| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 2146, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004659832246039142, | |
| "grad_norm": 14.767192461253016, | |
| "learning_rate": 4.651162790697675e-07, | |
| "loss": 0.8454, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.009319664492078284, | |
| "grad_norm": 10.691157784855342, | |
| "learning_rate": 9.30232558139535e-07, | |
| "loss": 0.8295, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.013979496738117428, | |
| "grad_norm": 4.753711347195471, | |
| "learning_rate": 1.3953488372093025e-06, | |
| "loss": 0.789, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.01863932898415657, | |
| "grad_norm": 4.574744947370259, | |
| "learning_rate": 1.86046511627907e-06, | |
| "loss": 0.7631, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.023299161230195712, | |
| "grad_norm": 2.938678417992091, | |
| "learning_rate": 2.3255813953488376e-06, | |
| "loss": 0.6551, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.027958993476234855, | |
| "grad_norm": 0.6490882548563339, | |
| "learning_rate": 2.790697674418605e-06, | |
| "loss": 0.5661, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.032618825722273995, | |
| "grad_norm": 0.6582256367580213, | |
| "learning_rate": 3.2558139534883724e-06, | |
| "loss": 0.5106, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.03727865796831314, | |
| "grad_norm": 0.5913603077065485, | |
| "learning_rate": 3.72093023255814e-06, | |
| "loss": 0.4935, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.04193849021435228, | |
| "grad_norm": 0.3873817321001381, | |
| "learning_rate": 4.186046511627907e-06, | |
| "loss": 0.4726, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.046598322460391424, | |
| "grad_norm": 0.28334846470708785, | |
| "learning_rate": 4.651162790697675e-06, | |
| "loss": 0.4644, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05125815470643057, | |
| "grad_norm": 0.35730577672386, | |
| "learning_rate": 5.116279069767442e-06, | |
| "loss": 0.4477, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.05591798695246971, | |
| "grad_norm": 0.22651422858035802, | |
| "learning_rate": 5.58139534883721e-06, | |
| "loss": 0.432, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.06057781919850885, | |
| "grad_norm": 0.20330529851492612, | |
| "learning_rate": 6.046511627906977e-06, | |
| "loss": 0.4219, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.06523765144454799, | |
| "grad_norm": 0.21798246079136097, | |
| "learning_rate": 6.511627906976745e-06, | |
| "loss": 0.4257, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.06989748369058714, | |
| "grad_norm": 0.19806053096479337, | |
| "learning_rate": 6.976744186046513e-06, | |
| "loss": 0.4136, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.07455731593662628, | |
| "grad_norm": 0.21322493838033024, | |
| "learning_rate": 7.44186046511628e-06, | |
| "loss": 0.4081, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.07921714818266543, | |
| "grad_norm": 0.1988425295831395, | |
| "learning_rate": 7.906976744186048e-06, | |
| "loss": 0.4059, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.08387698042870456, | |
| "grad_norm": 0.17732217907185543, | |
| "learning_rate": 8.372093023255815e-06, | |
| "loss": 0.4139, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.08853681267474371, | |
| "grad_norm": 0.18197721641419914, | |
| "learning_rate": 8.837209302325582e-06, | |
| "loss": 0.3999, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.09319664492078285, | |
| "grad_norm": 0.18805703800459986, | |
| "learning_rate": 9.30232558139535e-06, | |
| "loss": 0.3929, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.097856477166822, | |
| "grad_norm": 0.17447861554875335, | |
| "learning_rate": 9.767441860465117e-06, | |
| "loss": 0.3899, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.10251630941286113, | |
| "grad_norm": 0.17823985753713623, | |
| "learning_rate": 1.0232558139534884e-05, | |
| "loss": 0.3917, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.10717614165890028, | |
| "grad_norm": 0.1843804114434655, | |
| "learning_rate": 1.0697674418604651e-05, | |
| "loss": 0.3908, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.11183597390493942, | |
| "grad_norm": 0.1813727816186044, | |
| "learning_rate": 1.116279069767442e-05, | |
| "loss": 0.3847, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.11649580615097857, | |
| "grad_norm": 0.18362844363862849, | |
| "learning_rate": 1.1627906976744187e-05, | |
| "loss": 0.3798, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.1211556383970177, | |
| "grad_norm": 0.1976501054723186, | |
| "learning_rate": 1.2093023255813954e-05, | |
| "loss": 0.3869, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.12581547064305684, | |
| "grad_norm": 0.18549227048975853, | |
| "learning_rate": 1.2558139534883723e-05, | |
| "loss": 0.3837, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.13047530288909598, | |
| "grad_norm": 0.20270496221730072, | |
| "learning_rate": 1.302325581395349e-05, | |
| "loss": 0.3763, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.13513513513513514, | |
| "grad_norm": 0.20776510982034277, | |
| "learning_rate": 1.3488372093023257e-05, | |
| "loss": 0.3964, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.13979496738117428, | |
| "grad_norm": 0.2239648466821673, | |
| "learning_rate": 1.3953488372093025e-05, | |
| "loss": 0.3779, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.14445479962721341, | |
| "grad_norm": 0.2159680681931303, | |
| "learning_rate": 1.441860465116279e-05, | |
| "loss": 0.3895, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.14911463187325255, | |
| "grad_norm": 0.2230512293592664, | |
| "learning_rate": 1.488372093023256e-05, | |
| "loss": 0.3748, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.15377446411929171, | |
| "grad_norm": 0.19857008911014462, | |
| "learning_rate": 1.5348837209302328e-05, | |
| "loss": 0.379, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.15843429636533085, | |
| "grad_norm": 0.1876370992998556, | |
| "learning_rate": 1.5813953488372095e-05, | |
| "loss": 0.3599, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.16309412861137, | |
| "grad_norm": 0.19594094926576977, | |
| "learning_rate": 1.6279069767441862e-05, | |
| "loss": 0.3755, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.16775396085740912, | |
| "grad_norm": 0.19588958100246756, | |
| "learning_rate": 1.674418604651163e-05, | |
| "loss": 0.3714, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.1724137931034483, | |
| "grad_norm": 0.200501028483886, | |
| "learning_rate": 1.7209302325581396e-05, | |
| "loss": 0.3714, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.17707362534948742, | |
| "grad_norm": 0.21928747309088611, | |
| "learning_rate": 1.7674418604651163e-05, | |
| "loss": 0.3787, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.18173345759552656, | |
| "grad_norm": 0.24229931775136032, | |
| "learning_rate": 1.813953488372093e-05, | |
| "loss": 0.3725, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.1863932898415657, | |
| "grad_norm": 0.23749429090981541, | |
| "learning_rate": 1.86046511627907e-05, | |
| "loss": 0.375, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.19105312208760486, | |
| "grad_norm": 0.20750507399981963, | |
| "learning_rate": 1.9069767441860468e-05, | |
| "loss": 0.3682, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.195712954333644, | |
| "grad_norm": 0.247680918714745, | |
| "learning_rate": 1.9534883720930235e-05, | |
| "loss": 0.3766, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.20037278657968313, | |
| "grad_norm": 0.20635992002934314, | |
| "learning_rate": 2e-05, | |
| "loss": 0.3656, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.20503261882572227, | |
| "grad_norm": 0.21391797130718435, | |
| "learning_rate": 1.9948213360952874e-05, | |
| "loss": 0.371, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.2096924510717614, | |
| "grad_norm": 0.21826086561242994, | |
| "learning_rate": 1.989642672190575e-05, | |
| "loss": 0.3575, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.21435228331780057, | |
| "grad_norm": 0.2058918265850917, | |
| "learning_rate": 1.9844640082858625e-05, | |
| "loss": 0.3678, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.2190121155638397, | |
| "grad_norm": 0.21256383645525231, | |
| "learning_rate": 1.9792853443811498e-05, | |
| "loss": 0.3652, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.22367194780987884, | |
| "grad_norm": 0.2091365846820082, | |
| "learning_rate": 1.9741066804764374e-05, | |
| "loss": 0.3664, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.22833178005591798, | |
| "grad_norm": 0.2379635390767414, | |
| "learning_rate": 1.9689280165717246e-05, | |
| "loss": 0.369, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.23299161230195714, | |
| "grad_norm": 0.21145510152697863, | |
| "learning_rate": 1.963749352667012e-05, | |
| "loss": 0.3639, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.23765144454799628, | |
| "grad_norm": 0.23770392971510992, | |
| "learning_rate": 1.9585706887622994e-05, | |
| "loss": 0.3679, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.2423112767940354, | |
| "grad_norm": 0.2432075399312, | |
| "learning_rate": 1.953392024857587e-05, | |
| "loss": 0.3692, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.24697110904007455, | |
| "grad_norm": 0.2625016034254708, | |
| "learning_rate": 1.9482133609528746e-05, | |
| "loss": 0.3672, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.2516309412861137, | |
| "grad_norm": 0.2145060352354748, | |
| "learning_rate": 1.9430346970481618e-05, | |
| "loss": 0.3641, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.25629077353215285, | |
| "grad_norm": 0.2473831855865487, | |
| "learning_rate": 1.937856033143449e-05, | |
| "loss": 0.3664, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.26095060577819196, | |
| "grad_norm": 0.22052729908882035, | |
| "learning_rate": 1.9326773692387366e-05, | |
| "loss": 0.3623, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.2656104380242311, | |
| "grad_norm": 0.23934016164519892, | |
| "learning_rate": 1.927498705334024e-05, | |
| "loss": 0.3655, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.2702702702702703, | |
| "grad_norm": 0.23940932697090694, | |
| "learning_rate": 1.9223200414293114e-05, | |
| "loss": 0.3701, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.2749301025163094, | |
| "grad_norm": 0.28885349201279675, | |
| "learning_rate": 1.917141377524599e-05, | |
| "loss": 0.3787, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.27958993476234856, | |
| "grad_norm": 0.2732454315192837, | |
| "learning_rate": 1.9119627136198862e-05, | |
| "loss": 0.3748, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.2842497670083877, | |
| "grad_norm": 0.287512960553777, | |
| "learning_rate": 1.9067840497151738e-05, | |
| "loss": 0.3648, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.28890959925442683, | |
| "grad_norm": 0.25791283197780085, | |
| "learning_rate": 1.901605385810461e-05, | |
| "loss": 0.3605, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.293569431500466, | |
| "grad_norm": 0.22578196901427042, | |
| "learning_rate": 1.8964267219057483e-05, | |
| "loss": 0.3673, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.2982292637465051, | |
| "grad_norm": 0.26398442111942083, | |
| "learning_rate": 1.891248058001036e-05, | |
| "loss": 0.3526, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.30288909599254427, | |
| "grad_norm": 0.2266594032196434, | |
| "learning_rate": 1.8860693940963234e-05, | |
| "loss": 0.3533, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.30754892823858343, | |
| "grad_norm": 0.2034558825684523, | |
| "learning_rate": 1.8808907301916107e-05, | |
| "loss": 0.3625, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.31220876048462254, | |
| "grad_norm": 0.21230490114636388, | |
| "learning_rate": 1.8757120662868982e-05, | |
| "loss": 0.3612, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.3168685927306617, | |
| "grad_norm": 0.21940627029273344, | |
| "learning_rate": 1.8705334023821855e-05, | |
| "loss": 0.3616, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.32152842497670087, | |
| "grad_norm": 0.24083673925107846, | |
| "learning_rate": 1.865354738477473e-05, | |
| "loss": 0.3602, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.32618825722274, | |
| "grad_norm": 0.2522525259887139, | |
| "learning_rate": 1.8601760745727603e-05, | |
| "loss": 0.3723, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.33084808946877914, | |
| "grad_norm": 0.2573006621319903, | |
| "learning_rate": 1.854997410668048e-05, | |
| "loss": 0.3655, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.33550792171481825, | |
| "grad_norm": 0.20987075999325422, | |
| "learning_rate": 1.8498187467633354e-05, | |
| "loss": 0.3583, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.3401677539608574, | |
| "grad_norm": 0.23358976426813405, | |
| "learning_rate": 1.8446400828586227e-05, | |
| "loss": 0.3566, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.3448275862068966, | |
| "grad_norm": 0.2399874778334468, | |
| "learning_rate": 1.83946141895391e-05, | |
| "loss": 0.3657, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.3494874184529357, | |
| "grad_norm": 0.2320299659051788, | |
| "learning_rate": 1.8342827550491975e-05, | |
| "loss": 0.3572, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.35414725069897485, | |
| "grad_norm": 0.2141358345167075, | |
| "learning_rate": 1.8291040911444847e-05, | |
| "loss": 0.3685, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.35880708294501396, | |
| "grad_norm": 0.2394313640207702, | |
| "learning_rate": 1.8239254272397723e-05, | |
| "loss": 0.3667, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.3634669151910531, | |
| "grad_norm": 0.2535784041705651, | |
| "learning_rate": 1.81874676333506e-05, | |
| "loss": 0.3595, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.3681267474370923, | |
| "grad_norm": 0.267215419394161, | |
| "learning_rate": 1.813568099430347e-05, | |
| "loss": 0.3627, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.3727865796831314, | |
| "grad_norm": 0.22563546555156724, | |
| "learning_rate": 1.8083894355256347e-05, | |
| "loss": 0.3514, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.37744641192917056, | |
| "grad_norm": 0.26175999603352973, | |
| "learning_rate": 1.803210771620922e-05, | |
| "loss": 0.3713, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.3821062441752097, | |
| "grad_norm": 0.2557512981709262, | |
| "learning_rate": 1.798032107716209e-05, | |
| "loss": 0.3641, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.38676607642124883, | |
| "grad_norm": 0.2618244822212724, | |
| "learning_rate": 1.7928534438114967e-05, | |
| "loss": 0.3546, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.391425908667288, | |
| "grad_norm": 0.24363536974797073, | |
| "learning_rate": 1.7876747799067843e-05, | |
| "loss": 0.3685, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.3960857409133271, | |
| "grad_norm": 0.23592834779032065, | |
| "learning_rate": 1.7824961160020715e-05, | |
| "loss": 0.37, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.40074557315936626, | |
| "grad_norm": 0.22757209094122816, | |
| "learning_rate": 1.777317452097359e-05, | |
| "loss": 0.3618, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.40540540540540543, | |
| "grad_norm": 0.2247986656928316, | |
| "learning_rate": 1.7721387881926463e-05, | |
| "loss": 0.3703, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.41006523765144454, | |
| "grad_norm": 0.22949520459408446, | |
| "learning_rate": 1.766960124287934e-05, | |
| "loss": 0.3622, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.4147250698974837, | |
| "grad_norm": 0.2429836044789581, | |
| "learning_rate": 1.761781460383221e-05, | |
| "loss": 0.3597, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.4193849021435228, | |
| "grad_norm": 0.2220744501031802, | |
| "learning_rate": 1.7566027964785087e-05, | |
| "loss": 0.3517, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.424044734389562, | |
| "grad_norm": 0.24960146421273202, | |
| "learning_rate": 1.7514241325737963e-05, | |
| "loss": 0.3594, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.42870456663560114, | |
| "grad_norm": 0.23282637573412696, | |
| "learning_rate": 1.7462454686690835e-05, | |
| "loss": 0.3572, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.43336439888164024, | |
| "grad_norm": 0.2583784107904954, | |
| "learning_rate": 1.7410668047643708e-05, | |
| "loss": 0.3536, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.4380242311276794, | |
| "grad_norm": 0.24668787220594343, | |
| "learning_rate": 1.7358881408596583e-05, | |
| "loss": 0.3651, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.4426840633737186, | |
| "grad_norm": 0.23401991763098598, | |
| "learning_rate": 1.7307094769549456e-05, | |
| "loss": 0.3556, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.4473438956197577, | |
| "grad_norm": 0.2568440187662543, | |
| "learning_rate": 1.725530813050233e-05, | |
| "loss": 0.3702, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.45200372786579684, | |
| "grad_norm": 0.24082990346217126, | |
| "learning_rate": 1.7203521491455207e-05, | |
| "loss": 0.3612, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.45666356011183595, | |
| "grad_norm": 0.23359205783172027, | |
| "learning_rate": 1.715173485240808e-05, | |
| "loss": 0.3557, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.4613233923578751, | |
| "grad_norm": 0.2759487778473591, | |
| "learning_rate": 1.7099948213360955e-05, | |
| "loss": 0.3592, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.4659832246039143, | |
| "grad_norm": 0.24487068910640383, | |
| "learning_rate": 1.7048161574313828e-05, | |
| "loss": 0.3676, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.4706430568499534, | |
| "grad_norm": 0.23375869795982646, | |
| "learning_rate": 1.6996374935266703e-05, | |
| "loss": 0.3563, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.47530288909599255, | |
| "grad_norm": 0.23138038119764745, | |
| "learning_rate": 1.6944588296219576e-05, | |
| "loss": 0.3593, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.47996272134203166, | |
| "grad_norm": 0.2317039737462331, | |
| "learning_rate": 1.689280165717245e-05, | |
| "loss": 0.3585, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.4846225535880708, | |
| "grad_norm": 0.24700148883977308, | |
| "learning_rate": 1.6841015018125324e-05, | |
| "loss": 0.3631, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.48928238583411, | |
| "grad_norm": 0.24885633026286735, | |
| "learning_rate": 1.67892283790782e-05, | |
| "loss": 0.3629, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.4939422180801491, | |
| "grad_norm": 0.24542063920879958, | |
| "learning_rate": 1.6737441740031072e-05, | |
| "loss": 0.3579, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.49860205032618826, | |
| "grad_norm": 0.23117849321213937, | |
| "learning_rate": 1.6685655100983948e-05, | |
| "loss": 0.362, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.5032618825722274, | |
| "grad_norm": 0.22596894744470808, | |
| "learning_rate": 1.6633868461936823e-05, | |
| "loss": 0.3492, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.5079217148182665, | |
| "grad_norm": 0.25472474761261105, | |
| "learning_rate": 1.6582081822889696e-05, | |
| "loss": 0.3561, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.5125815470643057, | |
| "grad_norm": 0.24056074693134755, | |
| "learning_rate": 1.653029518384257e-05, | |
| "loss": 0.3604, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.5172413793103449, | |
| "grad_norm": 0.24945961942773268, | |
| "learning_rate": 1.6478508544795444e-05, | |
| "loss": 0.3569, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.5219012115563839, | |
| "grad_norm": 0.23366003314070544, | |
| "learning_rate": 1.6426721905748316e-05, | |
| "loss": 0.3708, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.5265610438024231, | |
| "grad_norm": 0.20558481716658702, | |
| "learning_rate": 1.6374935266701192e-05, | |
| "loss": 0.3478, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.5312208760484622, | |
| "grad_norm": 0.2439084307260332, | |
| "learning_rate": 1.6323148627654068e-05, | |
| "loss": 0.3514, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.5358807082945014, | |
| "grad_norm": 0.24465930880345313, | |
| "learning_rate": 1.6271361988606944e-05, | |
| "loss": 0.3554, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.5405405405405406, | |
| "grad_norm": 0.2612036108887042, | |
| "learning_rate": 1.6219575349559816e-05, | |
| "loss": 0.3553, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.5452003727865797, | |
| "grad_norm": 0.22670651805729272, | |
| "learning_rate": 1.6167788710512688e-05, | |
| "loss": 0.3575, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.5498602050326188, | |
| "grad_norm": 0.20453871875485685, | |
| "learning_rate": 1.6116002071465564e-05, | |
| "loss": 0.3507, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.554520037278658, | |
| "grad_norm": 0.23906772066616808, | |
| "learning_rate": 1.6064215432418436e-05, | |
| "loss": 0.3545, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.5591798695246971, | |
| "grad_norm": 0.2251010058012801, | |
| "learning_rate": 1.6012428793371312e-05, | |
| "loss": 0.3548, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.5638397017707363, | |
| "grad_norm": 0.2042911139628111, | |
| "learning_rate": 1.5960642154324188e-05, | |
| "loss": 0.3545, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.5684995340167754, | |
| "grad_norm": 0.3084593401305696, | |
| "learning_rate": 1.590885551527706e-05, | |
| "loss": 0.3634, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.5731593662628145, | |
| "grad_norm": 0.24668947749199188, | |
| "learning_rate": 1.5857068876229933e-05, | |
| "loss": 0.3534, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.5778191985088537, | |
| "grad_norm": 0.23815820060623935, | |
| "learning_rate": 1.5805282237182808e-05, | |
| "loss": 0.3473, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.5824790307548928, | |
| "grad_norm": 0.2581921333300729, | |
| "learning_rate": 1.575349559813568e-05, | |
| "loss": 0.365, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.587138863000932, | |
| "grad_norm": 0.22365105248877565, | |
| "learning_rate": 1.5701708959088556e-05, | |
| "loss": 0.3605, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.5917986952469712, | |
| "grad_norm": 0.2497896525527162, | |
| "learning_rate": 1.5649922320041432e-05, | |
| "loss": 0.3621, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.5964585274930102, | |
| "grad_norm": 0.2076670286228493, | |
| "learning_rate": 1.5598135680994304e-05, | |
| "loss": 0.3589, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.6011183597390494, | |
| "grad_norm": 0.21352599605542535, | |
| "learning_rate": 1.554634904194718e-05, | |
| "loss": 0.3571, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.6057781919850885, | |
| "grad_norm": 0.23960611526431302, | |
| "learning_rate": 1.5494562402900053e-05, | |
| "loss": 0.3613, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.6104380242311277, | |
| "grad_norm": 0.232800207968001, | |
| "learning_rate": 1.5442775763852925e-05, | |
| "loss": 0.3627, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.6150978564771669, | |
| "grad_norm": 0.2258791731716894, | |
| "learning_rate": 1.53909891248058e-05, | |
| "loss": 0.3459, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.6197576887232059, | |
| "grad_norm": 0.2658353866044793, | |
| "learning_rate": 1.5339202485758676e-05, | |
| "loss": 0.3483, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.6244175209692451, | |
| "grad_norm": 0.23305204875594826, | |
| "learning_rate": 1.5287415846711552e-05, | |
| "loss": 0.3479, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.6290773532152842, | |
| "grad_norm": 0.2384841430708915, | |
| "learning_rate": 1.5235629207664425e-05, | |
| "loss": 0.3518, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.6337371854613234, | |
| "grad_norm": 0.2316602358596732, | |
| "learning_rate": 1.5183842568617297e-05, | |
| "loss": 0.3576, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.6383970177073626, | |
| "grad_norm": 0.26925303516021515, | |
| "learning_rate": 1.5132055929570173e-05, | |
| "loss": 0.3626, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.6430568499534017, | |
| "grad_norm": 0.273521638688401, | |
| "learning_rate": 1.5080269290523047e-05, | |
| "loss": 0.3512, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.6477166821994408, | |
| "grad_norm": 0.2060802961004555, | |
| "learning_rate": 1.5028482651475919e-05, | |
| "loss": 0.3572, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.65237651444548, | |
| "grad_norm": 0.2288622125880128, | |
| "learning_rate": 1.4976696012428795e-05, | |
| "loss": 0.3634, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.6570363466915191, | |
| "grad_norm": 0.20752277736678076, | |
| "learning_rate": 1.4924909373381669e-05, | |
| "loss": 0.3533, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.6616961789375583, | |
| "grad_norm": 0.2501395274655127, | |
| "learning_rate": 1.4873122734334541e-05, | |
| "loss": 0.3546, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.6663560111835974, | |
| "grad_norm": 0.22880446120118658, | |
| "learning_rate": 1.4821336095287417e-05, | |
| "loss": 0.3472, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.6710158434296365, | |
| "grad_norm": 0.2594277925537233, | |
| "learning_rate": 1.4769549456240291e-05, | |
| "loss": 0.3534, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.6756756756756757, | |
| "grad_norm": 0.27706152102194753, | |
| "learning_rate": 1.4717762817193167e-05, | |
| "loss": 0.3564, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.6803355079217148, | |
| "grad_norm": 0.23322150825276064, | |
| "learning_rate": 1.4665976178146039e-05, | |
| "loss": 0.352, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.684995340167754, | |
| "grad_norm": 0.2301436336123601, | |
| "learning_rate": 1.4614189539098913e-05, | |
| "loss": 0.3535, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.6896551724137931, | |
| "grad_norm": 0.21931396895361877, | |
| "learning_rate": 1.4562402900051789e-05, | |
| "loss": 0.3486, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.6943150046598322, | |
| "grad_norm": 0.24163388236718913, | |
| "learning_rate": 1.4510616261004661e-05, | |
| "loss": 0.3477, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.6989748369058714, | |
| "grad_norm": 0.20365681061091856, | |
| "learning_rate": 1.4458829621957535e-05, | |
| "loss": 0.3488, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.7036346691519105, | |
| "grad_norm": 0.2337817758782465, | |
| "learning_rate": 1.4407042982910411e-05, | |
| "loss": 0.3527, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.7082945013979497, | |
| "grad_norm": 0.2405868217410285, | |
| "learning_rate": 1.4355256343863283e-05, | |
| "loss": 0.3542, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.7129543336439889, | |
| "grad_norm": 0.22716685518132887, | |
| "learning_rate": 1.430346970481616e-05, | |
| "loss": 0.3433, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.7176141658900279, | |
| "grad_norm": 0.2296954748413165, | |
| "learning_rate": 1.4251683065769033e-05, | |
| "loss": 0.351, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.7222739981360671, | |
| "grad_norm": 0.2527911740889576, | |
| "learning_rate": 1.4199896426721906e-05, | |
| "loss": 0.3599, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.7269338303821062, | |
| "grad_norm": 0.237870162257353, | |
| "learning_rate": 1.4148109787674781e-05, | |
| "loss": 0.3467, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.7315936626281454, | |
| "grad_norm": 0.23987588369726734, | |
| "learning_rate": 1.4096323148627655e-05, | |
| "loss": 0.3517, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.7362534948741846, | |
| "grad_norm": 0.27125345543342416, | |
| "learning_rate": 1.404453650958053e-05, | |
| "loss": 0.3565, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.7409133271202236, | |
| "grad_norm": 0.22038142310417874, | |
| "learning_rate": 1.3992749870533403e-05, | |
| "loss": 0.3532, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.7455731593662628, | |
| "grad_norm": 0.21707991862427248, | |
| "learning_rate": 1.3940963231486278e-05, | |
| "loss": 0.3516, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.750232991612302, | |
| "grad_norm": 0.22006541459822285, | |
| "learning_rate": 1.3889176592439152e-05, | |
| "loss": 0.3514, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.7548928238583411, | |
| "grad_norm": 0.2532121724442404, | |
| "learning_rate": 1.3837389953392026e-05, | |
| "loss": 0.3636, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.7595526561043803, | |
| "grad_norm": 0.23171770126485144, | |
| "learning_rate": 1.37856033143449e-05, | |
| "loss": 0.3475, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.7642124883504194, | |
| "grad_norm": 0.27888374187325565, | |
| "learning_rate": 1.3733816675297775e-05, | |
| "loss": 0.3577, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.7688723205964585, | |
| "grad_norm": 0.2880404787911838, | |
| "learning_rate": 1.3682030036250648e-05, | |
| "loss": 0.3499, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.7735321528424977, | |
| "grad_norm": 0.24672798067284002, | |
| "learning_rate": 1.3630243397203522e-05, | |
| "loss": 0.3474, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.7781919850885368, | |
| "grad_norm": 0.22126054050968327, | |
| "learning_rate": 1.3578456758156398e-05, | |
| "loss": 0.3499, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.782851817334576, | |
| "grad_norm": 0.25020155421404316, | |
| "learning_rate": 1.3526670119109272e-05, | |
| "loss": 0.3563, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.7875116495806151, | |
| "grad_norm": 0.2327271527985915, | |
| "learning_rate": 1.3474883480062144e-05, | |
| "loss": 0.3639, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.7921714818266542, | |
| "grad_norm": 0.2518085892560592, | |
| "learning_rate": 1.342309684101502e-05, | |
| "loss": 0.3506, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.7968313140726934, | |
| "grad_norm": 0.24951300339031132, | |
| "learning_rate": 1.3371310201967894e-05, | |
| "loss": 0.3563, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.8014911463187325, | |
| "grad_norm": 0.19924194513914378, | |
| "learning_rate": 1.3319523562920766e-05, | |
| "loss": 0.3629, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.8061509785647717, | |
| "grad_norm": 0.25003552030143006, | |
| "learning_rate": 1.3267736923873642e-05, | |
| "loss": 0.348, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.8108108108108109, | |
| "grad_norm": 0.24307899627814475, | |
| "learning_rate": 1.3215950284826516e-05, | |
| "loss": 0.3571, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.8154706430568499, | |
| "grad_norm": 0.21078480288845156, | |
| "learning_rate": 1.3164163645779392e-05, | |
| "loss": 0.3545, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.8201304753028891, | |
| "grad_norm": 0.23496877363911495, | |
| "learning_rate": 1.3112377006732264e-05, | |
| "loss": 0.35, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.8247903075489282, | |
| "grad_norm": 0.23754438146142418, | |
| "learning_rate": 1.3060590367685138e-05, | |
| "loss": 0.3572, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.8294501397949674, | |
| "grad_norm": 0.26052536196367965, | |
| "learning_rate": 1.3008803728638014e-05, | |
| "loss": 0.3492, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.8341099720410066, | |
| "grad_norm": 0.20695762420053648, | |
| "learning_rate": 1.2957017089590886e-05, | |
| "loss": 0.3662, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.8387698042870456, | |
| "grad_norm": 0.2144227087561431, | |
| "learning_rate": 1.290523045054376e-05, | |
| "loss": 0.3515, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.8434296365330848, | |
| "grad_norm": 0.19660702935819654, | |
| "learning_rate": 1.2853443811496636e-05, | |
| "loss": 0.3505, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.848089468779124, | |
| "grad_norm": 0.24655418170934199, | |
| "learning_rate": 1.2801657172449508e-05, | |
| "loss": 0.3549, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.8527493010251631, | |
| "grad_norm": 0.24959870558662023, | |
| "learning_rate": 1.2749870533402384e-05, | |
| "loss": 0.3542, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.8574091332712023, | |
| "grad_norm": 0.24939202340018593, | |
| "learning_rate": 1.2698083894355258e-05, | |
| "loss": 0.3529, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.8620689655172413, | |
| "grad_norm": 0.2516391276986144, | |
| "learning_rate": 1.264629725530813e-05, | |
| "loss": 0.3446, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.8667287977632805, | |
| "grad_norm": 0.21057329816992565, | |
| "learning_rate": 1.2594510616261006e-05, | |
| "loss": 0.3436, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.8713886300093197, | |
| "grad_norm": 0.252888553135217, | |
| "learning_rate": 1.254272397721388e-05, | |
| "loss": 0.3516, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.8760484622553588, | |
| "grad_norm": 0.24242068069037045, | |
| "learning_rate": 1.2490937338166753e-05, | |
| "loss": 0.352, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.880708294501398, | |
| "grad_norm": 0.24295740023564055, | |
| "learning_rate": 1.2439150699119628e-05, | |
| "loss": 0.3506, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.8853681267474371, | |
| "grad_norm": 0.22863631282733143, | |
| "learning_rate": 1.2387364060072502e-05, | |
| "loss": 0.3613, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.8900279589934762, | |
| "grad_norm": 0.2191581333955033, | |
| "learning_rate": 1.2335577421025375e-05, | |
| "loss": 0.3559, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.8946877912395154, | |
| "grad_norm": 0.19128532949024207, | |
| "learning_rate": 1.228379078197825e-05, | |
| "loss": 0.3573, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.8993476234855545, | |
| "grad_norm": 0.2750308888146951, | |
| "learning_rate": 1.2232004142931125e-05, | |
| "loss": 0.352, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.9040074557315937, | |
| "grad_norm": 0.2379831965995701, | |
| "learning_rate": 1.2180217503884e-05, | |
| "loss": 0.3447, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.9086672879776329, | |
| "grad_norm": 0.20411721240775887, | |
| "learning_rate": 1.2128430864836873e-05, | |
| "loss": 0.3606, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.9133271202236719, | |
| "grad_norm": 0.24726535607310818, | |
| "learning_rate": 1.2076644225789747e-05, | |
| "loss": 0.3509, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.9179869524697111, | |
| "grad_norm": 0.25484526322525997, | |
| "learning_rate": 1.2024857586742623e-05, | |
| "loss": 0.3505, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.9226467847157502, | |
| "grad_norm": 0.2313975203490289, | |
| "learning_rate": 1.1973070947695495e-05, | |
| "loss": 0.3485, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.9273066169617894, | |
| "grad_norm": 0.21915958189471746, | |
| "learning_rate": 1.1921284308648369e-05, | |
| "loss": 0.3493, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.9319664492078286, | |
| "grad_norm": 0.21919124866455636, | |
| "learning_rate": 1.1869497669601245e-05, | |
| "loss": 0.3497, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.9366262814538676, | |
| "grad_norm": 0.2028359200370513, | |
| "learning_rate": 1.1817711030554117e-05, | |
| "loss": 0.3537, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.9412861136999068, | |
| "grad_norm": 0.21945782585098136, | |
| "learning_rate": 1.1765924391506993e-05, | |
| "loss": 0.3508, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.9459459459459459, | |
| "grad_norm": 0.2523778581926582, | |
| "learning_rate": 1.1714137752459867e-05, | |
| "loss": 0.3557, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.9506057781919851, | |
| "grad_norm": 0.1984715311531445, | |
| "learning_rate": 1.166235111341274e-05, | |
| "loss": 0.3499, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.9552656104380243, | |
| "grad_norm": 0.20598435799914297, | |
| "learning_rate": 1.1610564474365615e-05, | |
| "loss": 0.3573, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.9599254426840633, | |
| "grad_norm": 0.19822137487156943, | |
| "learning_rate": 1.1558777835318489e-05, | |
| "loss": 0.3537, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.9645852749301025, | |
| "grad_norm": 0.2115159681810218, | |
| "learning_rate": 1.1506991196271361e-05, | |
| "loss": 0.3529, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.9692451071761417, | |
| "grad_norm": 0.22399685831955474, | |
| "learning_rate": 1.1455204557224237e-05, | |
| "loss": 0.3501, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.9739049394221808, | |
| "grad_norm": 0.21539710641856932, | |
| "learning_rate": 1.1403417918177111e-05, | |
| "loss": 0.3438, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.97856477166822, | |
| "grad_norm": 0.2525250092102095, | |
| "learning_rate": 1.1351631279129983e-05, | |
| "loss": 0.3669, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.983224603914259, | |
| "grad_norm": 0.22688066232449872, | |
| "learning_rate": 1.129984464008286e-05, | |
| "loss": 0.3474, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.9878844361602982, | |
| "grad_norm": 0.22444671482154194, | |
| "learning_rate": 1.1248058001035733e-05, | |
| "loss": 0.3492, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.9925442684063374, | |
| "grad_norm": 0.2494658035397173, | |
| "learning_rate": 1.1196271361988609e-05, | |
| "loss": 0.3521, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.9972041006523765, | |
| "grad_norm": 0.22375408565725688, | |
| "learning_rate": 1.1144484722941481e-05, | |
| "loss": 0.3504, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.0018639328984156, | |
| "grad_norm": 0.23743123795247612, | |
| "learning_rate": 1.1092698083894355e-05, | |
| "loss": 0.3269, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.0065237651444547, | |
| "grad_norm": 0.23313660102890352, | |
| "learning_rate": 1.1040911444847231e-05, | |
| "loss": 0.3147, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.011183597390494, | |
| "grad_norm": 0.2222831311931088, | |
| "learning_rate": 1.0989124805800104e-05, | |
| "loss": 0.3195, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 1.015843429636533, | |
| "grad_norm": 0.22590081130460746, | |
| "learning_rate": 1.0937338166752978e-05, | |
| "loss": 0.3263, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.0205032618825722, | |
| "grad_norm": 0.23070778966886485, | |
| "learning_rate": 1.0885551527705853e-05, | |
| "loss": 0.3143, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 1.0251630941286114, | |
| "grad_norm": 0.2037603982755685, | |
| "learning_rate": 1.0833764888658726e-05, | |
| "loss": 0.3053, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.0298229263746506, | |
| "grad_norm": 0.24097367185952903, | |
| "learning_rate": 1.0781978249611601e-05, | |
| "loss": 0.313, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 1.0344827586206897, | |
| "grad_norm": 0.20433112740141812, | |
| "learning_rate": 1.0730191610564476e-05, | |
| "loss": 0.3114, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.0391425908667289, | |
| "grad_norm": 0.21622277499075848, | |
| "learning_rate": 1.067840497151735e-05, | |
| "loss": 0.3145, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 1.0438024231127678, | |
| "grad_norm": 0.23195997985450958, | |
| "learning_rate": 1.0626618332470224e-05, | |
| "loss": 0.3101, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.048462255358807, | |
| "grad_norm": 0.19687727843381017, | |
| "learning_rate": 1.0574831693423098e-05, | |
| "loss": 0.3144, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.0531220876048462, | |
| "grad_norm": 0.19398268395045887, | |
| "learning_rate": 1.0523045054375972e-05, | |
| "loss": 0.3143, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.0577819198508853, | |
| "grad_norm": 0.21618958840342323, | |
| "learning_rate": 1.0471258415328846e-05, | |
| "loss": 0.3244, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 1.0624417520969245, | |
| "grad_norm": 0.21854767532013517, | |
| "learning_rate": 1.041947177628172e-05, | |
| "loss": 0.3196, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.0671015843429636, | |
| "grad_norm": 0.21752939753907818, | |
| "learning_rate": 1.0367685137234594e-05, | |
| "loss": 0.3128, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 1.0717614165890028, | |
| "grad_norm": 0.2024866876281859, | |
| "learning_rate": 1.031589849818747e-05, | |
| "loss": 0.3066, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.076421248835042, | |
| "grad_norm": 0.23303951985635973, | |
| "learning_rate": 1.0264111859140342e-05, | |
| "loss": 0.3126, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 1.0810810810810811, | |
| "grad_norm": 0.18836388691330083, | |
| "learning_rate": 1.0212325220093218e-05, | |
| "loss": 0.312, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.0857409133271203, | |
| "grad_norm": 0.2569075921016222, | |
| "learning_rate": 1.0160538581046092e-05, | |
| "loss": 0.3164, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 1.0904007455731595, | |
| "grad_norm": 0.20845453486936005, | |
| "learning_rate": 1.0108751941998964e-05, | |
| "loss": 0.3225, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.0950605778191984, | |
| "grad_norm": 0.19885909880508598, | |
| "learning_rate": 1.005696530295184e-05, | |
| "loss": 0.3125, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.0997204100652376, | |
| "grad_norm": 0.2669275870745111, | |
| "learning_rate": 1.0005178663904714e-05, | |
| "loss": 0.3167, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.1043802423112767, | |
| "grad_norm": 0.2232374749321823, | |
| "learning_rate": 9.953392024857588e-06, | |
| "loss": 0.3129, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 1.109040074557316, | |
| "grad_norm": 0.19527295898259664, | |
| "learning_rate": 9.901605385810462e-06, | |
| "loss": 0.317, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.113699906803355, | |
| "grad_norm": 0.22072636669010462, | |
| "learning_rate": 9.849818746763336e-06, | |
| "loss": 0.3228, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 1.1183597390493942, | |
| "grad_norm": 0.1949253986918304, | |
| "learning_rate": 9.79803210771621e-06, | |
| "loss": 0.3162, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.1230195712954334, | |
| "grad_norm": 0.2168403291673588, | |
| "learning_rate": 9.746245468669084e-06, | |
| "loss": 0.3164, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 1.1276794035414726, | |
| "grad_norm": 0.20528695943678604, | |
| "learning_rate": 9.694458829621958e-06, | |
| "loss": 0.3158, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.1323392357875117, | |
| "grad_norm": 0.20323454642905758, | |
| "learning_rate": 9.642672190574832e-06, | |
| "loss": 0.3166, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 1.1369990680335509, | |
| "grad_norm": 0.21352753091729954, | |
| "learning_rate": 9.590885551527706e-06, | |
| "loss": 0.3166, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.14165890027959, | |
| "grad_norm": 0.2270069255362695, | |
| "learning_rate": 9.53909891248058e-06, | |
| "loss": 0.3152, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.146318732525629, | |
| "grad_norm": 0.1933121328903308, | |
| "learning_rate": 9.487312273433454e-06, | |
| "loss": 0.317, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.1509785647716682, | |
| "grad_norm": 0.20309447114485268, | |
| "learning_rate": 9.435525634386328e-06, | |
| "loss": 0.3193, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 1.1556383970177073, | |
| "grad_norm": 0.21202399279436465, | |
| "learning_rate": 9.383738995339203e-06, | |
| "loss": 0.3131, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.1602982292637465, | |
| "grad_norm": 0.20262516559107904, | |
| "learning_rate": 9.331952356292078e-06, | |
| "loss": 0.3205, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 1.1649580615097856, | |
| "grad_norm": 0.1988930651334672, | |
| "learning_rate": 9.28016571724495e-06, | |
| "loss": 0.3145, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.1696178937558248, | |
| "grad_norm": 0.1976164571878271, | |
| "learning_rate": 9.228379078197825e-06, | |
| "loss": 0.2993, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 1.174277726001864, | |
| "grad_norm": 0.20859209970330467, | |
| "learning_rate": 9.1765924391507e-06, | |
| "loss": 0.3165, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.1789375582479031, | |
| "grad_norm": 0.21856326174939153, | |
| "learning_rate": 9.124805800103574e-06, | |
| "loss": 0.3112, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 1.1835973904939423, | |
| "grad_norm": 0.2139917663911859, | |
| "learning_rate": 9.073019161056449e-06, | |
| "loss": 0.3222, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.1882572227399812, | |
| "grad_norm": 0.20444481893780622, | |
| "learning_rate": 9.021232522009323e-06, | |
| "loss": 0.3156, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.1929170549860204, | |
| "grad_norm": 0.21896418458298256, | |
| "learning_rate": 8.969445882962197e-06, | |
| "loss": 0.3192, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.1975768872320596, | |
| "grad_norm": 0.20828809222865366, | |
| "learning_rate": 8.91765924391507e-06, | |
| "loss": 0.3225, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 1.2022367194780987, | |
| "grad_norm": 0.19945387268340256, | |
| "learning_rate": 8.865872604867945e-06, | |
| "loss": 0.3113, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.206896551724138, | |
| "grad_norm": 0.19091098433364817, | |
| "learning_rate": 8.814085965820819e-06, | |
| "loss": 0.3143, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 1.211556383970177, | |
| "grad_norm": 0.2444756340798008, | |
| "learning_rate": 8.762299326773693e-06, | |
| "loss": 0.3224, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.2162162162162162, | |
| "grad_norm": 0.21335472533568636, | |
| "learning_rate": 8.710512687726569e-06, | |
| "loss": 0.3197, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 1.2208760484622554, | |
| "grad_norm": 0.2011901037726579, | |
| "learning_rate": 8.658726048679441e-06, | |
| "loss": 0.3099, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.2255358807082946, | |
| "grad_norm": 0.19939445022217508, | |
| "learning_rate": 8.606939409632315e-06, | |
| "loss": 0.3271, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 1.2301957129543337, | |
| "grad_norm": 0.20616420365814805, | |
| "learning_rate": 8.55515277058519e-06, | |
| "loss": 0.3162, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.2348555452003729, | |
| "grad_norm": 0.19727962222530154, | |
| "learning_rate": 8.503366131538063e-06, | |
| "loss": 0.3216, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.2395153774464118, | |
| "grad_norm": 0.19280265908630698, | |
| "learning_rate": 8.451579492490937e-06, | |
| "loss": 0.3148, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.244175209692451, | |
| "grad_norm": 0.20437690863854796, | |
| "learning_rate": 8.399792853443813e-06, | |
| "loss": 0.3137, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 1.2488350419384902, | |
| "grad_norm": 0.21131034344431252, | |
| "learning_rate": 8.348006214396687e-06, | |
| "loss": 0.3187, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.2534948741845293, | |
| "grad_norm": 0.19470646825002752, | |
| "learning_rate": 8.29621957534956e-06, | |
| "loss": 0.3064, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 1.2581547064305685, | |
| "grad_norm": 0.2093800867383876, | |
| "learning_rate": 8.244432936302435e-06, | |
| "loss": 0.3139, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.2628145386766076, | |
| "grad_norm": 0.19557894552720384, | |
| "learning_rate": 8.192646297255309e-06, | |
| "loss": 0.3258, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 1.2674743709226468, | |
| "grad_norm": 0.22037610848319983, | |
| "learning_rate": 8.140859658208183e-06, | |
| "loss": 0.3167, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.272134203168686, | |
| "grad_norm": 0.19614411951293328, | |
| "learning_rate": 8.089073019161057e-06, | |
| "loss": 0.315, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 1.2767940354147251, | |
| "grad_norm": 0.2184836880400932, | |
| "learning_rate": 8.037286380113931e-06, | |
| "loss": 0.3156, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.281453867660764, | |
| "grad_norm": 0.21389038369245703, | |
| "learning_rate": 7.985499741066805e-06, | |
| "loss": 0.308, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.2861136999068035, | |
| "grad_norm": 0.20539585029211851, | |
| "learning_rate": 7.93371310201968e-06, | |
| "loss": 0.3192, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.2907735321528424, | |
| "grad_norm": 0.19934518228998827, | |
| "learning_rate": 7.881926462972553e-06, | |
| "loss": 0.3215, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 1.2954333643988816, | |
| "grad_norm": 0.20741086597019026, | |
| "learning_rate": 7.830139823925427e-06, | |
| "loss": 0.314, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.3000931966449207, | |
| "grad_norm": 0.22065169448140684, | |
| "learning_rate": 7.778353184878302e-06, | |
| "loss": 0.3199, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 1.30475302889096, | |
| "grad_norm": 0.19477745041465316, | |
| "learning_rate": 7.726566545831177e-06, | |
| "loss": 0.3105, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.309412861136999, | |
| "grad_norm": 0.19600332577375437, | |
| "learning_rate": 7.67477990678405e-06, | |
| "loss": 0.3201, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 1.3140726933830382, | |
| "grad_norm": 0.20148776251738024, | |
| "learning_rate": 7.6229932677369245e-06, | |
| "loss": 0.314, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.3187325256290774, | |
| "grad_norm": 0.2064755512233211, | |
| "learning_rate": 7.571206628689799e-06, | |
| "loss": 0.3209, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 1.3233923578751166, | |
| "grad_norm": 0.2031518775941825, | |
| "learning_rate": 7.519419989642673e-06, | |
| "loss": 0.3049, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.3280521901211557, | |
| "grad_norm": 0.19891962178529504, | |
| "learning_rate": 7.467633350595547e-06, | |
| "loss": 0.316, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.3327120223671947, | |
| "grad_norm": 0.20917352538781994, | |
| "learning_rate": 7.415846711548422e-06, | |
| "loss": 0.3197, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.337371854613234, | |
| "grad_norm": 0.18876283067477193, | |
| "learning_rate": 7.364060072501296e-06, | |
| "loss": 0.3231, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 1.342031686859273, | |
| "grad_norm": 0.19676783428533887, | |
| "learning_rate": 7.312273433454169e-06, | |
| "loss": 0.3177, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.3466915191053122, | |
| "grad_norm": 0.18572514729909312, | |
| "learning_rate": 7.260486794407044e-06, | |
| "loss": 0.3211, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 1.3513513513513513, | |
| "grad_norm": 0.19146579294733382, | |
| "learning_rate": 7.208700155359918e-06, | |
| "loss": 0.3118, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.3560111835973905, | |
| "grad_norm": 0.18852660444469807, | |
| "learning_rate": 7.156913516312793e-06, | |
| "loss": 0.3162, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 1.3606710158434296, | |
| "grad_norm": 0.19324215253779436, | |
| "learning_rate": 7.105126877265666e-06, | |
| "loss": 0.3233, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.3653308480894688, | |
| "grad_norm": 0.21358938430110477, | |
| "learning_rate": 7.05334023821854e-06, | |
| "loss": 0.3187, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 1.369990680335508, | |
| "grad_norm": 0.19555774419476726, | |
| "learning_rate": 7.001553599171415e-06, | |
| "loss": 0.3148, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.3746505125815471, | |
| "grad_norm": 0.17959759954123644, | |
| "learning_rate": 6.949766960124289e-06, | |
| "loss": 0.3139, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.3793103448275863, | |
| "grad_norm": 0.18964982625602786, | |
| "learning_rate": 6.897980321077162e-06, | |
| "loss": 0.302, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.3839701770736252, | |
| "grad_norm": 0.18930534523885367, | |
| "learning_rate": 6.846193682030037e-06, | |
| "loss": 0.3149, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 1.3886300093196646, | |
| "grad_norm": 0.19639297256742036, | |
| "learning_rate": 6.794407042982911e-06, | |
| "loss": 0.3236, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.3932898415657036, | |
| "grad_norm": 0.21467988374746413, | |
| "learning_rate": 6.742620403935786e-06, | |
| "loss": 0.3284, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 1.3979496738117427, | |
| "grad_norm": 0.22648269852701486, | |
| "learning_rate": 6.690833764888659e-06, | |
| "loss": 0.3248, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.402609506057782, | |
| "grad_norm": 0.2122629193310997, | |
| "learning_rate": 6.639047125841533e-06, | |
| "loss": 0.3181, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 1.407269338303821, | |
| "grad_norm": 0.21439399946575624, | |
| "learning_rate": 6.587260486794408e-06, | |
| "loss": 0.3111, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.4119291705498602, | |
| "grad_norm": 0.21805459861989804, | |
| "learning_rate": 6.535473847747281e-06, | |
| "loss": 0.3194, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 1.4165890027958994, | |
| "grad_norm": 0.18437558883318525, | |
| "learning_rate": 6.483687208700155e-06, | |
| "loss": 0.3186, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.4212488350419386, | |
| "grad_norm": 0.19383647738248216, | |
| "learning_rate": 6.43190056965303e-06, | |
| "loss": 0.316, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.4259086672879777, | |
| "grad_norm": 0.19238369432331215, | |
| "learning_rate": 6.380113930605904e-06, | |
| "loss": 0.3119, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.4305684995340169, | |
| "grad_norm": 0.20161511281163794, | |
| "learning_rate": 6.3283272915587775e-06, | |
| "loss": 0.3173, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 1.4352283317800558, | |
| "grad_norm": 0.18321974524601511, | |
| "learning_rate": 6.276540652511652e-06, | |
| "loss": 0.3138, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.439888164026095, | |
| "grad_norm": 0.21285415491504464, | |
| "learning_rate": 6.2247540134645265e-06, | |
| "loss": 0.3152, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 1.4445479962721341, | |
| "grad_norm": 0.19238925372724244, | |
| "learning_rate": 6.172967374417401e-06, | |
| "loss": 0.3148, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.4492078285181733, | |
| "grad_norm": 0.21136751157123412, | |
| "learning_rate": 6.1211807353702746e-06, | |
| "loss": 0.3111, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 1.4538676607642125, | |
| "grad_norm": 0.1979462412561175, | |
| "learning_rate": 6.069394096323149e-06, | |
| "loss": 0.3127, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.4585274930102516, | |
| "grad_norm": 0.18992603381080705, | |
| "learning_rate": 6.0176074572760235e-06, | |
| "loss": 0.3154, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 1.4631873252562908, | |
| "grad_norm": 0.18356342906034864, | |
| "learning_rate": 5.9658208182288976e-06, | |
| "loss": 0.3169, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.46784715750233, | |
| "grad_norm": 0.18698833501026985, | |
| "learning_rate": 5.914034179181772e-06, | |
| "loss": 0.3192, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.4725069897483691, | |
| "grad_norm": 0.18659658551949185, | |
| "learning_rate": 5.862247540134646e-06, | |
| "loss": 0.3202, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.477166821994408, | |
| "grad_norm": 0.20607394729321243, | |
| "learning_rate": 5.81046090108752e-06, | |
| "loss": 0.3179, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 1.4818266542404475, | |
| "grad_norm": 0.18871598623396943, | |
| "learning_rate": 5.758674262040395e-06, | |
| "loss": 0.32, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.4864864864864864, | |
| "grad_norm": 0.19724974014550137, | |
| "learning_rate": 5.706887622993268e-06, | |
| "loss": 0.3143, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 1.4911463187325256, | |
| "grad_norm": 0.19442309714878647, | |
| "learning_rate": 5.655100983946143e-06, | |
| "loss": 0.3155, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.4958061509785647, | |
| "grad_norm": 0.2290940185613949, | |
| "learning_rate": 5.603314344899017e-06, | |
| "loss": 0.3241, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 1.500465983224604, | |
| "grad_norm": 0.20552300203050755, | |
| "learning_rate": 5.55152770585189e-06, | |
| "loss": 0.3197, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.505125815470643, | |
| "grad_norm": 0.2138342700269858, | |
| "learning_rate": 5.499741066804765e-06, | |
| "loss": 0.3149, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 1.5097856477166822, | |
| "grad_norm": 0.1773024842853527, | |
| "learning_rate": 5.447954427757639e-06, | |
| "loss": 0.3143, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.5144454799627214, | |
| "grad_norm": 0.1782828676659382, | |
| "learning_rate": 5.396167788710514e-06, | |
| "loss": 0.3085, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.5191053122087603, | |
| "grad_norm": 0.20151066828643008, | |
| "learning_rate": 5.344381149663387e-06, | |
| "loss": 0.3092, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.5237651444547997, | |
| "grad_norm": 0.1936112629982006, | |
| "learning_rate": 5.292594510616261e-06, | |
| "loss": 0.3092, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 1.5284249767008387, | |
| "grad_norm": 0.20031650907945875, | |
| "learning_rate": 5.240807871569136e-06, | |
| "loss": 0.3079, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.533084808946878, | |
| "grad_norm": 0.19466786072761738, | |
| "learning_rate": 5.18902123252201e-06, | |
| "loss": 0.325, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 1.537744641192917, | |
| "grad_norm": 0.2014645297149083, | |
| "learning_rate": 5.137234593474883e-06, | |
| "loss": 0.3193, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.5424044734389561, | |
| "grad_norm": 0.18800764754924237, | |
| "learning_rate": 5.085447954427758e-06, | |
| "loss": 0.3223, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 1.5470643056849953, | |
| "grad_norm": 0.22145725970374222, | |
| "learning_rate": 5.033661315380632e-06, | |
| "loss": 0.3226, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.5517241379310345, | |
| "grad_norm": 0.2264737586413433, | |
| "learning_rate": 4.981874676333506e-06, | |
| "loss": 0.3098, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 1.5563839701770736, | |
| "grad_norm": 0.19843349190924084, | |
| "learning_rate": 4.930088037286381e-06, | |
| "loss": 0.3173, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 1.5610438024231128, | |
| "grad_norm": 0.21010850815608753, | |
| "learning_rate": 4.878301398239254e-06, | |
| "loss": 0.3242, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.565703634669152, | |
| "grad_norm": 0.2064935277967238, | |
| "learning_rate": 4.826514759192129e-06, | |
| "loss": 0.3184, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.570363466915191, | |
| "grad_norm": 0.19388277530401346, | |
| "learning_rate": 4.774728120145003e-06, | |
| "loss": 0.3096, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 1.5750232991612303, | |
| "grad_norm": 0.2009122264808897, | |
| "learning_rate": 4.7229414810978765e-06, | |
| "loss": 0.3233, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 1.5796831314072692, | |
| "grad_norm": 0.1847210364496899, | |
| "learning_rate": 4.671154842050751e-06, | |
| "loss": 0.3229, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 1.5843429636533086, | |
| "grad_norm": 0.19999440492528722, | |
| "learning_rate": 4.6193682030036255e-06, | |
| "loss": 0.3126, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.5890027958993476, | |
| "grad_norm": 0.1967714404409244, | |
| "learning_rate": 4.5675815639564995e-06, | |
| "loss": 0.3161, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 1.5936626281453867, | |
| "grad_norm": 0.19056228994472138, | |
| "learning_rate": 4.5157949249093736e-06, | |
| "loss": 0.3179, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 1.598322460391426, | |
| "grad_norm": 0.19780527100077158, | |
| "learning_rate": 4.464008285862248e-06, | |
| "loss": 0.3107, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 1.602982292637465, | |
| "grad_norm": 0.20087968051198601, | |
| "learning_rate": 4.412221646815122e-06, | |
| "loss": 0.3194, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.6076421248835042, | |
| "grad_norm": 0.18138579556889722, | |
| "learning_rate": 4.3604350077679966e-06, | |
| "loss": 0.3167, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.6123019571295434, | |
| "grad_norm": 0.19116121249902454, | |
| "learning_rate": 4.308648368720871e-06, | |
| "loss": 0.3119, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 1.6169617893755825, | |
| "grad_norm": 0.20253537385722165, | |
| "learning_rate": 4.256861729673745e-06, | |
| "loss": 0.3186, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 1.6216216216216215, | |
| "grad_norm": 0.19508207479386244, | |
| "learning_rate": 4.205075090626619e-06, | |
| "loss": 0.3175, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.6262814538676609, | |
| "grad_norm": 0.18521359809563662, | |
| "learning_rate": 4.153288451579493e-06, | |
| "loss": 0.3168, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 1.6309412861136998, | |
| "grad_norm": 0.21769599978818913, | |
| "learning_rate": 4.101501812532367e-06, | |
| "loss": 0.3192, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.6356011183597392, | |
| "grad_norm": 0.1951979997638157, | |
| "learning_rate": 4.049715173485242e-06, | |
| "loss": 0.3151, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 1.6402609506057781, | |
| "grad_norm": 0.20115642532305716, | |
| "learning_rate": 3.997928534438115e-06, | |
| "loss": 0.3221, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.6449207828518173, | |
| "grad_norm": 0.18954273661107576, | |
| "learning_rate": 3.94614189539099e-06, | |
| "loss": 0.319, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 1.6495806150978565, | |
| "grad_norm": 0.1898263373664628, | |
| "learning_rate": 3.894355256343864e-06, | |
| "loss": 0.3126, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 1.6542404473438956, | |
| "grad_norm": 0.18770630530008564, | |
| "learning_rate": 3.842568617296738e-06, | |
| "loss": 0.3077, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.6589002795899348, | |
| "grad_norm": 0.20920463370343462, | |
| "learning_rate": 3.790781978249612e-06, | |
| "loss": 0.319, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.6635601118359737, | |
| "grad_norm": 0.2035848113749002, | |
| "learning_rate": 3.7389953392024856e-06, | |
| "loss": 0.3179, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 1.6682199440820131, | |
| "grad_norm": 0.1838396556826729, | |
| "learning_rate": 3.68720870015536e-06, | |
| "loss": 0.316, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 1.672879776328052, | |
| "grad_norm": 0.21088430204050254, | |
| "learning_rate": 3.635422061108234e-06, | |
| "loss": 0.3086, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 1.6775396085740915, | |
| "grad_norm": 0.185482590257671, | |
| "learning_rate": 3.5836354220611086e-06, | |
| "loss": 0.3063, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.6821994408201304, | |
| "grad_norm": 0.19312227185115607, | |
| "learning_rate": 3.5318487830139827e-06, | |
| "loss": 0.3083, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 1.6868592730661698, | |
| "grad_norm": 0.1960317573244206, | |
| "learning_rate": 3.480062143966857e-06, | |
| "loss": 0.3122, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 1.6915191053122087, | |
| "grad_norm": 0.19789243390438088, | |
| "learning_rate": 3.4282755049197308e-06, | |
| "loss": 0.3087, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 1.696178937558248, | |
| "grad_norm": 0.18112744143370826, | |
| "learning_rate": 3.3764888658726052e-06, | |
| "loss": 0.3106, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.700838769804287, | |
| "grad_norm": 0.21548231654348435, | |
| "learning_rate": 3.3247022268254793e-06, | |
| "loss": 0.3217, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.7054986020503262, | |
| "grad_norm": 0.1814574471570304, | |
| "learning_rate": 3.2729155877783538e-06, | |
| "loss": 0.3193, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 1.7101584342963654, | |
| "grad_norm": 0.22078602305323497, | |
| "learning_rate": 3.2211289487312274e-06, | |
| "loss": 0.3206, | |
| "step": 1835 | |
| }, | |
| { | |
| "epoch": 1.7148182665424043, | |
| "grad_norm": 0.18658908941872768, | |
| "learning_rate": 3.169342309684102e-06, | |
| "loss": 0.3141, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.7194780987884437, | |
| "grad_norm": 0.1782787618724745, | |
| "learning_rate": 3.117555670636976e-06, | |
| "loss": 0.3149, | |
| "step": 1845 | |
| }, | |
| { | |
| "epoch": 1.7241379310344827, | |
| "grad_norm": 0.17932135518844652, | |
| "learning_rate": 3.0657690315898504e-06, | |
| "loss": 0.3017, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.728797763280522, | |
| "grad_norm": 0.19474093174359786, | |
| "learning_rate": 3.013982392542724e-06, | |
| "loss": 0.3101, | |
| "step": 1855 | |
| }, | |
| { | |
| "epoch": 1.733457595526561, | |
| "grad_norm": 0.19278349416418183, | |
| "learning_rate": 2.9621957534955985e-06, | |
| "loss": 0.311, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.7381174277726001, | |
| "grad_norm": 0.18373595675432944, | |
| "learning_rate": 2.9104091144484725e-06, | |
| "loss": 0.314, | |
| "step": 1865 | |
| }, | |
| { | |
| "epoch": 1.7427772600186393, | |
| "grad_norm": 0.1832866163272324, | |
| "learning_rate": 2.858622475401347e-06, | |
| "loss": 0.3099, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 1.7474370922646785, | |
| "grad_norm": 0.1700878976485856, | |
| "learning_rate": 2.8068358363542206e-06, | |
| "loss": 0.3177, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.7520969245107176, | |
| "grad_norm": 0.19630790934750134, | |
| "learning_rate": 2.7550491973070947e-06, | |
| "loss": 0.3147, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.7567567567567568, | |
| "grad_norm": 0.1935390479382422, | |
| "learning_rate": 2.703262558259969e-06, | |
| "loss": 0.3191, | |
| "step": 1885 | |
| }, | |
| { | |
| "epoch": 1.761416589002796, | |
| "grad_norm": 0.18024055440625963, | |
| "learning_rate": 2.6514759192128432e-06, | |
| "loss": 0.3125, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 1.766076421248835, | |
| "grad_norm": 0.1851607633561796, | |
| "learning_rate": 2.5996892801657177e-06, | |
| "loss": 0.3135, | |
| "step": 1895 | |
| }, | |
| { | |
| "epoch": 1.7707362534948743, | |
| "grad_norm": 0.18693937618463963, | |
| "learning_rate": 2.5479026411185913e-06, | |
| "loss": 0.3079, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.7753960857409132, | |
| "grad_norm": 0.18306356346123398, | |
| "learning_rate": 2.496116002071466e-06, | |
| "loss": 0.3087, | |
| "step": 1905 | |
| }, | |
| { | |
| "epoch": 1.7800559179869526, | |
| "grad_norm": 0.1750201835993522, | |
| "learning_rate": 2.44432936302434e-06, | |
| "loss": 0.316, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 1.7847157502329916, | |
| "grad_norm": 0.18396826364198557, | |
| "learning_rate": 2.3925427239772143e-06, | |
| "loss": 0.303, | |
| "step": 1915 | |
| }, | |
| { | |
| "epoch": 1.7893755824790307, | |
| "grad_norm": 0.18494502790887468, | |
| "learning_rate": 2.340756084930088e-06, | |
| "loss": 0.3195, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 1.7940354147250699, | |
| "grad_norm": 0.17779334157355717, | |
| "learning_rate": 2.2889694458829624e-06, | |
| "loss": 0.3119, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.798695246971109, | |
| "grad_norm": 0.18374129599917147, | |
| "learning_rate": 2.2371828068358365e-06, | |
| "loss": 0.3099, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 1.8033550792171482, | |
| "grad_norm": 0.18305924460244258, | |
| "learning_rate": 2.1853961677887105e-06, | |
| "loss": 0.3156, | |
| "step": 1935 | |
| }, | |
| { | |
| "epoch": 1.8080149114631874, | |
| "grad_norm": 0.17348252725460908, | |
| "learning_rate": 2.1336095287415846e-06, | |
| "loss": 0.3077, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 1.8126747437092265, | |
| "grad_norm": 0.18002736644951658, | |
| "learning_rate": 2.081822889694459e-06, | |
| "loss": 0.3217, | |
| "step": 1945 | |
| }, | |
| { | |
| "epoch": 1.8173345759552655, | |
| "grad_norm": 0.19097984942265778, | |
| "learning_rate": 2.030036250647333e-06, | |
| "loss": 0.3057, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.8219944082013049, | |
| "grad_norm": 0.19397089209826424, | |
| "learning_rate": 1.978249611600207e-06, | |
| "loss": 0.312, | |
| "step": 1955 | |
| }, | |
| { | |
| "epoch": 1.8266542404473438, | |
| "grad_norm": 0.1737258474947117, | |
| "learning_rate": 1.9264629725530816e-06, | |
| "loss": 0.3144, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 1.8313140726933832, | |
| "grad_norm": 0.17866826131161348, | |
| "learning_rate": 1.8746763335059557e-06, | |
| "loss": 0.3158, | |
| "step": 1965 | |
| }, | |
| { | |
| "epoch": 1.8359739049394221, | |
| "grad_norm": 0.1847963686612644, | |
| "learning_rate": 1.8228896944588297e-06, | |
| "loss": 0.3135, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 1.8406337371854613, | |
| "grad_norm": 0.17225040204469855, | |
| "learning_rate": 1.771103055411704e-06, | |
| "loss": 0.3201, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.8452935694315005, | |
| "grad_norm": 0.1784754273129112, | |
| "learning_rate": 1.719316416364578e-06, | |
| "loss": 0.3116, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 1.8499534016775396, | |
| "grad_norm": 0.1873529018355137, | |
| "learning_rate": 1.6675297773174523e-06, | |
| "loss": 0.3068, | |
| "step": 1985 | |
| }, | |
| { | |
| "epoch": 1.8546132339235788, | |
| "grad_norm": 0.18994212675755204, | |
| "learning_rate": 1.6157431382703264e-06, | |
| "loss": 0.3281, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 1.8592730661696177, | |
| "grad_norm": 0.17280992253062147, | |
| "learning_rate": 1.5639564992232006e-06, | |
| "loss": 0.3137, | |
| "step": 1995 | |
| }, | |
| { | |
| "epoch": 1.8639328984156571, | |
| "grad_norm": 0.18082977178380857, | |
| "learning_rate": 1.512169860176075e-06, | |
| "loss": 0.311, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.868592730661696, | |
| "grad_norm": 0.1865537678103653, | |
| "learning_rate": 1.460383221128949e-06, | |
| "loss": 0.3136, | |
| "step": 2005 | |
| }, | |
| { | |
| "epoch": 1.8732525629077355, | |
| "grad_norm": 0.18177081525856895, | |
| "learning_rate": 1.4085965820818228e-06, | |
| "loss": 0.3179, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 1.8779123951537744, | |
| "grad_norm": 0.17820198629127978, | |
| "learning_rate": 1.356809943034697e-06, | |
| "loss": 0.308, | |
| "step": 2015 | |
| }, | |
| { | |
| "epoch": 1.8825722273998136, | |
| "grad_norm": 0.185863894794329, | |
| "learning_rate": 1.305023303987571e-06, | |
| "loss": 0.3065, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 1.8872320596458527, | |
| "grad_norm": 0.18770653417781014, | |
| "learning_rate": 1.2532366649404454e-06, | |
| "loss": 0.3101, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.8918918918918919, | |
| "grad_norm": 0.17409257528881936, | |
| "learning_rate": 1.2014500258933196e-06, | |
| "loss": 0.3121, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 1.896551724137931, | |
| "grad_norm": 0.17868880226457717, | |
| "learning_rate": 1.1496633868461937e-06, | |
| "loss": 0.3127, | |
| "step": 2035 | |
| }, | |
| { | |
| "epoch": 1.9012115563839702, | |
| "grad_norm": 0.18133974763519492, | |
| "learning_rate": 1.097876747799068e-06, | |
| "loss": 0.3136, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 1.9058713886300094, | |
| "grad_norm": 0.18239370217368586, | |
| "learning_rate": 1.046090108751942e-06, | |
| "loss": 0.3241, | |
| "step": 2045 | |
| }, | |
| { | |
| "epoch": 1.9105312208760483, | |
| "grad_norm": 0.17690424687928452, | |
| "learning_rate": 9.943034697048163e-07, | |
| "loss": 0.3186, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.9151910531220877, | |
| "grad_norm": 0.17365540631574325, | |
| "learning_rate": 9.425168306576904e-07, | |
| "loss": 0.3092, | |
| "step": 2055 | |
| }, | |
| { | |
| "epoch": 1.9198508853681266, | |
| "grad_norm": 0.1760791742900446, | |
| "learning_rate": 8.907301916105646e-07, | |
| "loss": 0.3182, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 1.924510717614166, | |
| "grad_norm": 0.19407129934862996, | |
| "learning_rate": 8.389435525634387e-07, | |
| "loss": 0.313, | |
| "step": 2065 | |
| }, | |
| { | |
| "epoch": 1.929170549860205, | |
| "grad_norm": 0.17237038499017496, | |
| "learning_rate": 7.871569135163129e-07, | |
| "loss": 0.316, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 1.9338303821062441, | |
| "grad_norm": 0.18568643784585992, | |
| "learning_rate": 7.35370274469187e-07, | |
| "loss": 0.3232, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.9384902143522833, | |
| "grad_norm": 0.18031325235805645, | |
| "learning_rate": 6.835836354220611e-07, | |
| "loss": 0.311, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 1.9431500465983225, | |
| "grad_norm": 0.19850008316008122, | |
| "learning_rate": 6.317969963749353e-07, | |
| "loss": 0.313, | |
| "step": 2085 | |
| }, | |
| { | |
| "epoch": 1.9478098788443616, | |
| "grad_norm": 0.18169347646543055, | |
| "learning_rate": 5.800103573278094e-07, | |
| "loss": 0.3148, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 1.9524697110904008, | |
| "grad_norm": 0.18682984344018164, | |
| "learning_rate": 5.282237182806836e-07, | |
| "loss": 0.3016, | |
| "step": 2095 | |
| }, | |
| { | |
| "epoch": 1.95712954333644, | |
| "grad_norm": 0.19368082644307608, | |
| "learning_rate": 4.764370792335578e-07, | |
| "loss": 0.3175, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.961789375582479, | |
| "grad_norm": 0.17122020769064838, | |
| "learning_rate": 4.2465044018643194e-07, | |
| "loss": 0.3152, | |
| "step": 2105 | |
| }, | |
| { | |
| "epoch": 1.9664492078285183, | |
| "grad_norm": 0.17219123727385735, | |
| "learning_rate": 3.728638011393061e-07, | |
| "loss": 0.318, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 1.9711090400745572, | |
| "grad_norm": 0.16986195254826772, | |
| "learning_rate": 3.210771620921802e-07, | |
| "loss": 0.3211, | |
| "step": 2115 | |
| }, | |
| { | |
| "epoch": 1.9757688723205966, | |
| "grad_norm": 0.1676620379896325, | |
| "learning_rate": 2.692905230450544e-07, | |
| "loss": 0.3147, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 1.9804287045666356, | |
| "grad_norm": 0.1784700476744974, | |
| "learning_rate": 2.1750388399792857e-07, | |
| "loss": 0.3102, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.9850885368126747, | |
| "grad_norm": 0.17753084232759492, | |
| "learning_rate": 1.657172449508027e-07, | |
| "loss": 0.3154, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 1.9897483690587139, | |
| "grad_norm": 0.16810901109199483, | |
| "learning_rate": 1.1393060590367686e-07, | |
| "loss": 0.3064, | |
| "step": 2135 | |
| }, | |
| { | |
| "epoch": 1.994408201304753, | |
| "grad_norm": 0.16989060528581737, | |
| "learning_rate": 6.214396685655102e-08, | |
| "loss": 0.3126, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 1.9990680335507922, | |
| "grad_norm": 0.1916011537219613, | |
| "learning_rate": 1.0357327809425169e-08, | |
| "loss": 0.317, | |
| "step": 2145 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 2146, | |
| "total_flos": 1.8367600516462019e+18, | |
| "train_loss": 0.34532702185789893, | |
| "train_runtime": 73160.0405, | |
| "train_samples_per_second": 0.469, | |
| "train_steps_per_second": 0.029 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 2146, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.8367600516462019e+18, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |