| { |
| "best_metric": 22.283451007404857, |
| "best_model_checkpoint": "./whisper-large-v2/second2/checkpoint-500", |
| "epoch": 0.2133902373966391, |
| "eval_steps": 500, |
| "global_step": 1000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0010669511869831954, |
| "grad_norm": 19.418699264526367, |
| "learning_rate": 3.3333333333333334e-08, |
| "loss": 2.4187, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.002133902373966391, |
| "grad_norm": 21.75494956970215, |
| "learning_rate": 6.666666666666667e-08, |
| "loss": 2.3362, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.0032008535609495867, |
| "grad_norm": 21.226938247680664, |
| "learning_rate": 1e-07, |
| "loss": 2.4288, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.004267804747932782, |
| "grad_norm": 27.21125602722168, |
| "learning_rate": 1.3333333333333334e-07, |
| "loss": 2.4004, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.005334755934915978, |
| "grad_norm": 13.559316635131836, |
| "learning_rate": 1.6666666666666665e-07, |
| "loss": 2.1607, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.0064017071218991735, |
| "grad_norm": 13.044219970703125, |
| "learning_rate": 1.9333333333333332e-07, |
| "loss": 2.1591, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.007468658308882369, |
| "grad_norm": 13.89301872253418, |
| "learning_rate": 2.2666666666666663e-07, |
| "loss": 2.0855, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.008535609495865563, |
| "grad_norm": 14.154369354248047, |
| "learning_rate": 2.6e-07, |
| "loss": 1.9557, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.00960256068284876, |
| "grad_norm": 9.126311302185059, |
| "learning_rate": 2.933333333333333e-07, |
| "loss": 1.8661, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.010669511869831956, |
| "grad_norm": 13.619270324707031, |
| "learning_rate": 3.2666666666666663e-07, |
| "loss": 1.8424, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.01173646305681515, |
| "grad_norm": 9.875323295593262, |
| "learning_rate": 3.6e-07, |
| "loss": 1.6208, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.012803414243798347, |
| "grad_norm": 11.143688201904297, |
| "learning_rate": 3.933333333333333e-07, |
| "loss": 1.5396, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.013870365430781541, |
| "grad_norm": 10.829161643981934, |
| "learning_rate": 4.266666666666667e-07, |
| "loss": 1.4174, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.014937316617764738, |
| "grad_norm": 7.393435001373291, |
| "learning_rate": 4.6e-07, |
| "loss": 1.1693, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.016004267804747934, |
| "grad_norm": 4.182456016540527, |
| "learning_rate": 4.933333333333333e-07, |
| "loss": 1.1779, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.017071218991731127, |
| "grad_norm": 8.4662504196167, |
| "learning_rate": 5.266666666666666e-07, |
| "loss": 1.1281, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.018138170178714323, |
| "grad_norm": 4.043624401092529, |
| "learning_rate": 5.6e-07, |
| "loss": 1.0614, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.01920512136569752, |
| "grad_norm": 3.8517227172851562, |
| "learning_rate": 5.933333333333334e-07, |
| "loss": 1.0066, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.020272072552680716, |
| "grad_norm": 4.19964075088501, |
| "learning_rate": 6.266666666666667e-07, |
| "loss": 1.0659, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.021339023739663912, |
| "grad_norm": 4.7087178230285645, |
| "learning_rate": 6.6e-07, |
| "loss": 0.9793, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.022405974926647105, |
| "grad_norm": 4.864803314208984, |
| "learning_rate": 6.933333333333333e-07, |
| "loss": 0.9675, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.0234729261136303, |
| "grad_norm": 3.8939764499664307, |
| "learning_rate": 7.266666666666667e-07, |
| "loss": 0.9513, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.024539877300613498, |
| "grad_norm": 3.557497978210449, |
| "learning_rate": 7.599999999999999e-07, |
| "loss": 0.9164, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.025606828487596694, |
| "grad_norm": 4.033596038818359, |
| "learning_rate": 7.933333333333333e-07, |
| "loss": 0.9069, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.026673779674579887, |
| "grad_norm": 4.726081371307373, |
| "learning_rate": 8.266666666666667e-07, |
| "loss": 0.8976, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.027740730861563083, |
| "grad_norm": 18.711572647094727, |
| "learning_rate": 8.599999999999999e-07, |
| "loss": 0.9157, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.02880768204854628, |
| "grad_norm": 4.13236141204834, |
| "learning_rate": 8.933333333333333e-07, |
| "loss": 0.8808, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.029874633235529476, |
| "grad_norm": 3.7335939407348633, |
| "learning_rate": 9.266666666666665e-07, |
| "loss": 0.7707, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.03094158442251267, |
| "grad_norm": 4.5814313888549805, |
| "learning_rate": 9.6e-07, |
| "loss": 0.8349, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.03200853560949587, |
| "grad_norm": 4.993039608001709, |
| "learning_rate": 9.933333333333333e-07, |
| "loss": 0.7002, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.033075486796479064, |
| "grad_norm": 5.326510906219482, |
| "learning_rate": 9.996791443850267e-07, |
| "loss": 0.6418, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.034142437983462254, |
| "grad_norm": 4.4001994132995605, |
| "learning_rate": 9.99144385026738e-07, |
| "loss": 0.5632, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.03520938917044545, |
| "grad_norm": 71.39772033691406, |
| "learning_rate": 9.98716577540107e-07, |
| "loss": 0.6374, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.036276340357428646, |
| "grad_norm": 9.561308860778809, |
| "learning_rate": 9.98288770053476e-07, |
| "loss": 0.5493, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.03734329154441184, |
| "grad_norm": 2.866039991378784, |
| "learning_rate": 9.97754010695187e-07, |
| "loss": 0.473, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.03841024273139504, |
| "grad_norm": 3.0788004398345947, |
| "learning_rate": 9.972192513368983e-07, |
| "loss": 0.4903, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.039477193918378235, |
| "grad_norm": 3.4059643745422363, |
| "learning_rate": 9.966844919786097e-07, |
| "loss": 0.5017, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.04054414510536143, |
| "grad_norm": 2.9747097492218018, |
| "learning_rate": 9.961497326203208e-07, |
| "loss": 0.4269, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.04161109629234463, |
| "grad_norm": 3.476583242416382, |
| "learning_rate": 9.95614973262032e-07, |
| "loss": 0.4214, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.042678047479327824, |
| "grad_norm": 3.0810964107513428, |
| "learning_rate": 9.950802139037432e-07, |
| "loss": 0.4291, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.043744998666311014, |
| "grad_norm": 87.03646850585938, |
| "learning_rate": 9.945454545454544e-07, |
| "loss": 0.4035, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.04481194985329421, |
| "grad_norm": 2.6879165172576904, |
| "learning_rate": 9.940106951871658e-07, |
| "loss": 0.428, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.045878901040277406, |
| "grad_norm": 3.602217197418213, |
| "learning_rate": 9.93475935828877e-07, |
| "loss": 0.4534, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.0469458522272606, |
| "grad_norm": 2.9568989276885986, |
| "learning_rate": 9.929411764705881e-07, |
| "loss": 0.42, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.0480128034142438, |
| "grad_norm": 2.7602076530456543, |
| "learning_rate": 9.924064171122995e-07, |
| "loss": 0.448, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.049079754601226995, |
| "grad_norm": 2.7191200256347656, |
| "learning_rate": 9.918716577540107e-07, |
| "loss": 0.4046, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.05014670578821019, |
| "grad_norm": 2.7233424186706543, |
| "learning_rate": 9.913368983957219e-07, |
| "loss": 0.4028, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.05121365697519339, |
| "grad_norm": 2.730400323867798, |
| "learning_rate": 9.90802139037433e-07, |
| "loss": 0.4361, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.05228060816217658, |
| "grad_norm": 4.667561054229736, |
| "learning_rate": 9.902673796791442e-07, |
| "loss": 0.4183, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.05334755934915977, |
| "grad_norm": 2.981497049331665, |
| "learning_rate": 9.897326203208556e-07, |
| "loss": 0.4148, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.05441451053614297, |
| "grad_norm": 9.336112976074219, |
| "learning_rate": 9.891978609625668e-07, |
| "loss": 0.4386, |
| "step": 255 |
| }, |
| { |
| "epoch": 0.055481461723126166, |
| "grad_norm": 4.434041500091553, |
| "learning_rate": 9.88663101604278e-07, |
| "loss": 0.4235, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.05654841291010936, |
| "grad_norm": 2.695866346359253, |
| "learning_rate": 9.881283422459892e-07, |
| "loss": 0.4424, |
| "step": 265 |
| }, |
| { |
| "epoch": 0.05761536409709256, |
| "grad_norm": 2.8247873783111572, |
| "learning_rate": 9.875935828877004e-07, |
| "loss": 0.387, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.058682315284075755, |
| "grad_norm": 3.4680376052856445, |
| "learning_rate": 9.870588235294118e-07, |
| "loss": 0.4294, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.05974926647105895, |
| "grad_norm": 2.6994473934173584, |
| "learning_rate": 9.86524064171123e-07, |
| "loss": 0.4043, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.06081621765804215, |
| "grad_norm": 2.6048827171325684, |
| "learning_rate": 9.859893048128341e-07, |
| "loss": 0.4064, |
| "step": 285 |
| }, |
| { |
| "epoch": 0.06188316884502534, |
| "grad_norm": 3.028503179550171, |
| "learning_rate": 9.854545454545455e-07, |
| "loss": 0.4269, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.06295012003200853, |
| "grad_norm": 3.3406739234924316, |
| "learning_rate": 9.849197860962567e-07, |
| "loss": 0.4339, |
| "step": 295 |
| }, |
| { |
| "epoch": 0.06401707121899174, |
| "grad_norm": 3.1800878047943115, |
| "learning_rate": 9.843850267379679e-07, |
| "loss": 0.3972, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.06508402240597493, |
| "grad_norm": 2.818852424621582, |
| "learning_rate": 9.83850267379679e-07, |
| "loss": 0.3898, |
| "step": 305 |
| }, |
| { |
| "epoch": 0.06615097359295813, |
| "grad_norm": 3.256840705871582, |
| "learning_rate": 9.833155080213902e-07, |
| "loss": 0.4377, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.06721792477994132, |
| "grad_norm": 3.388472318649292, |
| "learning_rate": 9.827807486631016e-07, |
| "loss": 0.4187, |
| "step": 315 |
| }, |
| { |
| "epoch": 0.06828487596692451, |
| "grad_norm": 4.434413909912109, |
| "learning_rate": 9.822459893048128e-07, |
| "loss": 0.4009, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.06935182715390771, |
| "grad_norm": 3.08677077293396, |
| "learning_rate": 9.81711229946524e-07, |
| "loss": 0.3837, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.0704187783408909, |
| "grad_norm": 3.7852623462677, |
| "learning_rate": 9.811764705882352e-07, |
| "loss": 0.3999, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.0714857295278741, |
| "grad_norm": 2.947720766067505, |
| "learning_rate": 9.806417112299463e-07, |
| "loss": 0.3677, |
| "step": 335 |
| }, |
| { |
| "epoch": 0.07255268071485729, |
| "grad_norm": 3.1685659885406494, |
| "learning_rate": 9.801069518716577e-07, |
| "loss": 0.4035, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.0736196319018405, |
| "grad_norm": 2.661599636077881, |
| "learning_rate": 9.79572192513369e-07, |
| "loss": 0.3658, |
| "step": 345 |
| }, |
| { |
| "epoch": 0.07468658308882369, |
| "grad_norm": 2.6443653106689453, |
| "learning_rate": 9.7903743315508e-07, |
| "loss": 0.3805, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.07575353427580689, |
| "grad_norm": 3.653778314590454, |
| "learning_rate": 9.785026737967915e-07, |
| "loss": 0.407, |
| "step": 355 |
| }, |
| { |
| "epoch": 0.07682048546279008, |
| "grad_norm": 2.991504430770874, |
| "learning_rate": 9.779679144385027e-07, |
| "loss": 0.4135, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.07788743664977327, |
| "grad_norm": 2.7770955562591553, |
| "learning_rate": 9.774331550802139e-07, |
| "loss": 0.3934, |
| "step": 365 |
| }, |
| { |
| "epoch": 0.07895438783675647, |
| "grad_norm": 2.686368703842163, |
| "learning_rate": 9.76898395721925e-07, |
| "loss": 0.409, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.08002133902373966, |
| "grad_norm": 3.307638168334961, |
| "learning_rate": 9.763636363636362e-07, |
| "loss": 0.3872, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.08108829021072286, |
| "grad_norm": 8.76164722442627, |
| "learning_rate": 9.758288770053476e-07, |
| "loss": 0.3882, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.08215524139770605, |
| "grad_norm": 2.868077039718628, |
| "learning_rate": 9.752941176470588e-07, |
| "loss": 0.4123, |
| "step": 385 |
| }, |
| { |
| "epoch": 0.08322219258468926, |
| "grad_norm": 4.677827835083008, |
| "learning_rate": 9.7475935828877e-07, |
| "loss": 0.362, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.08428914377167245, |
| "grad_norm": 3.836914539337158, |
| "learning_rate": 9.742245989304812e-07, |
| "loss": 0.4023, |
| "step": 395 |
| }, |
| { |
| "epoch": 0.08535609495865565, |
| "grad_norm": 2.7120203971862793, |
| "learning_rate": 9.736898395721923e-07, |
| "loss": 0.419, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.08642304614563884, |
| "grad_norm": 3.1341233253479004, |
| "learning_rate": 9.731550802139037e-07, |
| "loss": 0.4275, |
| "step": 405 |
| }, |
| { |
| "epoch": 0.08748999733262203, |
| "grad_norm": 3.0901923179626465, |
| "learning_rate": 9.72620320855615e-07, |
| "loss": 0.3799, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.08855694851960523, |
| "grad_norm": 2.7124781608581543, |
| "learning_rate": 9.72085561497326e-07, |
| "loss": 0.3788, |
| "step": 415 |
| }, |
| { |
| "epoch": 0.08962389970658842, |
| "grad_norm": 2.8177292346954346, |
| "learning_rate": 9.715508021390375e-07, |
| "loss": 0.3436, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.09069085089357162, |
| "grad_norm": 3.063669204711914, |
| "learning_rate": 9.710160427807487e-07, |
| "loss": 0.4312, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.09175780208055481, |
| "grad_norm": 2.78320574760437, |
| "learning_rate": 9.704812834224598e-07, |
| "loss": 0.4122, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.09282475326753802, |
| "grad_norm": 3.080400228500366, |
| "learning_rate": 9.69946524064171e-07, |
| "loss": 0.3917, |
| "step": 435 |
| }, |
| { |
| "epoch": 0.0938917044545212, |
| "grad_norm": 3.02703857421875, |
| "learning_rate": 9.694117647058822e-07, |
| "loss": 0.3546, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.09495865564150441, |
| "grad_norm": 2.6559834480285645, |
| "learning_rate": 9.688770053475936e-07, |
| "loss": 0.3536, |
| "step": 445 |
| }, |
| { |
| "epoch": 0.0960256068284876, |
| "grad_norm": 3.939589738845825, |
| "learning_rate": 9.683422459893048e-07, |
| "loss": 0.3729, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.09709255801547079, |
| "grad_norm": 2.419799327850342, |
| "learning_rate": 9.67807486631016e-07, |
| "loss": 0.3474, |
| "step": 455 |
| }, |
| { |
| "epoch": 0.09815950920245399, |
| "grad_norm": 2.237165927886963, |
| "learning_rate": 9.672727272727271e-07, |
| "loss": 0.3387, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.09922646038943718, |
| "grad_norm": 2.6400506496429443, |
| "learning_rate": 9.667379679144385e-07, |
| "loss": 0.3462, |
| "step": 465 |
| }, |
| { |
| "epoch": 0.10029341157642038, |
| "grad_norm": 3.294222354888916, |
| "learning_rate": 9.662032085561497e-07, |
| "loss": 0.3724, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.10136036276340357, |
| "grad_norm": 2.5825111865997314, |
| "learning_rate": 9.656684491978609e-07, |
| "loss": 0.3652, |
| "step": 475 |
| }, |
| { |
| "epoch": 0.10242731395038678, |
| "grad_norm": 2.6807165145874023, |
| "learning_rate": 9.65133689839572e-07, |
| "loss": 0.3824, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.10349426513736996, |
| "grad_norm": 2.541398048400879, |
| "learning_rate": 9.645989304812835e-07, |
| "loss": 0.3509, |
| "step": 485 |
| }, |
| { |
| "epoch": 0.10456121632435315, |
| "grad_norm": 3.071420192718506, |
| "learning_rate": 9.640641711229946e-07, |
| "loss": 0.3511, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.10562816751133636, |
| "grad_norm": 3.117623805999756, |
| "learning_rate": 9.635294117647058e-07, |
| "loss": 0.3823, |
| "step": 495 |
| }, |
| { |
| "epoch": 0.10669511869831955, |
| "grad_norm": 2.9759390354156494, |
| "learning_rate": 9.62994652406417e-07, |
| "loss": 0.3595, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.10669511869831955, |
| "eval_bleu": 0.5858525890811711, |
| "eval_cer": 5.81035140115651, |
| "eval_loss": 0.35532379150390625, |
| "eval_runtime": 1038.2775, |
| "eval_samples_per_second": 0.884, |
| "eval_steps_per_second": 0.222, |
| "eval_wer": 22.283451007404857, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.10776206988530275, |
| "grad_norm": 2.9231984615325928, |
| "learning_rate": 9.475679090334807e-07, |
| "loss": 0.3783, |
| "step": 505 |
| }, |
| { |
| "epoch": 0.10882902107228594, |
| "grad_norm": 2.2739880084991455, |
| "learning_rate": 9.470414824173509e-07, |
| "loss": 0.3238, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.10989597225926914, |
| "grad_norm": 2.603306531906128, |
| "learning_rate": 9.465150558012212e-07, |
| "loss": 0.3538, |
| "step": 515 |
| }, |
| { |
| "epoch": 0.11096292344625233, |
| "grad_norm": 2.871011972427368, |
| "learning_rate": 9.459886291850916e-07, |
| "loss": 0.3955, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.11202987463323553, |
| "grad_norm": 3.134646415710449, |
| "learning_rate": 9.454622025689618e-07, |
| "loss": 0.3508, |
| "step": 525 |
| }, |
| { |
| "epoch": 0.11309682582021872, |
| "grad_norm": 3.014796733856201, |
| "learning_rate": 9.449357759528321e-07, |
| "loss": 0.3664, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.11416377700720191, |
| "grad_norm": 3.0675926208496094, |
| "learning_rate": 9.444093493367024e-07, |
| "loss": 0.3497, |
| "step": 535 |
| }, |
| { |
| "epoch": 0.11523072819418512, |
| "grad_norm": 4.508389472961426, |
| "learning_rate": 9.438829227205727e-07, |
| "loss": 0.3172, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.1162976793811683, |
| "grad_norm": 2.528317451477051, |
| "learning_rate": 9.43356496104443e-07, |
| "loss": 0.3495, |
| "step": 545 |
| }, |
| { |
| "epoch": 0.11736463056815151, |
| "grad_norm": 2.759575843811035, |
| "learning_rate": 9.428300694883133e-07, |
| "loss": 0.3326, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.1184315817551347, |
| "grad_norm": 6.699812889099121, |
| "learning_rate": 9.423036428721835e-07, |
| "loss": 0.4124, |
| "step": 555 |
| }, |
| { |
| "epoch": 0.1194985329421179, |
| "grad_norm": 2.358922243118286, |
| "learning_rate": 9.417772162560539e-07, |
| "loss": 0.3292, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.12056548412910109, |
| "grad_norm": 2.3635692596435547, |
| "learning_rate": 9.412507896399241e-07, |
| "loss": 0.3715, |
| "step": 565 |
| }, |
| { |
| "epoch": 0.1216324353160843, |
| "grad_norm": 2.8696706295013428, |
| "learning_rate": 9.407243630237944e-07, |
| "loss": 0.3348, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.12269938650306748, |
| "grad_norm": 3.717510223388672, |
| "learning_rate": 9.401979364076647e-07, |
| "loss": 0.355, |
| "step": 575 |
| }, |
| { |
| "epoch": 0.12376633769005067, |
| "grad_norm": 3.9285202026367188, |
| "learning_rate": 9.39671509791535e-07, |
| "loss": 0.3752, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.12483328887703388, |
| "grad_norm": 3.560582399368286, |
| "learning_rate": 9.391450831754053e-07, |
| "loss": 0.3422, |
| "step": 585 |
| }, |
| { |
| "epoch": 0.12590024006401707, |
| "grad_norm": 6.333406925201416, |
| "learning_rate": 9.386186565592757e-07, |
| "loss": 0.3436, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.12696719125100026, |
| "grad_norm": 2.94331431388855, |
| "learning_rate": 9.380922299431458e-07, |
| "loss": 0.3346, |
| "step": 595 |
| }, |
| { |
| "epoch": 0.12803414243798347, |
| "grad_norm": 2.933142900466919, |
| "learning_rate": 9.375658033270162e-07, |
| "loss": 0.3414, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.12910109362496666, |
| "grad_norm": 3.2017970085144043, |
| "learning_rate": 9.370393767108865e-07, |
| "loss": 0.3731, |
| "step": 605 |
| }, |
| { |
| "epoch": 0.13016804481194985, |
| "grad_norm": 3.2127702236175537, |
| "learning_rate": 9.365129500947567e-07, |
| "loss": 0.3399, |
| "step": 610 |
| }, |
| { |
| "epoch": 0.13123499599893304, |
| "grad_norm": 3.1433818340301514, |
| "learning_rate": 9.35986523478627e-07, |
| "loss": 0.342, |
| "step": 615 |
| }, |
| { |
| "epoch": 0.13230194718591626, |
| "grad_norm": 3.0913007259368896, |
| "learning_rate": 9.354600968624973e-07, |
| "loss": 0.3543, |
| "step": 620 |
| }, |
| { |
| "epoch": 0.13336889837289945, |
| "grad_norm": 3.459428310394287, |
| "learning_rate": 9.349336702463676e-07, |
| "loss": 0.292, |
| "step": 625 |
| }, |
| { |
| "epoch": 0.13443584955988264, |
| "grad_norm": 2.965162754058838, |
| "learning_rate": 9.34407243630238e-07, |
| "loss": 0.3591, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.13550280074686583, |
| "grad_norm": 2.476099967956543, |
| "learning_rate": 9.338808170141081e-07, |
| "loss": 0.3351, |
| "step": 635 |
| }, |
| { |
| "epoch": 0.13656975193384902, |
| "grad_norm": 2.853848695755005, |
| "learning_rate": 9.333543903979785e-07, |
| "loss": 0.3828, |
| "step": 640 |
| }, |
| { |
| "epoch": 0.13763670312083223, |
| "grad_norm": 2.560877561569214, |
| "learning_rate": 9.328279637818488e-07, |
| "loss": 0.3183, |
| "step": 645 |
| }, |
| { |
| "epoch": 0.13870365430781542, |
| "grad_norm": 2.7191262245178223, |
| "learning_rate": 9.32301537165719e-07, |
| "loss": 0.3572, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.1397706054947986, |
| "grad_norm": 3.388456106185913, |
| "learning_rate": 9.317751105495893e-07, |
| "loss": 0.3207, |
| "step": 655 |
| }, |
| { |
| "epoch": 0.1408375566817818, |
| "grad_norm": 2.827470302581787, |
| "learning_rate": 9.312486839334597e-07, |
| "loss": 0.3576, |
| "step": 660 |
| }, |
| { |
| "epoch": 0.14190450786876502, |
| "grad_norm": 2.630094528198242, |
| "learning_rate": 9.307222573173299e-07, |
| "loss": 0.3616, |
| "step": 665 |
| }, |
| { |
| "epoch": 0.1429714590557482, |
| "grad_norm": 2.7217891216278076, |
| "learning_rate": 9.301958307012003e-07, |
| "loss": 0.3434, |
| "step": 670 |
| }, |
| { |
| "epoch": 0.1440384102427314, |
| "grad_norm": 2.558335065841675, |
| "learning_rate": 9.296694040850705e-07, |
| "loss": 0.3516, |
| "step": 675 |
| }, |
| { |
| "epoch": 0.14510536142971459, |
| "grad_norm": 3.1991679668426514, |
| "learning_rate": 9.291429774689408e-07, |
| "loss": 0.338, |
| "step": 680 |
| }, |
| { |
| "epoch": 0.14617231261669777, |
| "grad_norm": 4.686666011810303, |
| "learning_rate": 9.286165508528111e-07, |
| "loss": 0.3539, |
| "step": 685 |
| }, |
| { |
| "epoch": 0.147239263803681, |
| "grad_norm": 3.0826447010040283, |
| "learning_rate": 9.280901242366814e-07, |
| "loss": 0.3313, |
| "step": 690 |
| }, |
| { |
| "epoch": 0.14830621499066418, |
| "grad_norm": 2.656141757965088, |
| "learning_rate": 9.275636976205516e-07, |
| "loss": 0.3454, |
| "step": 695 |
| }, |
| { |
| "epoch": 0.14937316617764737, |
| "grad_norm": 4.859818935394287, |
| "learning_rate": 9.27037271004422e-07, |
| "loss": 0.3331, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.15044011736463056, |
| "grad_norm": 5.376903057098389, |
| "learning_rate": 9.265108443882922e-07, |
| "loss": 0.3377, |
| "step": 705 |
| }, |
| { |
| "epoch": 0.15150706855161378, |
| "grad_norm": 2.7976577281951904, |
| "learning_rate": 9.259844177721626e-07, |
| "loss": 0.3768, |
| "step": 710 |
| }, |
| { |
| "epoch": 0.15257401973859697, |
| "grad_norm": 2.993427038192749, |
| "learning_rate": 9.254579911560328e-07, |
| "loss": 0.3509, |
| "step": 715 |
| }, |
| { |
| "epoch": 0.15364097092558016, |
| "grad_norm": 2.8581597805023193, |
| "learning_rate": 9.249315645399031e-07, |
| "loss": 0.3338, |
| "step": 720 |
| }, |
| { |
| "epoch": 0.15470792211256335, |
| "grad_norm": 2.8063721656799316, |
| "learning_rate": 9.244051379237734e-07, |
| "loss": 0.3597, |
| "step": 725 |
| }, |
| { |
| "epoch": 0.15577487329954653, |
| "grad_norm": 2.7333686351776123, |
| "learning_rate": 9.238787113076438e-07, |
| "loss": 0.3563, |
| "step": 730 |
| }, |
| { |
| "epoch": 0.15684182448652975, |
| "grad_norm": 3.4143598079681396, |
| "learning_rate": 9.233522846915139e-07, |
| "loss": 0.336, |
| "step": 735 |
| }, |
| { |
| "epoch": 0.15790877567351294, |
| "grad_norm": 2.84298038482666, |
| "learning_rate": 9.228258580753843e-07, |
| "loss": 0.3651, |
| "step": 740 |
| }, |
| { |
| "epoch": 0.15897572686049613, |
| "grad_norm": 5.912104606628418, |
| "learning_rate": 9.222994314592545e-07, |
| "loss": 0.3531, |
| "step": 745 |
| }, |
| { |
| "epoch": 0.16004267804747932, |
| "grad_norm": 3.0039525032043457, |
| "learning_rate": 9.217730048431249e-07, |
| "loss": 0.3212, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.16110962923446254, |
| "grad_norm": 3.2109341621398926, |
| "learning_rate": 9.212465782269951e-07, |
| "loss": 0.3206, |
| "step": 755 |
| }, |
| { |
| "epoch": 0.16217658042144573, |
| "grad_norm": 2.7238008975982666, |
| "learning_rate": 9.207201516108654e-07, |
| "loss": 0.3696, |
| "step": 760 |
| }, |
| { |
| "epoch": 0.16324353160842892, |
| "grad_norm": 2.6420886516571045, |
| "learning_rate": 9.201937249947357e-07, |
| "loss": 0.3339, |
| "step": 765 |
| }, |
| { |
| "epoch": 0.1643104827954121, |
| "grad_norm": 2.8822038173675537, |
| "learning_rate": 9.196672983786061e-07, |
| "loss": 0.3611, |
| "step": 770 |
| }, |
| { |
| "epoch": 0.1653774339823953, |
| "grad_norm": 2.558979034423828, |
| "learning_rate": 9.191408717624762e-07, |
| "loss": 0.3247, |
| "step": 775 |
| }, |
| { |
| "epoch": 0.1664443851693785, |
| "grad_norm": 2.649867057800293, |
| "learning_rate": 9.186144451463465e-07, |
| "loss": 0.3146, |
| "step": 780 |
| }, |
| { |
| "epoch": 0.1675113363563617, |
| "grad_norm": 2.6174542903900146, |
| "learning_rate": 9.180880185302169e-07, |
| "loss": 0.3319, |
| "step": 785 |
| }, |
| { |
| "epoch": 0.1685782875433449, |
| "grad_norm": 2.65977144241333, |
| "learning_rate": 9.175615919140871e-07, |
| "loss": 0.3643, |
| "step": 790 |
| }, |
| { |
| "epoch": 0.16964523873032808, |
| "grad_norm": 3.2722222805023193, |
| "learning_rate": 9.170351652979574e-07, |
| "loss": 0.3349, |
| "step": 795 |
| }, |
| { |
| "epoch": 0.1707121899173113, |
| "grad_norm": 2.346200466156006, |
| "learning_rate": 9.165087386818277e-07, |
| "loss": 0.3028, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.17177914110429449, |
| "grad_norm": 2.675050735473633, |
| "learning_rate": 9.15982312065698e-07, |
| "loss": 0.3256, |
| "step": 805 |
| }, |
| { |
| "epoch": 0.17284609229127768, |
| "grad_norm": 2.4576005935668945, |
| "learning_rate": 9.154558854495683e-07, |
| "loss": 0.3127, |
| "step": 810 |
| }, |
| { |
| "epoch": 0.17391304347826086, |
| "grad_norm": 2.73311710357666, |
| "learning_rate": 9.150347441566645e-07, |
| "loss": 0.3508, |
| "step": 815 |
| }, |
| { |
| "epoch": 0.17497999466524405, |
| "grad_norm": 2.6573374271392822, |
| "learning_rate": 9.145083175405348e-07, |
| "loss": 0.3643, |
| "step": 820 |
| }, |
| { |
| "epoch": 0.17604694585222727, |
| "grad_norm": 2.7585701942443848, |
| "learning_rate": 9.139818909244052e-07, |
| "loss": 0.3321, |
| "step": 825 |
| }, |
| { |
| "epoch": 0.17711389703921046, |
| "grad_norm": 2.8242616653442383, |
| "learning_rate": 9.134554643082753e-07, |
| "loss": 0.3175, |
| "step": 830 |
| }, |
| { |
| "epoch": 0.17818084822619365, |
| "grad_norm": 3.1244609355926514, |
| "learning_rate": 9.129290376921457e-07, |
| "loss": 0.3641, |
| "step": 835 |
| }, |
| { |
| "epoch": 0.17924779941317684, |
| "grad_norm": 2.4807212352752686, |
| "learning_rate": 9.12402611076016e-07, |
| "loss": 0.3464, |
| "step": 840 |
| }, |
| { |
| "epoch": 0.18031475060016006, |
| "grad_norm": 4.587203502655029, |
| "learning_rate": 9.118761844598863e-07, |
| "loss": 0.3199, |
| "step": 845 |
| }, |
| { |
| "epoch": 0.18138170178714325, |
| "grad_norm": 3.1150426864624023, |
| "learning_rate": 9.113497578437566e-07, |
| "loss": 0.3129, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.18244865297412644, |
| "grad_norm": 2.80098557472229, |
| "learning_rate": 9.108233312276268e-07, |
| "loss": 0.3296, |
| "step": 855 |
| }, |
| { |
| "epoch": 0.18351560416110962, |
| "grad_norm": 2.6048800945281982, |
| "learning_rate": 9.102969046114971e-07, |
| "loss": 0.3426, |
| "step": 860 |
| }, |
| { |
| "epoch": 0.1845825553480928, |
| "grad_norm": 2.4776954650878906, |
| "learning_rate": 9.097704779953675e-07, |
| "loss": 0.3389, |
| "step": 865 |
| }, |
| { |
| "epoch": 0.18564950653507603, |
| "grad_norm": 3.662856101989746, |
| "learning_rate": 9.092440513792377e-07, |
| "loss": 0.3191, |
| "step": 870 |
| }, |
| { |
| "epoch": 0.18671645772205922, |
| "grad_norm": 5.855990886688232, |
| "learning_rate": 9.08717624763108e-07, |
| "loss": 0.3491, |
| "step": 875 |
| }, |
| { |
| "epoch": 0.1877834089090424, |
| "grad_norm": 2.6081950664520264, |
| "learning_rate": 9.081911981469783e-07, |
| "loss": 0.3256, |
| "step": 880 |
| }, |
| { |
| "epoch": 0.1888503600960256, |
| "grad_norm": 3.206470251083374, |
| "learning_rate": 9.076647715308485e-07, |
| "loss": 0.3419, |
| "step": 885 |
| }, |
| { |
| "epoch": 0.18991731128300882, |
| "grad_norm": 3.7217066287994385, |
| "learning_rate": 9.071383449147189e-07, |
| "loss": 0.3369, |
| "step": 890 |
| }, |
| { |
| "epoch": 0.190984262469992, |
| "grad_norm": 4.7247633934021, |
| "learning_rate": 9.066119182985891e-07, |
| "loss": 0.3664, |
| "step": 895 |
| }, |
| { |
| "epoch": 0.1920512136569752, |
| "grad_norm": 3.455446481704712, |
| "learning_rate": 9.060854916824594e-07, |
| "loss": 0.3171, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.19311816484395838, |
| "grad_norm": 2.6066224575042725, |
| "learning_rate": 9.055590650663297e-07, |
| "loss": 0.3266, |
| "step": 905 |
| }, |
| { |
| "epoch": 0.19418511603094157, |
| "grad_norm": 3.2545228004455566, |
| "learning_rate": 9.050326384502e-07, |
| "loss": 0.3201, |
| "step": 910 |
| }, |
| { |
| "epoch": 0.1952520672179248, |
| "grad_norm": 3.8174829483032227, |
| "learning_rate": 9.045062118340702e-07, |
| "loss": 0.3537, |
| "step": 915 |
| }, |
| { |
| "epoch": 0.19631901840490798, |
| "grad_norm": 3.725991725921631, |
| "learning_rate": 9.039797852179406e-07, |
| "loss": 0.3373, |
| "step": 920 |
| }, |
| { |
| "epoch": 0.19738596959189117, |
| "grad_norm": 2.5402047634124756, |
| "learning_rate": 9.034533586018108e-07, |
| "loss": 0.3107, |
| "step": 925 |
| }, |
| { |
| "epoch": 0.19845292077887436, |
| "grad_norm": 2.3887853622436523, |
| "learning_rate": 9.029269319856812e-07, |
| "loss": 0.3255, |
| "step": 930 |
| }, |
| { |
| "epoch": 0.19951987196585758, |
| "grad_norm": 2.83178973197937, |
| "learning_rate": 9.024005053695514e-07, |
| "loss": 0.3192, |
| "step": 935 |
| }, |
| { |
| "epoch": 0.20058682315284077, |
| "grad_norm": 2.540933132171631, |
| "learning_rate": 9.018740787534217e-07, |
| "loss": 0.3414, |
| "step": 940 |
| }, |
| { |
| "epoch": 0.20165377433982395, |
| "grad_norm": 2.799577474594116, |
| "learning_rate": 9.01347652137292e-07, |
| "loss": 0.3383, |
| "step": 945 |
| }, |
| { |
| "epoch": 0.20272072552680714, |
| "grad_norm": 2.5945661067962646, |
| "learning_rate": 9.008212255211624e-07, |
| "loss": 0.3414, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.20378767671379033, |
| "grad_norm": 3.102372407913208, |
| "learning_rate": 9.002947989050325e-07, |
| "loss": 0.33, |
| "step": 955 |
| }, |
| { |
| "epoch": 0.20485462790077355, |
| "grad_norm": 3.1992921829223633, |
| "learning_rate": 8.997683722889029e-07, |
| "loss": 0.3719, |
| "step": 960 |
| }, |
| { |
| "epoch": 0.20592157908775674, |
| "grad_norm": 3.082578420639038, |
| "learning_rate": 8.992419456727732e-07, |
| "loss": 0.3412, |
| "step": 965 |
| }, |
| { |
| "epoch": 0.20698853027473993, |
| "grad_norm": 2.406475067138672, |
| "learning_rate": 8.987155190566435e-07, |
| "loss": 0.3069, |
| "step": 970 |
| }, |
| { |
| "epoch": 0.20805548146172312, |
| "grad_norm": 3.033055067062378, |
| "learning_rate": 8.981890924405137e-07, |
| "loss": 0.3683, |
| "step": 975 |
| }, |
| { |
| "epoch": 0.2091224326487063, |
| "grad_norm": 2.807874917984009, |
| "learning_rate": 8.97662665824384e-07, |
| "loss": 0.3024, |
| "step": 980 |
| }, |
| { |
| "epoch": 0.21018938383568952, |
| "grad_norm": 2.6849870681762695, |
| "learning_rate": 8.971362392082543e-07, |
| "loss": 0.3248, |
| "step": 985 |
| }, |
| { |
| "epoch": 0.21125633502267271, |
| "grad_norm": 3.144150495529175, |
| "learning_rate": 8.966098125921247e-07, |
| "loss": 0.3158, |
| "step": 990 |
| }, |
| { |
| "epoch": 0.2123232862096559, |
| "grad_norm": 2.495368003845215, |
| "learning_rate": 8.960833859759948e-07, |
| "loss": 0.312, |
| "step": 995 |
| }, |
| { |
| "epoch": 0.2133902373966391, |
| "grad_norm": 2.2597124576568604, |
| "learning_rate": 8.955569593598652e-07, |
| "loss": 0.3053, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.2133902373966391, |
| "eval_bleu": 0.6173395426659279, |
| "eval_cer": 5.420346953040605, |
| "eval_loss": 0.3327235281467438, |
| "eval_runtime": 1022.0953, |
| "eval_samples_per_second": 0.898, |
| "eval_steps_per_second": 0.225, |
| "eval_wer": 20.46380804775845, |
| "step": 1000 |
| } |
| ], |
| "logging_steps": 5, |
| "max_steps": 9500, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "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": 1.358829453312e+20, |
| "train_batch_size": 8, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
