| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 1084, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0009225092250922509, | |
| "grad_norm": 9.870990956795538, | |
| "learning_rate": 1.8348623853211012e-07, | |
| "loss": 1.1607, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.004612546125461255, | |
| "grad_norm": 7.654279412229945, | |
| "learning_rate": 9.174311926605506e-07, | |
| "loss": 1.1303, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.00922509225092251, | |
| "grad_norm": 3.8667009276426327, | |
| "learning_rate": 1.8348623853211011e-06, | |
| "loss": 1.0569, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.013837638376383764, | |
| "grad_norm": 2.534713969460468, | |
| "learning_rate": 2.7522935779816517e-06, | |
| "loss": 1.0237, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.01845018450184502, | |
| "grad_norm": 2.136778830664697, | |
| "learning_rate": 3.6697247706422022e-06, | |
| "loss": 0.993, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.023062730627306273, | |
| "grad_norm": 1.9089432945409115, | |
| "learning_rate": 4.587155963302753e-06, | |
| "loss": 0.9822, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.027675276752767528, | |
| "grad_norm": 2.483736099993364, | |
| "learning_rate": 5.504587155963303e-06, | |
| "loss": 0.9799, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03228782287822878, | |
| "grad_norm": 2.7399782424767936, | |
| "learning_rate": 6.422018348623854e-06, | |
| "loss": 0.9811, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.03690036900369004, | |
| "grad_norm": 2.5580299604649968, | |
| "learning_rate": 7.3394495412844045e-06, | |
| "loss": 0.9806, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.04151291512915129, | |
| "grad_norm": 2.25360816928206, | |
| "learning_rate": 8.256880733944956e-06, | |
| "loss": 0.9867, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.046125461254612546, | |
| "grad_norm": 2.225590488447417, | |
| "learning_rate": 9.174311926605506e-06, | |
| "loss": 0.9966, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0507380073800738, | |
| "grad_norm": 2.0775964396198394, | |
| "learning_rate": 1.0091743119266055e-05, | |
| "loss": 0.991, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.055350553505535055, | |
| "grad_norm": 2.3870280823080825, | |
| "learning_rate": 1.1009174311926607e-05, | |
| "loss": 0.9927, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.05996309963099631, | |
| "grad_norm": 1.9832001883028227, | |
| "learning_rate": 1.1926605504587156e-05, | |
| "loss": 1.0028, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.06457564575645756, | |
| "grad_norm": 2.494647363270376, | |
| "learning_rate": 1.2844036697247708e-05, | |
| "loss": 0.9875, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.06918819188191883, | |
| "grad_norm": 2.145757053008219, | |
| "learning_rate": 1.3761467889908258e-05, | |
| "loss": 1.004, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.07380073800738007, | |
| "grad_norm": 2.4778285872304453, | |
| "learning_rate": 1.4678899082568809e-05, | |
| "loss": 1.0176, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.07841328413284133, | |
| "grad_norm": 2.378650902592371, | |
| "learning_rate": 1.559633027522936e-05, | |
| "loss": 1.0029, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.08302583025830258, | |
| "grad_norm": 2.6973951224982935, | |
| "learning_rate": 1.6513761467889912e-05, | |
| "loss": 0.9945, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.08763837638376384, | |
| "grad_norm": 2.7900934520086405, | |
| "learning_rate": 1.743119266055046e-05, | |
| "loss": 0.9874, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.09225092250922509, | |
| "grad_norm": 1.9716018629774277, | |
| "learning_rate": 1.834862385321101e-05, | |
| "loss": 1.0383, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.09686346863468635, | |
| "grad_norm": 2.4571675806519564, | |
| "learning_rate": 1.9266055045871563e-05, | |
| "loss": 1.0108, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.1014760147601476, | |
| "grad_norm": 2.1433576022986025, | |
| "learning_rate": 1.9999948088910656e-05, | |
| "loss": 1.014, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.10608856088560886, | |
| "grad_norm": 2.2070794225590524, | |
| "learning_rate": 1.9998131257372878e-05, | |
| "loss": 1.0238, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.11070110701107011, | |
| "grad_norm": 1.797407393598312, | |
| "learning_rate": 1.999371941029485e-05, | |
| "loss": 1.0143, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.11531365313653137, | |
| "grad_norm": 1.9527368920731798, | |
| "learning_rate": 1.9986713692771732e-05, | |
| "loss": 1.0372, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.11992619926199262, | |
| "grad_norm": 2.065891476564467, | |
| "learning_rate": 1.9977115923137912e-05, | |
| "loss": 1.03, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.12453874538745388, | |
| "grad_norm": 1.8362510554036173, | |
| "learning_rate": 1.9964928592495046e-05, | |
| "loss": 1.0307, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.12915129151291513, | |
| "grad_norm": 2.054566764209732, | |
| "learning_rate": 1.9950154864065497e-05, | |
| "loss": 1.0292, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.13376383763837638, | |
| "grad_norm": 1.9408930683317647, | |
| "learning_rate": 1.993279857237133e-05, | |
| "loss": 1.0332, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.13837638376383765, | |
| "grad_norm": 2.046730114559884, | |
| "learning_rate": 1.9912864222239045e-05, | |
| "loss": 1.0408, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.1429889298892989, | |
| "grad_norm": 2.146975729863243, | |
| "learning_rate": 1.9890356987630362e-05, | |
| "loss": 1.0212, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.14760147601476015, | |
| "grad_norm": 1.8119879337927547, | |
| "learning_rate": 1.986528271029931e-05, | |
| "loss": 1.0241, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.1522140221402214, | |
| "grad_norm": 1.812857293149788, | |
| "learning_rate": 1.9837647898276008e-05, | |
| "loss": 1.0248, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.15682656826568267, | |
| "grad_norm": 2.148000093226172, | |
| "learning_rate": 1.9807459724177497e-05, | |
| "loss": 1.0268, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.16143911439114392, | |
| "grad_norm": 1.9129262356991528, | |
| "learning_rate": 1.977472602334609e-05, | |
| "loss": 1.033, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.16605166051660517, | |
| "grad_norm": 1.944891593472809, | |
| "learning_rate": 1.973945529181572e-05, | |
| "loss": 1.0346, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.1706642066420664, | |
| "grad_norm": 1.939828197754357, | |
| "learning_rate": 1.9701656684106764e-05, | |
| "loss": 1.0404, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.1752767527675277, | |
| "grad_norm": 1.8683452140679382, | |
| "learning_rate": 1.9661340010850025e-05, | |
| "loss": 1.0296, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.17988929889298894, | |
| "grad_norm": 1.7548439453154032, | |
| "learning_rate": 1.9618515736240353e-05, | |
| "loss": 1.0511, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.18450184501845018, | |
| "grad_norm": 1.8566237273323596, | |
| "learning_rate": 1.9573194975320672e-05, | |
| "loss": 1.0381, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.18911439114391143, | |
| "grad_norm": 1.7903039830605652, | |
| "learning_rate": 1.952538949109708e-05, | |
| "loss": 1.0451, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.1937269372693727, | |
| "grad_norm": 1.731821818321784, | |
| "learning_rate": 1.9475111691485737e-05, | |
| "loss": 1.0224, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.19833948339483395, | |
| "grad_norm": 1.6546119685109435, | |
| "learning_rate": 1.9422374626092414e-05, | |
| "loss": 1.0312, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.2029520295202952, | |
| "grad_norm": 1.6376019106503288, | |
| "learning_rate": 1.936719198282545e-05, | |
| "loss": 1.0307, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.20756457564575645, | |
| "grad_norm": 1.5223498231652732, | |
| "learning_rate": 1.930957808434307e-05, | |
| "loss": 1.0276, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.21217712177121772, | |
| "grad_norm": 1.5819696965290848, | |
| "learning_rate": 1.9249547884335917e-05, | |
| "loss": 1.0303, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.21678966789667897, | |
| "grad_norm": 1.7581116028501704, | |
| "learning_rate": 1.9187116963645845e-05, | |
| "loss": 1.0357, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.22140221402214022, | |
| "grad_norm": 1.7533150380634188, | |
| "learning_rate": 1.912230152622189e-05, | |
| "loss": 1.0329, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.22601476014760147, | |
| "grad_norm": 1.838564355654299, | |
| "learning_rate": 1.9055118394914545e-05, | |
| "loss": 1.0361, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.23062730627306274, | |
| "grad_norm": 1.584225431462834, | |
| "learning_rate": 1.898558500710939e-05, | |
| "loss": 1.0328, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.235239852398524, | |
| "grad_norm": 1.665130350791042, | |
| "learning_rate": 1.891371941020121e-05, | |
| "loss": 1.0532, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.23985239852398524, | |
| "grad_norm": 1.5868432052387649, | |
| "learning_rate": 1.88395402569098e-05, | |
| "loss": 1.0162, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.2444649446494465, | |
| "grad_norm": 1.612399555276925, | |
| "learning_rate": 1.8763066800438638e-05, | |
| "loss": 1.0252, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.24907749077490776, | |
| "grad_norm": 1.743031080914046, | |
| "learning_rate": 1.868431888947773e-05, | |
| "loss": 1.0367, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.253690036900369, | |
| "grad_norm": 1.5784408195964303, | |
| "learning_rate": 1.860331696305188e-05, | |
| "loss": 1.0271, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.25830258302583026, | |
| "grad_norm": 1.654736307660285, | |
| "learning_rate": 1.852008204521572e-05, | |
| "loss": 1.0197, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.2629151291512915, | |
| "grad_norm": 1.5194226783030391, | |
| "learning_rate": 1.8434635739596945e-05, | |
| "loss": 1.0364, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.26752767527675275, | |
| "grad_norm": 1.5778882827801204, | |
| "learning_rate": 1.834700022378907e-05, | |
| "loss": 1.0543, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.272140221402214, | |
| "grad_norm": 1.6004388094053688, | |
| "learning_rate": 1.825719824359524e-05, | |
| "loss": 1.0403, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.2767527675276753, | |
| "grad_norm": 1.5745761589908656, | |
| "learning_rate": 1.816525310712456e-05, | |
| "loss": 1.0098, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.28136531365313655, | |
| "grad_norm": 1.6069217146531487, | |
| "learning_rate": 1.8071188678742457e-05, | |
| "loss": 1.0168, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.2859778597785978, | |
| "grad_norm": 1.5646672793401115, | |
| "learning_rate": 1.7975029372876706e-05, | |
| "loss": 1.0215, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.29059040590405905, | |
| "grad_norm": 1.543486819243714, | |
| "learning_rate": 1.787680014768065e-05, | |
| "loss": 1.0439, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.2952029520295203, | |
| "grad_norm": 1.7337050309315247, | |
| "learning_rate": 1.777652649855531e-05, | |
| "loss": 1.0183, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.29981549815498154, | |
| "grad_norm": 1.680429731562412, | |
| "learning_rate": 1.7674234451532065e-05, | |
| "loss": 1.031, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.3044280442804428, | |
| "grad_norm": 1.6116817757891433, | |
| "learning_rate": 1.7569950556517566e-05, | |
| "loss": 1.0513, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.30904059040590404, | |
| "grad_norm": 1.5206192596129697, | |
| "learning_rate": 1.7463701880402738e-05, | |
| "loss": 1.0204, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.31365313653136534, | |
| "grad_norm": 1.5945571759720971, | |
| "learning_rate": 1.7355516000037555e-05, | |
| "loss": 1.0278, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.3182656826568266, | |
| "grad_norm": 1.5938814901923302, | |
| "learning_rate": 1.7245420995073453e-05, | |
| "loss": 1.0401, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.32287822878228783, | |
| "grad_norm": 1.6084910280435076, | |
| "learning_rate": 1.7133445440675268e-05, | |
| "loss": 1.0269, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.3274907749077491, | |
| "grad_norm": 1.6256132765843339, | |
| "learning_rate": 1.7019618400104572e-05, | |
| "loss": 1.0259, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.33210332103321033, | |
| "grad_norm": 1.5411807350221058, | |
| "learning_rate": 1.6903969417176244e-05, | |
| "loss": 1.0186, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.3367158671586716, | |
| "grad_norm": 1.52016944094053, | |
| "learning_rate": 1.6786528508590436e-05, | |
| "loss": 1.033, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.3413284132841328, | |
| "grad_norm": 1.8803691215532436, | |
| "learning_rate": 1.666732615614169e-05, | |
| "loss": 1.02, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.3459409594095941, | |
| "grad_norm": 1.499271413842361, | |
| "learning_rate": 1.6546393298807405e-05, | |
| "loss": 1.0178, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.3505535055350554, | |
| "grad_norm": 1.6023328682377094, | |
| "learning_rate": 1.6423761324717636e-05, | |
| "loss": 1.023, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.3551660516605166, | |
| "grad_norm": 1.5304826653478296, | |
| "learning_rate": 1.6299462063008272e-05, | |
| "loss": 1.0155, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.35977859778597787, | |
| "grad_norm": 1.5394673610887177, | |
| "learning_rate": 1.61735277755598e-05, | |
| "loss": 1.023, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.3643911439114391, | |
| "grad_norm": 1.813007395614376, | |
| "learning_rate": 1.6045991148623752e-05, | |
| "loss": 1.0168, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.36900369003690037, | |
| "grad_norm": 1.5526018219731565, | |
| "learning_rate": 1.5916885284338937e-05, | |
| "loss": 1.0045, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.3736162361623616, | |
| "grad_norm": 1.4318690915846115, | |
| "learning_rate": 1.5786243692139826e-05, | |
| "loss": 1.0265, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.37822878228782286, | |
| "grad_norm": 1.65542416947911, | |
| "learning_rate": 1.5654100280059155e-05, | |
| "loss": 1.0038, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.3828413284132841, | |
| "grad_norm": 1.5036498809187224, | |
| "learning_rate": 1.5520489345927095e-05, | |
| "loss": 1.003, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.3874538745387454, | |
| "grad_norm": 1.571209743645404, | |
| "learning_rate": 1.538544556846925e-05, | |
| "loss": 1.0251, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.39206642066420666, | |
| "grad_norm": 1.4496764368758055, | |
| "learning_rate": 1.5249003998305787e-05, | |
| "loss": 1.0136, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.3966789667896679, | |
| "grad_norm": 1.5466821761987914, | |
| "learning_rate": 1.5111200048854055e-05, | |
| "loss": 1.0194, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.40129151291512916, | |
| "grad_norm": 1.4987449489819178, | |
| "learning_rate": 1.4972069487137024e-05, | |
| "loss": 1.009, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.4059040590405904, | |
| "grad_norm": 1.4694129512198164, | |
| "learning_rate": 1.4831648424499953e-05, | |
| "loss": 1.0149, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.41051660516605165, | |
| "grad_norm": 1.434796826955623, | |
| "learning_rate": 1.4689973307237687e-05, | |
| "loss": 1.0285, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.4151291512915129, | |
| "grad_norm": 1.510041122963638, | |
| "learning_rate": 1.4547080907135024e-05, | |
| "loss": 1.013, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.41974169741697415, | |
| "grad_norm": 1.4273777036410285, | |
| "learning_rate": 1.4403008311922593e-05, | |
| "loss": 1.0106, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.42435424354243545, | |
| "grad_norm": 1.434029200696217, | |
| "learning_rate": 1.4257792915650728e-05, | |
| "loss": 1.0126, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.4289667896678967, | |
| "grad_norm": 1.3526682049403003, | |
| "learning_rate": 1.4111472408983843e-05, | |
| "loss": 1.0056, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.43357933579335795, | |
| "grad_norm": 1.560807386777768, | |
| "learning_rate": 1.3964084769417823e-05, | |
| "loss": 1.0293, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.4381918819188192, | |
| "grad_norm": 1.6384296659793862, | |
| "learning_rate": 1.3815668251422953e-05, | |
| "loss": 0.9952, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.44280442804428044, | |
| "grad_norm": 1.4643281809184674, | |
| "learning_rate": 1.3666261376514978e-05, | |
| "loss": 0.9927, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.4474169741697417, | |
| "grad_norm": 1.4762693155800783, | |
| "learning_rate": 1.3515902923256832e-05, | |
| "loss": 1.0082, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.45202952029520294, | |
| "grad_norm": 1.575453671421441, | |
| "learning_rate": 1.3364631917193671e-05, | |
| "loss": 1.0018, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.4566420664206642, | |
| "grad_norm": 1.7509884043580597, | |
| "learning_rate": 1.321248762072377e-05, | |
| "loss": 1.0031, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.4612546125461255, | |
| "grad_norm": 1.4633205407763117, | |
| "learning_rate": 1.3059509522907998e-05, | |
| "loss": 0.9977, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.46586715867158673, | |
| "grad_norm": 1.4464913424686388, | |
| "learning_rate": 1.2905737329220394e-05, | |
| "loss": 0.9962, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.470479704797048, | |
| "grad_norm": 1.5019368087709983, | |
| "learning_rate": 1.2751210951242636e-05, | |
| "loss": 1.0047, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.47509225092250923, | |
| "grad_norm": 1.424993301093826, | |
| "learning_rate": 1.2595970496304975e-05, | |
| "loss": 0.9935, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.4797047970479705, | |
| "grad_norm": 1.397860945200415, | |
| "learning_rate": 1.2440056257076376e-05, | |
| "loss": 1.0072, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.4843173431734317, | |
| "grad_norm": 1.4591151077748796, | |
| "learning_rate": 1.2283508701106559e-05, | |
| "loss": 0.9785, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.488929889298893, | |
| "grad_norm": 1.3985403114867758, | |
| "learning_rate": 1.2126368460322637e-05, | |
| "loss": 1.0073, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.4935424354243542, | |
| "grad_norm": 1.3622030450903666, | |
| "learning_rate": 1.1968676320483103e-05, | |
| "loss": 0.9679, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.4981549815498155, | |
| "grad_norm": 1.4585144558991643, | |
| "learning_rate": 1.1810473210591882e-05, | |
| "loss": 0.9846, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.5027675276752768, | |
| "grad_norm": 1.4502741925244078, | |
| "learning_rate": 1.1651800192275197e-05, | |
| "loss": 0.9792, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.507380073800738, | |
| "grad_norm": 1.4112975394158198, | |
| "learning_rate": 1.1492698449124042e-05, | |
| "loss": 0.9882, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.5119926199261993, | |
| "grad_norm": 1.3902737511333096, | |
| "learning_rate": 1.1333209276004959e-05, | |
| "loss": 0.996, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.5166051660516605, | |
| "grad_norm": 1.4114090100980268, | |
| "learning_rate": 1.1173374068341962e-05, | |
| "loss": 1.0053, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.5212177121771218, | |
| "grad_norm": 1.4743164096907508, | |
| "learning_rate": 1.1013234311372353e-05, | |
| "loss": 0.9835, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.525830258302583, | |
| "grad_norm": 1.519429603617901, | |
| "learning_rate": 1.0852831569379217e-05, | |
| "loss": 0.9908, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.5304428044280443, | |
| "grad_norm": 1.453522178819256, | |
| "learning_rate": 1.0692207474903421e-05, | |
| "loss": 0.991, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.5350553505535055, | |
| "grad_norm": 1.37300728807636, | |
| "learning_rate": 1.0531403717937888e-05, | |
| "loss": 0.9887, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.5396678966789668, | |
| "grad_norm": 1.4658227361639529, | |
| "learning_rate": 1.037046203510694e-05, | |
| "loss": 0.975, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.544280442804428, | |
| "grad_norm": 1.3920165357631773, | |
| "learning_rate": 1.0209424198833571e-05, | |
| "loss": 0.987, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.5488929889298892, | |
| "grad_norm": 1.5056794833077605, | |
| "learning_rate": 1.0048332006497406e-05, | |
| "loss": 0.9936, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.5535055350553506, | |
| "grad_norm": 1.3882676355943753, | |
| "learning_rate": 9.887227269586184e-06, | |
| "loss": 0.9807, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.5581180811808119, | |
| "grad_norm": 1.466557983615673, | |
| "learning_rate": 9.7261518028436e-06, | |
| "loss": 0.99, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.5627306273062731, | |
| "grad_norm": 1.4193893381047582, | |
| "learning_rate": 9.565147413416266e-06, | |
| "loss": 0.9782, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.5673431734317343, | |
| "grad_norm": 1.3465535691560198, | |
| "learning_rate": 9.404255890002677e-06, | |
| "loss": 0.9963, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.5719557195571956, | |
| "grad_norm": 1.2838850891021258, | |
| "learning_rate": 9.243518992006944e-06, | |
| "loss": 0.9735, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.5765682656826568, | |
| "grad_norm": 1.3173792395018515, | |
| "learning_rate": 9.082978438700138e-06, | |
| "loss": 0.968, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.5811808118081181, | |
| "grad_norm": 1.4339058777471723, | |
| "learning_rate": 8.922675898392072e-06, | |
| "loss": 0.9663, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.5857933579335793, | |
| "grad_norm": 1.3593794949774558, | |
| "learning_rate": 8.762652977616258e-06, | |
| "loss": 0.9861, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.5904059040590406, | |
| "grad_norm": 1.3750915173575393, | |
| "learning_rate": 8.602951210330942e-06, | |
| "loss": 0.9658, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.5950184501845018, | |
| "grad_norm": 1.3517116878942004, | |
| "learning_rate": 8.443612047138965e-06, | |
| "loss": 0.982, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.5996309963099631, | |
| "grad_norm": 1.4091397156737253, | |
| "learning_rate": 8.284676844529258e-06, | |
| "loss": 0.945, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.6042435424354243, | |
| "grad_norm": 1.2934337488666847, | |
| "learning_rate": 8.126186854142752e-06, | |
| "loss": 0.9526, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.6088560885608856, | |
| "grad_norm": 1.336680447919326, | |
| "learning_rate": 7.968183212065537e-06, | |
| "loss": 0.9648, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.6134686346863468, | |
| "grad_norm": 1.3335063906222555, | |
| "learning_rate": 7.81070692815195e-06, | |
| "loss": 0.9616, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.6180811808118081, | |
| "grad_norm": 1.4028504727483062, | |
| "learning_rate": 7.6537988753805e-06, | |
| "loss": 0.9559, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.6226937269372693, | |
| "grad_norm": 1.3423760518443784, | |
| "learning_rate": 7.497499779245268e-06, | |
| "loss": 0.9649, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.6273062730627307, | |
| "grad_norm": 1.3763883937348929, | |
| "learning_rate": 7.3418502071856004e-06, | |
| "loss": 0.9727, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.6319188191881919, | |
| "grad_norm": 1.3538188475466573, | |
| "learning_rate": 7.186890558056836e-06, | |
| "loss": 0.9525, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.6365313653136532, | |
| "grad_norm": 1.4363152298279036, | |
| "learning_rate": 7.0326610516447825e-06, | |
| "loss": 0.9519, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.6411439114391144, | |
| "grad_norm": 1.4065045912643077, | |
| "learning_rate": 6.879201718226658e-06, | |
| "loss": 0.9433, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.6457564575645757, | |
| "grad_norm": 1.2689078033695567, | |
| "learning_rate": 6.7265523881812335e-06, | |
| "loss": 0.9567, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.6503690036900369, | |
| "grad_norm": 1.2774626635023132, | |
| "learning_rate": 6.574752681650864e-06, | |
| "loss": 0.9423, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.6549815498154982, | |
| "grad_norm": 1.2857657790671198, | |
| "learning_rate": 6.423841998258069e-06, | |
| "loss": 0.9488, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.6595940959409594, | |
| "grad_norm": 1.3218825385433717, | |
| "learning_rate": 6.273859506879365e-06, | |
| "loss": 0.9675, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.6642066420664207, | |
| "grad_norm": 1.3140499267955912, | |
| "learning_rate": 6.124844135478971e-06, | |
| "loss": 0.9645, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.6688191881918819, | |
| "grad_norm": 1.2553616859013494, | |
| "learning_rate": 5.976834561005069e-06, | |
| "loss": 0.9593, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.6734317343173432, | |
| "grad_norm": 1.3433064911689856, | |
| "learning_rate": 5.829869199351188e-06, | |
| "loss": 0.9491, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.6780442804428044, | |
| "grad_norm": 1.3274312519554137, | |
| "learning_rate": 5.68398619538536e-06, | |
| "loss": 0.9425, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.6826568265682657, | |
| "grad_norm": 1.3743238375183062, | |
| "learning_rate": 5.53922341304961e-06, | |
| "loss": 0.9378, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.6872693726937269, | |
| "grad_norm": 1.3880330372119891, | |
| "learning_rate": 5.39561842553239e-06, | |
| "loss": 0.9759, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.6918819188191881, | |
| "grad_norm": 1.3337273623917134, | |
| "learning_rate": 5.2532085055164205e-06, | |
| "loss": 0.9493, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.6964944649446494, | |
| "grad_norm": 1.3077178571915684, | |
| "learning_rate": 5.112030615504601e-06, | |
| "loss": 0.9521, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.7011070110701108, | |
| "grad_norm": 1.2742302447784248, | |
| "learning_rate": 4.972121398226371e-06, | |
| "loss": 0.9382, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.705719557195572, | |
| "grad_norm": 1.3201506076517637, | |
| "learning_rate": 4.833517167127077e-06, | |
| "loss": 0.9525, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.7103321033210332, | |
| "grad_norm": 1.3060162706383742, | |
| "learning_rate": 4.6962538969428416e-06, | |
| "loss": 0.9518, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.7149446494464945, | |
| "grad_norm": 1.2666833397246693, | |
| "learning_rate": 4.560367214363295e-06, | |
| "loss": 0.9459, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.7195571955719557, | |
| "grad_norm": 1.3752391118024025, | |
| "learning_rate": 4.425892388784681e-06, | |
| "loss": 0.9481, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.724169741697417, | |
| "grad_norm": 1.381288016089436, | |
| "learning_rate": 4.292864323155684e-06, | |
| "loss": 0.9351, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.7287822878228782, | |
| "grad_norm": 1.3075298809556533, | |
| "learning_rate": 4.161317544918345e-06, | |
| "loss": 0.9382, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.7333948339483395, | |
| "grad_norm": 1.2816897642857799, | |
| "learning_rate": 4.031286197046493e-06, | |
| "loss": 0.9423, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.7380073800738007, | |
| "grad_norm": 1.2968384739032883, | |
| "learning_rate": 3.902804029183907e-06, | |
| "loss": 0.9446, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.742619926199262, | |
| "grad_norm": 1.329056519068419, | |
| "learning_rate": 3.775904388884618e-06, | |
| "loss": 0.9422, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.7472324723247232, | |
| "grad_norm": 1.2876813358003367, | |
| "learning_rate": 3.650620212957524e-06, | |
| "loss": 0.9404, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.7518450184501845, | |
| "grad_norm": 1.4667315877372276, | |
| "learning_rate": 3.5269840189176616e-06, | |
| "loss": 0.9282, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.7564575645756457, | |
| "grad_norm": 1.301616249350314, | |
| "learning_rate": 3.405027896546277e-06, | |
| "loss": 0.9259, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.761070110701107, | |
| "grad_norm": 1.3223905832430802, | |
| "learning_rate": 3.2847834995619067e-06, | |
| "loss": 0.9447, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.7656826568265682, | |
| "grad_norm": 1.262448560449103, | |
| "learning_rate": 3.1662820374046776e-06, | |
| "loss": 0.9353, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.7702952029520295, | |
| "grad_norm": 1.2951473418615072, | |
| "learning_rate": 3.0495542671358745e-06, | |
| "loss": 0.9291, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.7749077490774908, | |
| "grad_norm": 1.2820260584603826, | |
| "learning_rate": 2.934630485454948e-06, | |
| "loss": 0.9356, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.7795202952029521, | |
| "grad_norm": 1.2899359050310029, | |
| "learning_rate": 2.8215405208360237e-06, | |
| "loss": 0.9394, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.7841328413284133, | |
| "grad_norm": 1.3428587589811845, | |
| "learning_rate": 2.7103137257858867e-06, | |
| "loss": 0.9355, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.7887453874538746, | |
| "grad_norm": 1.3675117810995234, | |
| "learning_rate": 2.600978969225558e-06, | |
| "loss": 0.9406, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.7933579335793358, | |
| "grad_norm": 1.285843469838144, | |
| "learning_rate": 2.493564628997369e-06, | |
| "loss": 0.9472, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.7979704797047971, | |
| "grad_norm": 1.308062981287854, | |
| "learning_rate": 2.3880985844994674e-06, | |
| "loss": 0.9284, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.8025830258302583, | |
| "grad_norm": 1.311817269933356, | |
| "learning_rate": 2.284608209449746e-06, | |
| "loss": 0.9357, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.8071955719557196, | |
| "grad_norm": 1.3458370437390148, | |
| "learning_rate": 2.183120364780975e-06, | |
| "loss": 0.9387, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.8118081180811808, | |
| "grad_norm": 1.2723012259493913, | |
| "learning_rate": 2.083661391669043e-06, | |
| "loss": 0.9252, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.816420664206642, | |
| "grad_norm": 1.3187842301744654, | |
| "learning_rate": 1.986257104696121e-06, | |
| "loss": 0.9368, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.8210332103321033, | |
| "grad_norm": 1.3542997263964067, | |
| "learning_rate": 1.8909327851504633e-06, | |
| "loss": 0.9526, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.8256457564575646, | |
| "grad_norm": 1.3403454711197158, | |
| "learning_rate": 1.7977131744646724e-06, | |
| "loss": 0.9327, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.8302583025830258, | |
| "grad_norm": 1.3077988340377713, | |
| "learning_rate": 1.7066224677940313e-06, | |
| "loss": 0.9431, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.834870848708487, | |
| "grad_norm": 1.2548232997350641, | |
| "learning_rate": 1.6176843077366755e-06, | |
| "loss": 0.9136, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.8394833948339483, | |
| "grad_norm": 1.2897227556284836, | |
| "learning_rate": 1.5309217781971419e-06, | |
| "loss": 0.9134, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.8440959409594095, | |
| "grad_norm": 1.2995199761283964, | |
| "learning_rate": 1.446357398394934e-06, | |
| "loss": 0.9231, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.8487084870848709, | |
| "grad_norm": 1.3112786455057477, | |
| "learning_rate": 1.3640131170196758e-06, | |
| "loss": 0.9157, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.8533210332103321, | |
| "grad_norm": 1.3030169209685578, | |
| "learning_rate": 1.2839103065343084e-06, | |
| "loss": 0.9278, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.8579335793357934, | |
| "grad_norm": 1.3094669969598054, | |
| "learning_rate": 1.2060697576278812e-06, | |
| "loss": 0.9329, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.8625461254612546, | |
| "grad_norm": 1.276775688304098, | |
| "learning_rate": 1.1305116738193211e-06, | |
| "loss": 0.9346, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.8671586715867159, | |
| "grad_norm": 1.2892434172054825, | |
| "learning_rate": 1.0572556662136036e-06, | |
| "loss": 0.9301, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.8717712177121771, | |
| "grad_norm": 1.3021701449964453, | |
| "learning_rate": 9.863207484116987e-07, | |
| "loss": 0.9177, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.8763837638376384, | |
| "grad_norm": 1.2852161940710303, | |
| "learning_rate": 9.177253315755796e-07, | |
| "loss": 0.9378, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.8809963099630996, | |
| "grad_norm": 1.2451978022882328, | |
| "learning_rate": 8.514872196496182e-07, | |
| "loss": 0.9348, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.8856088560885609, | |
| "grad_norm": 1.2141597430813968, | |
| "learning_rate": 7.876236047395525e-07, | |
| "loss": 0.9265, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.8902214022140221, | |
| "grad_norm": 1.296008639609934, | |
| "learning_rate": 7.26151062650291e-07, | |
| "loss": 0.9338, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.8948339483394834, | |
| "grad_norm": 1.3124661816253649, | |
| "learning_rate": 6.670855485836525e-07, | |
| "loss": 0.9281, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.8994464944649446, | |
| "grad_norm": 1.2666059464350587, | |
| "learning_rate": 6.104423929971948e-07, | |
| "loss": 0.9327, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.9040590405904059, | |
| "grad_norm": 1.2784756107848798, | |
| "learning_rate": 5.562362976251901e-07, | |
| "loss": 0.938, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.9086715867158671, | |
| "grad_norm": 1.2246240779069228, | |
| "learning_rate": 5.044813316627994e-07, | |
| "loss": 0.9268, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.9132841328413284, | |
| "grad_norm": 1.2911170040738167, | |
| "learning_rate": 4.5519092811439627e-07, | |
| "loss": 0.9248, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.9178966789667896, | |
| "grad_norm": 1.320133517164497, | |
| "learning_rate": 4.083778803070504e-07, | |
| "loss": 0.9437, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.922509225092251, | |
| "grad_norm": 1.2549143119794002, | |
| "learning_rate": 3.6405433856999684e-07, | |
| "loss": 0.9565, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.9271217712177122, | |
| "grad_norm": 1.324909180737576, | |
| "learning_rate": 3.2223180708102933e-07, | |
| "loss": 0.9322, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.9317343173431735, | |
| "grad_norm": 1.2882570015971073, | |
| "learning_rate": 2.829211408805932e-07, | |
| "loss": 0.9173, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.9363468634686347, | |
| "grad_norm": 1.67906584184866, | |
| "learning_rate": 2.461325430543482e-07, | |
| "loss": 0.9319, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.940959409594096, | |
| "grad_norm": 1.2543153354274186, | |
| "learning_rate": 2.1187556208496885e-07, | |
| "loss": 0.9212, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.9455719557195572, | |
| "grad_norm": 1.230292220248059, | |
| "learning_rate": 1.8015908937382587e-07, | |
| "loss": 0.9329, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.9501845018450185, | |
| "grad_norm": 1.2685375441760296, | |
| "learning_rate": 1.5099135693322776e-07, | |
| "loss": 0.9335, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.9547970479704797, | |
| "grad_norm": 1.2252210378690607, | |
| "learning_rate": 1.2437993524979984e-07, | |
| "loss": 0.9029, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.959409594095941, | |
| "grad_norm": 1.2888608840481186, | |
| "learning_rate": 1.0033173131956175e-07, | |
| "loss": 0.9211, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.9640221402214022, | |
| "grad_norm": 1.2495839916442792, | |
| "learning_rate": 7.885298685522235e-08, | |
| "loss": 0.9152, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.9686346863468634, | |
| "grad_norm": 1.279639118993962, | |
| "learning_rate": 5.99492766661347e-08, | |
| "loss": 0.9212, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.9732472324723247, | |
| "grad_norm": 1.2775107158396932, | |
| "learning_rate": 4.362550721136338e-08, | |
| "loss": 0.9252, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.977859778597786, | |
| "grad_norm": 1.2306806429728543, | |
| "learning_rate": 2.988591532620322e-08, | |
| "loss": 0.9151, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.9824723247232472, | |
| "grad_norm": 1.2648118187129338, | |
| "learning_rate": 1.8734067122514464e-08, | |
| "loss": 0.9222, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.9870848708487084, | |
| "grad_norm": 1.3467629616768353, | |
| "learning_rate": 1.0172857063137643e-08, | |
| "loss": 0.9251, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.9916974169741697, | |
| "grad_norm": 1.281124121994121, | |
| "learning_rate": 4.204507210633368e-09, | |
| "loss": 0.9465, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.996309963099631, | |
| "grad_norm": 1.2725002065016535, | |
| "learning_rate": 8.30566650548148e-10, | |
| "loss": 0.9402, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.9401968121528625, | |
| "eval_runtime": 156.1033, | |
| "eval_samples_per_second": 98.332, | |
| "eval_steps_per_second": 1.537, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1084, | |
| "total_flos": 453935093514240.0, | |
| "train_loss": 0.982664321841349, | |
| "train_runtime": 6266.1955, | |
| "train_samples_per_second": 22.133, | |
| "train_steps_per_second": 0.173 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 1084, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 453935093514240.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |