_rate": 9.928735632183909e-06, | |
| "loss": 0.8915, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.5226960110041265, | |
| "grad_norm": 3.9594309329986572, | |
| "learning_rate": 9.917241379310347e-06, | |
| "loss": 0.9415, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.5364511691884457, | |
| "grad_norm": 3.3678927421569824, | |
| "learning_rate": 9.905747126436782e-06, | |
| "loss": 0.8949, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.5502063273727648, | |
| "grad_norm": 3.1255943775177, | |
| "learning_rate": 9.89425287356322e-06, | |
| "loss": 0.9411, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5639614855570839, | |
| "grad_norm": 3.340153694152832, | |
| "learning_rate": 9.882758620689657e-06, | |
| "loss": 0.9468, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.5777166437414031, | |
| "grad_norm": 2.989037275314331, | |
| "learning_rate": 9.871264367816093e-06, | |
| "loss": 0.8987, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.5914718019257221, | |
| "grad_norm": 3.296373128890991, | |
| "learning_rate": 9.85977011494253e-06, | |
| "loss": 0.9369, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.6052269601100413, | |
| "grad_norm": 4.3851189613342285, | |
| "learning_rate": 9.848275862068966e-06, | |
| "loss": 0.8394, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.6189821182943603, | |
| "grad_norm": 3.8930516242980957, | |
| "learning_rate": 9.836781609195403e-06, | |
| "loss": 0.7941, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.6327372764786795, | |
| "grad_norm": 3.3190417289733887, | |
| "learning_rate": 9.82528735632184e-06, | |
| "loss": 0.9271, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.6464924346629987, | |
| "grad_norm": 2.888249397277832, | |
| "learning_rate": 9.813793103448276e-06, | |
| "loss": 0.8505, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.6602475928473177, | |
| "grad_norm": 2.989713668823242, | |
| "learning_rate": 9.802298850574713e-06, | |
| "loss": 0.833, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.6740027510316369, | |
| "grad_norm": 3.96846866607666, | |
| "learning_rate": 9.79080459770115e-06, | |
| "loss": 0.8411, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.687757909215956, | |
| "grad_norm": 2.885392665863037, | |
| "learning_rate": 9.779310344827588e-06, | |
| "loss": 0.8882, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.7015130674002751, | |
| "grad_norm": 3.2487101554870605, | |
| "learning_rate": 9.767816091954022e-06, | |
| "loss": 0.8684, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.7152682255845942, | |
| "grad_norm": 3.253427505493164, | |
| "learning_rate": 9.75632183908046e-06, | |
| "loss": 0.8147, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.7290233837689133, | |
| "grad_norm": 6.355091094970703, | |
| "learning_rate": 9.744827586206897e-06, | |
| "loss": 0.7958, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.7427785419532325, | |
| "grad_norm": 4.210458278656006, | |
| "learning_rate": 9.733333333333334e-06, | |
| "loss": 0.7896, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.7565337001375516, | |
| "grad_norm": 3.744110584259033, | |
| "learning_rate": 9.72183908045977e-06, | |
| "loss": 0.8067, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.7702888583218707, | |
| "grad_norm": 3.912106513977051, | |
| "learning_rate": 9.710344827586207e-06, | |
| "loss": 0.8419, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.7840440165061898, | |
| "grad_norm": 10.62960433959961, | |
| "learning_rate": 9.698850574712645e-06, | |
| "loss": 0.7707, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.797799174690509, | |
| "grad_norm": 3.216627836227417, | |
| "learning_rate": 9.68735632183908e-06, | |
| "loss": 0.8188, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.811554332874828, | |
| "grad_norm": 16.345537185668945, | |
| "learning_rate": 9.675862068965518e-06, | |
| "loss": 0.7763, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.8253094910591472, | |
| "grad_norm": 3.217355489730835, | |
| "learning_rate": 9.664367816091955e-06, | |
| "loss": 0.8194, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.8390646492434664, | |
| "grad_norm": 3.4968042373657227, | |
| "learning_rate": 9.652873563218392e-06, | |
| "loss": 0.8462, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.8528198074277854, | |
| "grad_norm": 3.1585259437561035, | |
| "learning_rate": 9.641379310344828e-06, | |
| "loss": 0.7598, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.8665749656121046, | |
| "grad_norm": 2.978710651397705, | |
| "learning_rate": 9.629885057471265e-06, | |
| "loss": 0.788, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.8803301237964236, | |
| "grad_norm": 3.119516134262085, | |
| "learning_rate": 9.618390804597701e-06, | |
| "loss": 0.761, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.8940852819807428, | |
| "grad_norm": 2.88628888130188, | |
| "learning_rate": 9.606896551724138e-06, | |
| "loss": 0.7441, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.9078404401650619, | |
| "grad_norm": 2.832181453704834, | |
| "learning_rate": 9.595402298850576e-06, | |
| "loss": 0.7509, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.921595598349381, | |
| "grad_norm": 3.500303268432617, | |
| "learning_rate": 9.583908045977011e-06, | |
| "loss": 0.7589, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.9353507565337001, | |
| "grad_norm": 2.673982858657837, | |
| "learning_rate": 9.57241379310345e-06, | |
| "loss": 0.7677, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.9491059147180193, | |
| "grad_norm": 3.101490020751953, | |
| "learning_rate": 9.560919540229886e-06, | |
| "loss": 0.796, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.9628610729023384, | |
| "grad_norm": 2.6055097579956055, | |
| "learning_rate": 9.549425287356323e-06, | |
| "loss": 0.75, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.9766162310866575, | |
| "grad_norm": 4.560893535614014, | |
| "learning_rate": 9.537931034482759e-06, | |
| "loss": 0.7543, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.9903713892709766, | |
| "grad_norm": 2.834599733352661, | |
| "learning_rate": 9.526436781609196e-06, | |
| "loss": 0.7355, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.0027510316368637, | |
| "grad_norm": 3.335203170776367, | |
| "learning_rate": 9.514942528735634e-06, | |
| "loss": 0.6308, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.016506189821183, | |
| "grad_norm": 2.8525352478027344, | |
| "learning_rate": 9.503448275862069e-06, | |
| "loss": 0.6704, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.030261348005502, | |
| "grad_norm": 3.6981310844421387, | |
| "learning_rate": 9.491954022988507e-06, | |
| "loss": 0.6066, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.0440165061898212, | |
| "grad_norm": 2.5168659687042236, | |
| "learning_rate": 9.480459770114944e-06, | |
| "loss": 0.6544, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.0577716643741404, | |
| "grad_norm": 3.087484836578369, | |
| "learning_rate": 9.46896551724138e-06, | |
| "loss": 0.6235, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.0715268225584593, | |
| "grad_norm": 2.891552448272705, | |
| "learning_rate": 9.457471264367817e-06, | |
| "loss": 0.5911, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.0852819807427785, | |
| "grad_norm": 3.162491798400879, | |
| "learning_rate": 9.445977011494253e-06, | |
| "loss": 0.6187, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.0990371389270976, | |
| "grad_norm": 2.9897255897521973, | |
| "learning_rate": 9.43448275862069e-06, | |
| "loss": 0.622, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.1127922971114168, | |
| "grad_norm": 2.488380193710327, | |
| "learning_rate": 9.422988505747127e-06, | |
| "loss": 0.5951, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.126547455295736, | |
| "grad_norm": 4.275421619415283, | |
| "learning_rate": 9.411494252873565e-06, | |
| "loss": 0.5707, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.140302613480055, | |
| "grad_norm": 2.8604931831359863, | |
| "learning_rate": 9.4e-06, | |
| "loss": 0.5814, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.154057771664374, | |
| "grad_norm": 3.4569485187530518, | |
| "learning_rate": 9.388505747126438e-06, | |
| "loss": 0.6335, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.1678129298486932, | |
| "grad_norm": 3.236696481704712, | |
| "learning_rate": 9.377011494252875e-06, | |
| "loss": 0.5976, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.1815680880330124, | |
| "grad_norm": 2.5323238372802734, | |
| "learning_rate": 9.365517241379311e-06, | |
| "loss": 0.5809, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.1953232462173315, | |
| "grad_norm": 3.0921990871429443, | |
| "learning_rate": 9.354022988505748e-06, | |
| "loss": 0.5672, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.2090784044016507, | |
| "grad_norm": 2.991499185562134, | |
| "learning_rate": 9.342528735632184e-06, | |
| "loss": 0.6058, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.2228335625859696, | |
| "grad_norm": 2.8206140995025635, | |
| "learning_rate": 9.331034482758623e-06, | |
| "loss": 0.641, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.2365887207702888, | |
| "grad_norm": 2.5404281616210938, | |
| "learning_rate": 9.319540229885058e-06, | |
| "loss": 0.5627, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.250343878954608, | |
| "grad_norm": 2.977123737335205, | |
| "learning_rate": 9.308045977011496e-06, | |
| "loss": 0.5843, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.2640990371389271, | |
| "grad_norm": 3.053406000137329, | |
| "learning_rate": 9.296551724137932e-06, | |
| "loss": 0.6054, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.2778541953232463, | |
| "grad_norm": 2.8271589279174805, | |
| "learning_rate": 9.285057471264369e-06, | |
| "loss": 0.623, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.2916093535075652, | |
| "grad_norm": 2.7028932571411133, | |
| "learning_rate": 9.273563218390806e-06, | |
| "loss": 0.5977, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.3053645116918844, | |
| "grad_norm": 2.9352943897247314, | |
| "learning_rate": 9.262068965517242e-06, | |
| "loss": 0.517, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.3191196698762035, | |
| "grad_norm": 2.8113648891448975, | |
| "learning_rate": 9.250574712643679e-06, | |
| "loss": 0.5724, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.3328748280605227, | |
| "grad_norm": 2.876746892929077, | |
| "learning_rate": 9.239080459770115e-06, | |
| "loss": 0.5633, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.3466299862448419, | |
| "grad_norm": 3.2946653366088867, | |
| "learning_rate": 9.227586206896552e-06, | |
| "loss": 0.6288, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.360385144429161, | |
| "grad_norm": 3.5094716548919678, | |
| "learning_rate": 9.216091954022988e-06, | |
| "loss": 0.5127, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.3741403026134802, | |
| "grad_norm": 2.5474300384521484, | |
| "learning_rate": 9.204597701149425e-06, | |
| "loss": 0.5787, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.3741403026134802, | |
| "eval_bleu": 0.4206719346931395, | |
| "eval_cer": 18.365245374094933, | |
| "eval_loss": 0.7084596753120422, | |
| "eval_runtime": 733.6947, | |
| "eval_samples_per_second": 0.664, | |
| "eval_steps_per_second": 0.333, | |
| "eval_wer": 41.43696797730036, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.3878954607977991, | |
| "grad_norm": 2.845057249069214, | |
| "learning_rate": 9.193103448275863e-06, | |
| "loss": 0.5554, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.4016506189821183, | |
| "grad_norm": 3.106311082839966, | |
| "learning_rate": 9.1816091954023e-06, | |
| "loss": 0.5587, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.4154057771664375, | |
| "grad_norm": 3.01326322555542, | |
| "learning_rate": 9.170114942528736e-06, | |
| "loss": 0.5574, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.4291609353507566, | |
| "grad_norm": 2.919343948364258, | |
| "learning_rate": 9.158620689655173e-06, | |
| "loss": 0.5855, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.4429160935350756, | |
| "grad_norm": 2.971627712249756, | |
| "learning_rate": 9.14712643678161e-06, | |
| "loss": 0.5749, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.4566712517193947, | |
| "grad_norm": 12.000533103942871, | |
| "learning_rate": 9.135632183908046e-06, | |
| "loss": 0.5129, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.4704264099037139, | |
| "grad_norm": 2.927503824234009, | |
| "learning_rate": 9.124137931034483e-06, | |
| "loss": 0.5323, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.484181568088033, | |
| "grad_norm": 5.160599708557129, | |
| "learning_rate": 9.112643678160921e-06, | |
| "loss": 0.5212, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.4979367262723522, | |
| "grad_norm": 2.3872811794281006, | |
| "learning_rate": 9.101149425287356e-06, | |
| "loss": 0.5705, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.5116918844566714, | |
| "grad_norm": 2.8588502407073975, | |
| "learning_rate": 9.089655172413794e-06, | |
| "loss": 0.5803, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.5254470426409905, | |
| "grad_norm": 2.890172004699707, | |
| "learning_rate": 9.07816091954023e-06, | |
| "loss": 0.5466, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.5392022008253095, | |
| "grad_norm": 2.831617593765259, | |
| "learning_rate": 9.066666666666667e-06, | |
| "loss": 0.5848, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.5529573590096286, | |
| "grad_norm": 2.6025524139404297, | |
| "learning_rate": 9.055172413793104e-06, | |
| "loss": 0.5126, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.5667125171939478, | |
| "grad_norm": 3.2757222652435303, | |
| "learning_rate": 9.04367816091954e-06, | |
| "loss": 0.5366, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.5804676753782667, | |
| "grad_norm": 2.5900487899780273, | |
| "learning_rate": 9.032183908045977e-06, | |
| "loss": 0.5046, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.5942228335625859, | |
| "grad_norm": 2.6956920623779297, | |
| "learning_rate": 9.020689655172414e-06, | |
| "loss": 0.5223, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.607977991746905, | |
| "grad_norm": 3.1659083366394043, | |
| "learning_rate": 9.009195402298852e-06, | |
| "loss": 0.5554, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.6217331499312242, | |
| "grad_norm": 2.5289273262023926, | |
| "learning_rate": 8.997701149425289e-06, | |
| "loss": 0.508, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.6354883081155434, | |
| "grad_norm": 2.5216193199157715, | |
| "learning_rate": 8.986206896551725e-06, | |
| "loss": 0.4941, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.6492434662998625, | |
| "grad_norm": 2.9627370834350586, | |
| "learning_rate": 8.974712643678162e-06, | |
| "loss": 0.5181, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.6629986244841817, | |
| "grad_norm": 2.7418017387390137, | |
| "learning_rate": 8.963218390804598e-06, | |
| "loss": 0.5704, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.6767537826685008, | |
| "grad_norm": 2.909184694290161, | |
| "learning_rate": 8.951724137931035e-06, | |
| "loss": 0.5253, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.6905089408528198, | |
| "grad_norm": 3.2147293090820312, | |
| "learning_rate": 8.940229885057471e-06, | |
| "loss": 0.5127, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.704264099037139, | |
| "grad_norm": 2.9844889640808105, | |
| "learning_rate": 8.92873563218391e-06, | |
| "loss": 0.5295, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.718019257221458, | |
| "grad_norm": 2.8547401428222656, | |
| "learning_rate": 8.917241379310345e-06, | |
| "loss": 0.5191, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.731774415405777, | |
| "grad_norm": 2.84828782081604, | |
| "learning_rate": 8.905747126436783e-06, | |
| "loss": 0.517, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.7455295735900962, | |
| "grad_norm": 2.946942090988159, | |
| "learning_rate": 8.89425287356322e-06, | |
| "loss": 0.5044, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.7592847317744154, | |
| "grad_norm": 2.5053775310516357, | |
| "learning_rate": 8.882758620689656e-06, | |
| "loss": 0.4814, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.7730398899587345, | |
| "grad_norm": 3.0368754863739014, | |
| "learning_rate": 8.871264367816093e-06, | |
| "loss": 0.5065, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.7867950481430537, | |
| "grad_norm": 2.697561264038086, | |
| "learning_rate": 8.85977011494253e-06, | |
| "loss": 0.5209, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.8005502063273728, | |
| "grad_norm": 2.584867238998413, | |
| "learning_rate": 8.848275862068966e-06, | |
| "loss": 0.4833, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.814305364511692, | |
| "grad_norm": 3.0252091884613037, | |
| "learning_rate": 8.836781609195402e-06, | |
| "loss": 0.4799, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.8280605226960112, | |
| "grad_norm": 2.6939878463745117, | |
| "learning_rate": 8.82528735632184e-06, | |
| "loss": 0.4721, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.84181568088033, | |
| "grad_norm": 2.83384108543396, | |
| "learning_rate": 8.813793103448277e-06, | |
| "loss": 0.471, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.8555708390646493, | |
| "grad_norm": 2.3436243534088135, | |
| "learning_rate": 8.802298850574714e-06, | |
| "loss": 0.5411, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.8693259972489684, | |
| "grad_norm": 3.018484115600586, | |
| "learning_rate": 8.79080459770115e-06, | |
| "loss": 0.4651, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.8830811554332874, | |
| "grad_norm": 3.808323860168457, | |
| "learning_rate": 8.779310344827587e-06, | |
| "loss": 0.4797, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.8968363136176065, | |
| "grad_norm": 2.987705945968628, | |
| "learning_rate": 8.767816091954024e-06, | |
| "loss": 0.4793, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.9105914718019257, | |
| "grad_norm": 2.7502503395080566, | |
| "learning_rate": 8.75632183908046e-06, | |
| "loss": 0.4851, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.9243466299862448, | |
| "grad_norm": 2.606917381286621, | |
| "learning_rate": 8.744827586206898e-06, | |
| "loss": 0.5304, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.938101788170564, | |
| "grad_norm": 2.6112170219421387, | |
| "learning_rate": 8.733333333333333e-06, | |
| "loss": 0.4538, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.9518569463548832, | |
| "grad_norm": 2.6846561431884766, | |
| "learning_rate": 8.721839080459772e-06, | |
| "loss": 0.4557, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.9656121045392023, | |
| "grad_norm": 2.7344939708709717, | |
| "learning_rate": 8.710344827586208e-06, | |
| "loss": 0.4682, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.9793672627235215, | |
| "grad_norm": 2.805279493331909, | |
| "learning_rate": 8.698850574712645e-06, | |
| "loss": 0.4505, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.9931224209078404, | |
| "grad_norm": 3.22121000289917, | |
| "learning_rate": 8.687356321839081e-06, | |
| "loss": 0.4693, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 2.0055020632737275, | |
| "grad_norm": 2.3558602333068848, | |
| "learning_rate": 8.675862068965518e-06, | |
| "loss": 0.3819, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.0192572214580466, | |
| "grad_norm": 2.5739705562591553, | |
| "learning_rate": 8.664367816091954e-06, | |
| "loss": 0.3373, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 2.033012379642366, | |
| "grad_norm": 3.0432779788970947, | |
| "learning_rate": 8.652873563218391e-06, | |
| "loss": 0.3862, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 2.046767537826685, | |
| "grad_norm": 2.4599671363830566, | |
| "learning_rate": 8.641379310344828e-06, | |
| "loss": 0.3276, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 2.060522696011004, | |
| "grad_norm": 2.53336501121521, | |
| "learning_rate": 8.629885057471266e-06, | |
| "loss": 0.3393, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.0742778541953233, | |
| "grad_norm": 2.4407801628112793, | |
| "learning_rate": 8.6183908045977e-06, | |
| "loss": 0.3815, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 2.0880330123796425, | |
| "grad_norm": 3.65480899810791, | |
| "learning_rate": 8.606896551724139e-06, | |
| "loss": 0.3488, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.1017881705639616, | |
| "grad_norm": 21.605276107788086, | |
| "learning_rate": 8.595402298850576e-06, | |
| "loss": 0.358, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 2.1155433287482808, | |
| "grad_norm": 3.20198917388916, | |
| "learning_rate": 8.583908045977012e-06, | |
| "loss": 0.3401, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 2.1292984869326, | |
| "grad_norm": 13.919597625732422, | |
| "learning_rate": 8.572413793103449e-06, | |
| "loss": 0.3487, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 2.1430536451169186, | |
| "grad_norm": 2.4645321369171143, | |
| "learning_rate": 8.560919540229885e-06, | |
| "loss": 0.334, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.156808803301238, | |
| "grad_norm": 3.1687004566192627, | |
| "learning_rate": 8.549425287356322e-06, | |
| "loss": 0.3467, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 2.170563961485557, | |
| "grad_norm": 11.212874412536621, | |
| "learning_rate": 8.537931034482759e-06, | |
| "loss": 0.3275, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 2.184319119669876, | |
| "grad_norm": 2.719883441925049, | |
| "learning_rate": 8.526436781609197e-06, | |
| "loss": 0.3548, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 2.1980742778541953, | |
| "grad_norm": 2.6431100368499756, | |
| "learning_rate": 8.514942528735632e-06, | |
| "loss": 0.3257, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.2118294360385145, | |
| "grad_norm": 2.410521984100342, | |
| "learning_rate": 8.50344827586207e-06, | |
| "loss": 0.3428, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 2.2255845942228336, | |
| "grad_norm": 3.244124174118042, | |
| "learning_rate": 8.491954022988507e-06, | |
| "loss": 0.3311, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.2393397524071528, | |
| "grad_norm": 3.0431458950042725, | |
| "learning_rate": 8.480459770114943e-06, | |
| "loss": 0.3684, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 2.253094910591472, | |
| "grad_norm": 2.8572866916656494, | |
| "learning_rate": 8.46896551724138e-06, | |
| "loss": 0.359, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.266850068775791, | |
| "grad_norm": 2.5997812747955322, | |
| "learning_rate": 8.457471264367816e-06, | |
| "loss": 0.3391, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 2.28060522696011, | |
| "grad_norm": 2.5643837451934814, | |
| "learning_rate": 8.445977011494255e-06, | |
| "loss": 0.3347, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 2.294360385144429, | |
| "grad_norm": 2.5857532024383545, | |
| "learning_rate": 8.43448275862069e-06, | |
| "loss": 0.3346, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 2.308115543328748, | |
| "grad_norm": 2.9407107830047607, | |
| "learning_rate": 8.422988505747128e-06, | |
| "loss": 0.3213, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.3218707015130673, | |
| "grad_norm": 2.9194135665893555, | |
| "learning_rate": 8.411494252873564e-06, | |
| "loss": 0.2954, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 2.3356258596973865, | |
| "grad_norm": 2.4436566829681396, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.343, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.3493810178817056, | |
| "grad_norm": 2.485232353210449, | |
| "learning_rate": 8.388505747126437e-06, | |
| "loss": 0.3565, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 2.3631361760660248, | |
| "grad_norm": 2.972649335861206, | |
| "learning_rate": 8.377011494252874e-06, | |
| "loss": 0.3461, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 2.376891334250344, | |
| "grad_norm": 2.566713571548462, | |
| "learning_rate": 8.36551724137931e-06, | |
| "loss": 0.3372, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 2.390646492434663, | |
| "grad_norm": 2.9575984477996826, | |
| "learning_rate": 8.354022988505747e-06, | |
| "loss": 0.3427, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 2.4044016506189823, | |
| "grad_norm": 2.6989376544952393, | |
| "learning_rate": 8.342528735632185e-06, | |
| "loss": 0.3391, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 2.4181568088033014, | |
| "grad_norm": 2.6465554237365723, | |
| "learning_rate": 8.33103448275862e-06, | |
| "loss": 0.3128, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 2.43191196698762, | |
| "grad_norm": 2.6031441688537598, | |
| "learning_rate": 8.319540229885059e-06, | |
| "loss": 0.3227, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 2.4456671251719393, | |
| "grad_norm": 2.6808745861053467, | |
| "learning_rate": 8.308045977011495e-06, | |
| "loss": 0.3072, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 2.4594222833562585, | |
| "grad_norm": 2.9904232025146484, | |
| "learning_rate": 8.296551724137932e-06, | |
| "loss": 0.3265, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 2.4731774415405776, | |
| "grad_norm": 2.7612838745117188, | |
| "learning_rate": 8.285057471264368e-06, | |
| "loss": 0.3245, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.4869325997248968, | |
| "grad_norm": 2.7029314041137695, | |
| "learning_rate": 8.273563218390805e-06, | |
| "loss": 0.3273, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 2.500687757909216, | |
| "grad_norm": 2.87737774848938, | |
| "learning_rate": 8.262068965517243e-06, | |
| "loss": 0.308, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.514442916093535, | |
| "grad_norm": 3.425462245941162, | |
| "learning_rate": 8.250574712643678e-06, | |
| "loss": 0.3026, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 2.5281980742778543, | |
| "grad_norm": 2.760190725326538, | |
| "learning_rate": 8.239080459770116e-06, | |
| "loss": 0.3116, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.5419532324621734, | |
| "grad_norm": 2.7421491146087646, | |
| "learning_rate": 8.227586206896553e-06, | |
| "loss": 0.3149, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.5557083906464926, | |
| "grad_norm": 2.5775585174560547, | |
| "learning_rate": 8.21609195402299e-06, | |
| "loss": 0.3198, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.5694635488308117, | |
| "grad_norm": 2.3301403522491455, | |
| "learning_rate": 8.204597701149426e-06, | |
| "loss": 0.2859, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 2.5832187070151305, | |
| "grad_norm": 2.293553590774536, | |
| "learning_rate": 8.193103448275863e-06, | |
| "loss": 0.3045, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.59697386519945, | |
| "grad_norm": 10.349946022033691, | |
| "learning_rate": 8.1816091954023e-06, | |
| "loss": 0.312, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 2.6107290233837688, | |
| "grad_norm": 2.5628931522369385, | |
| "learning_rate": 8.170114942528736e-06, | |
| "loss": 0.3153, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.624484181568088, | |
| "grad_norm": 3.248706340789795, | |
| "learning_rate": 8.158620689655174e-06, | |
| "loss": 0.3039, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 2.638239339752407, | |
| "grad_norm": 16.6440486907959, | |
| "learning_rate": 8.147126436781609e-06, | |
| "loss": 0.3417, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.6519944979367263, | |
| "grad_norm": 2.431967258453369, | |
| "learning_rate": 8.135632183908047e-06, | |
| "loss": 0.3354, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 2.6657496561210454, | |
| "grad_norm": 2.516252279281616, | |
| "learning_rate": 8.124137931034484e-06, | |
| "loss": 0.2806, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.6795048143053646, | |
| "grad_norm": 2.657532215118408, | |
| "learning_rate": 8.11264367816092e-06, | |
| "loss": 0.3043, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.6932599724896837, | |
| "grad_norm": 2.3438785076141357, | |
| "learning_rate": 8.101149425287357e-06, | |
| "loss": 0.3201, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.707015130674003, | |
| "grad_norm": 3.0601348876953125, | |
| "learning_rate": 8.089655172413794e-06, | |
| "loss": 0.3047, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 2.720770288858322, | |
| "grad_norm": 2.5481855869293213, | |
| "learning_rate": 8.078160919540232e-06, | |
| "loss": 0.3035, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.7345254470426408, | |
| "grad_norm": 2.3380320072174072, | |
| "learning_rate": 8.066666666666667e-06, | |
| "loss": 0.2845, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 2.7482806052269604, | |
| "grad_norm": 2.7484054565429688, | |
| "learning_rate": 8.055172413793103e-06, | |
| "loss": 0.298, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.7482806052269604, | |
| "eval_bleu": 0.44979726292769356, | |
| "eval_cer": 13.05229283990346, | |
| "eval_loss": 0.7391562461853027, | |
| "eval_runtime": 676.7738, | |
| "eval_samples_per_second": 0.72, | |
| "eval_steps_per_second": 0.361, | |
| "eval_wer": 35.75192541548439, | |
| "step": 1000 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 4500, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 13, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.35713091649536e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |