| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 1624, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0006157635467980296, | |
| "grad_norm": 7.724584736238262, | |
| "learning_rate": 1.8404907975460124e-08, | |
| "loss": 0.9586, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.003078817733990148, | |
| "grad_norm": 8.79591767451402, | |
| "learning_rate": 9.202453987730061e-08, | |
| "loss": 1.0017, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.006157635467980296, | |
| "grad_norm": 7.712094187091605, | |
| "learning_rate": 1.8404907975460122e-07, | |
| "loss": 1.0138, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.009236453201970444, | |
| "grad_norm": 6.686984831397533, | |
| "learning_rate": 2.7607361963190183e-07, | |
| "loss": 0.9935, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.012315270935960592, | |
| "grad_norm": 2.9705498717633567, | |
| "learning_rate": 3.6809815950920245e-07, | |
| "loss": 0.9693, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01539408866995074, | |
| "grad_norm": 5.230925406141613, | |
| "learning_rate": 4.601226993865031e-07, | |
| "loss": 0.9517, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01847290640394089, | |
| "grad_norm": 2.0252746766726393, | |
| "learning_rate": 5.521472392638037e-07, | |
| "loss": 0.9302, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.021551724137931036, | |
| "grad_norm": 3.420398960371135, | |
| "learning_rate": 6.441717791411044e-07, | |
| "loss": 0.917, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.024630541871921183, | |
| "grad_norm": 2.4920184562580787, | |
| "learning_rate": 7.361963190184049e-07, | |
| "loss": 0.9173, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.02770935960591133, | |
| "grad_norm": 1.7523436431972144, | |
| "learning_rate": 8.282208588957056e-07, | |
| "loss": 0.8939, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.03078817733990148, | |
| "grad_norm": 1.5172341196305177, | |
| "learning_rate": 9.202453987730062e-07, | |
| "loss": 0.9125, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.033866995073891626, | |
| "grad_norm": 1.2637609259029952, | |
| "learning_rate": 1.0122699386503068e-06, | |
| "loss": 0.8659, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.03694581280788178, | |
| "grad_norm": 1.405211777798568, | |
| "learning_rate": 1.1042944785276073e-06, | |
| "loss": 0.8606, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.04002463054187192, | |
| "grad_norm": 1.400488357062075, | |
| "learning_rate": 1.1963190184049078e-06, | |
| "loss": 0.8914, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.04310344827586207, | |
| "grad_norm": 1.3965560372062285, | |
| "learning_rate": 1.2883435582822088e-06, | |
| "loss": 0.8638, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.046182266009852216, | |
| "grad_norm": 1.397481181493605, | |
| "learning_rate": 1.3803680981595093e-06, | |
| "loss": 0.8872, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.04926108374384237, | |
| "grad_norm": 1.3920893415057805, | |
| "learning_rate": 1.4723926380368098e-06, | |
| "loss": 0.8637, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.05233990147783251, | |
| "grad_norm": 1.4291285911903127, | |
| "learning_rate": 1.5644171779141105e-06, | |
| "loss": 0.8743, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.05541871921182266, | |
| "grad_norm": 1.3542212432257583, | |
| "learning_rate": 1.6564417177914112e-06, | |
| "loss": 0.8523, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.058497536945812806, | |
| "grad_norm": 1.4178159427780204, | |
| "learning_rate": 1.7484662576687115e-06, | |
| "loss": 0.8696, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.06157635467980296, | |
| "grad_norm": 1.3221606943995998, | |
| "learning_rate": 1.8404907975460124e-06, | |
| "loss": 0.8772, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06465517241379311, | |
| "grad_norm": 1.4394077687183169, | |
| "learning_rate": 1.932515337423313e-06, | |
| "loss": 0.8695, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.06773399014778325, | |
| "grad_norm": 1.3808456562566545, | |
| "learning_rate": 2.0245398773006137e-06, | |
| "loss": 0.8829, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0708128078817734, | |
| "grad_norm": 1.4593385524523184, | |
| "learning_rate": 2.1165644171779144e-06, | |
| "loss": 0.8715, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.07389162561576355, | |
| "grad_norm": 1.4179679608588616, | |
| "learning_rate": 2.2085889570552147e-06, | |
| "loss": 0.8614, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.0769704433497537, | |
| "grad_norm": 1.4261678544854806, | |
| "learning_rate": 2.3006134969325154e-06, | |
| "loss": 0.8509, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.08004926108374384, | |
| "grad_norm": 1.5460976196498386, | |
| "learning_rate": 2.3926380368098157e-06, | |
| "loss": 0.8593, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.08312807881773399, | |
| "grad_norm": 1.481503657824365, | |
| "learning_rate": 2.4846625766871164e-06, | |
| "loss": 0.8706, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.08620689655172414, | |
| "grad_norm": 1.4651088227604514, | |
| "learning_rate": 2.5766871165644175e-06, | |
| "loss": 0.8782, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.08928571428571429, | |
| "grad_norm": 1.4218537682816195, | |
| "learning_rate": 2.668711656441718e-06, | |
| "loss": 0.858, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.09236453201970443, | |
| "grad_norm": 1.5217714939469515, | |
| "learning_rate": 2.7607361963190186e-06, | |
| "loss": 0.858, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.09544334975369458, | |
| "grad_norm": 1.4463329417962474, | |
| "learning_rate": 2.852760736196319e-06, | |
| "loss": 0.8901, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.09852216748768473, | |
| "grad_norm": 1.2738127091018583, | |
| "learning_rate": 2.9447852760736196e-06, | |
| "loss": 0.851, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.10160098522167488, | |
| "grad_norm": 1.4212078637635643, | |
| "learning_rate": 2.999986128612664e-06, | |
| "loss": 0.8679, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.10467980295566502, | |
| "grad_norm": 1.4528112933674504, | |
| "learning_rate": 2.9998300784514776e-06, | |
| "loss": 0.8659, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.10775862068965517, | |
| "grad_norm": 1.3669524004218505, | |
| "learning_rate": 2.99950065699352e-06, | |
| "loss": 0.8814, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.11083743842364532, | |
| "grad_norm": 1.2653588333266155, | |
| "learning_rate": 2.9989979023179235e-06, | |
| "loss": 0.8585, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.11391625615763547, | |
| "grad_norm": 1.29184019357291, | |
| "learning_rate": 2.9983218725400893e-06, | |
| "loss": 0.8508, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.11699507389162561, | |
| "grad_norm": 1.2992358486439284, | |
| "learning_rate": 2.9974726458049776e-06, | |
| "loss": 0.8633, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.12007389162561577, | |
| "grad_norm": 1.250920595649257, | |
| "learning_rate": 2.9964503202780676e-06, | |
| "loss": 0.8468, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.12315270935960591, | |
| "grad_norm": 1.3222082464366094, | |
| "learning_rate": 2.9952550141340154e-06, | |
| "loss": 0.8722, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.12623152709359606, | |
| "grad_norm": 1.5229405032439445, | |
| "learning_rate": 2.99388686554299e-06, | |
| "loss": 0.8768, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.12931034482758622, | |
| "grad_norm": 1.3539647424442673, | |
| "learning_rate": 2.9923460326547038e-06, | |
| "loss": 0.8416, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.13238916256157635, | |
| "grad_norm": 2.1043494775741354, | |
| "learning_rate": 2.990632693580131e-06, | |
| "loss": 0.8574, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.1354679802955665, | |
| "grad_norm": 1.4039066575750099, | |
| "learning_rate": 2.988747046370918e-06, | |
| "loss": 0.8629, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.13854679802955666, | |
| "grad_norm": 1.4041519802548768, | |
| "learning_rate": 2.986689308996492e-06, | |
| "loss": 0.8723, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.1416256157635468, | |
| "grad_norm": 2.804730066260758, | |
| "learning_rate": 2.984459719318862e-06, | |
| "loss": 0.8578, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.14470443349753695, | |
| "grad_norm": 1.3160437065661477, | |
| "learning_rate": 2.9820585350651265e-06, | |
| "loss": 0.8485, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.1477832512315271, | |
| "grad_norm": 1.2698953454345538, | |
| "learning_rate": 2.9794860337976802e-06, | |
| "loss": 0.8587, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.15086206896551724, | |
| "grad_norm": 1.9733415697734495, | |
| "learning_rate": 2.9767425128821288e-06, | |
| "loss": 0.8506, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.1539408866995074, | |
| "grad_norm": 1.3672039273596184, | |
| "learning_rate": 2.9738282894529177e-06, | |
| "loss": 0.8665, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.15701970443349753, | |
| "grad_norm": 1.3462677423269411, | |
| "learning_rate": 2.9707437003766704e-06, | |
| "loss": 0.8573, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.16009852216748768, | |
| "grad_norm": 1.395722464948162, | |
| "learning_rate": 2.96748910221325e-06, | |
| "loss": 0.8646, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.16317733990147784, | |
| "grad_norm": 1.2949662167486484, | |
| "learning_rate": 2.964064871174544e-06, | |
| "loss": 0.8594, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.16625615763546797, | |
| "grad_norm": 1.3986879793877978, | |
| "learning_rate": 2.9604714030809755e-06, | |
| "loss": 0.85, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.16933497536945813, | |
| "grad_norm": 1.3544807529329361, | |
| "learning_rate": 2.9567091133157484e-06, | |
| "loss": 0.8615, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.1724137931034483, | |
| "grad_norm": 1.25058167504149, | |
| "learning_rate": 2.9527784367768305e-06, | |
| "loss": 0.8561, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.17549261083743842, | |
| "grad_norm": 1.2827316707560608, | |
| "learning_rate": 2.948679827826687e-06, | |
| "loss": 0.8632, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.17857142857142858, | |
| "grad_norm": 1.3580739609598804, | |
| "learning_rate": 2.9444137602397515e-06, | |
| "loss": 0.8823, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.1816502463054187, | |
| "grad_norm": 1.2894476992215949, | |
| "learning_rate": 2.9399807271476674e-06, | |
| "loss": 0.854, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.18472906403940886, | |
| "grad_norm": 1.3479537928129113, | |
| "learning_rate": 2.935381240982281e-06, | |
| "loss": 0.8585, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.18780788177339902, | |
| "grad_norm": 1.294266030285128, | |
| "learning_rate": 2.9306158334164066e-06, | |
| "loss": 0.8503, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.19088669950738915, | |
| "grad_norm": 1.2555453597545074, | |
| "learning_rate": 2.9256850553023724e-06, | |
| "loss": 0.868, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.1939655172413793, | |
| "grad_norm": 1.1788928865051034, | |
| "learning_rate": 2.920589476608343e-06, | |
| "loss": 0.8339, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.19704433497536947, | |
| "grad_norm": 1.2770185785021044, | |
| "learning_rate": 2.9153296863524315e-06, | |
| "loss": 0.8524, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.2001231527093596, | |
| "grad_norm": 1.394262786635702, | |
| "learning_rate": 2.90990629253462e-06, | |
| "loss": 0.8679, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.20320197044334976, | |
| "grad_norm": 1.275454411248854, | |
| "learning_rate": 2.9043199220664704e-06, | |
| "loss": 0.8456, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.2062807881773399, | |
| "grad_norm": 1.1750409774056632, | |
| "learning_rate": 2.8985712206986627e-06, | |
| "loss": 0.8614, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.20935960591133004, | |
| "grad_norm": 1.2682207809915187, | |
| "learning_rate": 2.8926608529463473e-06, | |
| "loss": 0.8547, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.2124384236453202, | |
| "grad_norm": 1.2484760929773475, | |
| "learning_rate": 2.8865895020123326e-06, | |
| "loss": 0.856, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.21551724137931033, | |
| "grad_norm": 1.218507223565624, | |
| "learning_rate": 2.880357869708111e-06, | |
| "loss": 0.8828, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.2185960591133005, | |
| "grad_norm": 1.2015850857754256, | |
| "learning_rate": 2.8739666763727315e-06, | |
| "loss": 0.8602, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.22167487684729065, | |
| "grad_norm": 1.3408884610418483, | |
| "learning_rate": 2.8674166607895357e-06, | |
| "loss": 0.8563, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.22475369458128078, | |
| "grad_norm": 1.4836285692174822, | |
| "learning_rate": 2.8607085801007565e-06, | |
| "loss": 0.8471, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.22783251231527094, | |
| "grad_norm": 1.3404651597215993, | |
| "learning_rate": 2.85384320972e-06, | |
| "loss": 0.8482, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.2309113300492611, | |
| "grad_norm": 1.5386066904678533, | |
| "learning_rate": 2.846821343242608e-06, | |
| "loss": 0.8389, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.23399014778325122, | |
| "grad_norm": 1.2754166515854897, | |
| "learning_rate": 2.839643792353928e-06, | |
| "loss": 0.8653, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.23706896551724138, | |
| "grad_norm": 1.3945185547743246, | |
| "learning_rate": 2.832311386735483e-06, | |
| "loss": 0.8808, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.24014778325123154, | |
| "grad_norm": 1.2744156648350857, | |
| "learning_rate": 2.824824973969069e-06, | |
| "loss": 0.8558, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.24322660098522167, | |
| "grad_norm": 1.405344843136268, | |
| "learning_rate": 2.817185419438777e-06, | |
| "loss": 0.856, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.24630541871921183, | |
| "grad_norm": 1.3542416852004673, | |
| "learning_rate": 2.8093936062309614e-06, | |
| "loss": 0.8386, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.24938423645320196, | |
| "grad_norm": 1.2804821522839576, | |
| "learning_rate": 2.8014504350321594e-06, | |
| "loss": 0.874, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.2524630541871921, | |
| "grad_norm": 1.403073424101837, | |
| "learning_rate": 2.7933568240249776e-06, | |
| "loss": 0.8525, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.2555418719211823, | |
| "grad_norm": 1.5114916625601154, | |
| "learning_rate": 2.7851137087819562e-06, | |
| "loss": 0.8657, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.25862068965517243, | |
| "grad_norm": 1.3288954210806725, | |
| "learning_rate": 2.776722042157421e-06, | |
| "loss": 0.8539, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.2616995073891626, | |
| "grad_norm": 1.2388602665884483, | |
| "learning_rate": 2.768182794177341e-06, | |
| "loss": 0.8492, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.2647783251231527, | |
| "grad_norm": 1.3100863403911909, | |
| "learning_rate": 2.7594969519271988e-06, | |
| "loss": 0.84, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.26785714285714285, | |
| "grad_norm": 1.3687219418507923, | |
| "learning_rate": 2.7506655194378874e-06, | |
| "loss": 0.8514, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.270935960591133, | |
| "grad_norm": 1.4724892872074864, | |
| "learning_rate": 2.7416895175696533e-06, | |
| "loss": 0.8641, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.27401477832512317, | |
| "grad_norm": 1.2885096742121314, | |
| "learning_rate": 2.7325699838940894e-06, | |
| "loss": 0.8671, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.2770935960591133, | |
| "grad_norm": 1.2454446036705014, | |
| "learning_rate": 2.723307972574199e-06, | |
| "loss": 0.8634, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.2801724137931034, | |
| "grad_norm": 1.2916820035058567, | |
| "learning_rate": 2.713904554242539e-06, | |
| "loss": 0.8541, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.2832512315270936, | |
| "grad_norm": 1.2593373941499395, | |
| "learning_rate": 2.7043608158774645e-06, | |
| "loss": 0.8535, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.28633004926108374, | |
| "grad_norm": 1.2902647841096262, | |
| "learning_rate": 2.6946778606774777e-06, | |
| "loss": 0.8534, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.2894088669950739, | |
| "grad_norm": 1.2821496105379782, | |
| "learning_rate": 2.684856807933706e-06, | |
| "loss": 0.858, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.29248768472906406, | |
| "grad_norm": 1.2703981820641408, | |
| "learning_rate": 2.6748987929005187e-06, | |
| "loss": 0.8494, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.2955665024630542, | |
| "grad_norm": 1.2473762414475023, | |
| "learning_rate": 2.664804966664298e-06, | |
| "loss": 0.872, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.2986453201970443, | |
| "grad_norm": 1.440650000265737, | |
| "learning_rate": 2.6545764960103823e-06, | |
| "loss": 0.8612, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.3017241379310345, | |
| "grad_norm": 1.2075401918303994, | |
| "learning_rate": 2.6442145632881894e-06, | |
| "loss": 0.8573, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.30480295566502463, | |
| "grad_norm": 1.1592052019959331, | |
| "learning_rate": 2.6337203662745465e-06, | |
| "loss": 0.8655, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.3078817733990148, | |
| "grad_norm": 1.2734539628875496, | |
| "learning_rate": 2.623095118035235e-06, | |
| "loss": 0.8672, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.31096059113300495, | |
| "grad_norm": 1.2334639159935359, | |
| "learning_rate": 2.612340046784765e-06, | |
| "loss": 0.8678, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.31403940886699505, | |
| "grad_norm": 1.2822243726091629, | |
| "learning_rate": 2.601456395744403e-06, | |
| "loss": 0.8604, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.3171182266009852, | |
| "grad_norm": 1.31932161927138, | |
| "learning_rate": 2.590445422998462e-06, | |
| "loss": 0.8774, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.32019704433497537, | |
| "grad_norm": 1.2448157244019422, | |
| "learning_rate": 2.579308401348876e-06, | |
| "loss": 0.8539, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.3232758620689655, | |
| "grad_norm": 1.276792729621205, | |
| "learning_rate": 2.56804661816807e-06, | |
| "loss": 0.8426, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.3263546798029557, | |
| "grad_norm": 1.1505036776360344, | |
| "learning_rate": 2.556661375250149e-06, | |
| "loss": 0.8579, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.3294334975369458, | |
| "grad_norm": 1.1370655194463635, | |
| "learning_rate": 2.5451539886604165e-06, | |
| "loss": 0.8439, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.33251231527093594, | |
| "grad_norm": 1.2780922824871068, | |
| "learning_rate": 2.533525788583248e-06, | |
| "loss": 0.8554, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.3355911330049261, | |
| "grad_norm": 1.2281556867963181, | |
| "learning_rate": 2.5217781191683284e-06, | |
| "loss": 0.8439, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.33866995073891626, | |
| "grad_norm": 1.1944897494951232, | |
| "learning_rate": 2.509912338375275e-06, | |
| "loss": 0.8389, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.3417487684729064, | |
| "grad_norm": 1.2161826106487, | |
| "learning_rate": 2.497929817816667e-06, | |
| "loss": 0.8658, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.3448275862068966, | |
| "grad_norm": 1.2343890699461388, | |
| "learning_rate": 2.4858319425994978e-06, | |
| "loss": 0.8615, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.3479064039408867, | |
| "grad_norm": 1.3096521558859537, | |
| "learning_rate": 2.4736201111650593e-06, | |
| "loss": 0.8627, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.35098522167487683, | |
| "grad_norm": 1.260420116901076, | |
| "learning_rate": 2.4612957351272963e-06, | |
| "loss": 0.8572, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.354064039408867, | |
| "grad_norm": 1.2711909674880961, | |
| "learning_rate": 2.448860239109627e-06, | |
| "loss": 0.85, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.35714285714285715, | |
| "grad_norm": 1.3045345199158591, | |
| "learning_rate": 2.4363150605802704e-06, | |
| "loss": 0.8242, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.3602216748768473, | |
| "grad_norm": 1.255053722802097, | |
| "learning_rate": 2.423661649686081e-06, | |
| "loss": 0.8664, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.3633004926108374, | |
| "grad_norm": 1.1808467142096963, | |
| "learning_rate": 2.41090146908492e-06, | |
| "loss": 0.8254, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.36637931034482757, | |
| "grad_norm": 1.2161111833295437, | |
| "learning_rate": 2.398035993776582e-06, | |
| "loss": 0.8303, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.3694581280788177, | |
| "grad_norm": 1.2018738392776112, | |
| "learning_rate": 2.385066710932294e-06, | |
| "loss": 0.8471, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.3725369458128079, | |
| "grad_norm": 1.3104602005273063, | |
| "learning_rate": 2.3719951197228068e-06, | |
| "loss": 0.8396, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.37561576354679804, | |
| "grad_norm": 1.3053912734059634, | |
| "learning_rate": 2.3588227311451007e-06, | |
| "loss": 0.8557, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.3786945812807882, | |
| "grad_norm": 1.1887756973639412, | |
| "learning_rate": 2.3455510678477217e-06, | |
| "loss": 0.8428, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.3817733990147783, | |
| "grad_norm": 1.3202665606343256, | |
| "learning_rate": 2.3321816639547747e-06, | |
| "loss": 0.8443, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.38485221674876846, | |
| "grad_norm": 1.481406724728173, | |
| "learning_rate": 2.3187160648885864e-06, | |
| "loss": 0.8455, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.3879310344827586, | |
| "grad_norm": 1.253872927504699, | |
| "learning_rate": 2.305155827191066e-06, | |
| "loss": 0.8424, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.3910098522167488, | |
| "grad_norm": 1.1863474981605258, | |
| "learning_rate": 2.291502518343774e-06, | |
| "loss": 0.8657, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.39408866995073893, | |
| "grad_norm": 1.3010619531118812, | |
| "learning_rate": 2.2777577165867354e-06, | |
| "loss": 0.8527, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.39716748768472904, | |
| "grad_norm": 1.2446913565784508, | |
| "learning_rate": 2.2639230107360033e-06, | |
| "loss": 0.8465, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.4002463054187192, | |
| "grad_norm": 1.2678544206625766, | |
| "learning_rate": 2.25e-06, | |
| "loss": 0.8398, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.40332512315270935, | |
| "grad_norm": 1.2805775413961258, | |
| "learning_rate": 2.235990293794659e-06, | |
| "loss": 0.8698, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.4064039408866995, | |
| "grad_norm": 1.2080762381092112, | |
| "learning_rate": 2.2218955115573864e-06, | |
| "loss": 0.8551, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.40948275862068967, | |
| "grad_norm": 1.31904016779892, | |
| "learning_rate": 2.2077172825598645e-06, | |
| "loss": 0.8429, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.4125615763546798, | |
| "grad_norm": 1.1884365693687557, | |
| "learning_rate": 2.1934572457197163e-06, | |
| "loss": 0.8592, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.41564039408866993, | |
| "grad_norm": 1.143061537231462, | |
| "learning_rate": 2.179117049411057e-06, | |
| "loss": 0.8619, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.4187192118226601, | |
| "grad_norm": 1.2386292522406654, | |
| "learning_rate": 2.164698351273952e-06, | |
| "loss": 0.8545, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.42179802955665024, | |
| "grad_norm": 1.2021340390709248, | |
| "learning_rate": 2.150202818022805e-06, | |
| "loss": 0.8501, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.4248768472906404, | |
| "grad_norm": 1.2781965578814818, | |
| "learning_rate": 2.1356321252536947e-06, | |
| "loss": 0.843, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.42795566502463056, | |
| "grad_norm": 1.2463234389480125, | |
| "learning_rate": 2.1209879572506836e-06, | |
| "loss": 0.8532, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.43103448275862066, | |
| "grad_norm": 1.2490176056060265, | |
| "learning_rate": 2.10627200679113e-06, | |
| "loss": 0.8512, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.4341133004926108, | |
| "grad_norm": 1.186593391495473, | |
| "learning_rate": 2.09148597495001e-06, | |
| "loss": 0.8642, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.437192118226601, | |
| "grad_norm": 1.3071569875293647, | |
| "learning_rate": 2.0766315709032837e-06, | |
| "loss": 0.8312, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.44027093596059114, | |
| "grad_norm": 1.2158014479042887, | |
| "learning_rate": 2.0617105117303247e-06, | |
| "loss": 0.8503, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.4433497536945813, | |
| "grad_norm": 1.2188719938005892, | |
| "learning_rate": 2.046724522215437e-06, | |
| "loss": 0.8381, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.44642857142857145, | |
| "grad_norm": 1.2816897559505145, | |
| "learning_rate": 2.03167533464848e-06, | |
| "loss": 0.8531, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.44950738916256155, | |
| "grad_norm": 1.1545846489999156, | |
| "learning_rate": 2.016564688624627e-06, | |
| "loss": 0.8556, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.4525862068965517, | |
| "grad_norm": 1.247432677442672, | |
| "learning_rate": 2.001394330843276e-06, | |
| "loss": 0.8273, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.45566502463054187, | |
| "grad_norm": 1.200421987011754, | |
| "learning_rate": 1.9861660149061435e-06, | |
| "loss": 0.8559, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.45874384236453203, | |
| "grad_norm": 1.242386621339582, | |
| "learning_rate": 1.9708815011145605e-06, | |
| "loss": 0.847, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.4618226600985222, | |
| "grad_norm": 1.2714576314445647, | |
| "learning_rate": 1.9555425562659878e-06, | |
| "loss": 0.8493, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.4649014778325123, | |
| "grad_norm": 1.199306019483452, | |
| "learning_rate": 1.940150953449791e-06, | |
| "loss": 0.8256, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.46798029556650245, | |
| "grad_norm": 1.176599940688529, | |
| "learning_rate": 1.924708471842276e-06, | |
| "loss": 0.8414, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.4710591133004926, | |
| "grad_norm": 1.1674248964748613, | |
| "learning_rate": 1.90921689650103e-06, | |
| "loss": 0.8429, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.47413793103448276, | |
| "grad_norm": 1.2904494723192372, | |
| "learning_rate": 1.8936780181585799e-06, | |
| "loss": 0.8361, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.4772167487684729, | |
| "grad_norm": 1.2504397812966646, | |
| "learning_rate": 1.8780936330153935e-06, | |
| "loss": 0.8677, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.4802955665024631, | |
| "grad_norm": 1.1554267622763779, | |
| "learning_rate": 1.86246554253225e-06, | |
| "loss": 0.8502, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.4833743842364532, | |
| "grad_norm": 1.2494756999598795, | |
| "learning_rate": 1.8467955532220017e-06, | |
| "loss": 0.8431, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.48645320197044334, | |
| "grad_norm": 1.1864925645414461, | |
| "learning_rate": 1.831085476440753e-06, | |
| "loss": 0.849, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.4895320197044335, | |
| "grad_norm": 1.230711251312914, | |
| "learning_rate": 1.8153371281784756e-06, | |
| "loss": 0.8516, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.49261083743842365, | |
| "grad_norm": 1.2701213116535017, | |
| "learning_rate": 1.7995523288490959e-06, | |
| "loss": 0.8649, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.4956896551724138, | |
| "grad_norm": 1.2920442137078154, | |
| "learning_rate": 1.783732903080062e-06, | |
| "loss": 0.8495, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.4987684729064039, | |
| "grad_norm": 1.2610776640382462, | |
| "learning_rate": 1.7678806795014293e-06, | |
| "loss": 0.8486, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.5018472906403941, | |
| "grad_norm": 1.2512197750575502, | |
| "learning_rate": 1.751997490534482e-06, | |
| "loss": 0.8433, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.5049261083743842, | |
| "grad_norm": 1.2693463293367837, | |
| "learning_rate": 1.7360851721799163e-06, | |
| "loss": 0.8159, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.5080049261083743, | |
| "grad_norm": 1.1915234367340757, | |
| "learning_rate": 1.7201455638056097e-06, | |
| "loss": 0.8367, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.5110837438423645, | |
| "grad_norm": 1.3350984034693405, | |
| "learning_rate": 1.7041805079340006e-06, | |
| "loss": 0.8412, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.5141625615763546, | |
| "grad_norm": 1.1953853505995866, | |
| "learning_rate": 1.6881918500291052e-06, | |
| "loss": 0.8608, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.5172413793103449, | |
| "grad_norm": 1.2261829318707587, | |
| "learning_rate": 1.6721814382831911e-06, | |
| "loss": 0.8235, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.520320197044335, | |
| "grad_norm": 1.197786936841144, | |
| "learning_rate": 1.6561511234031394e-06, | |
| "loss": 0.8364, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.5233990147783252, | |
| "grad_norm": 1.2488334159401087, | |
| "learning_rate": 1.6401027583965135e-06, | |
| "loss": 0.847, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.5264778325123153, | |
| "grad_norm": 1.1650388115198271, | |
| "learning_rate": 1.624038198357361e-06, | |
| "loss": 0.852, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.5295566502463054, | |
| "grad_norm": 1.1509354844941455, | |
| "learning_rate": 1.6079593002517785e-06, | |
| "loss": 0.875, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.5326354679802956, | |
| "grad_norm": 1.2308644335652692, | |
| "learning_rate": 1.5918679227032564e-06, | |
| "loss": 0.8358, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.5357142857142857, | |
| "grad_norm": 1.1735480926744941, | |
| "learning_rate": 1.575765925777834e-06, | |
| "loss": 0.8408, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.5387931034482759, | |
| "grad_norm": 1.2034817032001748, | |
| "learning_rate": 1.559655170769084e-06, | |
| "loss": 0.8405, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.541871921182266, | |
| "grad_norm": 1.268978613523728, | |
| "learning_rate": 1.543537519982963e-06, | |
| "loss": 0.847, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.5449507389162561, | |
| "grad_norm": 1.2241899601063064, | |
| "learning_rate": 1.5274148365225372e-06, | |
| "loss": 0.8394, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.5480295566502463, | |
| "grad_norm": 1.1676337312724008, | |
| "learning_rate": 1.5112889840726194e-06, | |
| "loss": 0.8622, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.5511083743842364, | |
| "grad_norm": 1.1301543137797185, | |
| "learning_rate": 1.4951618266843384e-06, | |
| "loss": 0.8266, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.5541871921182266, | |
| "grad_norm": 1.30141347624095, | |
| "learning_rate": 1.4790352285596656e-06, | |
| "loss": 0.8408, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.5572660098522167, | |
| "grad_norm": 1.3012499210796624, | |
| "learning_rate": 1.4629110538359251e-06, | |
| "loss": 0.8329, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.5603448275862069, | |
| "grad_norm": 1.2275901091379853, | |
| "learning_rate": 1.4467911663703118e-06, | |
| "loss": 0.8341, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.5634236453201971, | |
| "grad_norm": 1.2421267817837207, | |
| "learning_rate": 1.4306774295244372e-06, | |
| "loss": 0.8484, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.5665024630541872, | |
| "grad_norm": 1.1849317684107135, | |
| "learning_rate": 1.4145717059489405e-06, | |
| "loss": 0.845, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.5695812807881774, | |
| "grad_norm": 1.1560168622553235, | |
| "learning_rate": 1.3984758573681741e-06, | |
| "loss": 0.8343, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.5726600985221675, | |
| "grad_norm": 1.303736696697277, | |
| "learning_rate": 1.3823917443649994e-06, | |
| "loss": 0.8146, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.5757389162561576, | |
| "grad_norm": 1.2387568449101967, | |
| "learning_rate": 1.3663212261657191e-06, | |
| "loss": 0.845, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.5788177339901478, | |
| "grad_norm": 1.2382013938085306, | |
| "learning_rate": 1.3502661604251562e-06, | |
| "loss": 0.8403, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.5818965517241379, | |
| "grad_norm": 1.1875665599598455, | |
| "learning_rate": 1.3342284030119233e-06, | |
| "loss": 0.8313, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.5849753694581281, | |
| "grad_norm": 1.1428289684636799, | |
| "learning_rate": 1.3182098077938954e-06, | |
| "loss": 0.8334, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.5880541871921182, | |
| "grad_norm": 1.2318379752097235, | |
| "learning_rate": 1.3022122264239134e-06, | |
| "loss": 0.8584, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.5911330049261084, | |
| "grad_norm": 1.1908093263154107, | |
| "learning_rate": 1.286237508125744e-06, | |
| "loss": 0.8573, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.5942118226600985, | |
| "grad_norm": 1.1684422268700045, | |
| "learning_rate": 1.2702874994803206e-06, | |
| "loss": 0.8376, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.5972906403940886, | |
| "grad_norm": 1.2315278737517545, | |
| "learning_rate": 1.254364044212291e-06, | |
| "loss": 0.8505, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.6003694581280788, | |
| "grad_norm": 1.2388539336501045, | |
| "learning_rate": 1.2384689829768929e-06, | |
| "loss": 0.824, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.603448275862069, | |
| "grad_norm": 1.246236595179206, | |
| "learning_rate": 1.2226041531471835e-06, | |
| "loss": 0.8553, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.6065270935960592, | |
| "grad_norm": 1.3256259396928607, | |
| "learning_rate": 1.2067713886016547e-06, | |
| "loss": 0.8266, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.6096059113300493, | |
| "grad_norm": 1.1362791976484927, | |
| "learning_rate": 1.1909725195122443e-06, | |
| "loss": 0.8394, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.6126847290640394, | |
| "grad_norm": 1.166355873497428, | |
| "learning_rate": 1.1752093721327784e-06, | |
| "loss": 0.8277, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.6157635467980296, | |
| "grad_norm": 1.1743428317131432, | |
| "learning_rate": 1.1594837685878725e-06, | |
| "loss": 0.8183, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.6188423645320197, | |
| "grad_norm": 1.2512344679138039, | |
| "learning_rate": 1.143797526662299e-06, | |
| "loss": 0.8492, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.6219211822660099, | |
| "grad_norm": 1.1784283535469242, | |
| "learning_rate": 1.1281524595908653e-06, | |
| "loss": 0.8268, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 1.1906881964438814, | |
| "learning_rate": 1.1125503758488158e-06, | |
| "loss": 0.8516, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.6280788177339901, | |
| "grad_norm": 1.2048898820900105, | |
| "learning_rate": 1.0969930789427798e-06, | |
| "loss": 0.8505, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.6311576354679803, | |
| "grad_norm": 1.2384331414468177, | |
| "learning_rate": 1.0814823672023007e-06, | |
| "loss": 0.8327, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.6342364532019704, | |
| "grad_norm": 1.2370825731431196, | |
| "learning_rate": 1.0660200335719569e-06, | |
| "loss": 0.8577, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.6373152709359606, | |
| "grad_norm": 1.1890634236003874, | |
| "learning_rate": 1.0506078654041095e-06, | |
| "loss": 0.8402, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.6403940886699507, | |
| "grad_norm": 1.1933853331056958, | |
| "learning_rate": 1.0352476442522963e-06, | |
| "loss": 0.8606, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.6434729064039408, | |
| "grad_norm": 1.2675635086457242, | |
| "learning_rate": 1.0199411456652932e-06, | |
| "loss": 0.8239, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.646551724137931, | |
| "grad_norm": 1.1317105500451243, | |
| "learning_rate": 1.004690138981871e-06, | |
| "loss": 0.8607, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.6496305418719212, | |
| "grad_norm": 1.1674980504418897, | |
| "learning_rate": 9.89496387126274e-07, | |
| "loss": 0.8396, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.6527093596059114, | |
| "grad_norm": 1.2991830801026518, | |
| "learning_rate": 9.74361646404432e-07, | |
| "loss": 0.8159, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.6557881773399015, | |
| "grad_norm": 1.2506097073385514, | |
| "learning_rate": 9.592876663009462e-07, | |
| "loss": 0.8434, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.6588669950738916, | |
| "grad_norm": 1.185318025796248, | |
| "learning_rate": 9.442761892768561e-07, | |
| "loss": 0.8189, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.6619458128078818, | |
| "grad_norm": 1.1985852310220102, | |
| "learning_rate": 9.293289505682223e-07, | |
| "loss": 0.8503, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.6650246305418719, | |
| "grad_norm": 1.1993709021835215, | |
| "learning_rate": 9.144476779855462e-07, | |
| "loss": 0.8644, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.6681034482758621, | |
| "grad_norm": 1.1947284213216074, | |
| "learning_rate": 8.99634091714042e-07, | |
| "loss": 0.8465, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.6711822660098522, | |
| "grad_norm": 1.1476627282557448, | |
| "learning_rate": 8.848899041147947e-07, | |
| "loss": 0.8372, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.6742610837438424, | |
| "grad_norm": 1.1027722973387861, | |
| "learning_rate": 8.702168195268239e-07, | |
| "loss": 0.8319, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.6773399014778325, | |
| "grad_norm": 1.1582755840301646, | |
| "learning_rate": 8.556165340700687e-07, | |
| "loss": 0.833, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.6804187192118226, | |
| "grad_norm": 1.2919754975931053, | |
| "learning_rate": 8.410907354493304e-07, | |
| "loss": 0.827, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.6834975369458128, | |
| "grad_norm": 1.1351158697269301, | |
| "learning_rate": 8.266411027591801e-07, | |
| "loss": 0.8461, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.6865763546798029, | |
| "grad_norm": 1.1950786235402677, | |
| "learning_rate": 8.12269306289868e-07, | |
| "loss": 0.8359, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.6896551724137931, | |
| "grad_norm": 1.2839040815642278, | |
| "learning_rate": 7.979770073342484e-07, | |
| "loss": 0.8451, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.6927339901477833, | |
| "grad_norm": 1.2286952134329439, | |
| "learning_rate": 7.837658579957422e-07, | |
| "loss": 0.8289, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.6958128078817734, | |
| "grad_norm": 1.1443853696759385, | |
| "learning_rate": 7.696375009973643e-07, | |
| "loss": 0.8217, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.6988916256157636, | |
| "grad_norm": 1.1889401437521319, | |
| "learning_rate": 7.555935694918371e-07, | |
| "loss": 0.8449, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.7019704433497537, | |
| "grad_norm": 1.1789396438426583, | |
| "learning_rate": 7.41635686872804e-07, | |
| "loss": 0.8419, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.7050492610837439, | |
| "grad_norm": 1.2331607811782397, | |
| "learning_rate": 7.277654665871785e-07, | |
| "loss": 0.8083, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.708128078817734, | |
| "grad_norm": 1.2248130017166632, | |
| "learning_rate": 7.139845119486371e-07, | |
| "loss": 0.8161, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.7112068965517241, | |
| "grad_norm": 1.1673298869963107, | |
| "learning_rate": 7.00294415952286e-07, | |
| "loss": 0.8377, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.7142857142857143, | |
| "grad_norm": 1.2028852767764227, | |
| "learning_rate": 6.866967610905234e-07, | |
| "loss": 0.8386, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.7173645320197044, | |
| "grad_norm": 1.2317394913921158, | |
| "learning_rate": 6.731931191701088e-07, | |
| "loss": 0.8309, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.7204433497536946, | |
| "grad_norm": 1.1573011339095831, | |
| "learning_rate": 6.597850511304739e-07, | |
| "loss": 0.8271, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.7235221674876847, | |
| "grad_norm": 1.2372202934272833, | |
| "learning_rate": 6.464741068632883e-07, | |
| "loss": 0.8266, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.7266009852216748, | |
| "grad_norm": 1.151607214843666, | |
| "learning_rate": 6.332618250332988e-07, | |
| "loss": 0.8136, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.729679802955665, | |
| "grad_norm": 1.9235923929261696, | |
| "learning_rate": 6.201497329004718e-07, | |
| "loss": 0.8591, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.7327586206896551, | |
| "grad_norm": 1.196046809628109, | |
| "learning_rate": 6.071393461434488e-07, | |
| "loss": 0.8331, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.7358374384236454, | |
| "grad_norm": 1.1981116626841852, | |
| "learning_rate": 5.942321686843444e-07, | |
| "loss": 0.8279, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.7389162561576355, | |
| "grad_norm": 1.1554442972192485, | |
| "learning_rate": 5.814296925149026e-07, | |
| "loss": 0.8507, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.7419950738916257, | |
| "grad_norm": 1.2386249887563705, | |
| "learning_rate": 5.687333975240304e-07, | |
| "loss": 0.8363, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.7450738916256158, | |
| "grad_norm": 1.2412252117484952, | |
| "learning_rate": 5.561447513267311e-07, | |
| "loss": 0.8452, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.7481527093596059, | |
| "grad_norm": 1.1498554373981902, | |
| "learning_rate": 5.436652090944596e-07, | |
| "loss": 0.8224, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.7512315270935961, | |
| "grad_norm": 1.1413456236749904, | |
| "learning_rate": 5.312962133869093e-07, | |
| "loss": 0.8401, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.7543103448275862, | |
| "grad_norm": 1.2683847422835812, | |
| "learning_rate": 5.19039193985266e-07, | |
| "loss": 0.8245, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.7573891625615764, | |
| "grad_norm": 1.1722812031534011, | |
| "learning_rate": 5.068955677269281e-07, | |
| "loss": 0.8383, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.7604679802955665, | |
| "grad_norm": 1.112650732199805, | |
| "learning_rate": 4.948667383417332e-07, | |
| "loss": 0.8508, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.7635467980295566, | |
| "grad_norm": 1.2242225043847113, | |
| "learning_rate": 4.829540962896927e-07, | |
| "loss": 0.8402, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.7666256157635468, | |
| "grad_norm": 1.1909265717652961, | |
| "learning_rate": 4.7115901860026634e-07, | |
| "loss": 0.8439, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.7697044334975369, | |
| "grad_norm": 1.1870700244025632, | |
| "learning_rate": 4.594828687131814e-07, | |
| "loss": 0.8339, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.7727832512315271, | |
| "grad_norm": 1.2048136606527728, | |
| "learning_rate": 4.4792699632083043e-07, | |
| "loss": 0.8238, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.7758620689655172, | |
| "grad_norm": 1.218329342603074, | |
| "learning_rate": 4.36492737212255e-07, | |
| "loss": 0.8395, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.7789408866995073, | |
| "grad_norm": 1.2254982042270262, | |
| "learning_rate": 4.251814131187349e-07, | |
| "loss": 0.8227, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.7820197044334976, | |
| "grad_norm": 1.201055974202684, | |
| "learning_rate": 4.13994331561004e-07, | |
| "loss": 0.8227, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.7850985221674877, | |
| "grad_norm": 1.1075936645522768, | |
| "learning_rate": 4.0293278569811197e-07, | |
| "loss": 0.821, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.7881773399014779, | |
| "grad_norm": 1.1826513329616417, | |
| "learning_rate": 3.9199805417793833e-07, | |
| "loss": 0.838, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.791256157635468, | |
| "grad_norm": 1.1905828475039764, | |
| "learning_rate": 3.8119140098939074e-07, | |
| "loss": 0.8275, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.7943349753694581, | |
| "grad_norm": 1.1823163936900083, | |
| "learning_rate": 3.705140753162973e-07, | |
| "loss": 0.8552, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.7974137931034483, | |
| "grad_norm": 1.1726764171251989, | |
| "learning_rate": 3.5996731139300476e-07, | |
| "loss": 0.8298, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.8004926108374384, | |
| "grad_norm": 1.1733827657285574, | |
| "learning_rate": 3.495523283617106e-07, | |
| "loss": 0.8385, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.8035714285714286, | |
| "grad_norm": 1.1548336967321098, | |
| "learning_rate": 3.39270330131538e-07, | |
| "loss": 0.8288, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.8066502463054187, | |
| "grad_norm": 1.3225748675948077, | |
| "learning_rate": 3.2912250523937e-07, | |
| "loss": 0.8431, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.8097290640394089, | |
| "grad_norm": 1.2203880304737789, | |
| "learning_rate": 3.1911002671246164e-07, | |
| "loss": 0.8359, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 0.812807881773399, | |
| "grad_norm": 1.1635933717026323, | |
| "learning_rate": 3.092340519328474e-07, | |
| "loss": 0.8692, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.8158866995073891, | |
| "grad_norm": 1.246775297732715, | |
| "learning_rate": 2.9949572250355176e-07, | |
| "loss": 0.8133, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.8189655172413793, | |
| "grad_norm": 1.0986912775205508, | |
| "learning_rate": 2.8989616411662826e-07, | |
| "loss": 0.8441, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.8220443349753694, | |
| "grad_norm": 1.169380610686736, | |
| "learning_rate": 2.8043648642303716e-07, | |
| "loss": 0.8247, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 0.8251231527093597, | |
| "grad_norm": 1.1817068971461555, | |
| "learning_rate": 2.7111778290437465e-07, | |
| "loss": 0.8469, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.8282019704433498, | |
| "grad_norm": 1.165591743669198, | |
| "learning_rate": 2.619411307464747e-07, | |
| "loss": 0.8349, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 0.8312807881773399, | |
| "grad_norm": 1.2640850790656615, | |
| "learning_rate": 2.529075907148916e-07, | |
| "loss": 0.8406, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.8343596059113301, | |
| "grad_norm": 1.1927946924138935, | |
| "learning_rate": 2.440182070322818e-07, | |
| "loss": 0.8403, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 0.8374384236453202, | |
| "grad_norm": 1.2480501406180602, | |
| "learning_rate": 2.352740072577002e-07, | |
| "loss": 0.8292, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.8405172413793104, | |
| "grad_norm": 1.181685915033771, | |
| "learning_rate": 2.2667600216781757e-07, | |
| "loss": 0.8202, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 0.8435960591133005, | |
| "grad_norm": 1.30446620812208, | |
| "learning_rate": 2.182251856400826e-07, | |
| "loss": 0.8334, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.8466748768472906, | |
| "grad_norm": 1.3663542827209243, | |
| "learning_rate": 2.0992253453783595e-07, | |
| "loss": 0.8557, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.8497536945812808, | |
| "grad_norm": 1.2325789044380537, | |
| "learning_rate": 2.0176900859738906e-07, | |
| "loss": 0.8417, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.8528325123152709, | |
| "grad_norm": 1.262532082314722, | |
| "learning_rate": 1.93765550317087e-07, | |
| "loss": 0.8197, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 0.8559113300492611, | |
| "grad_norm": 1.1713865158884151, | |
| "learning_rate": 1.8591308484835833e-07, | |
| "loss": 0.8603, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.8589901477832512, | |
| "grad_norm": 1.1577049466829008, | |
| "learning_rate": 1.7821251988877556e-07, | |
| "loss": 0.8388, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 0.8620689655172413, | |
| "grad_norm": 1.2469180931897679, | |
| "learning_rate": 1.706647455771302e-07, | |
| "loss": 0.8434, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.8651477832512315, | |
| "grad_norm": 1.236966787026265, | |
| "learning_rate": 1.6327063439053675e-07, | |
| "loss": 0.8381, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 0.8682266009852216, | |
| "grad_norm": 1.2012445829877945, | |
| "learning_rate": 1.56031041043582e-07, | |
| "loss": 0.838, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.8713054187192119, | |
| "grad_norm": 1.2587646510638395, | |
| "learning_rate": 1.4894680238952318e-07, | |
| "loss": 0.8465, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 0.874384236453202, | |
| "grad_norm": 1.1884863239984578, | |
| "learning_rate": 1.4201873732355343e-07, | |
| "loss": 0.8392, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.8774630541871922, | |
| "grad_norm": 1.2366520156475107, | |
| "learning_rate": 1.3524764668814305e-07, | |
| "loss": 0.849, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.8805418719211823, | |
| "grad_norm": 1.2477328107078396, | |
| "learning_rate": 1.2863431318046615e-07, | |
| "loss": 0.8666, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.8836206896551724, | |
| "grad_norm": 1.1572068680725016, | |
| "learning_rate": 1.2217950126192523e-07, | |
| "loss": 0.8382, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 0.8866995073891626, | |
| "grad_norm": 1.1300521771674885, | |
| "learning_rate": 1.158839570697861e-07, | |
| "loss": 0.845, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.8897783251231527, | |
| "grad_norm": 1.158079158149626, | |
| "learning_rate": 1.0974840833092603e-07, | |
| "loss": 0.8225, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 0.8928571428571429, | |
| "grad_norm": 1.24229865374301, | |
| "learning_rate": 1.0377356427771567e-07, | |
| "loss": 0.8397, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.895935960591133, | |
| "grad_norm": 1.1992240511175045, | |
| "learning_rate": 9.796011556603418e-08, | |
| "loss": 0.828, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 0.8990147783251231, | |
| "grad_norm": 1.2150252181899364, | |
| "learning_rate": 9.230873419543373e-08, | |
| "loss": 0.837, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.9020935960591133, | |
| "grad_norm": 1.232372254864163, | |
| "learning_rate": 8.682007343146198e-08, | |
| "loss": 0.8361, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 0.9051724137931034, | |
| "grad_norm": 1.1760952608145245, | |
| "learning_rate": 8.14947677301468e-08, | |
| "loss": 0.837, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.9082512315270936, | |
| "grad_norm": 1.1647514203380553, | |
| "learning_rate": 7.633343266465731e-08, | |
| "loss": 0.8263, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.9113300492610837, | |
| "grad_norm": 1.109297673263934, | |
| "learning_rate": 7.133666485414858e-08, | |
| "loss": 0.836, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.9144088669950738, | |
| "grad_norm": 1.2140360761876554, | |
| "learning_rate": 6.650504189479417e-08, | |
| "loss": 0.8109, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 0.9174876847290641, | |
| "grad_norm": 1.1851881393830357, | |
| "learning_rate": 6.183912229302135e-08, | |
| "loss": 0.8556, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.9205665024630542, | |
| "grad_norm": 1.2139054216696445, | |
| "learning_rate": 5.7339445400949763e-08, | |
| "loss": 0.82, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 0.9236453201970444, | |
| "grad_norm": 1.1684560907583603, | |
| "learning_rate": 5.3006531354045596e-08, | |
| "loss": 0.8435, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.9267241379310345, | |
| "grad_norm": 1.1344224210764826, | |
| "learning_rate": 4.8840881010998536e-08, | |
| "loss": 0.8245, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 0.9298029556650246, | |
| "grad_norm": 1.1550860518730384, | |
| "learning_rate": 4.4842975895823926e-08, | |
| "loss": 0.8538, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.9328817733990148, | |
| "grad_norm": 1.208021844650614, | |
| "learning_rate": 4.1013278142201904e-08, | |
| "loss": 0.8433, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 0.9359605911330049, | |
| "grad_norm": 1.1428227153271917, | |
| "learning_rate": 3.7352230440058534e-08, | |
| "loss": 0.848, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.9390394088669951, | |
| "grad_norm": 1.156453897693646, | |
| "learning_rate": 3.386025598439185e-08, | |
| "loss": 0.8419, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.9421182266009852, | |
| "grad_norm": 1.1782380725430572, | |
| "learning_rate": 3.053775842635453e-08, | |
| "loss": 0.8388, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.9451970443349754, | |
| "grad_norm": 1.2305725491852646, | |
| "learning_rate": 2.7385121826593363e-08, | |
| "loss": 0.8413, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 0.9482758620689655, | |
| "grad_norm": 1.2052501974714247, | |
| "learning_rate": 2.4402710610854582e-08, | |
| "loss": 0.8246, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.9513546798029556, | |
| "grad_norm": 1.1451901965086437, | |
| "learning_rate": 2.159086952785827e-08, | |
| "loss": 0.8216, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 0.9544334975369458, | |
| "grad_norm": 1.223720355889366, | |
| "learning_rate": 1.894992360944786e-08, | |
| "loss": 0.8243, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.9575123152709359, | |
| "grad_norm": 1.1654986160551557, | |
| "learning_rate": 1.648017813301739e-08, | |
| "loss": 0.8346, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 0.9605911330049262, | |
| "grad_norm": 1.185817379490152, | |
| "learning_rate": 1.4181918586225029e-08, | |
| "loss": 0.8289, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.9636699507389163, | |
| "grad_norm": 1.0858902044429564, | |
| "learning_rate": 1.2055410633990515e-08, | |
| "loss": 0.8487, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 0.9667487684729064, | |
| "grad_norm": 1.1956389597775392, | |
| "learning_rate": 1.0100900087787357e-08, | |
| "loss": 0.8301, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.9698275862068966, | |
| "grad_norm": 1.061385774108055, | |
| "learning_rate": 8.318612877227571e-09, | |
| "loss": 0.8522, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.9729064039408867, | |
| "grad_norm": 1.1543403896164468, | |
| "learning_rate": 6.708755023946245e-09, | |
| "loss": 0.8269, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.9759852216748769, | |
| "grad_norm": 1.1713134315197617, | |
| "learning_rate": 5.271512617786078e-09, | |
| "loss": 0.8472, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 0.979064039408867, | |
| "grad_norm": 1.1439940501727845, | |
| "learning_rate": 4.007051795287098e-09, | |
| "loss": 0.8214, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.9821428571428571, | |
| "grad_norm": 1.1517573735882696, | |
| "learning_rate": 2.9155187204820134e-09, | |
| "loss": 0.8461, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 0.9852216748768473, | |
| "grad_norm": 1.28422286593181, | |
| "learning_rate": 1.997039568000403e-09, | |
| "loss": 0.8107, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.9883004926108374, | |
| "grad_norm": 1.1825383865033323, | |
| "learning_rate": 1.2517205084837112e-09, | |
| "loss": 0.8388, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 0.9913793103448276, | |
| "grad_norm": 1.212203466455243, | |
| "learning_rate": 6.796476963130683e-10, | |
| "loss": 0.8353, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.9944581280788177, | |
| "grad_norm": 1.2340921903407838, | |
| "learning_rate": 2.8088725964958705e-10, | |
| "loss": 0.8475, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 0.9975369458128078, | |
| "grad_norm": 1.1861846694894687, | |
| "learning_rate": 5.548529279081338e-11, | |
| "loss": 0.8177, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.8360112905502319, | |
| "eval_runtime": 942.1064, | |
| "eval_samples_per_second": 24.529, | |
| "eval_steps_per_second": 1.534, | |
| "step": 1624 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1624, | |
| "total_flos": 1147278378860544.0, | |
| "train_loss": 0.8499554973737947, | |
| "train_runtime": 34684.9056, | |
| "train_samples_per_second": 5.993, | |
| "train_steps_per_second": 0.047 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 1624, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": false, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1147278378860544.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |