| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.9183673469387755, |
| "eval_steps": 500, |
| "global_step": 45000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0010204081632653062, |
| "grad_norm": 1.747214674949646, |
| "learning_rate": 9.600000000000001e-06, |
| "loss": 2.9007135009765626, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.0020408163265306124, |
| "grad_norm": 1.4239413738250732, |
| "learning_rate": 1.9600000000000002e-05, |
| "loss": 1.5238629150390626, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.003061224489795918, |
| "grad_norm": 1.2353929281234741, |
| "learning_rate": 2.96e-05, |
| "loss": 1.0528372955322265, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.004081632653061225, |
| "grad_norm": 1.1621766090393066, |
| "learning_rate": 3.960000000000001e-05, |
| "loss": 0.7107688903808593, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.00510204081632653, |
| "grad_norm": 1.2706868648529053, |
| "learning_rate": 4.96e-05, |
| "loss": 0.5750661849975586, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.006122448979591836, |
| "grad_norm": 0.9565913677215576, |
| "learning_rate": 5.96e-05, |
| "loss": 0.5003826904296875, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.007142857142857143, |
| "grad_norm": 0.7394572496414185, |
| "learning_rate": 6.96e-05, |
| "loss": 0.47811126708984375, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.00816326530612245, |
| "grad_norm": 0.7540650367736816, |
| "learning_rate": 7.960000000000001e-05, |
| "loss": 0.41311241149902345, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.009183673469387756, |
| "grad_norm": 0.5830128192901611, |
| "learning_rate": 8.960000000000001e-05, |
| "loss": 0.42279777526855467, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.01020408163265306, |
| "grad_norm": 0.45722952485084534, |
| "learning_rate": 9.960000000000001e-05, |
| "loss": 0.3503189468383789, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.011224489795918367, |
| "grad_norm": 0.3943341076374054, |
| "learning_rate": 0.00010960000000000001, |
| "loss": 0.37160335540771483, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.012244897959183673, |
| "grad_norm": 0.5774444937705994, |
| "learning_rate": 0.00011960000000000001, |
| "loss": 0.3780157852172852, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.013265306122448979, |
| "grad_norm": 0.384718656539917, |
| "learning_rate": 0.0001296, |
| "loss": 0.3615667724609375, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.014285714285714285, |
| "grad_norm": 0.470546692609787, |
| "learning_rate": 0.0001396, |
| "loss": 0.36237464904785155, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.015306122448979591, |
| "grad_norm": 0.4296679198741913, |
| "learning_rate": 0.0001496, |
| "loss": 0.3645317077636719, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.0163265306122449, |
| "grad_norm": 0.5170298218727112, |
| "learning_rate": 0.0001596, |
| "loss": 0.3482204818725586, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.017346938775510204, |
| "grad_norm": 0.28658953309059143, |
| "learning_rate": 0.0001696, |
| "loss": 0.3285676574707031, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.018367346938775512, |
| "grad_norm": 0.27739083766937256, |
| "learning_rate": 0.0001796, |
| "loss": 0.25687044143676757, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.019387755102040816, |
| "grad_norm": 0.31801655888557434, |
| "learning_rate": 0.0001896, |
| "loss": 0.18809181213378906, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.02040816326530612, |
| "grad_norm": 0.3564321994781494, |
| "learning_rate": 0.0001996, |
| "loss": 0.28171655654907224, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.02142857142857143, |
| "grad_norm": 0.34279024600982666, |
| "learning_rate": 0.0001998, |
| "loss": 0.3477919387817383, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.022448979591836733, |
| "grad_norm": 0.5220856070518494, |
| "learning_rate": 0.00019959166666666666, |
| "loss": 0.3521986770629883, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.02346938775510204, |
| "grad_norm": 0.3435756266117096, |
| "learning_rate": 0.00019938333333333335, |
| "loss": 0.34694950103759764, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.024489795918367346, |
| "grad_norm": 0.39061853289604187, |
| "learning_rate": 0.000199175, |
| "loss": 0.33362808227539065, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.025510204081632654, |
| "grad_norm": 0.6991215348243713, |
| "learning_rate": 0.0001989666666666667, |
| "loss": 0.3256123733520508, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.026530612244897958, |
| "grad_norm": 0.3511783480644226, |
| "learning_rate": 0.00019875833333333334, |
| "loss": 0.33866146087646487, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.027551020408163266, |
| "grad_norm": 0.3808313012123108, |
| "learning_rate": 0.00019855, |
| "loss": 0.30784854888916013, |
| "step": 1350 |
| }, |
| { |
| "epoch": 0.02857142857142857, |
| "grad_norm": 0.45190656185150146, |
| "learning_rate": 0.00019834166666666666, |
| "loss": 0.2862832832336426, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.02959183673469388, |
| "grad_norm": 0.3800064027309418, |
| "learning_rate": 0.00019813333333333334, |
| "loss": 0.32695507049560546, |
| "step": 1450 |
| }, |
| { |
| "epoch": 0.030612244897959183, |
| "grad_norm": 0.3821946084499359, |
| "learning_rate": 0.000197925, |
| "loss": 0.32072910308837893, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.03163265306122449, |
| "grad_norm": 0.356329083442688, |
| "learning_rate": 0.00019771666666666668, |
| "loss": 0.3068550682067871, |
| "step": 1550 |
| }, |
| { |
| "epoch": 0.0326530612244898, |
| "grad_norm": 0.26790183782577515, |
| "learning_rate": 0.00019750833333333334, |
| "loss": 0.28629026412963865, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.0336734693877551, |
| "grad_norm": 0.24051353335380554, |
| "learning_rate": 0.00019730000000000003, |
| "loss": 0.3116634559631348, |
| "step": 1650 |
| }, |
| { |
| "epoch": 0.03469387755102041, |
| "grad_norm": 0.2855391502380371, |
| "learning_rate": 0.00019709166666666668, |
| "loss": 0.2768780517578125, |
| "step": 1700 |
| }, |
| { |
| "epoch": 0.03571428571428571, |
| "grad_norm": 0.28543320298194885, |
| "learning_rate": 0.00019688333333333334, |
| "loss": 0.3128256034851074, |
| "step": 1750 |
| }, |
| { |
| "epoch": 0.036734693877551024, |
| "grad_norm": 0.3829735517501831, |
| "learning_rate": 0.000196675, |
| "loss": 0.2714161109924316, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.03775510204081633, |
| "grad_norm": 0.3017023503780365, |
| "learning_rate": 0.00019646666666666665, |
| "loss": 0.2929274368286133, |
| "step": 1850 |
| }, |
| { |
| "epoch": 0.03877551020408163, |
| "grad_norm": 0.24069251120090485, |
| "learning_rate": 0.00019625833333333334, |
| "loss": 0.27240211486816407, |
| "step": 1900 |
| }, |
| { |
| "epoch": 0.03979591836734694, |
| "grad_norm": 0.2976750135421753, |
| "learning_rate": 0.00019605, |
| "loss": 0.2899633026123047, |
| "step": 1950 |
| }, |
| { |
| "epoch": 0.04081632653061224, |
| "grad_norm": 0.3270716667175293, |
| "learning_rate": 0.00019584166666666668, |
| "loss": 0.2711215591430664, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.04183673469387755, |
| "grad_norm": 0.2565205991268158, |
| "learning_rate": 0.00019563333333333334, |
| "loss": 0.30275012969970705, |
| "step": 2050 |
| }, |
| { |
| "epoch": 0.04285714285714286, |
| "grad_norm": 0.3401188552379608, |
| "learning_rate": 0.00019542500000000002, |
| "loss": 0.28493772506713866, |
| "step": 2100 |
| }, |
| { |
| "epoch": 0.04387755102040816, |
| "grad_norm": 0.39523783326148987, |
| "learning_rate": 0.00019521666666666668, |
| "loss": 0.27601970672607423, |
| "step": 2150 |
| }, |
| { |
| "epoch": 0.044897959183673466, |
| "grad_norm": 0.20206202566623688, |
| "learning_rate": 0.00019500833333333336, |
| "loss": 0.27290882110595704, |
| "step": 2200 |
| }, |
| { |
| "epoch": 0.04591836734693878, |
| "grad_norm": 0.2421082854270935, |
| "learning_rate": 0.0001948, |
| "loss": 0.2691427040100098, |
| "step": 2250 |
| }, |
| { |
| "epoch": 0.04693877551020408, |
| "grad_norm": 0.30177634954452515, |
| "learning_rate": 0.00019459166666666668, |
| "loss": 0.26737031936645506, |
| "step": 2300 |
| }, |
| { |
| "epoch": 0.04795918367346939, |
| "grad_norm": 0.21533440053462982, |
| "learning_rate": 0.00019438333333333333, |
| "loss": 0.2751629066467285, |
| "step": 2350 |
| }, |
| { |
| "epoch": 0.04897959183673469, |
| "grad_norm": 0.2456338256597519, |
| "learning_rate": 0.00019417500000000002, |
| "loss": 0.2776307487487793, |
| "step": 2400 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 0.31056123971939087, |
| "learning_rate": 0.00019396666666666667, |
| "loss": 0.3090613555908203, |
| "step": 2450 |
| }, |
| { |
| "epoch": 0.05102040816326531, |
| "grad_norm": 0.2408883273601532, |
| "learning_rate": 0.00019375833333333336, |
| "loss": 0.28357997894287107, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.05204081632653061, |
| "grad_norm": 0.3348694145679474, |
| "learning_rate": 0.00019355000000000002, |
| "loss": 0.22965316772460936, |
| "step": 2550 |
| }, |
| { |
| "epoch": 0.053061224489795916, |
| "grad_norm": 0.18955211341381073, |
| "learning_rate": 0.00019334166666666667, |
| "loss": 0.25385122299194335, |
| "step": 2600 |
| }, |
| { |
| "epoch": 0.05408163265306123, |
| "grad_norm": 0.2340240627527237, |
| "learning_rate": 0.00019313333333333336, |
| "loss": 0.24012853622436522, |
| "step": 2650 |
| }, |
| { |
| "epoch": 0.05510204081632653, |
| "grad_norm": 0.3079259693622589, |
| "learning_rate": 0.000192925, |
| "loss": 0.2679576301574707, |
| "step": 2700 |
| }, |
| { |
| "epoch": 0.05612244897959184, |
| "grad_norm": 0.21826183795928955, |
| "learning_rate": 0.00019271666666666667, |
| "loss": 0.2620100975036621, |
| "step": 2750 |
| }, |
| { |
| "epoch": 0.05714285714285714, |
| "grad_norm": 0.23138746619224548, |
| "learning_rate": 0.00019250833333333333, |
| "loss": 0.2539723587036133, |
| "step": 2800 |
| }, |
| { |
| "epoch": 0.058163265306122446, |
| "grad_norm": 0.20887167751789093, |
| "learning_rate": 0.0001923, |
| "loss": 0.2468751335144043, |
| "step": 2850 |
| }, |
| { |
| "epoch": 0.05918367346938776, |
| "grad_norm": 0.268592894077301, |
| "learning_rate": 0.00019209166666666667, |
| "loss": 0.15909523010253906, |
| "step": 2900 |
| }, |
| { |
| "epoch": 0.06020408163265306, |
| "grad_norm": 0.25304126739501953, |
| "learning_rate": 0.00019188333333333335, |
| "loss": 0.27721269607543947, |
| "step": 2950 |
| }, |
| { |
| "epoch": 0.061224489795918366, |
| "grad_norm": 0.20484556257724762, |
| "learning_rate": 0.000191675, |
| "loss": 0.2851785469055176, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.06224489795918367, |
| "grad_norm": 0.23477648198604584, |
| "learning_rate": 0.0001914666666666667, |
| "loss": 0.2561204528808594, |
| "step": 3050 |
| }, |
| { |
| "epoch": 0.06326530612244897, |
| "grad_norm": 0.20200228691101074, |
| "learning_rate": 0.00019125833333333335, |
| "loss": 0.2533460807800293, |
| "step": 3100 |
| }, |
| { |
| "epoch": 0.06428571428571428, |
| "grad_norm": 0.2645045220851898, |
| "learning_rate": 0.00019105, |
| "loss": 0.2751885223388672, |
| "step": 3150 |
| }, |
| { |
| "epoch": 0.0653061224489796, |
| "grad_norm": 0.21196508407592773, |
| "learning_rate": 0.00019084166666666667, |
| "loss": 0.23478796005249022, |
| "step": 3200 |
| }, |
| { |
| "epoch": 0.0663265306122449, |
| "grad_norm": 0.2803957462310791, |
| "learning_rate": 0.00019063333333333335, |
| "loss": 0.2723574638366699, |
| "step": 3250 |
| }, |
| { |
| "epoch": 0.0673469387755102, |
| "grad_norm": 0.5901449918746948, |
| "learning_rate": 0.000190425, |
| "loss": 0.26311100006103516, |
| "step": 3300 |
| }, |
| { |
| "epoch": 0.06836734693877551, |
| "grad_norm": 0.38461631536483765, |
| "learning_rate": 0.00019021666666666666, |
| "loss": 0.2697314262390137, |
| "step": 3350 |
| }, |
| { |
| "epoch": 0.06938775510204082, |
| "grad_norm": 0.2759239971637726, |
| "learning_rate": 0.00019000833333333335, |
| "loss": 0.2560941505432129, |
| "step": 3400 |
| }, |
| { |
| "epoch": 0.07040816326530612, |
| "grad_norm": 0.2897125780582428, |
| "learning_rate": 0.0001898, |
| "loss": 0.2736033058166504, |
| "step": 3450 |
| }, |
| { |
| "epoch": 0.07142857142857142, |
| "grad_norm": 0.33106592297554016, |
| "learning_rate": 0.0001895916666666667, |
| "loss": 0.2775533676147461, |
| "step": 3500 |
| }, |
| { |
| "epoch": 0.07244897959183673, |
| "grad_norm": 0.1906435638666153, |
| "learning_rate": 0.00018938333333333335, |
| "loss": 0.26322145462036134, |
| "step": 3550 |
| }, |
| { |
| "epoch": 0.07346938775510205, |
| "grad_norm": 0.2763810157775879, |
| "learning_rate": 0.000189175, |
| "loss": 0.27947967529296874, |
| "step": 3600 |
| }, |
| { |
| "epoch": 0.07448979591836735, |
| "grad_norm": 0.19828622043132782, |
| "learning_rate": 0.00018896666666666666, |
| "loss": 0.26597856521606444, |
| "step": 3650 |
| }, |
| { |
| "epoch": 0.07551020408163266, |
| "grad_norm": 0.2526462972164154, |
| "learning_rate": 0.00018875833333333335, |
| "loss": 0.26246599197387693, |
| "step": 3700 |
| }, |
| { |
| "epoch": 0.07653061224489796, |
| "grad_norm": 0.2099679410457611, |
| "learning_rate": 0.00018855, |
| "loss": 0.26106624603271483, |
| "step": 3750 |
| }, |
| { |
| "epoch": 0.07755102040816327, |
| "grad_norm": 0.27196380496025085, |
| "learning_rate": 0.0001883416666666667, |
| "loss": 0.25065629959106445, |
| "step": 3800 |
| }, |
| { |
| "epoch": 0.07857142857142857, |
| "grad_norm": 0.22897310554981232, |
| "learning_rate": 0.00018813333333333334, |
| "loss": 0.2487507629394531, |
| "step": 3850 |
| }, |
| { |
| "epoch": 0.07959183673469387, |
| "grad_norm": 0.2111479938030243, |
| "learning_rate": 0.00018792500000000003, |
| "loss": 0.28081771850585935, |
| "step": 3900 |
| }, |
| { |
| "epoch": 0.08061224489795918, |
| "grad_norm": 0.41333141922950745, |
| "learning_rate": 0.00018771666666666668, |
| "loss": 0.2307835578918457, |
| "step": 3950 |
| }, |
| { |
| "epoch": 0.08163265306122448, |
| "grad_norm": 0.21056070923805237, |
| "learning_rate": 0.00018750833333333334, |
| "loss": 0.24496456146240234, |
| "step": 4000 |
| }, |
| { |
| "epoch": 0.0826530612244898, |
| "grad_norm": 0.2062331885099411, |
| "learning_rate": 0.0001873, |
| "loss": 0.23850811004638672, |
| "step": 4050 |
| }, |
| { |
| "epoch": 0.0836734693877551, |
| "grad_norm": 0.18119142949581146, |
| "learning_rate": 0.00018709166666666666, |
| "loss": 0.2702301406860352, |
| "step": 4100 |
| }, |
| { |
| "epoch": 0.08469387755102041, |
| "grad_norm": 0.21935491263866425, |
| "learning_rate": 0.00018688333333333334, |
| "loss": 0.2529887008666992, |
| "step": 4150 |
| }, |
| { |
| "epoch": 0.08571428571428572, |
| "grad_norm": 0.2520492374897003, |
| "learning_rate": 0.000186675, |
| "loss": 0.24680517196655274, |
| "step": 4200 |
| }, |
| { |
| "epoch": 0.08673469387755102, |
| "grad_norm": 0.21996288001537323, |
| "learning_rate": 0.00018646666666666668, |
| "loss": 0.260424861907959, |
| "step": 4250 |
| }, |
| { |
| "epoch": 0.08775510204081632, |
| "grad_norm": 0.18490734696388245, |
| "learning_rate": 0.00018625833333333334, |
| "loss": 0.2572420883178711, |
| "step": 4300 |
| }, |
| { |
| "epoch": 0.08877551020408163, |
| "grad_norm": 0.19968494772911072, |
| "learning_rate": 0.00018605000000000002, |
| "loss": 0.25690040588378904, |
| "step": 4350 |
| }, |
| { |
| "epoch": 0.08979591836734693, |
| "grad_norm": 0.18151628971099854, |
| "learning_rate": 0.00018584166666666668, |
| "loss": 0.23846803665161131, |
| "step": 4400 |
| }, |
| { |
| "epoch": 0.09081632653061225, |
| "grad_norm": 0.19587203860282898, |
| "learning_rate": 0.00018563333333333334, |
| "loss": 0.2594054985046387, |
| "step": 4450 |
| }, |
| { |
| "epoch": 0.09183673469387756, |
| "grad_norm": 0.32912376523017883, |
| "learning_rate": 0.000185425, |
| "loss": 0.2486746406555176, |
| "step": 4500 |
| }, |
| { |
| "epoch": 0.09285714285714286, |
| "grad_norm": 0.2364601194858551, |
| "learning_rate": 0.00018521666666666668, |
| "loss": 0.2627392768859863, |
| "step": 4550 |
| }, |
| { |
| "epoch": 0.09387755102040816, |
| "grad_norm": 0.11974240839481354, |
| "learning_rate": 0.00018500833333333334, |
| "loss": 0.21929288864135743, |
| "step": 4600 |
| }, |
| { |
| "epoch": 0.09489795918367347, |
| "grad_norm": 0.19204378128051758, |
| "learning_rate": 0.00018480000000000002, |
| "loss": 0.18641704559326172, |
| "step": 4650 |
| }, |
| { |
| "epoch": 0.09591836734693877, |
| "grad_norm": 0.2613525986671448, |
| "learning_rate": 0.00018459166666666668, |
| "loss": 0.2442936897277832, |
| "step": 4700 |
| }, |
| { |
| "epoch": 0.09693877551020408, |
| "grad_norm": 0.22148281335830688, |
| "learning_rate": 0.00018438333333333336, |
| "loss": 0.25667667388916016, |
| "step": 4750 |
| }, |
| { |
| "epoch": 0.09795918367346938, |
| "grad_norm": 0.2820097506046295, |
| "learning_rate": 0.00018417500000000002, |
| "loss": 0.21245412826538085, |
| "step": 4800 |
| }, |
| { |
| "epoch": 0.09897959183673469, |
| "grad_norm": 0.2048281878232956, |
| "learning_rate": 0.00018396666666666668, |
| "loss": 0.23076135635375977, |
| "step": 4850 |
| }, |
| { |
| "epoch": 0.1, |
| "grad_norm": 0.2739567160606384, |
| "learning_rate": 0.00018375833333333333, |
| "loss": 0.2856523323059082, |
| "step": 4900 |
| }, |
| { |
| "epoch": 0.10102040816326531, |
| "grad_norm": 0.26582998037338257, |
| "learning_rate": 0.00018355, |
| "loss": 0.2545345878601074, |
| "step": 4950 |
| }, |
| { |
| "epoch": 0.10204081632653061, |
| "grad_norm": 0.192887082695961, |
| "learning_rate": 0.00018334166666666667, |
| "loss": 0.2435542869567871, |
| "step": 5000 |
| }, |
| { |
| "epoch": 0.10306122448979592, |
| "grad_norm": 0.18641699850559235, |
| "learning_rate": 0.00018313333333333333, |
| "loss": 0.23780487060546876, |
| "step": 5050 |
| }, |
| { |
| "epoch": 0.10408163265306122, |
| "grad_norm": 0.26348331570625305, |
| "learning_rate": 0.00018292500000000001, |
| "loss": 0.25740249633789064, |
| "step": 5100 |
| }, |
| { |
| "epoch": 0.10510204081632653, |
| "grad_norm": 0.18192711472511292, |
| "learning_rate": 0.00018271666666666667, |
| "loss": 0.2640067672729492, |
| "step": 5150 |
| }, |
| { |
| "epoch": 0.10612244897959183, |
| "grad_norm": 0.24081537127494812, |
| "learning_rate": 0.00018250833333333336, |
| "loss": 0.24000905990600585, |
| "step": 5200 |
| }, |
| { |
| "epoch": 0.10714285714285714, |
| "grad_norm": 0.1570502072572708, |
| "learning_rate": 0.0001823, |
| "loss": 0.245651912689209, |
| "step": 5250 |
| }, |
| { |
| "epoch": 0.10816326530612246, |
| "grad_norm": 0.13232532143592834, |
| "learning_rate": 0.00018209166666666667, |
| "loss": 0.15869504928588868, |
| "step": 5300 |
| }, |
| { |
| "epoch": 0.10918367346938776, |
| "grad_norm": 0.23496298491954803, |
| "learning_rate": 0.00018188333333333333, |
| "loss": 0.1904274559020996, |
| "step": 5350 |
| }, |
| { |
| "epoch": 0.11020408163265306, |
| "grad_norm": 0.25786077976226807, |
| "learning_rate": 0.000181675, |
| "loss": 0.25172576904296873, |
| "step": 5400 |
| }, |
| { |
| "epoch": 0.11122448979591837, |
| "grad_norm": 0.16870424151420593, |
| "learning_rate": 0.00018146666666666667, |
| "loss": 0.22960330963134765, |
| "step": 5450 |
| }, |
| { |
| "epoch": 0.11224489795918367, |
| "grad_norm": 0.13842836022377014, |
| "learning_rate": 0.00018125833333333335, |
| "loss": 0.21614797592163085, |
| "step": 5500 |
| }, |
| { |
| "epoch": 0.11326530612244898, |
| "grad_norm": 0.1558876633644104, |
| "learning_rate": 0.00018105, |
| "loss": 0.1952862548828125, |
| "step": 5550 |
| }, |
| { |
| "epoch": 0.11428571428571428, |
| "grad_norm": 0.1988261342048645, |
| "learning_rate": 0.00018084166666666667, |
| "loss": 0.22747718811035156, |
| "step": 5600 |
| }, |
| { |
| "epoch": 0.11530612244897959, |
| "grad_norm": 0.2106727659702301, |
| "learning_rate": 0.00018063333333333335, |
| "loss": 0.22860513687133788, |
| "step": 5650 |
| }, |
| { |
| "epoch": 0.11632653061224489, |
| "grad_norm": 0.19396160542964935, |
| "learning_rate": 0.000180425, |
| "loss": 0.268677806854248, |
| "step": 5700 |
| }, |
| { |
| "epoch": 0.11734693877551021, |
| "grad_norm": 0.2126370370388031, |
| "learning_rate": 0.00018021666666666667, |
| "loss": 0.23296730041503907, |
| "step": 5750 |
| }, |
| { |
| "epoch": 0.11836734693877551, |
| "grad_norm": 0.23235207796096802, |
| "learning_rate": 0.00018000833333333332, |
| "loss": 0.25676736831665037, |
| "step": 5800 |
| }, |
| { |
| "epoch": 0.11938775510204082, |
| "grad_norm": 0.20349349081516266, |
| "learning_rate": 0.0001798, |
| "loss": 0.22073328018188476, |
| "step": 5850 |
| }, |
| { |
| "epoch": 0.12040816326530612, |
| "grad_norm": 0.21047183871269226, |
| "learning_rate": 0.00017959166666666666, |
| "loss": 0.23178285598754883, |
| "step": 5900 |
| }, |
| { |
| "epoch": 0.12142857142857143, |
| "grad_norm": 0.17973244190216064, |
| "learning_rate": 0.00017938333333333335, |
| "loss": 0.2536413955688477, |
| "step": 5950 |
| }, |
| { |
| "epoch": 0.12244897959183673, |
| "grad_norm": 0.3100139796733856, |
| "learning_rate": 0.000179175, |
| "loss": 0.22924301147460938, |
| "step": 6000 |
| }, |
| { |
| "epoch": 0.12346938775510204, |
| "grad_norm": 0.23539181053638458, |
| "learning_rate": 0.0001789666666666667, |
| "loss": 0.3015317726135254, |
| "step": 6050 |
| }, |
| { |
| "epoch": 0.12448979591836734, |
| "grad_norm": 0.2297438383102417, |
| "learning_rate": 0.00017875833333333335, |
| "loss": 0.23267417907714844, |
| "step": 6100 |
| }, |
| { |
| "epoch": 0.12551020408163266, |
| "grad_norm": 0.24020229279994965, |
| "learning_rate": 0.00017855000000000003, |
| "loss": 0.2438960838317871, |
| "step": 6150 |
| }, |
| { |
| "epoch": 0.12653061224489795, |
| "grad_norm": 0.19935473799705505, |
| "learning_rate": 0.00017834166666666666, |
| "loss": 0.24910968780517578, |
| "step": 6200 |
| }, |
| { |
| "epoch": 0.12755102040816327, |
| "grad_norm": 0.14035022258758545, |
| "learning_rate": 0.00017813333333333334, |
| "loss": 0.19773637771606445, |
| "step": 6250 |
| }, |
| { |
| "epoch": 0.12857142857142856, |
| "grad_norm": 0.18467874825000763, |
| "learning_rate": 0.000177925, |
| "loss": 0.23008169174194337, |
| "step": 6300 |
| }, |
| { |
| "epoch": 0.12959183673469388, |
| "grad_norm": 0.18327519297599792, |
| "learning_rate": 0.00017771666666666666, |
| "loss": 0.22650135040283204, |
| "step": 6350 |
| }, |
| { |
| "epoch": 0.1306122448979592, |
| "grad_norm": 0.1742118000984192, |
| "learning_rate": 0.00017750833333333334, |
| "loss": 0.24949928283691405, |
| "step": 6400 |
| }, |
| { |
| "epoch": 0.13163265306122449, |
| "grad_norm": 0.2683607041835785, |
| "learning_rate": 0.0001773, |
| "loss": 0.23931278228759767, |
| "step": 6450 |
| }, |
| { |
| "epoch": 0.1326530612244898, |
| "grad_norm": 0.22325648367404938, |
| "learning_rate": 0.00017709166666666668, |
| "loss": 0.2348464012145996, |
| "step": 6500 |
| }, |
| { |
| "epoch": 0.1336734693877551, |
| "grad_norm": 0.23074108362197876, |
| "learning_rate": 0.00017688333333333334, |
| "loss": 0.22793367385864258, |
| "step": 6550 |
| }, |
| { |
| "epoch": 0.1346938775510204, |
| "grad_norm": 0.2005387246608734, |
| "learning_rate": 0.00017667500000000003, |
| "loss": 0.22257186889648437, |
| "step": 6600 |
| }, |
| { |
| "epoch": 0.1357142857142857, |
| "grad_norm": 0.22492945194244385, |
| "learning_rate": 0.00017646666666666668, |
| "loss": 0.2518248176574707, |
| "step": 6650 |
| }, |
| { |
| "epoch": 0.13673469387755102, |
| "grad_norm": 0.19270016252994537, |
| "learning_rate": 0.00017625833333333334, |
| "loss": 0.23488008499145507, |
| "step": 6700 |
| }, |
| { |
| "epoch": 0.1377551020408163, |
| "grad_norm": 0.2650369703769684, |
| "learning_rate": 0.00017605, |
| "loss": 0.2485354995727539, |
| "step": 6750 |
| }, |
| { |
| "epoch": 0.13877551020408163, |
| "grad_norm": 0.16986238956451416, |
| "learning_rate": 0.00017584166666666668, |
| "loss": 0.2421465301513672, |
| "step": 6800 |
| }, |
| { |
| "epoch": 0.13979591836734695, |
| "grad_norm": 0.584934413433075, |
| "learning_rate": 0.00017563333333333334, |
| "loss": 0.21577610015869142, |
| "step": 6850 |
| }, |
| { |
| "epoch": 0.14081632653061224, |
| "grad_norm": 0.211620032787323, |
| "learning_rate": 0.00017542500000000002, |
| "loss": 0.2501153373718262, |
| "step": 6900 |
| }, |
| { |
| "epoch": 0.14183673469387756, |
| "grad_norm": 0.22148476541042328, |
| "learning_rate": 0.00017521666666666668, |
| "loss": 0.27484086990356443, |
| "step": 6950 |
| }, |
| { |
| "epoch": 0.14285714285714285, |
| "grad_norm": 0.19075964391231537, |
| "learning_rate": 0.00017500833333333336, |
| "loss": 0.22781301498413087, |
| "step": 7000 |
| }, |
| { |
| "epoch": 0.14387755102040817, |
| "grad_norm": 0.1916443407535553, |
| "learning_rate": 0.00017480000000000002, |
| "loss": 0.2180193519592285, |
| "step": 7050 |
| }, |
| { |
| "epoch": 0.14489795918367346, |
| "grad_norm": 0.15787392854690552, |
| "learning_rate": 0.00017459166666666668, |
| "loss": 0.24084676742553712, |
| "step": 7100 |
| }, |
| { |
| "epoch": 0.14591836734693878, |
| "grad_norm": 0.14521488547325134, |
| "learning_rate": 0.00017438333333333333, |
| "loss": 0.1710742950439453, |
| "step": 7150 |
| }, |
| { |
| "epoch": 0.1469387755102041, |
| "grad_norm": 0.2912120819091797, |
| "learning_rate": 0.000174175, |
| "loss": 0.19460933685302734, |
| "step": 7200 |
| }, |
| { |
| "epoch": 0.14795918367346939, |
| "grad_norm": 0.235012486577034, |
| "learning_rate": 0.00017396666666666668, |
| "loss": 0.24076326370239257, |
| "step": 7250 |
| }, |
| { |
| "epoch": 0.1489795918367347, |
| "grad_norm": 0.15365862846374512, |
| "learning_rate": 0.00017375833333333333, |
| "loss": 0.15890993118286134, |
| "step": 7300 |
| }, |
| { |
| "epoch": 0.15, |
| "grad_norm": 0.1753380447626114, |
| "learning_rate": 0.00017355000000000002, |
| "loss": 0.1807994270324707, |
| "step": 7350 |
| }, |
| { |
| "epoch": 0.1510204081632653, |
| "grad_norm": 0.1783948391675949, |
| "learning_rate": 0.00017334166666666667, |
| "loss": 0.1949660873413086, |
| "step": 7400 |
| }, |
| { |
| "epoch": 0.1520408163265306, |
| "grad_norm": 0.21300694346427917, |
| "learning_rate": 0.00017313333333333336, |
| "loss": 0.26889331817626955, |
| "step": 7450 |
| }, |
| { |
| "epoch": 0.15306122448979592, |
| "grad_norm": 0.17175327241420746, |
| "learning_rate": 0.00017292500000000002, |
| "loss": 0.2403140640258789, |
| "step": 7500 |
| }, |
| { |
| "epoch": 0.1540816326530612, |
| "grad_norm": 0.17446576058864594, |
| "learning_rate": 0.00017271666666666667, |
| "loss": 0.2203696060180664, |
| "step": 7550 |
| }, |
| { |
| "epoch": 0.15510204081632653, |
| "grad_norm": 0.18991930782794952, |
| "learning_rate": 0.00017250833333333333, |
| "loss": 0.21493810653686524, |
| "step": 7600 |
| }, |
| { |
| "epoch": 0.15612244897959185, |
| "grad_norm": 0.17720980942249298, |
| "learning_rate": 0.00017230000000000001, |
| "loss": 0.2457187843322754, |
| "step": 7650 |
| }, |
| { |
| "epoch": 0.15714285714285714, |
| "grad_norm": 0.22987200319766998, |
| "learning_rate": 0.00017209166666666667, |
| "loss": 0.24987890243530272, |
| "step": 7700 |
| }, |
| { |
| "epoch": 0.15816326530612246, |
| "grad_norm": 0.20722761750221252, |
| "learning_rate": 0.00017188333333333336, |
| "loss": 0.22456403732299804, |
| "step": 7750 |
| }, |
| { |
| "epoch": 0.15918367346938775, |
| "grad_norm": 0.17368385195732117, |
| "learning_rate": 0.000171675, |
| "loss": 0.23329797744750977, |
| "step": 7800 |
| }, |
| { |
| "epoch": 0.16020408163265307, |
| "grad_norm": 0.1477518379688263, |
| "learning_rate": 0.00017146666666666667, |
| "loss": 0.150212345123291, |
| "step": 7850 |
| }, |
| { |
| "epoch": 0.16122448979591836, |
| "grad_norm": 0.23035384714603424, |
| "learning_rate": 0.00017125833333333335, |
| "loss": 0.18162582397460938, |
| "step": 7900 |
| }, |
| { |
| "epoch": 0.16224489795918368, |
| "grad_norm": 0.2011357545852661, |
| "learning_rate": 0.00017105, |
| "loss": 0.23587108612060548, |
| "step": 7950 |
| }, |
| { |
| "epoch": 0.16326530612244897, |
| "grad_norm": 0.23610138893127441, |
| "learning_rate": 0.00017084166666666667, |
| "loss": 0.25925134658813476, |
| "step": 8000 |
| }, |
| { |
| "epoch": 0.16428571428571428, |
| "grad_norm": 0.33437424898147583, |
| "learning_rate": 0.00017063333333333332, |
| "loss": 0.2467789649963379, |
| "step": 8050 |
| }, |
| { |
| "epoch": 0.1653061224489796, |
| "grad_norm": 0.16571049392223358, |
| "learning_rate": 0.000170425, |
| "loss": 0.22530656814575195, |
| "step": 8100 |
| }, |
| { |
| "epoch": 0.1663265306122449, |
| "grad_norm": 0.19635702669620514, |
| "learning_rate": 0.00017021666666666667, |
| "loss": 0.2387130355834961, |
| "step": 8150 |
| }, |
| { |
| "epoch": 0.1673469387755102, |
| "grad_norm": 0.2154119610786438, |
| "learning_rate": 0.00017000833333333335, |
| "loss": 0.22704246520996094, |
| "step": 8200 |
| }, |
| { |
| "epoch": 0.1683673469387755, |
| "grad_norm": 0.177113875746727, |
| "learning_rate": 0.0001698, |
| "loss": 0.2153651237487793, |
| "step": 8250 |
| }, |
| { |
| "epoch": 0.16938775510204082, |
| "grad_norm": 0.16632798314094543, |
| "learning_rate": 0.0001695916666666667, |
| "loss": 0.22499719619750977, |
| "step": 8300 |
| }, |
| { |
| "epoch": 0.1704081632653061, |
| "grad_norm": 0.12536615133285522, |
| "learning_rate": 0.00016938333333333335, |
| "loss": 0.18060928344726562, |
| "step": 8350 |
| }, |
| { |
| "epoch": 0.17142857142857143, |
| "grad_norm": 0.17900407314300537, |
| "learning_rate": 0.000169175, |
| "loss": 0.1342845344543457, |
| "step": 8400 |
| }, |
| { |
| "epoch": 0.17244897959183675, |
| "grad_norm": 0.36062344908714294, |
| "learning_rate": 0.00016896666666666666, |
| "loss": 0.2387704086303711, |
| "step": 8450 |
| }, |
| { |
| "epoch": 0.17346938775510204, |
| "grad_norm": 0.21417365968227386, |
| "learning_rate": 0.00016875833333333335, |
| "loss": 0.2395523452758789, |
| "step": 8500 |
| }, |
| { |
| "epoch": 0.17448979591836736, |
| "grad_norm": 0.15625624358654022, |
| "learning_rate": 0.00016855, |
| "loss": 0.23403961181640626, |
| "step": 8550 |
| }, |
| { |
| "epoch": 0.17551020408163265, |
| "grad_norm": 0.2608756422996521, |
| "learning_rate": 0.00016834166666666666, |
| "loss": 0.20653511047363282, |
| "step": 8600 |
| }, |
| { |
| "epoch": 0.17653061224489797, |
| "grad_norm": 0.14978773891925812, |
| "learning_rate": 0.00016813333333333335, |
| "loss": 0.2276185607910156, |
| "step": 8650 |
| }, |
| { |
| "epoch": 0.17755102040816326, |
| "grad_norm": 0.164831280708313, |
| "learning_rate": 0.000167925, |
| "loss": 0.2196169090270996, |
| "step": 8700 |
| }, |
| { |
| "epoch": 0.17857142857142858, |
| "grad_norm": 0.2136063277721405, |
| "learning_rate": 0.0001677166666666667, |
| "loss": 0.22439205169677734, |
| "step": 8750 |
| }, |
| { |
| "epoch": 0.17959183673469387, |
| "grad_norm": 0.2042110115289688, |
| "learning_rate": 0.00016750833333333334, |
| "loss": 0.25702680587768556, |
| "step": 8800 |
| }, |
| { |
| "epoch": 0.18061224489795918, |
| "grad_norm": 0.15435971319675446, |
| "learning_rate": 0.0001673, |
| "loss": 0.21708799362182618, |
| "step": 8850 |
| }, |
| { |
| "epoch": 0.1816326530612245, |
| "grad_norm": 0.15783603489398956, |
| "learning_rate": 0.00016709166666666666, |
| "loss": 0.2066781997680664, |
| "step": 8900 |
| }, |
| { |
| "epoch": 0.1826530612244898, |
| "grad_norm": 0.15515024960041046, |
| "learning_rate": 0.00016688333333333334, |
| "loss": 0.20220188140869141, |
| "step": 8950 |
| }, |
| { |
| "epoch": 0.1836734693877551, |
| "grad_norm": 0.13879244029521942, |
| "learning_rate": 0.000166675, |
| "loss": 0.15490728378295898, |
| "step": 9000 |
| }, |
| { |
| "epoch": 0.1846938775510204, |
| "grad_norm": 0.17483779788017273, |
| "learning_rate": 0.00016646666666666668, |
| "loss": 0.19913965225219726, |
| "step": 9050 |
| }, |
| { |
| "epoch": 0.18571428571428572, |
| "grad_norm": 0.14338476955890656, |
| "learning_rate": 0.00016625833333333334, |
| "loss": 0.2302665901184082, |
| "step": 9100 |
| }, |
| { |
| "epoch": 0.186734693877551, |
| "grad_norm": 0.19108951091766357, |
| "learning_rate": 0.00016605000000000002, |
| "loss": 0.2207638168334961, |
| "step": 9150 |
| }, |
| { |
| "epoch": 0.18775510204081633, |
| "grad_norm": 0.2902897894382477, |
| "learning_rate": 0.00016584166666666668, |
| "loss": 0.26275455474853515, |
| "step": 9200 |
| }, |
| { |
| "epoch": 0.18877551020408162, |
| "grad_norm": 0.2468404471874237, |
| "learning_rate": 0.00016563333333333334, |
| "loss": 0.22652421951293944, |
| "step": 9250 |
| }, |
| { |
| "epoch": 0.18979591836734694, |
| "grad_norm": 0.15006621181964874, |
| "learning_rate": 0.000165425, |
| "loss": 0.234797420501709, |
| "step": 9300 |
| }, |
| { |
| "epoch": 0.19081632653061226, |
| "grad_norm": 0.3083290159702301, |
| "learning_rate": 0.00016521666666666665, |
| "loss": 0.2213102912902832, |
| "step": 9350 |
| }, |
| { |
| "epoch": 0.19183673469387755, |
| "grad_norm": 0.23879224061965942, |
| "learning_rate": 0.00016500833333333334, |
| "loss": 0.23831743240356446, |
| "step": 9400 |
| }, |
| { |
| "epoch": 0.19285714285714287, |
| "grad_norm": 0.24960699677467346, |
| "learning_rate": 0.0001648, |
| "loss": 0.23019039154052734, |
| "step": 9450 |
| }, |
| { |
| "epoch": 0.19387755102040816, |
| "grad_norm": 0.23694069683551788, |
| "learning_rate": 0.00016459166666666668, |
| "loss": 0.1808626174926758, |
| "step": 9500 |
| }, |
| { |
| "epoch": 0.19489795918367347, |
| "grad_norm": 0.1712574064731598, |
| "learning_rate": 0.00016438333333333334, |
| "loss": 0.23488353729248046, |
| "step": 9550 |
| }, |
| { |
| "epoch": 0.19591836734693877, |
| "grad_norm": 0.22367513179779053, |
| "learning_rate": 0.00016417500000000002, |
| "loss": 0.21684967041015624, |
| "step": 9600 |
| }, |
| { |
| "epoch": 0.19693877551020408, |
| "grad_norm": 0.16823172569274902, |
| "learning_rate": 0.00016396666666666668, |
| "loss": 0.23959226608276368, |
| "step": 9650 |
| }, |
| { |
| "epoch": 0.19795918367346937, |
| "grad_norm": 0.25998687744140625, |
| "learning_rate": 0.00016375833333333333, |
| "loss": 0.22377561569213866, |
| "step": 9700 |
| }, |
| { |
| "epoch": 0.1989795918367347, |
| "grad_norm": 0.18959476053714752, |
| "learning_rate": 0.00016355, |
| "loss": 0.2755254554748535, |
| "step": 9750 |
| }, |
| { |
| "epoch": 0.2, |
| "grad_norm": 0.26864951848983765, |
| "learning_rate": 0.00016334166666666668, |
| "loss": 0.239896240234375, |
| "step": 9800 |
| }, |
| { |
| "epoch": 0.2010204081632653, |
| "grad_norm": 0.16514073312282562, |
| "learning_rate": 0.00016313333333333333, |
| "loss": 0.20732919692993165, |
| "step": 9850 |
| }, |
| { |
| "epoch": 0.20204081632653062, |
| "grad_norm": 0.19930091500282288, |
| "learning_rate": 0.00016292500000000002, |
| "loss": 0.22147659301757813, |
| "step": 9900 |
| }, |
| { |
| "epoch": 0.2030612244897959, |
| "grad_norm": 0.20155663788318634, |
| "learning_rate": 0.00016271666666666667, |
| "loss": 0.21227607727050782, |
| "step": 9950 |
| }, |
| { |
| "epoch": 0.20408163265306123, |
| "grad_norm": 0.15768975019454956, |
| "learning_rate": 0.00016250833333333336, |
| "loss": 0.20134233474731444, |
| "step": 10000 |
| }, |
| { |
| "epoch": 0.20510204081632652, |
| "grad_norm": 0.2521868348121643, |
| "learning_rate": 0.00016230000000000001, |
| "loss": 0.22343421936035157, |
| "step": 10050 |
| }, |
| { |
| "epoch": 0.20612244897959184, |
| "grad_norm": 0.15740498900413513, |
| "learning_rate": 0.00016209166666666667, |
| "loss": 0.2235814094543457, |
| "step": 10100 |
| }, |
| { |
| "epoch": 0.20714285714285716, |
| "grad_norm": 0.2574876844882965, |
| "learning_rate": 0.00016188333333333333, |
| "loss": 0.22124135971069336, |
| "step": 10150 |
| }, |
| { |
| "epoch": 0.20816326530612245, |
| "grad_norm": 0.1711849570274353, |
| "learning_rate": 0.00016167499999999999, |
| "loss": 0.22449752807617188, |
| "step": 10200 |
| }, |
| { |
| "epoch": 0.20918367346938777, |
| "grad_norm": 0.16069597005844116, |
| "learning_rate": 0.00016146666666666667, |
| "loss": 0.22119144439697266, |
| "step": 10250 |
| }, |
| { |
| "epoch": 0.21020408163265306, |
| "grad_norm": 0.16962496936321259, |
| "learning_rate": 0.00016125833333333333, |
| "loss": 0.23080724716186524, |
| "step": 10300 |
| }, |
| { |
| "epoch": 0.21122448979591837, |
| "grad_norm": 0.22259433567523956, |
| "learning_rate": 0.00016105, |
| "loss": 0.19102418899536133, |
| "step": 10350 |
| }, |
| { |
| "epoch": 0.21224489795918366, |
| "grad_norm": 0.14817878603935242, |
| "learning_rate": 0.00016084166666666667, |
| "loss": 0.21070737838745118, |
| "step": 10400 |
| }, |
| { |
| "epoch": 0.21326530612244898, |
| "grad_norm": 0.3290054202079773, |
| "learning_rate": 0.00016063333333333335, |
| "loss": 0.18405279159545898, |
| "step": 10450 |
| }, |
| { |
| "epoch": 0.21428571428571427, |
| "grad_norm": 0.34895798563957214, |
| "learning_rate": 0.000160425, |
| "loss": 0.21118934631347655, |
| "step": 10500 |
| }, |
| { |
| "epoch": 0.2153061224489796, |
| "grad_norm": 0.18928605318069458, |
| "learning_rate": 0.0001602166666666667, |
| "loss": 0.2026797866821289, |
| "step": 10550 |
| }, |
| { |
| "epoch": 0.2163265306122449, |
| "grad_norm": 0.31760379672050476, |
| "learning_rate": 0.00016000833333333335, |
| "loss": 0.20241106033325196, |
| "step": 10600 |
| }, |
| { |
| "epoch": 0.2173469387755102, |
| "grad_norm": 0.21261534094810486, |
| "learning_rate": 0.0001598, |
| "loss": 0.2348382568359375, |
| "step": 10650 |
| }, |
| { |
| "epoch": 0.21836734693877552, |
| "grad_norm": 0.25979313254356384, |
| "learning_rate": 0.00015959166666666667, |
| "loss": 0.1868024253845215, |
| "step": 10700 |
| }, |
| { |
| "epoch": 0.2193877551020408, |
| "grad_norm": 0.18968409299850464, |
| "learning_rate": 0.00015938333333333335, |
| "loss": 0.23709041595458985, |
| "step": 10750 |
| }, |
| { |
| "epoch": 0.22040816326530613, |
| "grad_norm": 0.113967664539814, |
| "learning_rate": 0.000159175, |
| "loss": 0.18982641220092775, |
| "step": 10800 |
| }, |
| { |
| "epoch": 0.22142857142857142, |
| "grad_norm": 0.17538873851299286, |
| "learning_rate": 0.00015896666666666666, |
| "loss": 0.1972733497619629, |
| "step": 10850 |
| }, |
| { |
| "epoch": 0.22244897959183674, |
| "grad_norm": 0.22401735186576843, |
| "learning_rate": 0.00015875833333333335, |
| "loss": 0.21322017669677734, |
| "step": 10900 |
| }, |
| { |
| "epoch": 0.22346938775510203, |
| "grad_norm": 0.1772245168685913, |
| "learning_rate": 0.00015855, |
| "loss": 0.21429908752441407, |
| "step": 10950 |
| }, |
| { |
| "epoch": 0.22448979591836735, |
| "grad_norm": 0.16962389647960663, |
| "learning_rate": 0.0001583416666666667, |
| "loss": 0.22701108932495118, |
| "step": 11000 |
| }, |
| { |
| "epoch": 0.22551020408163266, |
| "grad_norm": 0.16838356852531433, |
| "learning_rate": 0.00015813333333333335, |
| "loss": 0.19800830841064454, |
| "step": 11050 |
| }, |
| { |
| "epoch": 0.22653061224489796, |
| "grad_norm": 0.14440977573394775, |
| "learning_rate": 0.000157925, |
| "loss": 0.2604854774475098, |
| "step": 11100 |
| }, |
| { |
| "epoch": 0.22755102040816327, |
| "grad_norm": 0.16742219030857086, |
| "learning_rate": 0.00015771666666666666, |
| "loss": 0.2227446937561035, |
| "step": 11150 |
| }, |
| { |
| "epoch": 0.22857142857142856, |
| "grad_norm": 0.21329857409000397, |
| "learning_rate": 0.00015750833333333334, |
| "loss": 0.2092646026611328, |
| "step": 11200 |
| }, |
| { |
| "epoch": 0.22959183673469388, |
| "grad_norm": 0.14716580510139465, |
| "learning_rate": 0.0001573, |
| "loss": 0.20734121322631835, |
| "step": 11250 |
| }, |
| { |
| "epoch": 0.23061224489795917, |
| "grad_norm": 0.17271864414215088, |
| "learning_rate": 0.00015709166666666669, |
| "loss": 0.14340096473693847, |
| "step": 11300 |
| }, |
| { |
| "epoch": 0.2316326530612245, |
| "grad_norm": 0.15021106600761414, |
| "learning_rate": 0.00015688333333333334, |
| "loss": 0.18479793548583984, |
| "step": 11350 |
| }, |
| { |
| "epoch": 0.23265306122448978, |
| "grad_norm": 0.22414959967136383, |
| "learning_rate": 0.00015667500000000003, |
| "loss": 0.2072342300415039, |
| "step": 11400 |
| }, |
| { |
| "epoch": 0.2336734693877551, |
| "grad_norm": 0.20402732491493225, |
| "learning_rate": 0.00015646666666666668, |
| "loss": 0.21692256927490233, |
| "step": 11450 |
| }, |
| { |
| "epoch": 0.23469387755102042, |
| "grad_norm": 0.15619000792503357, |
| "learning_rate": 0.00015625833333333334, |
| "loss": 0.22381683349609374, |
| "step": 11500 |
| }, |
| { |
| "epoch": 0.2357142857142857, |
| "grad_norm": 0.16317050158977509, |
| "learning_rate": 0.00015605, |
| "loss": 0.23182619094848633, |
| "step": 11550 |
| }, |
| { |
| "epoch": 0.23673469387755103, |
| "grad_norm": 0.18910205364227295, |
| "learning_rate": 0.00015584166666666666, |
| "loss": 0.25546884536743164, |
| "step": 11600 |
| }, |
| { |
| "epoch": 0.23775510204081632, |
| "grad_norm": 0.14334766566753387, |
| "learning_rate": 0.00015563333333333334, |
| "loss": 0.22352867126464843, |
| "step": 11650 |
| }, |
| { |
| "epoch": 0.23877551020408164, |
| "grad_norm": 0.18432728946208954, |
| "learning_rate": 0.000155425, |
| "loss": 0.1954519271850586, |
| "step": 11700 |
| }, |
| { |
| "epoch": 0.23979591836734693, |
| "grad_norm": 0.13672347366809845, |
| "learning_rate": 0.00015521666666666668, |
| "loss": 0.18877283096313477, |
| "step": 11750 |
| }, |
| { |
| "epoch": 0.24081632653061225, |
| "grad_norm": 0.20067130029201508, |
| "learning_rate": 0.00015500833333333334, |
| "loss": 0.1814111328125, |
| "step": 11800 |
| }, |
| { |
| "epoch": 0.24183673469387756, |
| "grad_norm": 0.11683396250009537, |
| "learning_rate": 0.00015480000000000002, |
| "loss": 0.21560895919799805, |
| "step": 11850 |
| }, |
| { |
| "epoch": 0.24285714285714285, |
| "grad_norm": 0.11114772409200668, |
| "learning_rate": 0.00015459166666666668, |
| "loss": 0.20422132492065428, |
| "step": 11900 |
| }, |
| { |
| "epoch": 0.24387755102040817, |
| "grad_norm": 0.17009824514389038, |
| "learning_rate": 0.00015438333333333334, |
| "loss": 0.23441919326782226, |
| "step": 11950 |
| }, |
| { |
| "epoch": 0.24489795918367346, |
| "grad_norm": 0.10565453767776489, |
| "learning_rate": 0.00015417916666666666, |
| "loss": 0.214456729888916, |
| "step": 12000 |
| }, |
| { |
| "epoch": 0.24591836734693878, |
| "grad_norm": 0.21939121186733246, |
| "learning_rate": 0.00015397083333333332, |
| "loss": 0.2204405975341797, |
| "step": 12050 |
| }, |
| { |
| "epoch": 0.24693877551020407, |
| "grad_norm": 0.1426985263824463, |
| "learning_rate": 0.0001537625, |
| "loss": 0.2276213264465332, |
| "step": 12100 |
| }, |
| { |
| "epoch": 0.2479591836734694, |
| "grad_norm": 0.16745182871818542, |
| "learning_rate": 0.00015355416666666666, |
| "loss": 0.21220468521118163, |
| "step": 12150 |
| }, |
| { |
| "epoch": 0.24897959183673468, |
| "grad_norm": 0.31632089614868164, |
| "learning_rate": 0.00015334583333333335, |
| "loss": 0.1934981346130371, |
| "step": 12200 |
| }, |
| { |
| "epoch": 0.25, |
| "grad_norm": 0.16067934036254883, |
| "learning_rate": 0.0001531375, |
| "loss": 0.20258354187011718, |
| "step": 12250 |
| }, |
| { |
| "epoch": 0.2510204081632653, |
| "grad_norm": 0.17517229914665222, |
| "learning_rate": 0.00015292916666666669, |
| "loss": 0.2431671905517578, |
| "step": 12300 |
| }, |
| { |
| "epoch": 0.25204081632653064, |
| "grad_norm": 0.13739600777626038, |
| "learning_rate": 0.00015272083333333334, |
| "loss": 0.2115278434753418, |
| "step": 12350 |
| }, |
| { |
| "epoch": 0.2530612244897959, |
| "grad_norm": 0.13237659633159637, |
| "learning_rate": 0.0001525125, |
| "loss": 0.14292340278625487, |
| "step": 12400 |
| }, |
| { |
| "epoch": 0.2540816326530612, |
| "grad_norm": 0.19046801328659058, |
| "learning_rate": 0.00015230416666666666, |
| "loss": 0.21214740753173827, |
| "step": 12450 |
| }, |
| { |
| "epoch": 0.25510204081632654, |
| "grad_norm": 0.1318289339542389, |
| "learning_rate": 0.00015209583333333334, |
| "loss": 0.21549345016479493, |
| "step": 12500 |
| }, |
| { |
| "epoch": 0.25612244897959185, |
| "grad_norm": 0.18249279260635376, |
| "learning_rate": 0.0001518875, |
| "loss": 0.2081346893310547, |
| "step": 12550 |
| }, |
| { |
| "epoch": 0.2571428571428571, |
| "grad_norm": 0.14595630764961243, |
| "learning_rate": 0.00015167916666666668, |
| "loss": 0.22366453170776368, |
| "step": 12600 |
| }, |
| { |
| "epoch": 0.25816326530612244, |
| "grad_norm": 0.1734742671251297, |
| "learning_rate": 0.000151475, |
| "loss": 0.2523087120056152, |
| "step": 12650 |
| }, |
| { |
| "epoch": 0.25918367346938775, |
| "grad_norm": 0.17593304812908173, |
| "learning_rate": 0.00015126666666666667, |
| "loss": 0.20111059188842773, |
| "step": 12700 |
| }, |
| { |
| "epoch": 0.2602040816326531, |
| "grad_norm": 0.1608940213918686, |
| "learning_rate": 0.00015105833333333335, |
| "loss": 0.20536836624145507, |
| "step": 12750 |
| }, |
| { |
| "epoch": 0.2612244897959184, |
| "grad_norm": 0.10997316241264343, |
| "learning_rate": 0.00015085, |
| "loss": 0.2102074432373047, |
| "step": 12800 |
| }, |
| { |
| "epoch": 0.26224489795918365, |
| "grad_norm": 0.18243081867694855, |
| "learning_rate": 0.0001506416666666667, |
| "loss": 0.21278114318847657, |
| "step": 12850 |
| }, |
| { |
| "epoch": 0.26326530612244897, |
| "grad_norm": 0.16395965218544006, |
| "learning_rate": 0.00015043333333333335, |
| "loss": 0.21783830642700194, |
| "step": 12900 |
| }, |
| { |
| "epoch": 0.2642857142857143, |
| "grad_norm": 0.1172817125916481, |
| "learning_rate": 0.000150225, |
| "loss": 0.20723285675048828, |
| "step": 12950 |
| }, |
| { |
| "epoch": 0.2653061224489796, |
| "grad_norm": 0.18447045981884003, |
| "learning_rate": 0.00015001666666666666, |
| "loss": 0.2183824348449707, |
| "step": 13000 |
| }, |
| { |
| "epoch": 0.26632653061224487, |
| "grad_norm": 0.1754911243915558, |
| "learning_rate": 0.00014980833333333335, |
| "loss": 0.21639692306518554, |
| "step": 13050 |
| }, |
| { |
| "epoch": 0.2673469387755102, |
| "grad_norm": 0.20869971811771393, |
| "learning_rate": 0.0001496, |
| "loss": 0.22688184738159178, |
| "step": 13100 |
| }, |
| { |
| "epoch": 0.2683673469387755, |
| "grad_norm": 0.2710501253604889, |
| "learning_rate": 0.00014939166666666666, |
| "loss": 0.20403726577758788, |
| "step": 13150 |
| }, |
| { |
| "epoch": 0.2693877551020408, |
| "grad_norm": 0.20377513766288757, |
| "learning_rate": 0.00014918333333333335, |
| "loss": 0.21639616012573243, |
| "step": 13200 |
| }, |
| { |
| "epoch": 0.27040816326530615, |
| "grad_norm": 0.20667989552021027, |
| "learning_rate": 0.000148975, |
| "loss": 0.20512529373168945, |
| "step": 13250 |
| }, |
| { |
| "epoch": 0.2714285714285714, |
| "grad_norm": 0.31901517510414124, |
| "learning_rate": 0.0001487666666666667, |
| "loss": 0.20101198196411132, |
| "step": 13300 |
| }, |
| { |
| "epoch": 0.2724489795918367, |
| "grad_norm": 0.3628154993057251, |
| "learning_rate": 0.00014855833333333334, |
| "loss": 0.1903107452392578, |
| "step": 13350 |
| }, |
| { |
| "epoch": 0.27346938775510204, |
| "grad_norm": 0.20295929908752441, |
| "learning_rate": 0.00014835, |
| "loss": 0.22166210174560547, |
| "step": 13400 |
| }, |
| { |
| "epoch": 0.27448979591836736, |
| "grad_norm": 0.19172048568725586, |
| "learning_rate": 0.00014814166666666666, |
| "loss": 0.20491140365600585, |
| "step": 13450 |
| }, |
| { |
| "epoch": 0.2755102040816326, |
| "grad_norm": 0.22426478564739227, |
| "learning_rate": 0.00014793333333333334, |
| "loss": 0.2474876594543457, |
| "step": 13500 |
| }, |
| { |
| "epoch": 0.27653061224489794, |
| "grad_norm": 0.15273892879486084, |
| "learning_rate": 0.000147725, |
| "loss": 0.21244457244873047, |
| "step": 13550 |
| }, |
| { |
| "epoch": 0.27755102040816326, |
| "grad_norm": 0.12474813312292099, |
| "learning_rate": 0.00014751666666666668, |
| "loss": 0.1948986053466797, |
| "step": 13600 |
| }, |
| { |
| "epoch": 0.2785714285714286, |
| "grad_norm": 0.1677834391593933, |
| "learning_rate": 0.00014730833333333334, |
| "loss": 0.21693151473999023, |
| "step": 13650 |
| }, |
| { |
| "epoch": 0.2795918367346939, |
| "grad_norm": 0.1882069855928421, |
| "learning_rate": 0.00014710000000000002, |
| "loss": 0.22328054428100585, |
| "step": 13700 |
| }, |
| { |
| "epoch": 0.28061224489795916, |
| "grad_norm": 0.21768781542778015, |
| "learning_rate": 0.00014689166666666668, |
| "loss": 0.24165786743164064, |
| "step": 13750 |
| }, |
| { |
| "epoch": 0.2816326530612245, |
| "grad_norm": 0.1738264113664627, |
| "learning_rate": 0.00014668333333333334, |
| "loss": 0.22251943588256837, |
| "step": 13800 |
| }, |
| { |
| "epoch": 0.2826530612244898, |
| "grad_norm": 0.15541969239711761, |
| "learning_rate": 0.000146475, |
| "loss": 0.21156435012817382, |
| "step": 13850 |
| }, |
| { |
| "epoch": 0.2836734693877551, |
| "grad_norm": 0.26248255372047424, |
| "learning_rate": 0.00014626666666666665, |
| "loss": 0.21172939300537108, |
| "step": 13900 |
| }, |
| { |
| "epoch": 0.28469387755102044, |
| "grad_norm": 0.1985601931810379, |
| "learning_rate": 0.00014605833333333334, |
| "loss": 0.2555007553100586, |
| "step": 13950 |
| }, |
| { |
| "epoch": 0.2857142857142857, |
| "grad_norm": 0.16674289107322693, |
| "learning_rate": 0.00014585, |
| "loss": 0.22686000823974609, |
| "step": 14000 |
| }, |
| { |
| "epoch": 0.286734693877551, |
| "grad_norm": 0.08805567771196365, |
| "learning_rate": 0.00014564166666666668, |
| "loss": 0.1840823173522949, |
| "step": 14050 |
| }, |
| { |
| "epoch": 0.28775510204081634, |
| "grad_norm": 0.14864468574523926, |
| "learning_rate": 0.00014543333333333334, |
| "loss": 0.20063446044921876, |
| "step": 14100 |
| }, |
| { |
| "epoch": 0.28877551020408165, |
| "grad_norm": 0.16803936660289764, |
| "learning_rate": 0.00014522500000000002, |
| "loss": 0.17357282638549804, |
| "step": 14150 |
| }, |
| { |
| "epoch": 0.2897959183673469, |
| "grad_norm": 0.17610159516334534, |
| "learning_rate": 0.00014501666666666668, |
| "loss": 0.2013835906982422, |
| "step": 14200 |
| }, |
| { |
| "epoch": 0.29081632653061223, |
| "grad_norm": 0.13486459851264954, |
| "learning_rate": 0.00014480833333333333, |
| "loss": 0.19870820999145508, |
| "step": 14250 |
| }, |
| { |
| "epoch": 0.29183673469387755, |
| "grad_norm": 0.17316874861717224, |
| "learning_rate": 0.0001446, |
| "loss": 0.2167327308654785, |
| "step": 14300 |
| }, |
| { |
| "epoch": 0.29285714285714287, |
| "grad_norm": 0.1329314410686493, |
| "learning_rate": 0.00014439166666666668, |
| "loss": 0.20319538116455077, |
| "step": 14350 |
| }, |
| { |
| "epoch": 0.2938775510204082, |
| "grad_norm": 0.12000191956758499, |
| "learning_rate": 0.00014418333333333333, |
| "loss": 0.12253636360168457, |
| "step": 14400 |
| }, |
| { |
| "epoch": 0.29489795918367345, |
| "grad_norm": 0.17667782306671143, |
| "learning_rate": 0.00014397500000000002, |
| "loss": 0.18405498504638673, |
| "step": 14450 |
| }, |
| { |
| "epoch": 0.29591836734693877, |
| "grad_norm": 0.18920040130615234, |
| "learning_rate": 0.00014376666666666667, |
| "loss": 0.21451929092407226, |
| "step": 14500 |
| }, |
| { |
| "epoch": 0.2969387755102041, |
| "grad_norm": 0.21450640261173248, |
| "learning_rate": 0.00014355833333333336, |
| "loss": 0.21518003463745117, |
| "step": 14550 |
| }, |
| { |
| "epoch": 0.2979591836734694, |
| "grad_norm": 0.1683834046125412, |
| "learning_rate": 0.00014335000000000001, |
| "loss": 0.20549680709838866, |
| "step": 14600 |
| }, |
| { |
| "epoch": 0.29897959183673467, |
| "grad_norm": 0.14073589444160461, |
| "learning_rate": 0.00014314166666666667, |
| "loss": 0.17846385955810548, |
| "step": 14650 |
| }, |
| { |
| "epoch": 0.3, |
| "grad_norm": 0.12049887329339981, |
| "learning_rate": 0.00014293333333333333, |
| "loss": 0.14772893905639647, |
| "step": 14700 |
| }, |
| { |
| "epoch": 0.3010204081632653, |
| "grad_norm": 0.1303739845752716, |
| "learning_rate": 0.00014272499999999999, |
| "loss": 0.12617334365844726, |
| "step": 14750 |
| }, |
| { |
| "epoch": 0.3020408163265306, |
| "grad_norm": 0.13680632412433624, |
| "learning_rate": 0.00014251666666666667, |
| "loss": 0.21211812973022462, |
| "step": 14800 |
| }, |
| { |
| "epoch": 0.30306122448979594, |
| "grad_norm": 0.19090735912322998, |
| "learning_rate": 0.00014230833333333333, |
| "loss": 0.1917464828491211, |
| "step": 14850 |
| }, |
| { |
| "epoch": 0.3040816326530612, |
| "grad_norm": 0.15105141699314117, |
| "learning_rate": 0.0001421, |
| "loss": 0.1993863868713379, |
| "step": 14900 |
| }, |
| { |
| "epoch": 0.3051020408163265, |
| "grad_norm": 0.177749902009964, |
| "learning_rate": 0.00014189166666666667, |
| "loss": 0.20117059707641602, |
| "step": 14950 |
| }, |
| { |
| "epoch": 0.30612244897959184, |
| "grad_norm": 0.20181018114089966, |
| "learning_rate": 0.00014168333333333335, |
| "loss": 0.18704294204711913, |
| "step": 15000 |
| }, |
| { |
| "epoch": 0.30714285714285716, |
| "grad_norm": 0.19316235184669495, |
| "learning_rate": 0.000141475, |
| "loss": 0.2391501235961914, |
| "step": 15050 |
| }, |
| { |
| "epoch": 0.3081632653061224, |
| "grad_norm": 0.24478937685489655, |
| "learning_rate": 0.00014126666666666667, |
| "loss": 0.2043419075012207, |
| "step": 15100 |
| }, |
| { |
| "epoch": 0.30918367346938774, |
| "grad_norm": 0.21967218816280365, |
| "learning_rate": 0.00014105833333333332, |
| "loss": 0.22933086395263672, |
| "step": 15150 |
| }, |
| { |
| "epoch": 0.31020408163265306, |
| "grad_norm": 0.16746099293231964, |
| "learning_rate": 0.00014085416666666668, |
| "loss": 0.21231243133544922, |
| "step": 15200 |
| }, |
| { |
| "epoch": 0.3112244897959184, |
| "grad_norm": 0.18917246162891388, |
| "learning_rate": 0.00014064583333333333, |
| "loss": 0.22859695434570312, |
| "step": 15250 |
| }, |
| { |
| "epoch": 0.3122448979591837, |
| "grad_norm": 0.2702839970588684, |
| "learning_rate": 0.0001404375, |
| "loss": 0.19720211029052734, |
| "step": 15300 |
| }, |
| { |
| "epoch": 0.31326530612244896, |
| "grad_norm": 0.1158054769039154, |
| "learning_rate": 0.00014022916666666668, |
| "loss": 0.19617162704467772, |
| "step": 15350 |
| }, |
| { |
| "epoch": 0.3142857142857143, |
| "grad_norm": 0.22025619447231293, |
| "learning_rate": 0.00014002083333333333, |
| "loss": 0.20038124084472655, |
| "step": 15400 |
| }, |
| { |
| "epoch": 0.3153061224489796, |
| "grad_norm": 0.16724291443824768, |
| "learning_rate": 0.00013981250000000002, |
| "loss": 0.20154199600219727, |
| "step": 15450 |
| }, |
| { |
| "epoch": 0.3163265306122449, |
| "grad_norm": 0.2141387015581131, |
| "learning_rate": 0.00013960416666666667, |
| "loss": 0.2146558380126953, |
| "step": 15500 |
| }, |
| { |
| "epoch": 0.3173469387755102, |
| "grad_norm": 0.13922351598739624, |
| "learning_rate": 0.00013939583333333336, |
| "loss": 0.15184280395507813, |
| "step": 15550 |
| }, |
| { |
| "epoch": 0.3183673469387755, |
| "grad_norm": 0.11073990911245346, |
| "learning_rate": 0.00013918750000000002, |
| "loss": 0.11846443176269532, |
| "step": 15600 |
| }, |
| { |
| "epoch": 0.3193877551020408, |
| "grad_norm": 0.2762114405632019, |
| "learning_rate": 0.00013897916666666667, |
| "loss": 0.1905454635620117, |
| "step": 15650 |
| }, |
| { |
| "epoch": 0.32040816326530613, |
| "grad_norm": 0.22723205387592316, |
| "learning_rate": 0.00013877083333333333, |
| "loss": 0.19393016815185546, |
| "step": 15700 |
| }, |
| { |
| "epoch": 0.32142857142857145, |
| "grad_norm": 0.24777233600616455, |
| "learning_rate": 0.00013856250000000001, |
| "loss": 0.20483108520507812, |
| "step": 15750 |
| }, |
| { |
| "epoch": 0.3224489795918367, |
| "grad_norm": 0.17726479470729828, |
| "learning_rate": 0.00013835416666666667, |
| "loss": 0.22141836166381837, |
| "step": 15800 |
| }, |
| { |
| "epoch": 0.32346938775510203, |
| "grad_norm": 0.20626652240753174, |
| "learning_rate": 0.00013814583333333333, |
| "loss": 0.22326576232910156, |
| "step": 15850 |
| }, |
| { |
| "epoch": 0.32448979591836735, |
| "grad_norm": 0.15777790546417236, |
| "learning_rate": 0.0001379375, |
| "loss": 0.2150650978088379, |
| "step": 15900 |
| }, |
| { |
| "epoch": 0.32551020408163267, |
| "grad_norm": 0.15837018191814423, |
| "learning_rate": 0.00013772916666666667, |
| "loss": 0.13302292823791503, |
| "step": 15950 |
| }, |
| { |
| "epoch": 0.32653061224489793, |
| "grad_norm": 0.12448123842477798, |
| "learning_rate": 0.00013752083333333335, |
| "loss": 0.1244437599182129, |
| "step": 16000 |
| }, |
| { |
| "epoch": 0.32755102040816325, |
| "grad_norm": 0.21877269446849823, |
| "learning_rate": 0.0001373125, |
| "loss": 0.19579626083374024, |
| "step": 16050 |
| }, |
| { |
| "epoch": 0.32857142857142857, |
| "grad_norm": 0.14148931205272675, |
| "learning_rate": 0.00013710416666666667, |
| "loss": 0.20735546112060546, |
| "step": 16100 |
| }, |
| { |
| "epoch": 0.3295918367346939, |
| "grad_norm": 0.17035670578479767, |
| "learning_rate": 0.00013689583333333332, |
| "loss": 0.1896759605407715, |
| "step": 16150 |
| }, |
| { |
| "epoch": 0.3306122448979592, |
| "grad_norm": 0.12121950089931488, |
| "learning_rate": 0.0001366875, |
| "loss": 0.1927642822265625, |
| "step": 16200 |
| }, |
| { |
| "epoch": 0.33163265306122447, |
| "grad_norm": 0.25413230061531067, |
| "learning_rate": 0.00013647916666666667, |
| "loss": 0.19843025207519532, |
| "step": 16250 |
| }, |
| { |
| "epoch": 0.3326530612244898, |
| "grad_norm": 0.08759265393018723, |
| "learning_rate": 0.00013627083333333335, |
| "loss": 0.18141515731811522, |
| "step": 16300 |
| }, |
| { |
| "epoch": 0.3336734693877551, |
| "grad_norm": 0.18287113308906555, |
| "learning_rate": 0.0001360625, |
| "loss": 0.22243984222412108, |
| "step": 16350 |
| }, |
| { |
| "epoch": 0.3346938775510204, |
| "grad_norm": 0.2288346141576767, |
| "learning_rate": 0.0001358541666666667, |
| "loss": 0.2296261215209961, |
| "step": 16400 |
| }, |
| { |
| "epoch": 0.3357142857142857, |
| "grad_norm": 0.19713585078716278, |
| "learning_rate": 0.00013564583333333335, |
| "loss": 0.2095242691040039, |
| "step": 16450 |
| }, |
| { |
| "epoch": 0.336734693877551, |
| "grad_norm": 0.08195829391479492, |
| "learning_rate": 0.0001354375, |
| "loss": 0.19855600357055664, |
| "step": 16500 |
| }, |
| { |
| "epoch": 0.3377551020408163, |
| "grad_norm": 0.14508482813835144, |
| "learning_rate": 0.00013522916666666666, |
| "loss": 0.19972381591796876, |
| "step": 16550 |
| }, |
| { |
| "epoch": 0.33877551020408164, |
| "grad_norm": 0.10825417190790176, |
| "learning_rate": 0.00013502083333333332, |
| "loss": 0.17222192764282226, |
| "step": 16600 |
| }, |
| { |
| "epoch": 0.33979591836734696, |
| "grad_norm": 0.13268212974071503, |
| "learning_rate": 0.0001348125, |
| "loss": 0.12606014251708986, |
| "step": 16650 |
| }, |
| { |
| "epoch": 0.3408163265306122, |
| "grad_norm": 0.1764751672744751, |
| "learning_rate": 0.00013460416666666666, |
| "loss": 0.183403377532959, |
| "step": 16700 |
| }, |
| { |
| "epoch": 0.34183673469387754, |
| "grad_norm": 0.21741646528244019, |
| "learning_rate": 0.00013439583333333335, |
| "loss": 0.19946996688842775, |
| "step": 16750 |
| }, |
| { |
| "epoch": 0.34285714285714286, |
| "grad_norm": 0.16558875143527985, |
| "learning_rate": 0.0001341875, |
| "loss": 0.21058109283447266, |
| "step": 16800 |
| }, |
| { |
| "epoch": 0.3438775510204082, |
| "grad_norm": 0.27157577872276306, |
| "learning_rate": 0.0001339791666666667, |
| "loss": 0.19955120086669922, |
| "step": 16850 |
| }, |
| { |
| "epoch": 0.3448979591836735, |
| "grad_norm": 0.2037670761346817, |
| "learning_rate": 0.00013377083333333334, |
| "loss": 0.1966884994506836, |
| "step": 16900 |
| }, |
| { |
| "epoch": 0.34591836734693876, |
| "grad_norm": 0.23232632875442505, |
| "learning_rate": 0.0001335625, |
| "loss": 0.20686569213867187, |
| "step": 16950 |
| }, |
| { |
| "epoch": 0.3469387755102041, |
| "grad_norm": 0.1624077707529068, |
| "learning_rate": 0.00013335416666666666, |
| "loss": 0.20075689315795897, |
| "step": 17000 |
| }, |
| { |
| "epoch": 0.3479591836734694, |
| "grad_norm": 0.20937331020832062, |
| "learning_rate": 0.00013314583333333334, |
| "loss": 0.18493799209594727, |
| "step": 17050 |
| }, |
| { |
| "epoch": 0.3489795918367347, |
| "grad_norm": 0.11835204064846039, |
| "learning_rate": 0.0001329375, |
| "loss": 0.19891904830932616, |
| "step": 17100 |
| }, |
| { |
| "epoch": 0.35, |
| "grad_norm": 0.13524866104125977, |
| "learning_rate": 0.00013272916666666668, |
| "loss": 0.20436981201171875, |
| "step": 17150 |
| }, |
| { |
| "epoch": 0.3510204081632653, |
| "grad_norm": 0.15755361318588257, |
| "learning_rate": 0.00013252083333333334, |
| "loss": 0.21048551559448242, |
| "step": 17200 |
| }, |
| { |
| "epoch": 0.3520408163265306, |
| "grad_norm": 0.22293004393577576, |
| "learning_rate": 0.00013231250000000002, |
| "loss": 0.20778837203979492, |
| "step": 17250 |
| }, |
| { |
| "epoch": 0.35306122448979593, |
| "grad_norm": 0.19373543560504913, |
| "learning_rate": 0.00013210416666666668, |
| "loss": 0.20396459579467774, |
| "step": 17300 |
| }, |
| { |
| "epoch": 0.35408163265306125, |
| "grad_norm": 0.3679802119731903, |
| "learning_rate": 0.00013189583333333334, |
| "loss": 0.19777801513671875, |
| "step": 17350 |
| }, |
| { |
| "epoch": 0.3551020408163265, |
| "grad_norm": 0.17816053330898285, |
| "learning_rate": 0.0001316875, |
| "loss": 0.176708927154541, |
| "step": 17400 |
| }, |
| { |
| "epoch": 0.35612244897959183, |
| "grad_norm": 0.17044921219348907, |
| "learning_rate": 0.00013147916666666665, |
| "loss": 0.20279596328735353, |
| "step": 17450 |
| }, |
| { |
| "epoch": 0.35714285714285715, |
| "grad_norm": 0.12418626993894577, |
| "learning_rate": 0.00013127083333333334, |
| "loss": 0.14005967140197753, |
| "step": 17500 |
| }, |
| { |
| "epoch": 0.35816326530612247, |
| "grad_norm": 0.13893789052963257, |
| "learning_rate": 0.0001310625, |
| "loss": 0.12252273559570312, |
| "step": 17550 |
| }, |
| { |
| "epoch": 0.35918367346938773, |
| "grad_norm": 0.12449125945568085, |
| "learning_rate": 0.00013085416666666668, |
| "loss": 0.1294875717163086, |
| "step": 17600 |
| }, |
| { |
| "epoch": 0.36020408163265305, |
| "grad_norm": 0.11940774321556091, |
| "learning_rate": 0.00013064583333333334, |
| "loss": 0.21808555603027344, |
| "step": 17650 |
| }, |
| { |
| "epoch": 0.36122448979591837, |
| "grad_norm": 0.16444045305252075, |
| "learning_rate": 0.00013043750000000002, |
| "loss": 0.18556377410888672, |
| "step": 17700 |
| }, |
| { |
| "epoch": 0.3622448979591837, |
| "grad_norm": 0.17309384047985077, |
| "learning_rate": 0.00013022916666666668, |
| "loss": 0.20985712051391603, |
| "step": 17750 |
| }, |
| { |
| "epoch": 0.363265306122449, |
| "grad_norm": 0.19860877096652985, |
| "learning_rate": 0.00013002083333333333, |
| "loss": 0.19437423706054688, |
| "step": 17800 |
| }, |
| { |
| "epoch": 0.36428571428571427, |
| "grad_norm": 0.18023747205734253, |
| "learning_rate": 0.0001298125, |
| "loss": 0.1935722541809082, |
| "step": 17850 |
| }, |
| { |
| "epoch": 0.3653061224489796, |
| "grad_norm": 0.18868409097194672, |
| "learning_rate": 0.00012960416666666668, |
| "loss": 0.2154340934753418, |
| "step": 17900 |
| }, |
| { |
| "epoch": 0.3663265306122449, |
| "grad_norm": 0.19930492341518402, |
| "learning_rate": 0.00012939583333333333, |
| "loss": 0.19957744598388671, |
| "step": 17950 |
| }, |
| { |
| "epoch": 0.3673469387755102, |
| "grad_norm": 0.18167056143283844, |
| "learning_rate": 0.00012918750000000002, |
| "loss": 0.15686763763427736, |
| "step": 18000 |
| }, |
| { |
| "epoch": 0.3683673469387755, |
| "grad_norm": 0.2629138231277466, |
| "learning_rate": 0.00012897916666666667, |
| "loss": 0.22448341369628907, |
| "step": 18050 |
| }, |
| { |
| "epoch": 0.3693877551020408, |
| "grad_norm": 0.20073753595352173, |
| "learning_rate": 0.00012877083333333333, |
| "loss": 0.2116405487060547, |
| "step": 18100 |
| }, |
| { |
| "epoch": 0.3704081632653061, |
| "grad_norm": 0.2358439415693283, |
| "learning_rate": 0.00012856250000000001, |
| "loss": 0.2034568214416504, |
| "step": 18150 |
| }, |
| { |
| "epoch": 0.37142857142857144, |
| "grad_norm": 0.1448625773191452, |
| "learning_rate": 0.00012835416666666667, |
| "loss": 0.20943548202514647, |
| "step": 18200 |
| }, |
| { |
| "epoch": 0.37244897959183676, |
| "grad_norm": 0.39532026648521423, |
| "learning_rate": 0.00012814583333333336, |
| "loss": 0.2011641502380371, |
| "step": 18250 |
| }, |
| { |
| "epoch": 0.373469387755102, |
| "grad_norm": 0.12298086285591125, |
| "learning_rate": 0.00012793749999999999, |
| "loss": 0.1635613250732422, |
| "step": 18300 |
| }, |
| { |
| "epoch": 0.37448979591836734, |
| "grad_norm": 0.16486021876335144, |
| "learning_rate": 0.00012772916666666667, |
| "loss": 0.19621923446655273, |
| "step": 18350 |
| }, |
| { |
| "epoch": 0.37551020408163266, |
| "grad_norm": 0.20563790202140808, |
| "learning_rate": 0.00012752083333333333, |
| "loss": 0.25450227737426756, |
| "step": 18400 |
| }, |
| { |
| "epoch": 0.376530612244898, |
| "grad_norm": 0.14039361476898193, |
| "learning_rate": 0.0001273125, |
| "loss": 0.20071006774902345, |
| "step": 18450 |
| }, |
| { |
| "epoch": 0.37755102040816324, |
| "grad_norm": 0.12278524041175842, |
| "learning_rate": 0.00012710416666666667, |
| "loss": 0.1823217010498047, |
| "step": 18500 |
| }, |
| { |
| "epoch": 0.37857142857142856, |
| "grad_norm": 0.17723079025745392, |
| "learning_rate": 0.00012689583333333335, |
| "loss": 0.18245498657226564, |
| "step": 18550 |
| }, |
| { |
| "epoch": 0.3795918367346939, |
| "grad_norm": 0.14125439524650574, |
| "learning_rate": 0.0001266875, |
| "loss": 0.22428083419799805, |
| "step": 18600 |
| }, |
| { |
| "epoch": 0.3806122448979592, |
| "grad_norm": 0.13816727697849274, |
| "learning_rate": 0.0001264791666666667, |
| "loss": 0.1764548873901367, |
| "step": 18650 |
| }, |
| { |
| "epoch": 0.3816326530612245, |
| "grad_norm": 0.2603978216648102, |
| "learning_rate": 0.00012627083333333335, |
| "loss": 0.20926473617553712, |
| "step": 18700 |
| }, |
| { |
| "epoch": 0.3826530612244898, |
| "grad_norm": 0.2258712649345398, |
| "learning_rate": 0.0001260625, |
| "loss": 0.19603876113891602, |
| "step": 18750 |
| }, |
| { |
| "epoch": 0.3836734693877551, |
| "grad_norm": 0.27304700016975403, |
| "learning_rate": 0.00012585416666666667, |
| "loss": 0.22457117080688477, |
| "step": 18800 |
| }, |
| { |
| "epoch": 0.3846938775510204, |
| "grad_norm": 0.21007055044174194, |
| "learning_rate": 0.00012564583333333332, |
| "loss": 0.1995391082763672, |
| "step": 18850 |
| }, |
| { |
| "epoch": 0.38571428571428573, |
| "grad_norm": 0.18496140837669373, |
| "learning_rate": 0.0001254375, |
| "loss": 0.18473093032836915, |
| "step": 18900 |
| }, |
| { |
| "epoch": 0.386734693877551, |
| "grad_norm": 0.13952362537384033, |
| "learning_rate": 0.00012522916666666666, |
| "loss": 0.22313592910766603, |
| "step": 18950 |
| }, |
| { |
| "epoch": 0.3877551020408163, |
| "grad_norm": 0.31036996841430664, |
| "learning_rate": 0.00012502083333333335, |
| "loss": 0.20665483474731444, |
| "step": 19000 |
| }, |
| { |
| "epoch": 0.38877551020408163, |
| "grad_norm": 0.19020035862922668, |
| "learning_rate": 0.0001248125, |
| "loss": 0.22109838485717773, |
| "step": 19050 |
| }, |
| { |
| "epoch": 0.38979591836734695, |
| "grad_norm": 0.22469305992126465, |
| "learning_rate": 0.0001246041666666667, |
| "loss": 0.21268951416015625, |
| "step": 19100 |
| }, |
| { |
| "epoch": 0.39081632653061227, |
| "grad_norm": 0.17466503381729126, |
| "learning_rate": 0.00012439583333333335, |
| "loss": 0.20178163528442383, |
| "step": 19150 |
| }, |
| { |
| "epoch": 0.39183673469387753, |
| "grad_norm": 0.14300084114074707, |
| "learning_rate": 0.0001241875, |
| "loss": 0.24622438430786134, |
| "step": 19200 |
| }, |
| { |
| "epoch": 0.39285714285714285, |
| "grad_norm": 0.1847544014453888, |
| "learning_rate": 0.00012397916666666666, |
| "loss": 0.2195741081237793, |
| "step": 19250 |
| }, |
| { |
| "epoch": 0.39387755102040817, |
| "grad_norm": 0.09982876479625702, |
| "learning_rate": 0.00012377083333333334, |
| "loss": 0.204034481048584, |
| "step": 19300 |
| }, |
| { |
| "epoch": 0.3948979591836735, |
| "grad_norm": 0.18855072557926178, |
| "learning_rate": 0.0001235625, |
| "loss": 0.18522701263427735, |
| "step": 19350 |
| }, |
| { |
| "epoch": 0.39591836734693875, |
| "grad_norm": 0.13189591467380524, |
| "learning_rate": 0.00012335416666666669, |
| "loss": 0.18743988037109374, |
| "step": 19400 |
| }, |
| { |
| "epoch": 0.39693877551020407, |
| "grad_norm": 0.09443347156047821, |
| "learning_rate": 0.00012314583333333334, |
| "loss": 0.18457218170166015, |
| "step": 19450 |
| }, |
| { |
| "epoch": 0.3979591836734694, |
| "grad_norm": 0.13609136641025543, |
| "learning_rate": 0.00012293750000000003, |
| "loss": 0.1992996597290039, |
| "step": 19500 |
| }, |
| { |
| "epoch": 0.3989795918367347, |
| "grad_norm": 0.1060328483581543, |
| "learning_rate": 0.00012272916666666668, |
| "loss": 0.18354759216308594, |
| "step": 19550 |
| }, |
| { |
| "epoch": 0.4, |
| "grad_norm": 0.1333695501089096, |
| "learning_rate": 0.00012252083333333334, |
| "loss": 0.19399749755859375, |
| "step": 19600 |
| }, |
| { |
| "epoch": 0.4010204081632653, |
| "grad_norm": 0.1901199072599411, |
| "learning_rate": 0.0001223125, |
| "loss": 0.2290254592895508, |
| "step": 19650 |
| }, |
| { |
| "epoch": 0.4020408163265306, |
| "grad_norm": 0.2177937775850296, |
| "learning_rate": 0.00012210416666666666, |
| "loss": 0.24125240325927735, |
| "step": 19700 |
| }, |
| { |
| "epoch": 0.4030612244897959, |
| "grad_norm": 0.12752966582775116, |
| "learning_rate": 0.00012189583333333334, |
| "loss": 0.22714794158935547, |
| "step": 19750 |
| }, |
| { |
| "epoch": 0.40408163265306124, |
| "grad_norm": 0.18761104345321655, |
| "learning_rate": 0.0001216875, |
| "loss": 0.18184953689575195, |
| "step": 19800 |
| }, |
| { |
| "epoch": 0.4051020408163265, |
| "grad_norm": 0.127014622092247, |
| "learning_rate": 0.00012147916666666668, |
| "loss": 0.19482044219970704, |
| "step": 19850 |
| }, |
| { |
| "epoch": 0.4061224489795918, |
| "grad_norm": 0.10520447790622711, |
| "learning_rate": 0.00012127083333333334, |
| "loss": 0.14860014915466307, |
| "step": 19900 |
| }, |
| { |
| "epoch": 0.40714285714285714, |
| "grad_norm": 0.12470293045043945, |
| "learning_rate": 0.00012106250000000001, |
| "loss": 0.16204519271850587, |
| "step": 19950 |
| }, |
| { |
| "epoch": 0.40816326530612246, |
| "grad_norm": 0.12424734234809875, |
| "learning_rate": 0.00012085416666666667, |
| "loss": 0.16527006149291992, |
| "step": 20000 |
| }, |
| { |
| "epoch": 0.4091836734693878, |
| "grad_norm": 0.14997172355651855, |
| "learning_rate": 0.00012064583333333335, |
| "loss": 0.18037498474121094, |
| "step": 20050 |
| }, |
| { |
| "epoch": 0.41020408163265304, |
| "grad_norm": 0.17725597321987152, |
| "learning_rate": 0.00012043750000000001, |
| "loss": 0.2216799736022949, |
| "step": 20100 |
| }, |
| { |
| "epoch": 0.41122448979591836, |
| "grad_norm": 0.1883675903081894, |
| "learning_rate": 0.00012022916666666668, |
| "loss": 0.18260782241821288, |
| "step": 20150 |
| }, |
| { |
| "epoch": 0.4122448979591837, |
| "grad_norm": 0.19166825711727142, |
| "learning_rate": 0.00012002083333333333, |
| "loss": 0.196572265625, |
| "step": 20200 |
| }, |
| { |
| "epoch": 0.413265306122449, |
| "grad_norm": 0.13306336104869843, |
| "learning_rate": 0.00011981250000000002, |
| "loss": 0.16494216918945312, |
| "step": 20250 |
| }, |
| { |
| "epoch": 0.4142857142857143, |
| "grad_norm": 0.22641228139400482, |
| "learning_rate": 0.00011960416666666668, |
| "loss": 0.20227954864501954, |
| "step": 20300 |
| }, |
| { |
| "epoch": 0.4153061224489796, |
| "grad_norm": 0.16950534284114838, |
| "learning_rate": 0.00011939583333333333, |
| "loss": 0.24698841094970703, |
| "step": 20350 |
| }, |
| { |
| "epoch": 0.4163265306122449, |
| "grad_norm": 0.1433454155921936, |
| "learning_rate": 0.0001191875, |
| "loss": 0.2045128631591797, |
| "step": 20400 |
| }, |
| { |
| "epoch": 0.4173469387755102, |
| "grad_norm": 0.1904461830854416, |
| "learning_rate": 0.00011897916666666666, |
| "loss": 0.1645878028869629, |
| "step": 20450 |
| }, |
| { |
| "epoch": 0.41836734693877553, |
| "grad_norm": 0.227298766374588, |
| "learning_rate": 0.00011877083333333334, |
| "loss": 0.16797447204589844, |
| "step": 20500 |
| }, |
| { |
| "epoch": 0.4193877551020408, |
| "grad_norm": 0.16995885968208313, |
| "learning_rate": 0.0001185625, |
| "loss": 0.193846435546875, |
| "step": 20550 |
| }, |
| { |
| "epoch": 0.4204081632653061, |
| "grad_norm": 0.14499473571777344, |
| "learning_rate": 0.00011835416666666667, |
| "loss": 0.18684532165527343, |
| "step": 20600 |
| }, |
| { |
| "epoch": 0.42142857142857143, |
| "grad_norm": 0.1715216338634491, |
| "learning_rate": 0.00011814583333333333, |
| "loss": 0.18035064697265624, |
| "step": 20650 |
| }, |
| { |
| "epoch": 0.42244897959183675, |
| "grad_norm": 0.14939017593860626, |
| "learning_rate": 0.00011793750000000001, |
| "loss": 0.1902223014831543, |
| "step": 20700 |
| }, |
| { |
| "epoch": 0.42346938775510207, |
| "grad_norm": 0.10494500398635864, |
| "learning_rate": 0.00011772916666666667, |
| "loss": 0.121597900390625, |
| "step": 20750 |
| }, |
| { |
| "epoch": 0.42448979591836733, |
| "grad_norm": 0.1562127023935318, |
| "learning_rate": 0.00011752083333333334, |
| "loss": 0.17706047058105467, |
| "step": 20800 |
| }, |
| { |
| "epoch": 0.42551020408163265, |
| "grad_norm": 0.14377187192440033, |
| "learning_rate": 0.0001173125, |
| "loss": 0.17805103302001954, |
| "step": 20850 |
| }, |
| { |
| "epoch": 0.42653061224489797, |
| "grad_norm": 0.12476065754890442, |
| "learning_rate": 0.00011710416666666668, |
| "loss": 0.18395469665527345, |
| "step": 20900 |
| }, |
| { |
| "epoch": 0.4275510204081633, |
| "grad_norm": 0.14680302143096924, |
| "learning_rate": 0.00011689583333333334, |
| "loss": 0.14980695724487306, |
| "step": 20950 |
| }, |
| { |
| "epoch": 0.42857142857142855, |
| "grad_norm": 0.44721558690071106, |
| "learning_rate": 0.00011668750000000001, |
| "loss": 0.18807683944702147, |
| "step": 21000 |
| }, |
| { |
| "epoch": 0.42959183673469387, |
| "grad_norm": 0.1630980372428894, |
| "learning_rate": 0.00011647916666666667, |
| "loss": 0.16178571701049804, |
| "step": 21050 |
| }, |
| { |
| "epoch": 0.4306122448979592, |
| "grad_norm": 0.18277651071548462, |
| "learning_rate": 0.00011627083333333332, |
| "loss": 0.19975339889526367, |
| "step": 21100 |
| }, |
| { |
| "epoch": 0.4316326530612245, |
| "grad_norm": 0.14588306844234467, |
| "learning_rate": 0.00011606250000000001, |
| "loss": 0.2004733657836914, |
| "step": 21150 |
| }, |
| { |
| "epoch": 0.4326530612244898, |
| "grad_norm": 0.30171841382980347, |
| "learning_rate": 0.00011585416666666667, |
| "loss": 0.21247049331665038, |
| "step": 21200 |
| }, |
| { |
| "epoch": 0.4336734693877551, |
| "grad_norm": 0.21481254696846008, |
| "learning_rate": 0.00011564583333333334, |
| "loss": 0.163582706451416, |
| "step": 21250 |
| }, |
| { |
| "epoch": 0.4346938775510204, |
| "grad_norm": 0.21892467141151428, |
| "learning_rate": 0.0001154375, |
| "loss": 0.17682687759399415, |
| "step": 21300 |
| }, |
| { |
| "epoch": 0.4357142857142857, |
| "grad_norm": 0.3684554100036621, |
| "learning_rate": 0.00011522916666666668, |
| "loss": 0.21702127456665038, |
| "step": 21350 |
| }, |
| { |
| "epoch": 0.43673469387755104, |
| "grad_norm": 0.1822352409362793, |
| "learning_rate": 0.00011502083333333334, |
| "loss": 0.20249832153320313, |
| "step": 21400 |
| }, |
| { |
| "epoch": 0.4377551020408163, |
| "grad_norm": 0.14486338198184967, |
| "learning_rate": 0.0001148125, |
| "loss": 0.16668758392333985, |
| "step": 21450 |
| }, |
| { |
| "epoch": 0.4387755102040816, |
| "grad_norm": 0.12427511811256409, |
| "learning_rate": 0.00011460416666666666, |
| "loss": 0.22259849548339844, |
| "step": 21500 |
| }, |
| { |
| "epoch": 0.43979591836734694, |
| "grad_norm": 0.12884387373924255, |
| "learning_rate": 0.00011439583333333335, |
| "loss": 0.19325353622436522, |
| "step": 21550 |
| }, |
| { |
| "epoch": 0.44081632653061226, |
| "grad_norm": 0.19641028344631195, |
| "learning_rate": 0.0001141875, |
| "loss": 0.18497804641723634, |
| "step": 21600 |
| }, |
| { |
| "epoch": 0.4418367346938776, |
| "grad_norm": 0.1729995608329773, |
| "learning_rate": 0.00011397916666666667, |
| "loss": 0.18649446487426757, |
| "step": 21650 |
| }, |
| { |
| "epoch": 0.44285714285714284, |
| "grad_norm": 0.15542583167552948, |
| "learning_rate": 0.00011377083333333333, |
| "loss": 0.18261802673339844, |
| "step": 21700 |
| }, |
| { |
| "epoch": 0.44387755102040816, |
| "grad_norm": 0.2611772119998932, |
| "learning_rate": 0.00011356250000000002, |
| "loss": 0.2124016761779785, |
| "step": 21750 |
| }, |
| { |
| "epoch": 0.4448979591836735, |
| "grad_norm": 0.18791942298412323, |
| "learning_rate": 0.00011335416666666667, |
| "loss": 0.18935688018798827, |
| "step": 21800 |
| }, |
| { |
| "epoch": 0.4459183673469388, |
| "grad_norm": 0.13250787556171417, |
| "learning_rate": 0.00011314583333333333, |
| "loss": 0.20374603271484376, |
| "step": 21850 |
| }, |
| { |
| "epoch": 0.44693877551020406, |
| "grad_norm": 0.3324398398399353, |
| "learning_rate": 0.0001129375, |
| "loss": 0.20623138427734375, |
| "step": 21900 |
| }, |
| { |
| "epoch": 0.4479591836734694, |
| "grad_norm": 0.1987542361021042, |
| "learning_rate": 0.00011272916666666666, |
| "loss": 0.1954686737060547, |
| "step": 21950 |
| }, |
| { |
| "epoch": 0.4489795918367347, |
| "grad_norm": 0.10962691158056259, |
| "learning_rate": 0.00011252083333333334, |
| "loss": 0.16852352142333984, |
| "step": 22000 |
| }, |
| { |
| "epoch": 0.45, |
| "grad_norm": 0.1651119887828827, |
| "learning_rate": 0.0001123125, |
| "loss": 0.1804306411743164, |
| "step": 22050 |
| }, |
| { |
| "epoch": 0.45102040816326533, |
| "grad_norm": 0.135285884141922, |
| "learning_rate": 0.00011210416666666668, |
| "loss": 0.1708018112182617, |
| "step": 22100 |
| }, |
| { |
| "epoch": 0.4520408163265306, |
| "grad_norm": 0.16042831540107727, |
| "learning_rate": 0.00011189583333333333, |
| "loss": 0.20246376037597658, |
| "step": 22150 |
| }, |
| { |
| "epoch": 0.4530612244897959, |
| "grad_norm": 0.1572142094373703, |
| "learning_rate": 0.00011168750000000001, |
| "loss": 0.20286598205566406, |
| "step": 22200 |
| }, |
| { |
| "epoch": 0.45408163265306123, |
| "grad_norm": 0.23938754200935364, |
| "learning_rate": 0.00011147916666666667, |
| "loss": 0.19506494522094728, |
| "step": 22250 |
| }, |
| { |
| "epoch": 0.45510204081632655, |
| "grad_norm": 0.09990780800580978, |
| "learning_rate": 0.00011127083333333335, |
| "loss": 0.18617202758789062, |
| "step": 22300 |
| }, |
| { |
| "epoch": 0.4561224489795918, |
| "grad_norm": 0.1848301887512207, |
| "learning_rate": 0.00011106250000000001, |
| "loss": 0.17628313064575196, |
| "step": 22350 |
| }, |
| { |
| "epoch": 0.45714285714285713, |
| "grad_norm": 0.10634121298789978, |
| "learning_rate": 0.00011085416666666668, |
| "loss": 0.20597057342529296, |
| "step": 22400 |
| }, |
| { |
| "epoch": 0.45816326530612245, |
| "grad_norm": 0.15891435742378235, |
| "learning_rate": 0.00011064583333333334, |
| "loss": 0.18053476333618165, |
| "step": 22450 |
| }, |
| { |
| "epoch": 0.45918367346938777, |
| "grad_norm": 0.13435883820056915, |
| "learning_rate": 0.00011043750000000002, |
| "loss": 0.17978906631469727, |
| "step": 22500 |
| }, |
| { |
| "epoch": 0.4602040816326531, |
| "grad_norm": 0.13572899997234344, |
| "learning_rate": 0.00011022916666666668, |
| "loss": 0.18978822708129883, |
| "step": 22550 |
| }, |
| { |
| "epoch": 0.46122448979591835, |
| "grad_norm": 0.17398372292518616, |
| "learning_rate": 0.00011002083333333332, |
| "loss": 0.2147346878051758, |
| "step": 22600 |
| }, |
| { |
| "epoch": 0.46224489795918366, |
| "grad_norm": 0.17112791538238525, |
| "learning_rate": 0.0001098125, |
| "loss": 0.17975505828857422, |
| "step": 22650 |
| }, |
| { |
| "epoch": 0.463265306122449, |
| "grad_norm": 0.19042028486728668, |
| "learning_rate": 0.00010960416666666666, |
| "loss": 0.1823933219909668, |
| "step": 22700 |
| }, |
| { |
| "epoch": 0.4642857142857143, |
| "grad_norm": 0.24797481298446655, |
| "learning_rate": 0.00010939583333333335, |
| "loss": 0.21492622375488282, |
| "step": 22750 |
| }, |
| { |
| "epoch": 0.46530612244897956, |
| "grad_norm": 0.3532654643058777, |
| "learning_rate": 0.0001091875, |
| "loss": 0.18178869247436524, |
| "step": 22800 |
| }, |
| { |
| "epoch": 0.4663265306122449, |
| "grad_norm": 0.11487548053264618, |
| "learning_rate": 0.00010897916666666668, |
| "loss": 0.2208608627319336, |
| "step": 22850 |
| }, |
| { |
| "epoch": 0.4673469387755102, |
| "grad_norm": 0.15501990914344788, |
| "learning_rate": 0.00010877083333333333, |
| "loss": 0.17604400634765624, |
| "step": 22900 |
| }, |
| { |
| "epoch": 0.4683673469387755, |
| "grad_norm": 0.1941416710615158, |
| "learning_rate": 0.00010856250000000002, |
| "loss": 0.18979434967041015, |
| "step": 22950 |
| }, |
| { |
| "epoch": 0.46938775510204084, |
| "grad_norm": 0.20254535973072052, |
| "learning_rate": 0.00010835416666666667, |
| "loss": 0.19501394271850586, |
| "step": 23000 |
| }, |
| { |
| "epoch": 0.4704081632653061, |
| "grad_norm": 0.1467224657535553, |
| "learning_rate": 0.00010814583333333334, |
| "loss": 0.18560121536254884, |
| "step": 23050 |
| }, |
| { |
| "epoch": 0.4714285714285714, |
| "grad_norm": 0.1789759397506714, |
| "learning_rate": 0.0001079375, |
| "loss": 0.20693172454833986, |
| "step": 23100 |
| }, |
| { |
| "epoch": 0.47244897959183674, |
| "grad_norm": 0.17060811817646027, |
| "learning_rate": 0.00010772916666666669, |
| "loss": 0.24144332885742187, |
| "step": 23150 |
| }, |
| { |
| "epoch": 0.47346938775510206, |
| "grad_norm": 0.13840524852275848, |
| "learning_rate": 0.00010752500000000001, |
| "loss": 0.22703798294067382, |
| "step": 23200 |
| }, |
| { |
| "epoch": 0.4744897959183674, |
| "grad_norm": 0.22518107295036316, |
| "learning_rate": 0.00010731666666666667, |
| "loss": 0.18032278060913087, |
| "step": 23250 |
| }, |
| { |
| "epoch": 0.47551020408163264, |
| "grad_norm": 0.14671215415000916, |
| "learning_rate": 0.00010710833333333334, |
| "loss": 0.19824394226074218, |
| "step": 23300 |
| }, |
| { |
| "epoch": 0.47653061224489796, |
| "grad_norm": 0.15659059584140778, |
| "learning_rate": 0.0001069, |
| "loss": 0.1770372772216797, |
| "step": 23350 |
| }, |
| { |
| "epoch": 0.4775510204081633, |
| "grad_norm": 0.29906657338142395, |
| "learning_rate": 0.00010669166666666668, |
| "loss": 0.16190267562866212, |
| "step": 23400 |
| }, |
| { |
| "epoch": 0.4785714285714286, |
| "grad_norm": 0.1921195089817047, |
| "learning_rate": 0.00010648333333333334, |
| "loss": 0.1724691581726074, |
| "step": 23450 |
| }, |
| { |
| "epoch": 0.47959183673469385, |
| "grad_norm": 0.22063606977462769, |
| "learning_rate": 0.00010627500000000001, |
| "loss": 0.18350271224975587, |
| "step": 23500 |
| }, |
| { |
| "epoch": 0.4806122448979592, |
| "grad_norm": 0.1806723177433014, |
| "learning_rate": 0.00010606666666666667, |
| "loss": 0.23431238174438476, |
| "step": 23550 |
| }, |
| { |
| "epoch": 0.4816326530612245, |
| "grad_norm": 0.14175860583782196, |
| "learning_rate": 0.00010585833333333335, |
| "loss": 0.1887098503112793, |
| "step": 23600 |
| }, |
| { |
| "epoch": 0.4826530612244898, |
| "grad_norm": 0.2510957717895508, |
| "learning_rate": 0.00010565, |
| "loss": 0.18034719467163085, |
| "step": 23650 |
| }, |
| { |
| "epoch": 0.48367346938775513, |
| "grad_norm": 0.2018708735704422, |
| "learning_rate": 0.00010544166666666666, |
| "loss": 0.18103116989135742, |
| "step": 23700 |
| }, |
| { |
| "epoch": 0.4846938775510204, |
| "grad_norm": 0.15439936518669128, |
| "learning_rate": 0.00010523333333333333, |
| "loss": 0.14322800636291505, |
| "step": 23750 |
| }, |
| { |
| "epoch": 0.4857142857142857, |
| "grad_norm": 0.15107983350753784, |
| "learning_rate": 0.00010502499999999999, |
| "loss": 0.19188661575317384, |
| "step": 23800 |
| }, |
| { |
| "epoch": 0.48673469387755103, |
| "grad_norm": 0.16207917034626007, |
| "learning_rate": 0.00010481666666666668, |
| "loss": 0.1676897621154785, |
| "step": 23850 |
| }, |
| { |
| "epoch": 0.48775510204081635, |
| "grad_norm": 0.17554081976413727, |
| "learning_rate": 0.00010460833333333333, |
| "loss": 0.1537460517883301, |
| "step": 23900 |
| }, |
| { |
| "epoch": 0.4887755102040816, |
| "grad_norm": 0.18597754836082458, |
| "learning_rate": 0.0001044, |
| "loss": 0.16681880950927735, |
| "step": 23950 |
| }, |
| { |
| "epoch": 0.4897959183673469, |
| "grad_norm": 0.15398620069026947, |
| "learning_rate": 0.00010419166666666666, |
| "loss": 0.19642560958862304, |
| "step": 24000 |
| }, |
| { |
| "epoch": 0.49081632653061225, |
| "grad_norm": 0.2529297173023224, |
| "learning_rate": 0.00010398333333333334, |
| "loss": 0.19098554611206053, |
| "step": 24050 |
| }, |
| { |
| "epoch": 0.49183673469387756, |
| "grad_norm": 0.1370229721069336, |
| "learning_rate": 0.000103775, |
| "loss": 0.18588760375976562, |
| "step": 24100 |
| }, |
| { |
| "epoch": 0.4928571428571429, |
| "grad_norm": 0.16766472160816193, |
| "learning_rate": 0.00010356666666666667, |
| "loss": 0.20669187545776369, |
| "step": 24150 |
| }, |
| { |
| "epoch": 0.49387755102040815, |
| "grad_norm": 0.1888003945350647, |
| "learning_rate": 0.00010335833333333333, |
| "loss": 0.1881814956665039, |
| "step": 24200 |
| }, |
| { |
| "epoch": 0.49489795918367346, |
| "grad_norm": 0.15162178874015808, |
| "learning_rate": 0.00010315000000000001, |
| "loss": 0.19838417053222657, |
| "step": 24250 |
| }, |
| { |
| "epoch": 0.4959183673469388, |
| "grad_norm": 0.12254944443702698, |
| "learning_rate": 0.00010294166666666667, |
| "loss": 0.20314399719238282, |
| "step": 24300 |
| }, |
| { |
| "epoch": 0.4969387755102041, |
| "grad_norm": 0.21737927198410034, |
| "learning_rate": 0.00010273333333333334, |
| "loss": 0.20496267318725586, |
| "step": 24350 |
| }, |
| { |
| "epoch": 0.49795918367346936, |
| "grad_norm": 0.17958377301692963, |
| "learning_rate": 0.000102525, |
| "loss": 0.17489883422851563, |
| "step": 24400 |
| }, |
| { |
| "epoch": 0.4989795918367347, |
| "grad_norm": 0.13511185348033905, |
| "learning_rate": 0.00010231666666666666, |
| "loss": 0.17418720245361327, |
| "step": 24450 |
| }, |
| { |
| "epoch": 0.5, |
| "grad_norm": 0.2787463665008545, |
| "learning_rate": 0.00010210833333333334, |
| "loss": 0.17176372528076173, |
| "step": 24500 |
| }, |
| { |
| "epoch": 0.5010204081632653, |
| "grad_norm": 0.18124207854270935, |
| "learning_rate": 0.0001019, |
| "loss": 0.1850296974182129, |
| "step": 24550 |
| }, |
| { |
| "epoch": 0.5020408163265306, |
| "grad_norm": 0.1749614179134369, |
| "learning_rate": 0.00010169166666666667, |
| "loss": 0.17426204681396484, |
| "step": 24600 |
| }, |
| { |
| "epoch": 0.503061224489796, |
| "grad_norm": 0.16055095195770264, |
| "learning_rate": 0.00010148333333333332, |
| "loss": 0.21457443237304688, |
| "step": 24650 |
| }, |
| { |
| "epoch": 0.5040816326530613, |
| "grad_norm": 0.25023123621940613, |
| "learning_rate": 0.00010127500000000001, |
| "loss": 0.2075656509399414, |
| "step": 24700 |
| }, |
| { |
| "epoch": 0.5051020408163265, |
| "grad_norm": 0.2181357741355896, |
| "learning_rate": 0.00010106666666666667, |
| "loss": 0.17456653594970703, |
| "step": 24750 |
| }, |
| { |
| "epoch": 0.5061224489795918, |
| "grad_norm": 0.17197753489017487, |
| "learning_rate": 0.00010085833333333335, |
| "loss": 0.1632823371887207, |
| "step": 24800 |
| }, |
| { |
| "epoch": 0.5071428571428571, |
| "grad_norm": 0.21415536105632782, |
| "learning_rate": 0.00010065, |
| "loss": 0.19841451644897462, |
| "step": 24850 |
| }, |
| { |
| "epoch": 0.5081632653061224, |
| "grad_norm": 0.20820344984531403, |
| "learning_rate": 0.00010044166666666668, |
| "loss": 0.17958457946777343, |
| "step": 24900 |
| }, |
| { |
| "epoch": 0.5091836734693878, |
| "grad_norm": 0.18609052896499634, |
| "learning_rate": 0.00010023333333333333, |
| "loss": 0.17983253479003905, |
| "step": 24950 |
| }, |
| { |
| "epoch": 0.5102040816326531, |
| "grad_norm": 0.2240186482667923, |
| "learning_rate": 0.00010002500000000002, |
| "loss": 0.1995364761352539, |
| "step": 25000 |
| }, |
| { |
| "epoch": 0.5112244897959184, |
| "grad_norm": 0.2212294191122055, |
| "learning_rate": 9.981666666666668e-05, |
| "loss": 0.1803879165649414, |
| "step": 25050 |
| }, |
| { |
| "epoch": 0.5122448979591837, |
| "grad_norm": 0.15916708111763, |
| "learning_rate": 9.960833333333333e-05, |
| "loss": 0.1730225944519043, |
| "step": 25100 |
| }, |
| { |
| "epoch": 0.513265306122449, |
| "grad_norm": 0.16600583493709564, |
| "learning_rate": 9.94e-05, |
| "loss": 0.1652672004699707, |
| "step": 25150 |
| }, |
| { |
| "epoch": 0.5142857142857142, |
| "grad_norm": 0.17638596892356873, |
| "learning_rate": 9.919166666666667e-05, |
| "loss": 0.202372989654541, |
| "step": 25200 |
| }, |
| { |
| "epoch": 0.5153061224489796, |
| "grad_norm": 0.14149129390716553, |
| "learning_rate": 9.898333333333335e-05, |
| "loss": 0.17594612121582032, |
| "step": 25250 |
| }, |
| { |
| "epoch": 0.5163265306122449, |
| "grad_norm": 0.1270650029182434, |
| "learning_rate": 9.8775e-05, |
| "loss": 0.18878219604492188, |
| "step": 25300 |
| }, |
| { |
| "epoch": 0.5173469387755102, |
| "grad_norm": 0.17733386158943176, |
| "learning_rate": 9.856666666666667e-05, |
| "loss": 0.2115381622314453, |
| "step": 25350 |
| }, |
| { |
| "epoch": 0.5183673469387755, |
| "grad_norm": 0.1744958609342575, |
| "learning_rate": 9.835833333333334e-05, |
| "loss": 0.16448768615722656, |
| "step": 25400 |
| }, |
| { |
| "epoch": 0.5193877551020408, |
| "grad_norm": 0.21373602747917175, |
| "learning_rate": 9.815000000000001e-05, |
| "loss": 0.1772871208190918, |
| "step": 25450 |
| }, |
| { |
| "epoch": 0.5204081632653061, |
| "grad_norm": 0.1898619532585144, |
| "learning_rate": 9.794166666666667e-05, |
| "loss": 0.2503120231628418, |
| "step": 25500 |
| }, |
| { |
| "epoch": 0.5214285714285715, |
| "grad_norm": 0.28234219551086426, |
| "learning_rate": 9.773333333333334e-05, |
| "loss": 0.17822504043579102, |
| "step": 25550 |
| }, |
| { |
| "epoch": 0.5224489795918368, |
| "grad_norm": 0.11259794235229492, |
| "learning_rate": 9.7525e-05, |
| "loss": 0.1666012191772461, |
| "step": 25600 |
| }, |
| { |
| "epoch": 0.523469387755102, |
| "grad_norm": 0.1492464244365692, |
| "learning_rate": 9.731666666666667e-05, |
| "loss": 0.17302135467529298, |
| "step": 25650 |
| }, |
| { |
| "epoch": 0.5244897959183673, |
| "grad_norm": 0.17087487876415253, |
| "learning_rate": 9.710833333333334e-05, |
| "loss": 0.17946313858032226, |
| "step": 25700 |
| }, |
| { |
| "epoch": 0.5255102040816326, |
| "grad_norm": 0.12258443981409073, |
| "learning_rate": 9.69e-05, |
| "loss": 0.18571706771850585, |
| "step": 25750 |
| }, |
| { |
| "epoch": 0.5265306122448979, |
| "grad_norm": 0.1733115315437317, |
| "learning_rate": 9.669166666666667e-05, |
| "loss": 0.1553201961517334, |
| "step": 25800 |
| }, |
| { |
| "epoch": 0.5275510204081633, |
| "grad_norm": 0.13839919865131378, |
| "learning_rate": 9.648333333333334e-05, |
| "loss": 0.19531139373779297, |
| "step": 25850 |
| }, |
| { |
| "epoch": 0.5285714285714286, |
| "grad_norm": 0.20602594316005707, |
| "learning_rate": 9.627500000000001e-05, |
| "loss": 0.16800386428833008, |
| "step": 25900 |
| }, |
| { |
| "epoch": 0.5295918367346939, |
| "grad_norm": 0.20680412650108337, |
| "learning_rate": 9.606666666666667e-05, |
| "loss": 0.1898546600341797, |
| "step": 25950 |
| }, |
| { |
| "epoch": 0.5306122448979592, |
| "grad_norm": 0.16150370240211487, |
| "learning_rate": 9.585833333333334e-05, |
| "loss": 0.15692392349243164, |
| "step": 26000 |
| }, |
| { |
| "epoch": 0.5316326530612245, |
| "grad_norm": 0.1982293277978897, |
| "learning_rate": 9.565000000000001e-05, |
| "loss": 0.19491943359375, |
| "step": 26050 |
| }, |
| { |
| "epoch": 0.5326530612244897, |
| "grad_norm": 0.1009984016418457, |
| "learning_rate": 9.544166666666668e-05, |
| "loss": 0.20024131774902343, |
| "step": 26100 |
| }, |
| { |
| "epoch": 0.5336734693877551, |
| "grad_norm": 0.20309703052043915, |
| "learning_rate": 9.523333333333334e-05, |
| "loss": 0.16962615966796876, |
| "step": 26150 |
| }, |
| { |
| "epoch": 0.5346938775510204, |
| "grad_norm": 0.17204195261001587, |
| "learning_rate": 9.5025e-05, |
| "loss": 0.1707878303527832, |
| "step": 26200 |
| }, |
| { |
| "epoch": 0.5357142857142857, |
| "grad_norm": 0.1919504553079605, |
| "learning_rate": 9.481666666666668e-05, |
| "loss": 0.17004173278808593, |
| "step": 26250 |
| }, |
| { |
| "epoch": 0.536734693877551, |
| "grad_norm": 0.15040870010852814, |
| "learning_rate": 9.460833333333335e-05, |
| "loss": 0.18361337661743163, |
| "step": 26300 |
| }, |
| { |
| "epoch": 0.5377551020408163, |
| "grad_norm": 0.1572142243385315, |
| "learning_rate": 9.44e-05, |
| "loss": 0.1793185615539551, |
| "step": 26350 |
| }, |
| { |
| "epoch": 0.5387755102040817, |
| "grad_norm": 0.2479061782360077, |
| "learning_rate": 9.419166666666666e-05, |
| "loss": 0.17476287841796875, |
| "step": 26400 |
| }, |
| { |
| "epoch": 0.539795918367347, |
| "grad_norm": 0.1852135807275772, |
| "learning_rate": 9.398333333333333e-05, |
| "loss": 0.16330987930297852, |
| "step": 26450 |
| }, |
| { |
| "epoch": 0.5408163265306123, |
| "grad_norm": 0.19276225566864014, |
| "learning_rate": 9.3775e-05, |
| "loss": 0.16047971725463867, |
| "step": 26500 |
| }, |
| { |
| "epoch": 0.5418367346938775, |
| "grad_norm": 0.1404666304588318, |
| "learning_rate": 9.356666666666667e-05, |
| "loss": 0.1972671127319336, |
| "step": 26550 |
| }, |
| { |
| "epoch": 0.5428571428571428, |
| "grad_norm": 0.10398276150226593, |
| "learning_rate": 9.335833333333333e-05, |
| "loss": 0.14255274772644044, |
| "step": 26600 |
| }, |
| { |
| "epoch": 0.5438775510204081, |
| "grad_norm": 0.14555589854717255, |
| "learning_rate": 9.315e-05, |
| "loss": 0.16434885025024415, |
| "step": 26650 |
| }, |
| { |
| "epoch": 0.5448979591836735, |
| "grad_norm": 0.1104368194937706, |
| "learning_rate": 9.294166666666667e-05, |
| "loss": 0.16342899322509766, |
| "step": 26700 |
| }, |
| { |
| "epoch": 0.5459183673469388, |
| "grad_norm": 0.11711753159761429, |
| "learning_rate": 9.273333333333334e-05, |
| "loss": 0.1949435806274414, |
| "step": 26750 |
| }, |
| { |
| "epoch": 0.5469387755102041, |
| "grad_norm": 0.12293335795402527, |
| "learning_rate": 9.252916666666667e-05, |
| "loss": 0.18458440780639648, |
| "step": 26800 |
| }, |
| { |
| "epoch": 0.5479591836734694, |
| "grad_norm": 0.23470911383628845, |
| "learning_rate": 9.232083333333334e-05, |
| "loss": 0.19848438262939452, |
| "step": 26850 |
| }, |
| { |
| "epoch": 0.5489795918367347, |
| "grad_norm": 0.16929446160793304, |
| "learning_rate": 9.21125e-05, |
| "loss": 0.1938508415222168, |
| "step": 26900 |
| }, |
| { |
| "epoch": 0.55, |
| "grad_norm": 0.22963696718215942, |
| "learning_rate": 9.190416666666667e-05, |
| "loss": 0.18044023513793944, |
| "step": 26950 |
| }, |
| { |
| "epoch": 0.5510204081632653, |
| "grad_norm": 0.21981900930404663, |
| "learning_rate": 9.169583333333334e-05, |
| "loss": 0.18905908584594727, |
| "step": 27000 |
| }, |
| { |
| "epoch": 0.5520408163265306, |
| "grad_norm": 0.10104738175868988, |
| "learning_rate": 9.148750000000001e-05, |
| "loss": 0.17186660766601564, |
| "step": 27050 |
| }, |
| { |
| "epoch": 0.5530612244897959, |
| "grad_norm": 0.27401968836784363, |
| "learning_rate": 9.127916666666667e-05, |
| "loss": 0.1650037384033203, |
| "step": 27100 |
| }, |
| { |
| "epoch": 0.5540816326530612, |
| "grad_norm": 0.20953841507434845, |
| "learning_rate": 9.107083333333334e-05, |
| "loss": 0.1946626091003418, |
| "step": 27150 |
| }, |
| { |
| "epoch": 0.5551020408163265, |
| "grad_norm": 0.20660588145256042, |
| "learning_rate": 9.08625e-05, |
| "loss": 0.2258216667175293, |
| "step": 27200 |
| }, |
| { |
| "epoch": 0.5561224489795918, |
| "grad_norm": 0.1534866839647293, |
| "learning_rate": 9.065416666666668e-05, |
| "loss": 0.20488595962524414, |
| "step": 27250 |
| }, |
| { |
| "epoch": 0.5571428571428572, |
| "grad_norm": 0.14898627996444702, |
| "learning_rate": 9.044583333333333e-05, |
| "loss": 0.16483705520629882, |
| "step": 27300 |
| }, |
| { |
| "epoch": 0.5581632653061225, |
| "grad_norm": 0.12369991093873978, |
| "learning_rate": 9.02375e-05, |
| "loss": 0.18049694061279298, |
| "step": 27350 |
| }, |
| { |
| "epoch": 0.5591836734693878, |
| "grad_norm": 0.20999008417129517, |
| "learning_rate": 9.002916666666668e-05, |
| "loss": 0.16304676055908204, |
| "step": 27400 |
| }, |
| { |
| "epoch": 0.560204081632653, |
| "grad_norm": 0.1313706636428833, |
| "learning_rate": 8.982083333333335e-05, |
| "loss": 0.19436006546020507, |
| "step": 27450 |
| }, |
| { |
| "epoch": 0.5612244897959183, |
| "grad_norm": 0.10687727481126785, |
| "learning_rate": 8.96125e-05, |
| "loss": 0.1658367919921875, |
| "step": 27500 |
| }, |
| { |
| "epoch": 0.5622448979591836, |
| "grad_norm": 0.19357901811599731, |
| "learning_rate": 8.940416666666666e-05, |
| "loss": 0.15278014183044433, |
| "step": 27550 |
| }, |
| { |
| "epoch": 0.563265306122449, |
| "grad_norm": 0.21208421885967255, |
| "learning_rate": 8.919583333333333e-05, |
| "loss": 0.14099541664123535, |
| "step": 27600 |
| }, |
| { |
| "epoch": 0.5642857142857143, |
| "grad_norm": 0.1888977587223053, |
| "learning_rate": 8.89875e-05, |
| "loss": 0.18072011947631836, |
| "step": 27650 |
| }, |
| { |
| "epoch": 0.5653061224489796, |
| "grad_norm": 0.12501735985279083, |
| "learning_rate": 8.877916666666667e-05, |
| "loss": 0.16086437225341796, |
| "step": 27700 |
| }, |
| { |
| "epoch": 0.5663265306122449, |
| "grad_norm": 0.18862837553024292, |
| "learning_rate": 8.857083333333334e-05, |
| "loss": 0.1829837989807129, |
| "step": 27750 |
| }, |
| { |
| "epoch": 0.5673469387755102, |
| "grad_norm": 0.15165726840496063, |
| "learning_rate": 8.83625e-05, |
| "loss": 0.1913384437561035, |
| "step": 27800 |
| }, |
| { |
| "epoch": 0.5683673469387756, |
| "grad_norm": 0.2055622786283493, |
| "learning_rate": 8.815416666666667e-05, |
| "loss": 0.18679183959960938, |
| "step": 27850 |
| }, |
| { |
| "epoch": 0.5693877551020409, |
| "grad_norm": 0.17202599346637726, |
| "learning_rate": 8.794583333333334e-05, |
| "loss": 0.1639151954650879, |
| "step": 27900 |
| }, |
| { |
| "epoch": 0.5704081632653061, |
| "grad_norm": 0.20725397765636444, |
| "learning_rate": 8.773750000000001e-05, |
| "loss": 0.165450439453125, |
| "step": 27950 |
| }, |
| { |
| "epoch": 0.5714285714285714, |
| "grad_norm": 0.1799977719783783, |
| "learning_rate": 8.752916666666667e-05, |
| "loss": 0.15639174461364747, |
| "step": 28000 |
| }, |
| { |
| "epoch": 0.5724489795918367, |
| "grad_norm": 0.2165798842906952, |
| "learning_rate": 8.732083333333334e-05, |
| "loss": 0.17581111907958985, |
| "step": 28050 |
| }, |
| { |
| "epoch": 0.573469387755102, |
| "grad_norm": 0.15645107626914978, |
| "learning_rate": 8.711250000000001e-05, |
| "loss": 0.184375057220459, |
| "step": 28100 |
| }, |
| { |
| "epoch": 0.5744897959183674, |
| "grad_norm": 0.16599629819393158, |
| "learning_rate": 8.690416666666668e-05, |
| "loss": 0.17481328964233397, |
| "step": 28150 |
| }, |
| { |
| "epoch": 0.5755102040816327, |
| "grad_norm": 0.13364972174167633, |
| "learning_rate": 8.669583333333334e-05, |
| "loss": 0.19754388809204101, |
| "step": 28200 |
| }, |
| { |
| "epoch": 0.576530612244898, |
| "grad_norm": 0.14583703875541687, |
| "learning_rate": 8.64875e-05, |
| "loss": 0.21159143447875978, |
| "step": 28250 |
| }, |
| { |
| "epoch": 0.5775510204081633, |
| "grad_norm": 0.20671476423740387, |
| "learning_rate": 8.627916666666667e-05, |
| "loss": 0.16632991790771484, |
| "step": 28300 |
| }, |
| { |
| "epoch": 0.5785714285714286, |
| "grad_norm": 0.21721985936164856, |
| "learning_rate": 8.607083333333334e-05, |
| "loss": 0.17269231796264647, |
| "step": 28350 |
| }, |
| { |
| "epoch": 0.5795918367346938, |
| "grad_norm": 0.15544889867305756, |
| "learning_rate": 8.58625e-05, |
| "loss": 0.17736480712890626, |
| "step": 28400 |
| }, |
| { |
| "epoch": 0.5806122448979592, |
| "grad_norm": 0.13705679774284363, |
| "learning_rate": 8.565416666666666e-05, |
| "loss": 0.2022319984436035, |
| "step": 28450 |
| }, |
| { |
| "epoch": 0.5816326530612245, |
| "grad_norm": 0.1865527480840683, |
| "learning_rate": 8.544583333333333e-05, |
| "loss": 0.18072795867919922, |
| "step": 28500 |
| }, |
| { |
| "epoch": 0.5826530612244898, |
| "grad_norm": 0.1536557823419571, |
| "learning_rate": 8.52375e-05, |
| "loss": 0.1788094711303711, |
| "step": 28550 |
| }, |
| { |
| "epoch": 0.5836734693877551, |
| "grad_norm": 0.21039696037769318, |
| "learning_rate": 8.502916666666668e-05, |
| "loss": 0.16440990447998047, |
| "step": 28600 |
| }, |
| { |
| "epoch": 0.5846938775510204, |
| "grad_norm": 0.12752863764762878, |
| "learning_rate": 8.482083333333333e-05, |
| "loss": 0.18658769607543946, |
| "step": 28650 |
| }, |
| { |
| "epoch": 0.5857142857142857, |
| "grad_norm": 0.2432672679424286, |
| "learning_rate": 8.46125e-05, |
| "loss": 0.17903919219970704, |
| "step": 28700 |
| }, |
| { |
| "epoch": 0.5867346938775511, |
| "grad_norm": 0.1309838593006134, |
| "learning_rate": 8.440416666666667e-05, |
| "loss": 0.16778242111206054, |
| "step": 28750 |
| }, |
| { |
| "epoch": 0.5877551020408164, |
| "grad_norm": 0.17199347913265228, |
| "learning_rate": 8.419583333333334e-05, |
| "loss": 0.19217905044555664, |
| "step": 28800 |
| }, |
| { |
| "epoch": 0.5887755102040816, |
| "grad_norm": 0.1479269117116928, |
| "learning_rate": 8.39875e-05, |
| "loss": 0.14942014694213868, |
| "step": 28850 |
| }, |
| { |
| "epoch": 0.5897959183673469, |
| "grad_norm": 0.15467581152915955, |
| "learning_rate": 8.377916666666667e-05, |
| "loss": 0.1766769027709961, |
| "step": 28900 |
| }, |
| { |
| "epoch": 0.5908163265306122, |
| "grad_norm": 0.14890123903751373, |
| "learning_rate": 8.357083333333334e-05, |
| "loss": 0.18058040618896484, |
| "step": 28950 |
| }, |
| { |
| "epoch": 0.5918367346938775, |
| "grad_norm": 0.17233826220035553, |
| "learning_rate": 8.33625e-05, |
| "loss": 0.16066322326660157, |
| "step": 29000 |
| }, |
| { |
| "epoch": 0.5928571428571429, |
| "grad_norm": 0.11463279277086258, |
| "learning_rate": 8.315416666666667e-05, |
| "loss": 0.14516469955444336, |
| "step": 29050 |
| }, |
| { |
| "epoch": 0.5938775510204082, |
| "grad_norm": 0.19112053513526917, |
| "learning_rate": 8.294583333333333e-05, |
| "loss": 0.18832321166992189, |
| "step": 29100 |
| }, |
| { |
| "epoch": 0.5948979591836735, |
| "grad_norm": 0.10975775122642517, |
| "learning_rate": 8.274166666666667e-05, |
| "loss": 0.17870872497558593, |
| "step": 29150 |
| }, |
| { |
| "epoch": 0.5959183673469388, |
| "grad_norm": 0.1835976541042328, |
| "learning_rate": 8.253333333333334e-05, |
| "loss": 0.15557955741882323, |
| "step": 29200 |
| }, |
| { |
| "epoch": 0.5969387755102041, |
| "grad_norm": 0.2260265052318573, |
| "learning_rate": 8.232500000000001e-05, |
| "loss": 0.16450485229492187, |
| "step": 29250 |
| }, |
| { |
| "epoch": 0.5979591836734693, |
| "grad_norm": 0.1235552653670311, |
| "learning_rate": 8.211666666666668e-05, |
| "loss": 0.16741344451904297, |
| "step": 29300 |
| }, |
| { |
| "epoch": 0.5989795918367347, |
| "grad_norm": 0.1380196213722229, |
| "learning_rate": 8.190833333333334e-05, |
| "loss": 0.18348945617675783, |
| "step": 29350 |
| }, |
| { |
| "epoch": 0.6, |
| "grad_norm": 0.1534113883972168, |
| "learning_rate": 8.17e-05, |
| "loss": 0.17626720428466797, |
| "step": 29400 |
| }, |
| { |
| "epoch": 0.6010204081632653, |
| "grad_norm": 0.17966432869434357, |
| "learning_rate": 8.149166666666666e-05, |
| "loss": 0.18582574844360353, |
| "step": 29450 |
| }, |
| { |
| "epoch": 0.6020408163265306, |
| "grad_norm": 0.17597244679927826, |
| "learning_rate": 8.128333333333333e-05, |
| "loss": 0.1618438720703125, |
| "step": 29500 |
| }, |
| { |
| "epoch": 0.6030612244897959, |
| "grad_norm": 0.1751350611448288, |
| "learning_rate": 8.1075e-05, |
| "loss": 0.17317716598510743, |
| "step": 29550 |
| }, |
| { |
| "epoch": 0.6040816326530613, |
| "grad_norm": 0.2299734205007553, |
| "learning_rate": 8.086666666666666e-05, |
| "loss": 0.19614763259887696, |
| "step": 29600 |
| }, |
| { |
| "epoch": 0.6051020408163266, |
| "grad_norm": 0.19768504798412323, |
| "learning_rate": 8.065833333333333e-05, |
| "loss": 0.18232860565185546, |
| "step": 29650 |
| }, |
| { |
| "epoch": 0.6061224489795919, |
| "grad_norm": 0.14797262847423553, |
| "learning_rate": 8.045e-05, |
| "loss": 0.18202657699584962, |
| "step": 29700 |
| }, |
| { |
| "epoch": 0.6071428571428571, |
| "grad_norm": 0.18453800678253174, |
| "learning_rate": 8.024166666666667e-05, |
| "loss": 0.1945005989074707, |
| "step": 29750 |
| }, |
| { |
| "epoch": 0.6081632653061224, |
| "grad_norm": 0.24138791859149933, |
| "learning_rate": 8.003333333333333e-05, |
| "loss": 0.15666120529174804, |
| "step": 29800 |
| }, |
| { |
| "epoch": 0.6091836734693877, |
| "grad_norm": 0.09741450846195221, |
| "learning_rate": 7.9825e-05, |
| "loss": 0.13373960494995119, |
| "step": 29850 |
| }, |
| { |
| "epoch": 0.610204081632653, |
| "grad_norm": 0.1770496666431427, |
| "learning_rate": 7.961666666666667e-05, |
| "loss": 0.16969673156738282, |
| "step": 29900 |
| }, |
| { |
| "epoch": 0.6112244897959184, |
| "grad_norm": 0.13566668331623077, |
| "learning_rate": 7.940833333333334e-05, |
| "loss": 0.1536888599395752, |
| "step": 29950 |
| }, |
| { |
| "epoch": 0.6122448979591837, |
| "grad_norm": 0.2294454276561737, |
| "learning_rate": 7.920000000000001e-05, |
| "loss": 0.1650291633605957, |
| "step": 30000 |
| }, |
| { |
| "epoch": 0.613265306122449, |
| "grad_norm": 0.1496036797761917, |
| "learning_rate": 7.899166666666667e-05, |
| "loss": 0.1761037826538086, |
| "step": 30050 |
| }, |
| { |
| "epoch": 0.6142857142857143, |
| "grad_norm": 0.08214724063873291, |
| "learning_rate": 7.878333333333334e-05, |
| "loss": 0.19763166427612305, |
| "step": 30100 |
| }, |
| { |
| "epoch": 0.6153061224489796, |
| "grad_norm": 0.10310641676187515, |
| "learning_rate": 7.8575e-05, |
| "loss": 0.1637880516052246, |
| "step": 30150 |
| }, |
| { |
| "epoch": 0.6163265306122448, |
| "grad_norm": 0.1486400067806244, |
| "learning_rate": 7.836666666666667e-05, |
| "loss": 0.1754462242126465, |
| "step": 30200 |
| }, |
| { |
| "epoch": 0.6173469387755102, |
| "grad_norm": 0.1679656207561493, |
| "learning_rate": 7.815833333333333e-05, |
| "loss": 0.14919873237609863, |
| "step": 30250 |
| }, |
| { |
| "epoch": 0.6183673469387755, |
| "grad_norm": 0.18345525860786438, |
| "learning_rate": 7.795e-05, |
| "loss": 0.19217906951904296, |
| "step": 30300 |
| }, |
| { |
| "epoch": 0.6193877551020408, |
| "grad_norm": 0.15979202091693878, |
| "learning_rate": 7.774166666666667e-05, |
| "loss": 0.18305580139160157, |
| "step": 30350 |
| }, |
| { |
| "epoch": 0.6204081632653061, |
| "grad_norm": 0.15286119282245636, |
| "learning_rate": 7.753333333333334e-05, |
| "loss": 0.14841639518737793, |
| "step": 30400 |
| }, |
| { |
| "epoch": 0.6214285714285714, |
| "grad_norm": 0.20205309987068176, |
| "learning_rate": 7.732500000000001e-05, |
| "loss": 0.1803289222717285, |
| "step": 30450 |
| }, |
| { |
| "epoch": 0.6224489795918368, |
| "grad_norm": 0.10329090803861618, |
| "learning_rate": 7.711666666666667e-05, |
| "loss": 0.19099008560180664, |
| "step": 30500 |
| }, |
| { |
| "epoch": 0.6234693877551021, |
| "grad_norm": 0.2586289346218109, |
| "learning_rate": 7.690833333333334e-05, |
| "loss": 0.194241943359375, |
| "step": 30550 |
| }, |
| { |
| "epoch": 0.6244897959183674, |
| "grad_norm": 0.18437634408473969, |
| "learning_rate": 7.670000000000001e-05, |
| "loss": 0.19446161270141601, |
| "step": 30600 |
| }, |
| { |
| "epoch": 0.6255102040816326, |
| "grad_norm": 0.20497523248195648, |
| "learning_rate": 7.649166666666668e-05, |
| "loss": 0.16262346267700195, |
| "step": 30650 |
| }, |
| { |
| "epoch": 0.6265306122448979, |
| "grad_norm": 0.15078936517238617, |
| "learning_rate": 7.628333333333334e-05, |
| "loss": 0.1567188549041748, |
| "step": 30700 |
| }, |
| { |
| "epoch": 0.6275510204081632, |
| "grad_norm": 0.13786426186561584, |
| "learning_rate": 7.6075e-05, |
| "loss": 0.14639299392700195, |
| "step": 30750 |
| }, |
| { |
| "epoch": 0.6285714285714286, |
| "grad_norm": 0.16654115915298462, |
| "learning_rate": 7.586666666666668e-05, |
| "loss": 0.1610732650756836, |
| "step": 30800 |
| }, |
| { |
| "epoch": 0.6295918367346939, |
| "grad_norm": 0.16476523876190186, |
| "learning_rate": 7.565833333333335e-05, |
| "loss": 0.17198530197143555, |
| "step": 30850 |
| }, |
| { |
| "epoch": 0.6306122448979592, |
| "grad_norm": 0.29812154173851013, |
| "learning_rate": 7.545e-05, |
| "loss": 0.18083591461181642, |
| "step": 30900 |
| }, |
| { |
| "epoch": 0.6316326530612245, |
| "grad_norm": 0.18219386041164398, |
| "learning_rate": 7.524166666666666e-05, |
| "loss": 0.1330328941345215, |
| "step": 30950 |
| }, |
| { |
| "epoch": 0.6326530612244898, |
| "grad_norm": 0.1219499409198761, |
| "learning_rate": 7.503333333333333e-05, |
| "loss": 0.1631675910949707, |
| "step": 31000 |
| }, |
| { |
| "epoch": 0.6336734693877552, |
| "grad_norm": 0.16549046337604523, |
| "learning_rate": 7.4825e-05, |
| "loss": 0.1869835662841797, |
| "step": 31050 |
| }, |
| { |
| "epoch": 0.6346938775510204, |
| "grad_norm": 0.13592897355556488, |
| "learning_rate": 7.461666666666667e-05, |
| "loss": 0.16354337692260743, |
| "step": 31100 |
| }, |
| { |
| "epoch": 0.6357142857142857, |
| "grad_norm": 0.25532266497612, |
| "learning_rate": 7.440833333333333e-05, |
| "loss": 0.1795151138305664, |
| "step": 31150 |
| }, |
| { |
| "epoch": 0.636734693877551, |
| "grad_norm": 0.13497677445411682, |
| "learning_rate": 7.42e-05, |
| "loss": 0.17722156524658203, |
| "step": 31200 |
| }, |
| { |
| "epoch": 0.6377551020408163, |
| "grad_norm": 0.10550568997859955, |
| "learning_rate": 7.399166666666667e-05, |
| "loss": 0.16642751693725585, |
| "step": 31250 |
| }, |
| { |
| "epoch": 0.6387755102040816, |
| "grad_norm": 0.1571359634399414, |
| "learning_rate": 7.378333333333334e-05, |
| "loss": 0.14314020156860352, |
| "step": 31300 |
| }, |
| { |
| "epoch": 0.639795918367347, |
| "grad_norm": 0.16488172113895416, |
| "learning_rate": 7.3575e-05, |
| "loss": 0.17789432525634766, |
| "step": 31350 |
| }, |
| { |
| "epoch": 0.6408163265306123, |
| "grad_norm": 0.15754637122154236, |
| "learning_rate": 7.336666666666667e-05, |
| "loss": 0.1632952880859375, |
| "step": 31400 |
| }, |
| { |
| "epoch": 0.6418367346938776, |
| "grad_norm": 0.16609132289886475, |
| "learning_rate": 7.315833333333334e-05, |
| "loss": 0.17927698135375977, |
| "step": 31450 |
| }, |
| { |
| "epoch": 0.6428571428571429, |
| "grad_norm": 0.16577483713626862, |
| "learning_rate": 7.295000000000001e-05, |
| "loss": 0.16777738571166992, |
| "step": 31500 |
| }, |
| { |
| "epoch": 0.6438775510204081, |
| "grad_norm": 0.1324053853750229, |
| "learning_rate": 7.274166666666667e-05, |
| "loss": 0.17467840194702147, |
| "step": 31550 |
| }, |
| { |
| "epoch": 0.6448979591836734, |
| "grad_norm": 0.15986141562461853, |
| "learning_rate": 7.253333333333334e-05, |
| "loss": 0.1784464454650879, |
| "step": 31600 |
| }, |
| { |
| "epoch": 0.6459183673469387, |
| "grad_norm": 0.1035812571644783, |
| "learning_rate": 7.2325e-05, |
| "loss": 0.17713926315307618, |
| "step": 31650 |
| }, |
| { |
| "epoch": 0.6469387755102041, |
| "grad_norm": 0.20758819580078125, |
| "learning_rate": 7.211666666666667e-05, |
| "loss": 0.15218260765075683, |
| "step": 31700 |
| }, |
| { |
| "epoch": 0.6479591836734694, |
| "grad_norm": 0.16112469136714935, |
| "learning_rate": 7.190833333333334e-05, |
| "loss": 0.14512678146362304, |
| "step": 31750 |
| }, |
| { |
| "epoch": 0.6489795918367347, |
| "grad_norm": 0.1377130150794983, |
| "learning_rate": 7.17e-05, |
| "loss": 0.15772339820861817, |
| "step": 31800 |
| }, |
| { |
| "epoch": 0.65, |
| "grad_norm": 0.16242578625679016, |
| "learning_rate": 7.149166666666667e-05, |
| "loss": 0.15441008567810058, |
| "step": 31850 |
| }, |
| { |
| "epoch": 0.6510204081632653, |
| "grad_norm": 0.19400998950004578, |
| "learning_rate": 7.128333333333334e-05, |
| "loss": 0.14529372215270997, |
| "step": 31900 |
| }, |
| { |
| "epoch": 0.6520408163265307, |
| "grad_norm": 0.13732214272022247, |
| "learning_rate": 7.1075e-05, |
| "loss": 0.20341796875, |
| "step": 31950 |
| }, |
| { |
| "epoch": 0.6530612244897959, |
| "grad_norm": 0.1565335988998413, |
| "learning_rate": 7.086666666666666e-05, |
| "loss": 0.15315020561218262, |
| "step": 32000 |
| }, |
| { |
| "epoch": 0.6540816326530612, |
| "grad_norm": 0.1519925892353058, |
| "learning_rate": 7.065833333333333e-05, |
| "loss": 0.17672977447509766, |
| "step": 32050 |
| }, |
| { |
| "epoch": 0.6551020408163265, |
| "grad_norm": 0.10438583046197891, |
| "learning_rate": 7.045e-05, |
| "loss": 0.1952037239074707, |
| "step": 32100 |
| }, |
| { |
| "epoch": 0.6561224489795918, |
| "grad_norm": 0.12770964205265045, |
| "learning_rate": 7.024166666666668e-05, |
| "loss": 0.14889378547668458, |
| "step": 32150 |
| }, |
| { |
| "epoch": 0.6571428571428571, |
| "grad_norm": 0.15392087399959564, |
| "learning_rate": 7.003333333333335e-05, |
| "loss": 0.19392810821533202, |
| "step": 32200 |
| }, |
| { |
| "epoch": 0.6581632653061225, |
| "grad_norm": 0.16497986018657684, |
| "learning_rate": 6.9825e-05, |
| "loss": 0.14279278755187988, |
| "step": 32250 |
| }, |
| { |
| "epoch": 0.6591836734693878, |
| "grad_norm": 0.16945810616016388, |
| "learning_rate": 6.961666666666667e-05, |
| "loss": 0.1629298973083496, |
| "step": 32300 |
| }, |
| { |
| "epoch": 0.6602040816326531, |
| "grad_norm": 0.08853259682655334, |
| "learning_rate": 6.940833333333334e-05, |
| "loss": 0.1372718906402588, |
| "step": 32350 |
| }, |
| { |
| "epoch": 0.6612244897959184, |
| "grad_norm": 0.10184511542320251, |
| "learning_rate": 6.92e-05, |
| "loss": 0.12464181900024414, |
| "step": 32400 |
| }, |
| { |
| "epoch": 0.6622448979591836, |
| "grad_norm": 0.1579914540052414, |
| "learning_rate": 6.899166666666667e-05, |
| "loss": 0.1898073959350586, |
| "step": 32450 |
| }, |
| { |
| "epoch": 0.6632653061224489, |
| "grad_norm": 0.23278109729290009, |
| "learning_rate": 6.878333333333333e-05, |
| "loss": 0.19114408493041993, |
| "step": 32500 |
| }, |
| { |
| "epoch": 0.6642857142857143, |
| "grad_norm": 0.18605078756809235, |
| "learning_rate": 6.8575e-05, |
| "loss": 0.14429281234741212, |
| "step": 32550 |
| }, |
| { |
| "epoch": 0.6653061224489796, |
| "grad_norm": 0.20744793117046356, |
| "learning_rate": 6.836666666666667e-05, |
| "loss": 0.15178895950317384, |
| "step": 32600 |
| }, |
| { |
| "epoch": 0.6663265306122449, |
| "grad_norm": 0.1467064917087555, |
| "learning_rate": 6.816250000000001e-05, |
| "loss": 0.17128976821899414, |
| "step": 32650 |
| }, |
| { |
| "epoch": 0.6673469387755102, |
| "grad_norm": 0.21961116790771484, |
| "learning_rate": 6.795416666666668e-05, |
| "loss": 0.14988475799560547, |
| "step": 32700 |
| }, |
| { |
| "epoch": 0.6683673469387755, |
| "grad_norm": 0.19205953180789948, |
| "learning_rate": 6.774583333333334e-05, |
| "loss": 0.15868806838989258, |
| "step": 32750 |
| }, |
| { |
| "epoch": 0.6693877551020408, |
| "grad_norm": 0.19790537655353546, |
| "learning_rate": 6.75375e-05, |
| "loss": 0.22682212829589843, |
| "step": 32800 |
| }, |
| { |
| "epoch": 0.6704081632653062, |
| "grad_norm": 0.12448656558990479, |
| "learning_rate": 6.732916666666667e-05, |
| "loss": 0.16850042343139648, |
| "step": 32850 |
| }, |
| { |
| "epoch": 0.6714285714285714, |
| "grad_norm": 0.20378604531288147, |
| "learning_rate": 6.712083333333334e-05, |
| "loss": 0.16008363723754881, |
| "step": 32900 |
| }, |
| { |
| "epoch": 0.6724489795918367, |
| "grad_norm": 0.23952776193618774, |
| "learning_rate": 6.691250000000001e-05, |
| "loss": 0.20326755523681642, |
| "step": 32950 |
| }, |
| { |
| "epoch": 0.673469387755102, |
| "grad_norm": 0.17432567477226257, |
| "learning_rate": 6.670416666666666e-05, |
| "loss": 0.18787288665771484, |
| "step": 33000 |
| }, |
| { |
| "epoch": 0.6744897959183673, |
| "grad_norm": 0.20003508031368256, |
| "learning_rate": 6.649583333333333e-05, |
| "loss": 0.17611515045166015, |
| "step": 33050 |
| }, |
| { |
| "epoch": 0.6755102040816326, |
| "grad_norm": 0.19111096858978271, |
| "learning_rate": 6.62875e-05, |
| "loss": 0.16515064239501953, |
| "step": 33100 |
| }, |
| { |
| "epoch": 0.676530612244898, |
| "grad_norm": 0.1962498277425766, |
| "learning_rate": 6.607916666666668e-05, |
| "loss": 0.2007039260864258, |
| "step": 33150 |
| }, |
| { |
| "epoch": 0.6775510204081633, |
| "grad_norm": 0.21504424512386322, |
| "learning_rate": 6.587083333333333e-05, |
| "loss": 0.21071577072143555, |
| "step": 33200 |
| }, |
| { |
| "epoch": 0.6785714285714286, |
| "grad_norm": 0.19784651696681976, |
| "learning_rate": 6.56625e-05, |
| "loss": 0.16339536666870116, |
| "step": 33250 |
| }, |
| { |
| "epoch": 0.6795918367346939, |
| "grad_norm": 0.21860648691654205, |
| "learning_rate": 6.545416666666667e-05, |
| "loss": 0.17276811599731445, |
| "step": 33300 |
| }, |
| { |
| "epoch": 0.6806122448979591, |
| "grad_norm": 0.12416710704565048, |
| "learning_rate": 6.524583333333334e-05, |
| "loss": 0.17321861267089844, |
| "step": 33350 |
| }, |
| { |
| "epoch": 0.6816326530612244, |
| "grad_norm": 0.14108151197433472, |
| "learning_rate": 6.50375e-05, |
| "loss": 0.16422855377197265, |
| "step": 33400 |
| }, |
| { |
| "epoch": 0.6826530612244898, |
| "grad_norm": 0.12449406087398529, |
| "learning_rate": 6.482916666666667e-05, |
| "loss": 0.1222480583190918, |
| "step": 33450 |
| }, |
| { |
| "epoch": 0.6836734693877551, |
| "grad_norm": 0.1724015772342682, |
| "learning_rate": 6.462083333333334e-05, |
| "loss": 0.15967099189758302, |
| "step": 33500 |
| }, |
| { |
| "epoch": 0.6846938775510204, |
| "grad_norm": 0.27706989645957947, |
| "learning_rate": 6.44125e-05, |
| "loss": 0.21222850799560547, |
| "step": 33550 |
| }, |
| { |
| "epoch": 0.6857142857142857, |
| "grad_norm": 0.21608980000019073, |
| "learning_rate": 6.420416666666667e-05, |
| "loss": 0.19895681381225586, |
| "step": 33600 |
| }, |
| { |
| "epoch": 0.686734693877551, |
| "grad_norm": 0.17284168303012848, |
| "learning_rate": 6.399583333333333e-05, |
| "loss": 0.16315589904785155, |
| "step": 33650 |
| }, |
| { |
| "epoch": 0.6877551020408164, |
| "grad_norm": 0.2790461480617523, |
| "learning_rate": 6.37875e-05, |
| "loss": 0.17992326736450195, |
| "step": 33700 |
| }, |
| { |
| "epoch": 0.6887755102040817, |
| "grad_norm": 0.1978480964899063, |
| "learning_rate": 6.357916666666667e-05, |
| "loss": 0.19150606155395508, |
| "step": 33750 |
| }, |
| { |
| "epoch": 0.689795918367347, |
| "grad_norm": 0.22785209119319916, |
| "learning_rate": 6.337083333333334e-05, |
| "loss": 0.1392769718170166, |
| "step": 33800 |
| }, |
| { |
| "epoch": 0.6908163265306122, |
| "grad_norm": 0.15017133951187134, |
| "learning_rate": 6.31625e-05, |
| "loss": 0.1485874366760254, |
| "step": 33850 |
| }, |
| { |
| "epoch": 0.6918367346938775, |
| "grad_norm": 0.15515482425689697, |
| "learning_rate": 6.295416666666667e-05, |
| "loss": 0.13293993949890137, |
| "step": 33900 |
| }, |
| { |
| "epoch": 0.6928571428571428, |
| "grad_norm": 0.1759190857410431, |
| "learning_rate": 6.274583333333334e-05, |
| "loss": 0.15329167366027832, |
| "step": 33950 |
| }, |
| { |
| "epoch": 0.6938775510204082, |
| "grad_norm": 0.2007198929786682, |
| "learning_rate": 6.253750000000001e-05, |
| "loss": 0.16196441650390625, |
| "step": 34000 |
| }, |
| { |
| "epoch": 0.6948979591836735, |
| "grad_norm": 0.07480475306510925, |
| "learning_rate": 6.232916666666667e-05, |
| "loss": 0.14109251976013185, |
| "step": 34050 |
| }, |
| { |
| "epoch": 0.6959183673469388, |
| "grad_norm": 0.12071823328733444, |
| "learning_rate": 6.212083333333334e-05, |
| "loss": 0.15877034187316894, |
| "step": 34100 |
| }, |
| { |
| "epoch": 0.6969387755102041, |
| "grad_norm": 0.175904780626297, |
| "learning_rate": 6.191250000000001e-05, |
| "loss": 0.14621192932128907, |
| "step": 34150 |
| }, |
| { |
| "epoch": 0.6979591836734694, |
| "grad_norm": 0.09303800016641617, |
| "learning_rate": 6.170416666666668e-05, |
| "loss": 0.16565973281860352, |
| "step": 34200 |
| }, |
| { |
| "epoch": 0.6989795918367347, |
| "grad_norm": 0.38047531247138977, |
| "learning_rate": 6.149583333333334e-05, |
| "loss": 0.167867488861084, |
| "step": 34250 |
| }, |
| { |
| "epoch": 0.7, |
| "grad_norm": 0.1675473004579544, |
| "learning_rate": 6.128749999999999e-05, |
| "loss": 0.17095783233642578, |
| "step": 34300 |
| }, |
| { |
| "epoch": 0.7010204081632653, |
| "grad_norm": 0.12302542477846146, |
| "learning_rate": 6.107916666666666e-05, |
| "loss": 0.17203062057495117, |
| "step": 34350 |
| }, |
| { |
| "epoch": 0.7020408163265306, |
| "grad_norm": 0.1513855755329132, |
| "learning_rate": 6.0870833333333333e-05, |
| "loss": 0.1459299373626709, |
| "step": 34400 |
| }, |
| { |
| "epoch": 0.7030612244897959, |
| "grad_norm": 0.1730256825685501, |
| "learning_rate": 6.06625e-05, |
| "loss": 0.15072107315063477, |
| "step": 34450 |
| }, |
| { |
| "epoch": 0.7040816326530612, |
| "grad_norm": 0.18623055517673492, |
| "learning_rate": 6.045416666666667e-05, |
| "loss": 0.12694622993469237, |
| "step": 34500 |
| }, |
| { |
| "epoch": 0.7051020408163265, |
| "grad_norm": 0.15233691036701202, |
| "learning_rate": 6.024583333333333e-05, |
| "loss": 0.17883399963378907, |
| "step": 34550 |
| }, |
| { |
| "epoch": 0.7061224489795919, |
| "grad_norm": 0.13142655789852142, |
| "learning_rate": 6.00375e-05, |
| "loss": 0.1831110382080078, |
| "step": 34600 |
| }, |
| { |
| "epoch": 0.7071428571428572, |
| "grad_norm": 0.17817802727222443, |
| "learning_rate": 5.9829166666666666e-05, |
| "loss": 0.16313209533691406, |
| "step": 34650 |
| }, |
| { |
| "epoch": 0.7081632653061225, |
| "grad_norm": 0.14485876262187958, |
| "learning_rate": 5.962083333333334e-05, |
| "loss": 0.16395410537719726, |
| "step": 34700 |
| }, |
| { |
| "epoch": 0.7091836734693877, |
| "grad_norm": 0.17738710343837738, |
| "learning_rate": 5.941250000000001e-05, |
| "loss": 0.14464349746704103, |
| "step": 34750 |
| }, |
| { |
| "epoch": 0.710204081632653, |
| "grad_norm": 0.12579292058944702, |
| "learning_rate": 5.920416666666667e-05, |
| "loss": 0.17616662979125977, |
| "step": 34800 |
| }, |
| { |
| "epoch": 0.7112244897959183, |
| "grad_norm": 0.09561062604188919, |
| "learning_rate": 5.899583333333334e-05, |
| "loss": 0.11348544120788574, |
| "step": 34850 |
| }, |
| { |
| "epoch": 0.7122448979591837, |
| "grad_norm": 0.14961975812911987, |
| "learning_rate": 5.8787500000000006e-05, |
| "loss": 0.17728403091430664, |
| "step": 34900 |
| }, |
| { |
| "epoch": 0.713265306122449, |
| "grad_norm": 0.18672503530979156, |
| "learning_rate": 5.8579166666666677e-05, |
| "loss": 0.16636423110961915, |
| "step": 34950 |
| }, |
| { |
| "epoch": 0.7142857142857143, |
| "grad_norm": 0.13996222615242004, |
| "learning_rate": 5.837083333333334e-05, |
| "loss": 0.14518523216247559, |
| "step": 35000 |
| }, |
| { |
| "epoch": 0.7153061224489796, |
| "grad_norm": 0.06967910379171371, |
| "learning_rate": 5.81625e-05, |
| "loss": 0.17640968322753905, |
| "step": 35050 |
| }, |
| { |
| "epoch": 0.7163265306122449, |
| "grad_norm": 0.07042952626943588, |
| "learning_rate": 5.795416666666667e-05, |
| "loss": 0.14964930534362794, |
| "step": 35100 |
| }, |
| { |
| "epoch": 0.7173469387755103, |
| "grad_norm": 0.12094786763191223, |
| "learning_rate": 5.774583333333333e-05, |
| "loss": 0.16445869445800781, |
| "step": 35150 |
| }, |
| { |
| "epoch": 0.7183673469387755, |
| "grad_norm": 0.1543540060520172, |
| "learning_rate": 5.75375e-05, |
| "loss": 0.13257647514343263, |
| "step": 35200 |
| }, |
| { |
| "epoch": 0.7193877551020408, |
| "grad_norm": 0.14520111680030823, |
| "learning_rate": 5.7329166666666667e-05, |
| "loss": 0.16949567794799805, |
| "step": 35250 |
| }, |
| { |
| "epoch": 0.7204081632653061, |
| "grad_norm": 0.17580607533454895, |
| "learning_rate": 5.712083333333334e-05, |
| "loss": 0.16740028381347657, |
| "step": 35300 |
| }, |
| { |
| "epoch": 0.7214285714285714, |
| "grad_norm": 0.1910821497440338, |
| "learning_rate": 5.69125e-05, |
| "loss": 0.1391041946411133, |
| "step": 35350 |
| }, |
| { |
| "epoch": 0.7224489795918367, |
| "grad_norm": 0.0856727808713913, |
| "learning_rate": 5.670416666666667e-05, |
| "loss": 0.13688218116760253, |
| "step": 35400 |
| }, |
| { |
| "epoch": 0.7234693877551021, |
| "grad_norm": 0.22578872740268707, |
| "learning_rate": 5.6495833333333336e-05, |
| "loss": 0.13912582397460938, |
| "step": 35450 |
| }, |
| { |
| "epoch": 0.7244897959183674, |
| "grad_norm": 0.15911556780338287, |
| "learning_rate": 5.6287500000000006e-05, |
| "loss": 0.14832959175109864, |
| "step": 35500 |
| }, |
| { |
| "epoch": 0.7255102040816327, |
| "grad_norm": 0.18621928989887238, |
| "learning_rate": 5.607916666666667e-05, |
| "loss": 0.1669852066040039, |
| "step": 35550 |
| }, |
| { |
| "epoch": 0.726530612244898, |
| "grad_norm": 0.2119055539369583, |
| "learning_rate": 5.587083333333334e-05, |
| "loss": 0.1806819534301758, |
| "step": 35600 |
| }, |
| { |
| "epoch": 0.7275510204081632, |
| "grad_norm": 0.07961991429328918, |
| "learning_rate": 5.5662500000000005e-05, |
| "loss": 0.16743293762207032, |
| "step": 35650 |
| }, |
| { |
| "epoch": 0.7285714285714285, |
| "grad_norm": 0.1379423886537552, |
| "learning_rate": 5.5454166666666675e-05, |
| "loss": 0.15212342262268067, |
| "step": 35700 |
| }, |
| { |
| "epoch": 0.7295918367346939, |
| "grad_norm": 0.15476073324680328, |
| "learning_rate": 5.524583333333334e-05, |
| "loss": 0.16397836685180664, |
| "step": 35750 |
| }, |
| { |
| "epoch": 0.7306122448979592, |
| "grad_norm": 0.09884586185216904, |
| "learning_rate": 5.5037499999999996e-05, |
| "loss": 0.14825711250305176, |
| "step": 35800 |
| }, |
| { |
| "epoch": 0.7316326530612245, |
| "grad_norm": 0.13884589076042175, |
| "learning_rate": 5.482916666666667e-05, |
| "loss": 0.14152326583862304, |
| "step": 35850 |
| }, |
| { |
| "epoch": 0.7326530612244898, |
| "grad_norm": 0.21771541237831116, |
| "learning_rate": 5.462083333333333e-05, |
| "loss": 0.15861477851867675, |
| "step": 35900 |
| }, |
| { |
| "epoch": 0.7336734693877551, |
| "grad_norm": 0.14293690025806427, |
| "learning_rate": 5.44125e-05, |
| "loss": 0.15552971839904786, |
| "step": 35950 |
| }, |
| { |
| "epoch": 0.7346938775510204, |
| "grad_norm": 0.25441673398017883, |
| "learning_rate": 5.4204166666666665e-05, |
| "loss": 0.16837917327880858, |
| "step": 36000 |
| }, |
| { |
| "epoch": 0.7357142857142858, |
| "grad_norm": 0.15737579762935638, |
| "learning_rate": 5.3995833333333336e-05, |
| "loss": 0.15274593353271484, |
| "step": 36050 |
| }, |
| { |
| "epoch": 0.736734693877551, |
| "grad_norm": 0.15187661349773407, |
| "learning_rate": 5.37875e-05, |
| "loss": 0.1639406967163086, |
| "step": 36100 |
| }, |
| { |
| "epoch": 0.7377551020408163, |
| "grad_norm": 0.24913407862186432, |
| "learning_rate": 5.357916666666667e-05, |
| "loss": 0.12389989852905274, |
| "step": 36150 |
| }, |
| { |
| "epoch": 0.7387755102040816, |
| "grad_norm": 0.17340479791164398, |
| "learning_rate": 5.3370833333333334e-05, |
| "loss": 0.2008836555480957, |
| "step": 36200 |
| }, |
| { |
| "epoch": 0.7397959183673469, |
| "grad_norm": 0.2266087830066681, |
| "learning_rate": 5.3162500000000005e-05, |
| "loss": 0.15062840461730956, |
| "step": 36250 |
| }, |
| { |
| "epoch": 0.7408163265306122, |
| "grad_norm": 0.1296575516462326, |
| "learning_rate": 5.295416666666667e-05, |
| "loss": 0.14468287467956542, |
| "step": 36300 |
| }, |
| { |
| "epoch": 0.7418367346938776, |
| "grad_norm": 0.17865030467510223, |
| "learning_rate": 5.274583333333334e-05, |
| "loss": 0.13884951591491698, |
| "step": 36350 |
| }, |
| { |
| "epoch": 0.7428571428571429, |
| "grad_norm": 0.16539889574050903, |
| "learning_rate": 5.25375e-05, |
| "loss": 0.1667474365234375, |
| "step": 36400 |
| }, |
| { |
| "epoch": 0.7438775510204082, |
| "grad_norm": 0.17142266035079956, |
| "learning_rate": 5.2329166666666674e-05, |
| "loss": 0.14144489288330078, |
| "step": 36450 |
| }, |
| { |
| "epoch": 0.7448979591836735, |
| "grad_norm": 0.08687719702720642, |
| "learning_rate": 5.212083333333334e-05, |
| "loss": 0.14284836769104003, |
| "step": 36500 |
| }, |
| { |
| "epoch": 0.7459183673469387, |
| "grad_norm": 0.11344967782497406, |
| "learning_rate": 5.1912499999999995e-05, |
| "loss": 0.1589106559753418, |
| "step": 36550 |
| }, |
| { |
| "epoch": 0.746938775510204, |
| "grad_norm": 0.126994788646698, |
| "learning_rate": 5.1704166666666665e-05, |
| "loss": 0.1657052993774414, |
| "step": 36600 |
| }, |
| { |
| "epoch": 0.7479591836734694, |
| "grad_norm": 0.11640670150518417, |
| "learning_rate": 5.149583333333333e-05, |
| "loss": 0.15780337333679198, |
| "step": 36650 |
| }, |
| { |
| "epoch": 0.7489795918367347, |
| "grad_norm": 0.14036226272583008, |
| "learning_rate": 5.12875e-05, |
| "loss": 0.1352565383911133, |
| "step": 36700 |
| }, |
| { |
| "epoch": 0.75, |
| "grad_norm": 0.18937057256698608, |
| "learning_rate": 5.108333333333334e-05, |
| "loss": 0.15480207443237304, |
| "step": 36750 |
| }, |
| { |
| "epoch": 0.7510204081632653, |
| "grad_norm": 0.0924379974603653, |
| "learning_rate": 5.0875e-05, |
| "loss": 0.1446268367767334, |
| "step": 36800 |
| }, |
| { |
| "epoch": 0.7520408163265306, |
| "grad_norm": 0.19327092170715332, |
| "learning_rate": 5.0666666666666674e-05, |
| "loss": 0.13928128242492677, |
| "step": 36850 |
| }, |
| { |
| "epoch": 0.753061224489796, |
| "grad_norm": 0.17900340259075165, |
| "learning_rate": 5.045833333333334e-05, |
| "loss": 0.1468774127960205, |
| "step": 36900 |
| }, |
| { |
| "epoch": 0.7540816326530613, |
| "grad_norm": 0.11567780375480652, |
| "learning_rate": 5.0249999999999995e-05, |
| "loss": 0.14537551879882812, |
| "step": 36950 |
| }, |
| { |
| "epoch": 0.7551020408163265, |
| "grad_norm": 0.132771298289299, |
| "learning_rate": 5.0041666666666666e-05, |
| "loss": 0.18957576751708985, |
| "step": 37000 |
| }, |
| { |
| "epoch": 0.7561224489795918, |
| "grad_norm": 0.2857951819896698, |
| "learning_rate": 4.9833333333333336e-05, |
| "loss": 0.14247038841247558, |
| "step": 37050 |
| }, |
| { |
| "epoch": 0.7571428571428571, |
| "grad_norm": 0.24694068729877472, |
| "learning_rate": 4.962500000000001e-05, |
| "loss": 0.16788265228271484, |
| "step": 37100 |
| }, |
| { |
| "epoch": 0.7581632653061224, |
| "grad_norm": 0.14023227989673615, |
| "learning_rate": 4.9416666666666664e-05, |
| "loss": 0.1499267864227295, |
| "step": 37150 |
| }, |
| { |
| "epoch": 0.7591836734693878, |
| "grad_norm": 0.22128233313560486, |
| "learning_rate": 4.9208333333333335e-05, |
| "loss": 0.1711456298828125, |
| "step": 37200 |
| }, |
| { |
| "epoch": 0.7602040816326531, |
| "grad_norm": 0.16241143643856049, |
| "learning_rate": 4.9e-05, |
| "loss": 0.14633071899414063, |
| "step": 37250 |
| }, |
| { |
| "epoch": 0.7612244897959184, |
| "grad_norm": 0.0808044821023941, |
| "learning_rate": 4.879166666666667e-05, |
| "loss": 0.16124568939208983, |
| "step": 37300 |
| }, |
| { |
| "epoch": 0.7622448979591837, |
| "grad_norm": 0.16860656440258026, |
| "learning_rate": 4.858333333333333e-05, |
| "loss": 0.11210680961608886, |
| "step": 37350 |
| }, |
| { |
| "epoch": 0.763265306122449, |
| "grad_norm": 0.10932660102844238, |
| "learning_rate": 4.8375000000000004e-05, |
| "loss": 0.1334698486328125, |
| "step": 37400 |
| }, |
| { |
| "epoch": 0.7642857142857142, |
| "grad_norm": 0.16516442596912384, |
| "learning_rate": 4.8166666666666674e-05, |
| "loss": 0.15440690994262696, |
| "step": 37450 |
| }, |
| { |
| "epoch": 0.7653061224489796, |
| "grad_norm": 0.17629872262477875, |
| "learning_rate": 4.795833333333333e-05, |
| "loss": 0.15302536010742188, |
| "step": 37500 |
| }, |
| { |
| "epoch": 0.7663265306122449, |
| "grad_norm": 0.1887020468711853, |
| "learning_rate": 4.775e-05, |
| "loss": 0.18522512435913085, |
| "step": 37550 |
| }, |
| { |
| "epoch": 0.7673469387755102, |
| "grad_norm": 0.2348717302083969, |
| "learning_rate": 4.7541666666666666e-05, |
| "loss": 0.15498061180114747, |
| "step": 37600 |
| }, |
| { |
| "epoch": 0.7683673469387755, |
| "grad_norm": 0.12663526833057404, |
| "learning_rate": 4.73375e-05, |
| "loss": 0.17798063278198242, |
| "step": 37650 |
| }, |
| { |
| "epoch": 0.7693877551020408, |
| "grad_norm": 0.09127508103847504, |
| "learning_rate": 4.712916666666667e-05, |
| "loss": 0.15106979370117188, |
| "step": 37700 |
| }, |
| { |
| "epoch": 0.7704081632653061, |
| "grad_norm": 0.1976308673620224, |
| "learning_rate": 4.692083333333333e-05, |
| "loss": 0.1309073066711426, |
| "step": 37750 |
| }, |
| { |
| "epoch": 0.7714285714285715, |
| "grad_norm": 0.17086738348007202, |
| "learning_rate": 4.6712500000000004e-05, |
| "loss": 0.14809440612792968, |
| "step": 37800 |
| }, |
| { |
| "epoch": 0.7724489795918368, |
| "grad_norm": 0.19537247717380524, |
| "learning_rate": 4.650416666666667e-05, |
| "loss": 0.1556560516357422, |
| "step": 37850 |
| }, |
| { |
| "epoch": 0.773469387755102, |
| "grad_norm": 0.18661127984523773, |
| "learning_rate": 4.629583333333334e-05, |
| "loss": 0.1466090488433838, |
| "step": 37900 |
| }, |
| { |
| "epoch": 0.7744897959183673, |
| "grad_norm": 0.17013102769851685, |
| "learning_rate": 4.60875e-05, |
| "loss": 0.1757086181640625, |
| "step": 37950 |
| }, |
| { |
| "epoch": 0.7755102040816326, |
| "grad_norm": 0.1524754762649536, |
| "learning_rate": 4.5879166666666666e-05, |
| "loss": 0.16644367218017578, |
| "step": 38000 |
| }, |
| { |
| "epoch": 0.7765306122448979, |
| "grad_norm": 0.1900121122598648, |
| "learning_rate": 4.567083333333334e-05, |
| "loss": 0.16857398986816408, |
| "step": 38050 |
| }, |
| { |
| "epoch": 0.7775510204081633, |
| "grad_norm": 0.10611417144536972, |
| "learning_rate": 4.54625e-05, |
| "loss": 0.14672259330749512, |
| "step": 38100 |
| }, |
| { |
| "epoch": 0.7785714285714286, |
| "grad_norm": 0.18626046180725098, |
| "learning_rate": 4.525416666666667e-05, |
| "loss": 0.18764514923095704, |
| "step": 38150 |
| }, |
| { |
| "epoch": 0.7795918367346939, |
| "grad_norm": 0.24131523072719574, |
| "learning_rate": 4.5045833333333335e-05, |
| "loss": 0.1816444206237793, |
| "step": 38200 |
| }, |
| { |
| "epoch": 0.7806122448979592, |
| "grad_norm": 0.17299646139144897, |
| "learning_rate": 4.4837500000000006e-05, |
| "loss": 0.15335065841674805, |
| "step": 38250 |
| }, |
| { |
| "epoch": 0.7816326530612245, |
| "grad_norm": 0.17390669882297516, |
| "learning_rate": 4.462916666666667e-05, |
| "loss": 0.18706804275512695, |
| "step": 38300 |
| }, |
| { |
| "epoch": 0.7826530612244897, |
| "grad_norm": 0.09298673272132874, |
| "learning_rate": 4.442083333333333e-05, |
| "loss": 0.18538618087768555, |
| "step": 38350 |
| }, |
| { |
| "epoch": 0.7836734693877551, |
| "grad_norm": 0.15914246439933777, |
| "learning_rate": 4.4212500000000004e-05, |
| "loss": 0.1349315643310547, |
| "step": 38400 |
| }, |
| { |
| "epoch": 0.7846938775510204, |
| "grad_norm": 0.08660084754228592, |
| "learning_rate": 4.400416666666667e-05, |
| "loss": 0.12608378410339355, |
| "step": 38450 |
| }, |
| { |
| "epoch": 0.7857142857142857, |
| "grad_norm": 0.21280916035175323, |
| "learning_rate": 4.379583333333334e-05, |
| "loss": 0.20869346618652343, |
| "step": 38500 |
| }, |
| { |
| "epoch": 0.786734693877551, |
| "grad_norm": 0.11491119861602783, |
| "learning_rate": 4.35875e-05, |
| "loss": 0.1711391258239746, |
| "step": 38550 |
| }, |
| { |
| "epoch": 0.7877551020408163, |
| "grad_norm": 0.154352068901062, |
| "learning_rate": 4.337916666666667e-05, |
| "loss": 0.1766732406616211, |
| "step": 38600 |
| }, |
| { |
| "epoch": 0.7887755102040817, |
| "grad_norm": 0.16147321462631226, |
| "learning_rate": 4.317083333333334e-05, |
| "loss": 0.16638633728027344, |
| "step": 38650 |
| }, |
| { |
| "epoch": 0.789795918367347, |
| "grad_norm": 0.08217877149581909, |
| "learning_rate": 4.29625e-05, |
| "loss": 0.14051429748535157, |
| "step": 38700 |
| }, |
| { |
| "epoch": 0.7908163265306123, |
| "grad_norm": 0.4309787154197693, |
| "learning_rate": 4.2754166666666665e-05, |
| "loss": 0.15030960083007813, |
| "step": 38750 |
| }, |
| { |
| "epoch": 0.7918367346938775, |
| "grad_norm": 0.08852960914373398, |
| "learning_rate": 4.2545833333333335e-05, |
| "loss": 0.15894084930419922, |
| "step": 38800 |
| }, |
| { |
| "epoch": 0.7928571428571428, |
| "grad_norm": 0.20330609381198883, |
| "learning_rate": 4.23375e-05, |
| "loss": 0.16075292587280274, |
| "step": 38850 |
| }, |
| { |
| "epoch": 0.7938775510204081, |
| "grad_norm": 0.15123698115348816, |
| "learning_rate": 4.212916666666667e-05, |
| "loss": 0.14566773414611817, |
| "step": 38900 |
| }, |
| { |
| "epoch": 0.7948979591836735, |
| "grad_norm": 0.09001250565052032, |
| "learning_rate": 4.1920833333333334e-05, |
| "loss": 0.15867124557495116, |
| "step": 38950 |
| }, |
| { |
| "epoch": 0.7959183673469388, |
| "grad_norm": 0.1784236878156662, |
| "learning_rate": 4.1712500000000004e-05, |
| "loss": 0.1991616439819336, |
| "step": 39000 |
| }, |
| { |
| "epoch": 0.7969387755102041, |
| "grad_norm": 0.15025418996810913, |
| "learning_rate": 4.150416666666667e-05, |
| "loss": 0.151847505569458, |
| "step": 39050 |
| }, |
| { |
| "epoch": 0.7979591836734694, |
| "grad_norm": 0.21809156239032745, |
| "learning_rate": 4.129583333333333e-05, |
| "loss": 0.15395938873291015, |
| "step": 39100 |
| }, |
| { |
| "epoch": 0.7989795918367347, |
| "grad_norm": 0.19765830039978027, |
| "learning_rate": 4.10875e-05, |
| "loss": 0.16577257156372072, |
| "step": 39150 |
| }, |
| { |
| "epoch": 0.8, |
| "grad_norm": 0.12601009011268616, |
| "learning_rate": 4.0879166666666666e-05, |
| "loss": 0.17003007888793945, |
| "step": 39200 |
| }, |
| { |
| "epoch": 0.8010204081632653, |
| "grad_norm": 0.17706438899040222, |
| "learning_rate": 4.067083333333334e-05, |
| "loss": 0.18737960815429688, |
| "step": 39250 |
| }, |
| { |
| "epoch": 0.8020408163265306, |
| "grad_norm": 0.09074155986309052, |
| "learning_rate": 4.04625e-05, |
| "loss": 0.17569421768188476, |
| "step": 39300 |
| }, |
| { |
| "epoch": 0.8030612244897959, |
| "grad_norm": 0.12752097845077515, |
| "learning_rate": 4.025416666666667e-05, |
| "loss": 0.14584141731262207, |
| "step": 39350 |
| }, |
| { |
| "epoch": 0.8040816326530612, |
| "grad_norm": 0.18636369705200195, |
| "learning_rate": 4.0045833333333335e-05, |
| "loss": 0.1786494255065918, |
| "step": 39400 |
| }, |
| { |
| "epoch": 0.8051020408163265, |
| "grad_norm": 0.07327538728713989, |
| "learning_rate": 3.98375e-05, |
| "loss": 0.17225727081298828, |
| "step": 39450 |
| }, |
| { |
| "epoch": 0.8061224489795918, |
| "grad_norm": 0.14465896785259247, |
| "learning_rate": 3.962916666666667e-05, |
| "loss": 0.15083393096923828, |
| "step": 39500 |
| }, |
| { |
| "epoch": 0.8071428571428572, |
| "grad_norm": 0.2394832968711853, |
| "learning_rate": 3.9420833333333334e-05, |
| "loss": 0.19293224334716796, |
| "step": 39550 |
| }, |
| { |
| "epoch": 0.8081632653061225, |
| "grad_norm": 0.18358109891414642, |
| "learning_rate": 3.9212500000000004e-05, |
| "loss": 0.14710227966308595, |
| "step": 39600 |
| }, |
| { |
| "epoch": 0.8091836734693878, |
| "grad_norm": 0.18446484208106995, |
| "learning_rate": 3.900416666666667e-05, |
| "loss": 0.12325849533081054, |
| "step": 39650 |
| }, |
| { |
| "epoch": 0.810204081632653, |
| "grad_norm": 0.14655612409114838, |
| "learning_rate": 3.879583333333334e-05, |
| "loss": 0.1633004951477051, |
| "step": 39700 |
| }, |
| { |
| "epoch": 0.8112244897959183, |
| "grad_norm": 0.07564426213502884, |
| "learning_rate": 3.85875e-05, |
| "loss": 0.18752107620239258, |
| "step": 39750 |
| }, |
| { |
| "epoch": 0.8122448979591836, |
| "grad_norm": 0.11330200731754303, |
| "learning_rate": 3.837916666666667e-05, |
| "loss": 0.1585384750366211, |
| "step": 39800 |
| }, |
| { |
| "epoch": 0.813265306122449, |
| "grad_norm": 0.09287655353546143, |
| "learning_rate": 3.817083333333333e-05, |
| "loss": 0.13122535705566407, |
| "step": 39850 |
| }, |
| { |
| "epoch": 0.8142857142857143, |
| "grad_norm": 0.2616575360298157, |
| "learning_rate": 3.79625e-05, |
| "loss": 0.18795604705810548, |
| "step": 39900 |
| }, |
| { |
| "epoch": 0.8153061224489796, |
| "grad_norm": 0.11491675674915314, |
| "learning_rate": 3.7754166666666665e-05, |
| "loss": 0.13429009437561035, |
| "step": 39950 |
| }, |
| { |
| "epoch": 0.8163265306122449, |
| "grad_norm": 0.1451539248228073, |
| "learning_rate": 3.7545833333333336e-05, |
| "loss": 0.13625697135925294, |
| "step": 40000 |
| }, |
| { |
| "epoch": 0.8173469387755102, |
| "grad_norm": 0.2314739227294922, |
| "learning_rate": 3.7337500000000006e-05, |
| "loss": 0.16612844467163085, |
| "step": 40050 |
| }, |
| { |
| "epoch": 0.8183673469387756, |
| "grad_norm": 0.19042688608169556, |
| "learning_rate": 3.712916666666667e-05, |
| "loss": 0.1493498992919922, |
| "step": 40100 |
| }, |
| { |
| "epoch": 0.8193877551020409, |
| "grad_norm": 0.2371101975440979, |
| "learning_rate": 3.692083333333334e-05, |
| "loss": 0.17770370483398437, |
| "step": 40150 |
| }, |
| { |
| "epoch": 0.8204081632653061, |
| "grad_norm": 0.1424565613269806, |
| "learning_rate": 3.67125e-05, |
| "loss": 0.18499235153198243, |
| "step": 40200 |
| }, |
| { |
| "epoch": 0.8214285714285714, |
| "grad_norm": 0.1678062230348587, |
| "learning_rate": 3.650416666666667e-05, |
| "loss": 0.140090913772583, |
| "step": 40250 |
| }, |
| { |
| "epoch": 0.8224489795918367, |
| "grad_norm": 0.16148346662521362, |
| "learning_rate": 3.629583333333333e-05, |
| "loss": 0.19844322204589843, |
| "step": 40300 |
| }, |
| { |
| "epoch": 0.823469387755102, |
| "grad_norm": 0.1150701716542244, |
| "learning_rate": 3.60875e-05, |
| "loss": 0.1799906921386719, |
| "step": 40350 |
| }, |
| { |
| "epoch": 0.8244897959183674, |
| "grad_norm": 0.14582528173923492, |
| "learning_rate": 3.587916666666667e-05, |
| "loss": 0.1387934684753418, |
| "step": 40400 |
| }, |
| { |
| "epoch": 0.8255102040816327, |
| "grad_norm": 0.17526212334632874, |
| "learning_rate": 3.567083333333334e-05, |
| "loss": 0.13370927810668945, |
| "step": 40450 |
| }, |
| { |
| "epoch": 0.826530612244898, |
| "grad_norm": 0.13824090361595154, |
| "learning_rate": 3.54625e-05, |
| "loss": 0.17501338958740234, |
| "step": 40500 |
| }, |
| { |
| "epoch": 0.8275510204081633, |
| "grad_norm": 0.18103674054145813, |
| "learning_rate": 3.5254166666666665e-05, |
| "loss": 0.16202123641967772, |
| "step": 40550 |
| }, |
| { |
| "epoch": 0.8285714285714286, |
| "grad_norm": 0.10612452030181885, |
| "learning_rate": 3.5045833333333336e-05, |
| "loss": 0.1486236572265625, |
| "step": 40600 |
| }, |
| { |
| "epoch": 0.8295918367346938, |
| "grad_norm": 0.18228718638420105, |
| "learning_rate": 3.48375e-05, |
| "loss": 0.1618939781188965, |
| "step": 40650 |
| }, |
| { |
| "epoch": 0.8306122448979592, |
| "grad_norm": 0.2844036817550659, |
| "learning_rate": 3.462916666666667e-05, |
| "loss": 0.19489418029785155, |
| "step": 40700 |
| }, |
| { |
| "epoch": 0.8316326530612245, |
| "grad_norm": 0.17485639452934265, |
| "learning_rate": 3.4420833333333334e-05, |
| "loss": 0.16676589965820313, |
| "step": 40750 |
| }, |
| { |
| "epoch": 0.8326530612244898, |
| "grad_norm": 0.23534175753593445, |
| "learning_rate": 3.4212500000000005e-05, |
| "loss": 0.1604459762573242, |
| "step": 40800 |
| }, |
| { |
| "epoch": 0.8336734693877551, |
| "grad_norm": 0.31001633405685425, |
| "learning_rate": 3.400416666666667e-05, |
| "loss": 0.19572820663452148, |
| "step": 40850 |
| }, |
| { |
| "epoch": 0.8346938775510204, |
| "grad_norm": 0.14709772169589996, |
| "learning_rate": 3.379583333333334e-05, |
| "loss": 0.15393299102783203, |
| "step": 40900 |
| }, |
| { |
| "epoch": 0.8357142857142857, |
| "grad_norm": 0.2015165537595749, |
| "learning_rate": 3.3587499999999996e-05, |
| "loss": 0.16096088409423828, |
| "step": 40950 |
| }, |
| { |
| "epoch": 0.8367346938775511, |
| "grad_norm": 0.1936115324497223, |
| "learning_rate": 3.337916666666667e-05, |
| "loss": 0.13105152130126954, |
| "step": 41000 |
| }, |
| { |
| "epoch": 0.8377551020408164, |
| "grad_norm": 0.29701679944992065, |
| "learning_rate": 3.317083333333333e-05, |
| "loss": 0.15563657760620117, |
| "step": 41050 |
| }, |
| { |
| "epoch": 0.8387755102040816, |
| "grad_norm": 0.18862281739711761, |
| "learning_rate": 3.29625e-05, |
| "loss": 0.1588943386077881, |
| "step": 41100 |
| }, |
| { |
| "epoch": 0.8397959183673469, |
| "grad_norm": 0.19941005110740662, |
| "learning_rate": 3.275416666666667e-05, |
| "loss": 0.15184050559997558, |
| "step": 41150 |
| }, |
| { |
| "epoch": 0.8408163265306122, |
| "grad_norm": 0.16846665740013123, |
| "learning_rate": 3.2545833333333336e-05, |
| "loss": 0.19719802856445312, |
| "step": 41200 |
| }, |
| { |
| "epoch": 0.8418367346938775, |
| "grad_norm": 0.16473744809627533, |
| "learning_rate": 3.233750000000001e-05, |
| "loss": 0.17479293823242187, |
| "step": 41250 |
| }, |
| { |
| "epoch": 0.8428571428571429, |
| "grad_norm": 0.1598232537508011, |
| "learning_rate": 3.2129166666666664e-05, |
| "loss": 0.15978538513183593, |
| "step": 41300 |
| }, |
| { |
| "epoch": 0.8438775510204082, |
| "grad_norm": 0.17177850008010864, |
| "learning_rate": 3.1920833333333334e-05, |
| "loss": 0.1575559616088867, |
| "step": 41350 |
| }, |
| { |
| "epoch": 0.8448979591836735, |
| "grad_norm": 0.1356944888830185, |
| "learning_rate": 3.17125e-05, |
| "loss": 0.1598154830932617, |
| "step": 41400 |
| }, |
| { |
| "epoch": 0.8459183673469388, |
| "grad_norm": 0.12465784698724747, |
| "learning_rate": 3.150416666666667e-05, |
| "loss": 0.15516016960144044, |
| "step": 41450 |
| }, |
| { |
| "epoch": 0.8469387755102041, |
| "grad_norm": 0.2187613993883133, |
| "learning_rate": 3.129583333333333e-05, |
| "loss": 0.1535673427581787, |
| "step": 41500 |
| }, |
| { |
| "epoch": 0.8479591836734693, |
| "grad_norm": 0.134515643119812, |
| "learning_rate": 3.1087500000000003e-05, |
| "loss": 0.15695632934570314, |
| "step": 41550 |
| }, |
| { |
| "epoch": 0.8489795918367347, |
| "grad_norm": 0.19475696980953217, |
| "learning_rate": 3.087916666666667e-05, |
| "loss": 0.15306434631347657, |
| "step": 41600 |
| }, |
| { |
| "epoch": 0.85, |
| "grad_norm": 0.16016463935375214, |
| "learning_rate": 3.067500000000001e-05, |
| "loss": 0.1699496841430664, |
| "step": 41650 |
| }, |
| { |
| "epoch": 0.8510204081632653, |
| "grad_norm": 0.22221848368644714, |
| "learning_rate": 3.0466666666666664e-05, |
| "loss": 0.22820037841796875, |
| "step": 41700 |
| }, |
| { |
| "epoch": 0.8520408163265306, |
| "grad_norm": 0.13827036321163177, |
| "learning_rate": 3.025833333333333e-05, |
| "loss": 0.19656036376953126, |
| "step": 41750 |
| }, |
| { |
| "epoch": 0.8530612244897959, |
| "grad_norm": 0.19884911179542542, |
| "learning_rate": 3.0050000000000002e-05, |
| "loss": 0.17680858612060546, |
| "step": 41800 |
| }, |
| { |
| "epoch": 0.8540816326530613, |
| "grad_norm": 0.20734833180904388, |
| "learning_rate": 2.984166666666667e-05, |
| "loss": 0.2142172050476074, |
| "step": 41850 |
| }, |
| { |
| "epoch": 0.8551020408163266, |
| "grad_norm": 0.25795796513557434, |
| "learning_rate": 2.9633333333333336e-05, |
| "loss": 0.21104377746582031, |
| "step": 41900 |
| }, |
| { |
| "epoch": 0.8561224489795919, |
| "grad_norm": 0.33637145161628723, |
| "learning_rate": 2.9425000000000004e-05, |
| "loss": 0.20429901123046876, |
| "step": 41950 |
| }, |
| { |
| "epoch": 0.8571428571428571, |
| "grad_norm": 0.16273394227027893, |
| "learning_rate": 2.921666666666667e-05, |
| "loss": 0.17451427459716798, |
| "step": 42000 |
| }, |
| { |
| "epoch": 0.8581632653061224, |
| "grad_norm": 0.1235772967338562, |
| "learning_rate": 2.9008333333333338e-05, |
| "loss": 0.12104739189147949, |
| "step": 42050 |
| }, |
| { |
| "epoch": 0.8591836734693877, |
| "grad_norm": 0.20922167599201202, |
| "learning_rate": 2.88e-05, |
| "loss": 0.1750694465637207, |
| "step": 42100 |
| }, |
| { |
| "epoch": 0.860204081632653, |
| "grad_norm": 0.15653084218502045, |
| "learning_rate": 2.8591666666666666e-05, |
| "loss": 0.1647883415222168, |
| "step": 42150 |
| }, |
| { |
| "epoch": 0.8612244897959184, |
| "grad_norm": 0.21672451496124268, |
| "learning_rate": 2.8383333333333333e-05, |
| "loss": 0.1777618408203125, |
| "step": 42200 |
| }, |
| { |
| "epoch": 0.8622448979591837, |
| "grad_norm": 0.16229403018951416, |
| "learning_rate": 2.8175e-05, |
| "loss": 0.14013175010681153, |
| "step": 42250 |
| }, |
| { |
| "epoch": 0.863265306122449, |
| "grad_norm": 0.23231209814548492, |
| "learning_rate": 2.7966666666666668e-05, |
| "loss": 0.1481611442565918, |
| "step": 42300 |
| }, |
| { |
| "epoch": 0.8642857142857143, |
| "grad_norm": 0.11112941056489944, |
| "learning_rate": 2.7758333333333335e-05, |
| "loss": 0.1890835189819336, |
| "step": 42350 |
| }, |
| { |
| "epoch": 0.8653061224489796, |
| "grad_norm": 0.2170908898115158, |
| "learning_rate": 2.7550000000000002e-05, |
| "loss": 0.16120462417602538, |
| "step": 42400 |
| }, |
| { |
| "epoch": 0.8663265306122448, |
| "grad_norm": 0.1535445749759674, |
| "learning_rate": 2.7341666666666666e-05, |
| "loss": 0.15845059394836425, |
| "step": 42450 |
| }, |
| { |
| "epoch": 0.8673469387755102, |
| "grad_norm": 0.1733650416135788, |
| "learning_rate": 2.7133333333333333e-05, |
| "loss": 0.1984349822998047, |
| "step": 42500 |
| }, |
| { |
| "epoch": 0.8683673469387755, |
| "grad_norm": 0.13763122260570526, |
| "learning_rate": 2.6925e-05, |
| "loss": 0.17453062057495117, |
| "step": 42550 |
| }, |
| { |
| "epoch": 0.8693877551020408, |
| "grad_norm": 0.06716243177652359, |
| "learning_rate": 2.6716666666666668e-05, |
| "loss": 0.12827292442321778, |
| "step": 42600 |
| }, |
| { |
| "epoch": 0.8704081632653061, |
| "grad_norm": 0.06482294201850891, |
| "learning_rate": 2.6508333333333335e-05, |
| "loss": 0.14777050971984862, |
| "step": 42650 |
| }, |
| { |
| "epoch": 0.8714285714285714, |
| "grad_norm": 0.11893775314092636, |
| "learning_rate": 2.6300000000000002e-05, |
| "loss": 0.14279632568359374, |
| "step": 42700 |
| }, |
| { |
| "epoch": 0.8724489795918368, |
| "grad_norm": 0.11864366382360458, |
| "learning_rate": 2.609166666666667e-05, |
| "loss": 0.15069831848144533, |
| "step": 42750 |
| }, |
| { |
| "epoch": 0.8734693877551021, |
| "grad_norm": 0.11892051994800568, |
| "learning_rate": 2.5883333333333337e-05, |
| "loss": 0.17250820159912109, |
| "step": 42800 |
| }, |
| { |
| "epoch": 0.8744897959183674, |
| "grad_norm": 0.18396931886672974, |
| "learning_rate": 2.5675e-05, |
| "loss": 0.14836503982543944, |
| "step": 42850 |
| }, |
| { |
| "epoch": 0.8755102040816326, |
| "grad_norm": 0.1566888689994812, |
| "learning_rate": 2.5466666666666668e-05, |
| "loss": 0.16654870986938478, |
| "step": 42900 |
| }, |
| { |
| "epoch": 0.8765306122448979, |
| "grad_norm": 0.0947994589805603, |
| "learning_rate": 2.5258333333333335e-05, |
| "loss": 0.1264040946960449, |
| "step": 42950 |
| }, |
| { |
| "epoch": 0.8775510204081632, |
| "grad_norm": 0.18539394438266754, |
| "learning_rate": 2.5050000000000002e-05, |
| "loss": 0.1276088047027588, |
| "step": 43000 |
| }, |
| { |
| "epoch": 0.8785714285714286, |
| "grad_norm": 0.15481586754322052, |
| "learning_rate": 2.484166666666667e-05, |
| "loss": 0.17538042068481446, |
| "step": 43050 |
| }, |
| { |
| "epoch": 0.8795918367346939, |
| "grad_norm": 0.2394324243068695, |
| "learning_rate": 2.4633333333333334e-05, |
| "loss": 0.14613781929016112, |
| "step": 43100 |
| }, |
| { |
| "epoch": 0.8806122448979592, |
| "grad_norm": 0.1770860105752945, |
| "learning_rate": 2.4425e-05, |
| "loss": 0.14055556297302246, |
| "step": 43150 |
| }, |
| { |
| "epoch": 0.8816326530612245, |
| "grad_norm": 0.26251518726348877, |
| "learning_rate": 2.4216666666666668e-05, |
| "loss": 0.18869239807128907, |
| "step": 43200 |
| }, |
| { |
| "epoch": 0.8826530612244898, |
| "grad_norm": 0.24112340807914734, |
| "learning_rate": 2.4008333333333335e-05, |
| "loss": 0.16565069198608398, |
| "step": 43250 |
| }, |
| { |
| "epoch": 0.8836734693877552, |
| "grad_norm": 0.18587692081928253, |
| "learning_rate": 2.38e-05, |
| "loss": 0.1443570613861084, |
| "step": 43300 |
| }, |
| { |
| "epoch": 0.8846938775510204, |
| "grad_norm": 0.13177326321601868, |
| "learning_rate": 2.3591666666666666e-05, |
| "loss": 0.1775250816345215, |
| "step": 43350 |
| }, |
| { |
| "epoch": 0.8857142857142857, |
| "grad_norm": 0.2826145589351654, |
| "learning_rate": 2.3383333333333334e-05, |
| "loss": 0.19880430221557618, |
| "step": 43400 |
| }, |
| { |
| "epoch": 0.886734693877551, |
| "grad_norm": 0.15196163952350616, |
| "learning_rate": 2.3175e-05, |
| "loss": 0.1795181465148926, |
| "step": 43450 |
| }, |
| { |
| "epoch": 0.8877551020408163, |
| "grad_norm": 0.1676856428384781, |
| "learning_rate": 2.2966666666666668e-05, |
| "loss": 0.11898002624511719, |
| "step": 43500 |
| }, |
| { |
| "epoch": 0.8887755102040816, |
| "grad_norm": 0.22324547171592712, |
| "learning_rate": 2.2758333333333335e-05, |
| "loss": 0.15103040695190428, |
| "step": 43550 |
| }, |
| { |
| "epoch": 0.889795918367347, |
| "grad_norm": 0.18912465870380402, |
| "learning_rate": 2.2550000000000003e-05, |
| "loss": 0.17227333068847656, |
| "step": 43600 |
| }, |
| { |
| "epoch": 0.8908163265306123, |
| "grad_norm": 0.13828060030937195, |
| "learning_rate": 2.234166666666667e-05, |
| "loss": 0.1739276695251465, |
| "step": 43650 |
| }, |
| { |
| "epoch": 0.8918367346938776, |
| "grad_norm": 0.17106176912784576, |
| "learning_rate": 2.21375e-05, |
| "loss": 0.1273856544494629, |
| "step": 43700 |
| }, |
| { |
| "epoch": 0.8928571428571429, |
| "grad_norm": 0.17478223145008087, |
| "learning_rate": 2.1929166666666667e-05, |
| "loss": 0.11760704040527344, |
| "step": 43750 |
| }, |
| { |
| "epoch": 0.8938775510204081, |
| "grad_norm": 0.15052562952041626, |
| "learning_rate": 2.1720833333333334e-05, |
| "loss": 0.16121845245361327, |
| "step": 43800 |
| }, |
| { |
| "epoch": 0.8948979591836734, |
| "grad_norm": 0.2820034921169281, |
| "learning_rate": 2.15125e-05, |
| "loss": 0.1849671173095703, |
| "step": 43850 |
| }, |
| { |
| "epoch": 0.8959183673469387, |
| "grad_norm": 0.18933306634426117, |
| "learning_rate": 2.130416666666667e-05, |
| "loss": 0.19127286911010744, |
| "step": 43900 |
| }, |
| { |
| "epoch": 0.8969387755102041, |
| "grad_norm": 0.1372126191854477, |
| "learning_rate": 2.1095833333333336e-05, |
| "loss": 0.18945589065551757, |
| "step": 43950 |
| }, |
| { |
| "epoch": 0.8979591836734694, |
| "grad_norm": 0.22014056146144867, |
| "learning_rate": 2.0887500000000003e-05, |
| "loss": 0.1549460220336914, |
| "step": 44000 |
| }, |
| { |
| "epoch": 0.8989795918367347, |
| "grad_norm": 0.12489187717437744, |
| "learning_rate": 2.067916666666667e-05, |
| "loss": 0.1980477523803711, |
| "step": 44050 |
| }, |
| { |
| "epoch": 0.9, |
| "grad_norm": 0.19070874154567719, |
| "learning_rate": 2.0470833333333334e-05, |
| "loss": 0.12702374458312987, |
| "step": 44100 |
| }, |
| { |
| "epoch": 0.9010204081632653, |
| "grad_norm": 0.16794297099113464, |
| "learning_rate": 2.02625e-05, |
| "loss": 0.11833565711975097, |
| "step": 44150 |
| }, |
| { |
| "epoch": 0.9020408163265307, |
| "grad_norm": 0.2177940309047699, |
| "learning_rate": 2.005416666666667e-05, |
| "loss": 0.13561912536621093, |
| "step": 44200 |
| }, |
| { |
| "epoch": 0.9030612244897959, |
| "grad_norm": 0.14129720628261566, |
| "learning_rate": 1.9845833333333332e-05, |
| "loss": 0.15628966331481933, |
| "step": 44250 |
| }, |
| { |
| "epoch": 0.9040816326530612, |
| "grad_norm": 0.14769406616687775, |
| "learning_rate": 1.96375e-05, |
| "loss": 0.15256972312927247, |
| "step": 44300 |
| }, |
| { |
| "epoch": 0.9051020408163265, |
| "grad_norm": 0.2667403221130371, |
| "learning_rate": 1.9429166666666667e-05, |
| "loss": 0.14853564262390137, |
| "step": 44350 |
| }, |
| { |
| "epoch": 0.9061224489795918, |
| "grad_norm": 0.0967816635966301, |
| "learning_rate": 1.9220833333333334e-05, |
| "loss": 0.14923785209655763, |
| "step": 44400 |
| }, |
| { |
| "epoch": 0.9071428571428571, |
| "grad_norm": 0.16255813837051392, |
| "learning_rate": 1.90125e-05, |
| "loss": 0.1490318012237549, |
| "step": 44450 |
| }, |
| { |
| "epoch": 0.9081632653061225, |
| "grad_norm": 0.18926061689853668, |
| "learning_rate": 1.880416666666667e-05, |
| "loss": 0.149087553024292, |
| "step": 44500 |
| }, |
| { |
| "epoch": 0.9091836734693878, |
| "grad_norm": 0.1748647391796112, |
| "learning_rate": 1.8595833333333336e-05, |
| "loss": 0.14986876487731934, |
| "step": 44550 |
| }, |
| { |
| "epoch": 0.9102040816326531, |
| "grad_norm": 0.08740395307540894, |
| "learning_rate": 1.8387500000000003e-05, |
| "loss": 0.1448510456085205, |
| "step": 44600 |
| }, |
| { |
| "epoch": 0.9112244897959184, |
| "grad_norm": 0.1446746289730072, |
| "learning_rate": 1.8179166666666667e-05, |
| "loss": 0.15715752601623534, |
| "step": 44650 |
| }, |
| { |
| "epoch": 0.9122448979591836, |
| "grad_norm": 0.1801069974899292, |
| "learning_rate": 1.7970833333333334e-05, |
| "loss": 0.16352605819702148, |
| "step": 44700 |
| }, |
| { |
| "epoch": 0.9132653061224489, |
| "grad_norm": 0.18530972301959991, |
| "learning_rate": 1.77625e-05, |
| "loss": 0.1280909538269043, |
| "step": 44750 |
| }, |
| { |
| "epoch": 0.9142857142857143, |
| "grad_norm": 0.1628209501504898, |
| "learning_rate": 1.755416666666667e-05, |
| "loss": 0.14237693786621095, |
| "step": 44800 |
| }, |
| { |
| "epoch": 0.9153061224489796, |
| "grad_norm": 0.17065215110778809, |
| "learning_rate": 1.7345833333333333e-05, |
| "loss": 0.15755342483520507, |
| "step": 44850 |
| }, |
| { |
| "epoch": 0.9163265306122449, |
| "grad_norm": 0.20158743858337402, |
| "learning_rate": 1.71375e-05, |
| "loss": 0.15958144187927245, |
| "step": 44900 |
| }, |
| { |
| "epoch": 0.9173469387755102, |
| "grad_norm": 0.17208772897720337, |
| "learning_rate": 1.6929166666666667e-05, |
| "loss": 0.1588733959197998, |
| "step": 44950 |
| }, |
| { |
| "epoch": 0.9183673469387755, |
| "grad_norm": 0.24862460792064667, |
| "learning_rate": 1.6720833333333334e-05, |
| "loss": 0.1676609230041504, |
| "step": 45000 |
| } |
| ], |
| "logging_steps": 50, |
| "max_steps": 49000, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 9223372036854775807, |
| "save_steps": 5000, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 2.8072562688e+17, |
| "train_batch_size": 4, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
