diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,5999 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 2.0, + "eval_steps": 500, + "global_step": 17092, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "grad_norm": 11.189342496234838, + "learning_rate": 4.2000000000000006e-07, + "loss": 0.6334, + "step": 20 + }, + { + "epoch": 0.0, + "grad_norm": 12.102946252189938, + "learning_rate": 6.515789473684211e-07, + "loss": 0.6334, + "step": 40 + }, + { + "epoch": 0.01, + "grad_norm": 9.23868714422811, + "learning_rate": 8.831578947368421e-07, + "loss": 0.5919, + "step": 60 + }, + { + "epoch": 0.01, + "grad_norm": 10.163427827153294, + "learning_rate": 1.1147368421052632e-06, + "loss": 0.591, + "step": 80 + }, + { + "epoch": 0.01, + "grad_norm": 7.703301789458378, + "learning_rate": 1.3463157894736842e-06, + "loss": 0.5663, + "step": 100 + }, + { + "epoch": 0.01, + "grad_norm": 10.292384648986614, + "learning_rate": 1.5778947368421053e-06, + "loss": 0.5649, + "step": 120 + }, + { + "epoch": 0.02, + "grad_norm": 9.622475032798782, + "learning_rate": 1.8094736842105263e-06, + "loss": 0.5215, + "step": 140 + }, + { + "epoch": 0.02, + "grad_norm": 12.80974448742081, + "learning_rate": 2.0410526315789475e-06, + "loss": 0.5166, + "step": 160 + }, + { + "epoch": 0.02, + "grad_norm": 7.2494201428270495, + "learning_rate": 2.2726315789473687e-06, + "loss": 0.4708, + "step": 180 + }, + { + "epoch": 0.02, + "grad_norm": 7.525527996410342, + "learning_rate": 2.5042105263157895e-06, + "loss": 0.4809, + "step": 200 + }, + { + "epoch": 0.03, + "grad_norm": 9.553244905277726, + "learning_rate": 2.735789473684211e-06, + "loss": 0.4149, + "step": 220 + }, + { + "epoch": 0.03, + "grad_norm": 10.08934392514418, + "learning_rate": 2.967368421052632e-06, + "loss": 0.4709, + "step": 240 + }, + { + "epoch": 0.03, + "grad_norm": 8.77781615537256, + "learning_rate": 3.198947368421053e-06, + "loss": 0.4706, + "step": 260 + }, + { + "epoch": 0.03, + "grad_norm": 8.842231539252408, + "learning_rate": 3.430526315789474e-06, + "loss": 0.4033, + "step": 280 + }, + { + "epoch": 0.04, + "grad_norm": 10.007575408247385, + "learning_rate": 3.662105263157895e-06, + "loss": 0.4411, + "step": 300 + }, + { + "epoch": 0.04, + "grad_norm": 8.652379461692668, + "learning_rate": 3.893684210526316e-06, + "loss": 0.4011, + "step": 320 + }, + { + "epoch": 0.04, + "grad_norm": 9.03023367578369, + "learning_rate": 4.125263157894737e-06, + "loss": 0.4252, + "step": 340 + }, + { + "epoch": 0.04, + "grad_norm": 8.141636176102654, + "learning_rate": 4.356842105263159e-06, + "loss": 0.4364, + "step": 360 + }, + { + "epoch": 0.04, + "grad_norm": 8.361423248626743, + "learning_rate": 4.58842105263158e-06, + "loss": 0.4193, + "step": 380 + }, + { + "epoch": 0.05, + "grad_norm": 9.373028764113052, + "learning_rate": 4.8200000000000004e-06, + "loss": 0.4078, + "step": 400 + }, + { + "epoch": 0.05, + "grad_norm": 8.799844515033747, + "learning_rate": 5.051578947368422e-06, + "loss": 0.4291, + "step": 420 + }, + { + "epoch": 0.05, + "grad_norm": 8.642174028335415, + "learning_rate": 5.283157894736842e-06, + "loss": 0.382, + "step": 440 + }, + { + "epoch": 0.05, + "grad_norm": 9.451813409785364, + "learning_rate": 5.514736842105264e-06, + "loss": 0.4089, + "step": 460 + }, + { + "epoch": 0.06, + "grad_norm": 11.329024075313082, + "learning_rate": 5.7463157894736845e-06, + "loss": 0.4015, + "step": 480 + }, + { + "epoch": 0.06, + "grad_norm": 7.271736970582471, + "learning_rate": 5.977894736842106e-06, + "loss": 0.4028, + "step": 500 + }, + { + "epoch": 0.06, + "grad_norm": 10.856360235187129, + "learning_rate": 6.209473684210528e-06, + "loss": 0.3814, + "step": 520 + }, + { + "epoch": 0.06, + "grad_norm": 10.40735773829138, + "learning_rate": 6.441052631578948e-06, + "loss": 0.3868, + "step": 540 + }, + { + "epoch": 0.07, + "grad_norm": 7.176143531188922, + "learning_rate": 6.672631578947369e-06, + "loss": 0.4089, + "step": 560 + }, + { + "epoch": 0.07, + "grad_norm": 9.628282656024224, + "learning_rate": 6.90421052631579e-06, + "loss": 0.3771, + "step": 580 + }, + { + "epoch": 0.07, + "grad_norm": 11.199839686200068, + "learning_rate": 7.135789473684212e-06, + "loss": 0.383, + "step": 600 + }, + { + "epoch": 0.07, + "grad_norm": 8.042449860326105, + "learning_rate": 7.367368421052632e-06, + "loss": 0.3875, + "step": 620 + }, + { + "epoch": 0.07, + "grad_norm": 6.256777636127039, + "learning_rate": 7.598947368421053e-06, + "loss": 0.399, + "step": 640 + }, + { + "epoch": 0.08, + "grad_norm": 12.140400652487944, + "learning_rate": 7.830526315789473e-06, + "loss": 0.399, + "step": 660 + }, + { + "epoch": 0.08, + "grad_norm": 9.31740323216306, + "learning_rate": 8.062105263157896e-06, + "loss": 0.383, + "step": 680 + }, + { + "epoch": 0.08, + "grad_norm": 11.518507949844242, + "learning_rate": 8.293684210526317e-06, + "loss": 0.3673, + "step": 700 + }, + { + "epoch": 0.08, + "grad_norm": 12.23485079993539, + "learning_rate": 8.525263157894737e-06, + "loss": 0.4105, + "step": 720 + }, + { + "epoch": 0.09, + "grad_norm": 7.506176757455166, + "learning_rate": 8.756842105263158e-06, + "loss": 0.425, + "step": 740 + }, + { + "epoch": 0.09, + "grad_norm": 28.3327677352401, + "learning_rate": 8.988421052631579e-06, + "loss": 0.3874, + "step": 760 + }, + { + "epoch": 0.09, + "grad_norm": 10.006030362522525, + "learning_rate": 9.220000000000002e-06, + "loss": 0.4055, + "step": 780 + }, + { + "epoch": 0.09, + "grad_norm": 10.010446813711976, + "learning_rate": 9.451578947368422e-06, + "loss": 0.3599, + "step": 800 + }, + { + "epoch": 0.1, + "grad_norm": 7.043688000572858, + "learning_rate": 9.683157894736843e-06, + "loss": 0.3624, + "step": 820 + }, + { + "epoch": 0.1, + "grad_norm": 9.607021831412599, + "learning_rate": 9.914736842105264e-06, + "loss": 0.3952, + "step": 840 + }, + { + "epoch": 0.1, + "grad_norm": 9.147395412600936, + "learning_rate": 1.0146315789473686e-05, + "loss": 0.3819, + "step": 860 + }, + { + "epoch": 0.1, + "grad_norm": 7.993375419567414, + "learning_rate": 1.0377894736842107e-05, + "loss": 0.3678, + "step": 880 + }, + { + "epoch": 0.11, + "grad_norm": 7.52609874523302, + "learning_rate": 1.0609473684210528e-05, + "loss": 0.375, + "step": 900 + }, + { + "epoch": 0.11, + "grad_norm": 9.52738539095041, + "learning_rate": 1.084105263157895e-05, + "loss": 0.3681, + "step": 920 + }, + { + "epoch": 0.11, + "grad_norm": 11.370588516129013, + "learning_rate": 1.107263157894737e-05, + "loss": 0.3911, + "step": 940 + }, + { + "epoch": 0.11, + "grad_norm": 9.391054368953709, + "learning_rate": 1.130421052631579e-05, + "loss": 0.329, + "step": 960 + }, + { + "epoch": 0.11, + "grad_norm": 7.997549873904863, + "learning_rate": 1.1535789473684211e-05, + "loss": 0.359, + "step": 980 + }, + { + "epoch": 0.12, + "grad_norm": 10.755576173466588, + "learning_rate": 1.1767368421052634e-05, + "loss": 0.393, + "step": 1000 + }, + { + "epoch": 0.12, + "grad_norm": 6.458117606549104, + "learning_rate": 1.1998947368421054e-05, + "loss": 0.363, + "step": 1020 + }, + { + "epoch": 0.12, + "grad_norm": 9.937055469862932, + "learning_rate": 1.2230526315789475e-05, + "loss": 0.3766, + "step": 1040 + }, + { + "epoch": 0.12, + "grad_norm": 7.485816768913991, + "learning_rate": 1.2462105263157898e-05, + "loss": 0.3845, + "step": 1060 + }, + { + "epoch": 0.13, + "grad_norm": 8.01333127284645, + "learning_rate": 1.2693684210526319e-05, + "loss": 0.3617, + "step": 1080 + }, + { + "epoch": 0.13, + "grad_norm": 8.911089094498925, + "learning_rate": 1.292526315789474e-05, + "loss": 0.3599, + "step": 1100 + }, + { + "epoch": 0.13, + "grad_norm": 8.32079434857686, + "learning_rate": 1.3156842105263158e-05, + "loss": 0.3722, + "step": 1120 + }, + { + "epoch": 0.13, + "grad_norm": 9.806231592184165, + "learning_rate": 1.3388421052631581e-05, + "loss": 0.3755, + "step": 1140 + }, + { + "epoch": 0.14, + "grad_norm": 8.798076540805445, + "learning_rate": 1.3620000000000002e-05, + "loss": 0.3616, + "step": 1160 + }, + { + "epoch": 0.14, + "grad_norm": 8.343613755617508, + "learning_rate": 1.3851578947368423e-05, + "loss": 0.3631, + "step": 1180 + }, + { + "epoch": 0.14, + "grad_norm": 8.35814690526204, + "learning_rate": 1.4083157894736843e-05, + "loss": 0.3638, + "step": 1200 + }, + { + "epoch": 0.14, + "grad_norm": 7.706542458286787, + "learning_rate": 1.4314736842105266e-05, + "loss": 0.375, + "step": 1220 + }, + { + "epoch": 0.15, + "grad_norm": 10.89212756963322, + "learning_rate": 1.4546315789473687e-05, + "loss": 0.4049, + "step": 1240 + }, + { + "epoch": 0.15, + "grad_norm": 11.46198790559678, + "learning_rate": 1.4777894736842107e-05, + "loss": 0.332, + "step": 1260 + }, + { + "epoch": 0.15, + "grad_norm": 8.365156167551179, + "learning_rate": 1.500947368421053e-05, + "loss": 0.3586, + "step": 1280 + }, + { + "epoch": 0.15, + "grad_norm": 7.841036992292662, + "learning_rate": 1.5241052631578949e-05, + "loss": 0.368, + "step": 1300 + }, + { + "epoch": 0.15, + "grad_norm": 9.84275168548854, + "learning_rate": 1.547263157894737e-05, + "loss": 0.3767, + "step": 1320 + }, + { + "epoch": 0.16, + "grad_norm": 10.506254240871392, + "learning_rate": 1.5704210526315792e-05, + "loss": 0.378, + "step": 1340 + }, + { + "epoch": 0.16, + "grad_norm": 7.844581104033147, + "learning_rate": 1.5935789473684213e-05, + "loss": 0.3453, + "step": 1360 + }, + { + "epoch": 0.16, + "grad_norm": 8.193758138016792, + "learning_rate": 1.6167368421052634e-05, + "loss": 0.3752, + "step": 1380 + }, + { + "epoch": 0.16, + "grad_norm": 7.475620318034102, + "learning_rate": 1.6398947368421055e-05, + "loss": 0.3642, + "step": 1400 + }, + { + "epoch": 0.17, + "grad_norm": 7.4964850772334355, + "learning_rate": 1.6630526315789475e-05, + "loss": 0.3377, + "step": 1420 + }, + { + "epoch": 0.17, + "grad_norm": 7.903753436810423, + "learning_rate": 1.6862105263157896e-05, + "loss": 0.3586, + "step": 1440 + }, + { + "epoch": 0.17, + "grad_norm": 10.49594728277505, + "learning_rate": 1.7093684210526317e-05, + "loss": 0.3468, + "step": 1460 + }, + { + "epoch": 0.17, + "grad_norm": 8.027713457118994, + "learning_rate": 1.7325263157894738e-05, + "loss": 0.3491, + "step": 1480 + }, + { + "epoch": 0.18, + "grad_norm": 7.2748681125287415, + "learning_rate": 1.755684210526316e-05, + "loss": 0.3581, + "step": 1500 + }, + { + "epoch": 0.18, + "grad_norm": 6.077326940397276, + "learning_rate": 1.7788421052631583e-05, + "loss": 0.3356, + "step": 1520 + }, + { + "epoch": 0.18, + "grad_norm": 6.724980645701413, + "learning_rate": 1.802e-05, + "loss": 0.3656, + "step": 1540 + }, + { + "epoch": 0.18, + "grad_norm": 8.592315221136559, + "learning_rate": 1.825157894736842e-05, + "loss": 0.3467, + "step": 1560 + }, + { + "epoch": 0.18, + "grad_norm": 9.036253074638084, + "learning_rate": 1.8483157894736845e-05, + "loss": 0.3414, + "step": 1580 + }, + { + "epoch": 0.19, + "grad_norm": 11.349080832480995, + "learning_rate": 1.8714736842105266e-05, + "loss": 0.3655, + "step": 1600 + }, + { + "epoch": 0.19, + "grad_norm": 9.361475025905081, + "learning_rate": 1.8946315789473687e-05, + "loss": 0.3564, + "step": 1620 + }, + { + "epoch": 0.19, + "grad_norm": 6.53643855564322, + "learning_rate": 1.9177894736842108e-05, + "loss": 0.3714, + "step": 1640 + }, + { + "epoch": 0.19, + "grad_norm": 6.65699298275954, + "learning_rate": 1.940947368421053e-05, + "loss": 0.3546, + "step": 1660 + }, + { + "epoch": 0.2, + "grad_norm": 9.856127510930383, + "learning_rate": 1.964105263157895e-05, + "loss": 0.3308, + "step": 1680 + }, + { + "epoch": 0.2, + "grad_norm": 8.532146476622113, + "learning_rate": 1.987263157894737e-05, + "loss": 0.366, + "step": 1700 + }, + { + "epoch": 0.2, + "grad_norm": 8.327666937449752, + "learning_rate": 1.999999532098205e-05, + "loss": 0.3247, + "step": 1720 + }, + { + "epoch": 0.2, + "grad_norm": 10.876035443701728, + "learning_rate": 1.99999578888647e-05, + "loss": 0.3543, + "step": 1740 + }, + { + "epoch": 0.21, + "grad_norm": 8.263711805253918, + "learning_rate": 1.999988302477013e-05, + "loss": 0.3664, + "step": 1760 + }, + { + "epoch": 0.21, + "grad_norm": 7.575419338189409, + "learning_rate": 1.9999770728978607e-05, + "loss": 0.345, + "step": 1780 + }, + { + "epoch": 0.21, + "grad_norm": 7.814279338388738, + "learning_rate": 1.9999621001910508e-05, + "loss": 0.3573, + "step": 1800 + }, + { + "epoch": 0.21, + "grad_norm": 13.040107479471628, + "learning_rate": 1.999943384412636e-05, + "loss": 0.3344, + "step": 1820 + }, + { + "epoch": 0.22, + "grad_norm": 7.535441012066771, + "learning_rate": 1.99992092563268e-05, + "loss": 0.3552, + "step": 1840 + }, + { + "epoch": 0.22, + "grad_norm": 8.059491683038381, + "learning_rate": 1.9998947239352593e-05, + "loss": 0.3442, + "step": 1860 + }, + { + "epoch": 0.22, + "grad_norm": 6.700609139607897, + "learning_rate": 1.999864779418462e-05, + "loss": 0.3584, + "step": 1880 + }, + { + "epoch": 0.22, + "grad_norm": 6.831136707088909, + "learning_rate": 1.999831092194389e-05, + "loss": 0.3341, + "step": 1900 + }, + { + "epoch": 0.22, + "grad_norm": 10.30262237054047, + "learning_rate": 1.9997936623891498e-05, + "loss": 0.3704, + "step": 1920 + }, + { + "epoch": 0.23, + "grad_norm": 6.703691792033561, + "learning_rate": 1.9997524901428677e-05, + "loss": 0.3769, + "step": 1940 + }, + { + "epoch": 0.23, + "grad_norm": 7.251026212477321, + "learning_rate": 1.9997075756096734e-05, + "loss": 0.3234, + "step": 1960 + }, + { + "epoch": 0.23, + "grad_norm": 6.14471190830255, + "learning_rate": 1.9996589189577094e-05, + "loss": 0.3443, + "step": 1980 + }, + { + "epoch": 0.23, + "grad_norm": 7.254885638733633, + "learning_rate": 1.999606520369125e-05, + "loss": 0.3178, + "step": 2000 + }, + { + "epoch": 0.24, + "grad_norm": 7.330977870771951, + "learning_rate": 1.9995503800400805e-05, + "loss": 0.3309, + "step": 2020 + }, + { + "epoch": 0.24, + "grad_norm": 7.291895603037368, + "learning_rate": 1.9994904981807407e-05, + "loss": 0.3372, + "step": 2040 + }, + { + "epoch": 0.24, + "grad_norm": 12.772894555548028, + "learning_rate": 1.9994268750152793e-05, + "loss": 0.3747, + "step": 2060 + }, + { + "epoch": 0.24, + "grad_norm": 9.497788021161657, + "learning_rate": 1.999359510781875e-05, + "loss": 0.3529, + "step": 2080 + }, + { + "epoch": 0.25, + "grad_norm": 6.581748417513793, + "learning_rate": 1.9992884057327118e-05, + "loss": 0.3395, + "step": 2100 + }, + { + "epoch": 0.25, + "grad_norm": 9.523030924078585, + "learning_rate": 1.9992135601339776e-05, + "loss": 0.3469, + "step": 2120 + }, + { + "epoch": 0.25, + "grad_norm": 17.640537636796516, + "learning_rate": 1.999134974265864e-05, + "loss": 0.3365, + "step": 2140 + }, + { + "epoch": 0.25, + "grad_norm": 5.4385511763432985, + "learning_rate": 1.999052648422563e-05, + "loss": 0.3845, + "step": 2160 + }, + { + "epoch": 0.26, + "grad_norm": 7.585921156892153, + "learning_rate": 1.9989665829122704e-05, + "loss": 0.3464, + "step": 2180 + }, + { + "epoch": 0.26, + "grad_norm": 12.163408304702841, + "learning_rate": 1.9988767780571778e-05, + "loss": 0.3127, + "step": 2200 + }, + { + "epoch": 0.26, + "grad_norm": 8.246994395835456, + "learning_rate": 1.9987832341934793e-05, + "loss": 0.3424, + "step": 2220 + }, + { + "epoch": 0.26, + "grad_norm": 9.628684787640838, + "learning_rate": 1.9986859516713636e-05, + "loss": 0.3472, + "step": 2240 + }, + { + "epoch": 0.26, + "grad_norm": 10.912326639203211, + "learning_rate": 1.9985849308550167e-05, + "loss": 0.3326, + "step": 2260 + }, + { + "epoch": 0.27, + "grad_norm": 6.419410251137725, + "learning_rate": 1.9984801721226188e-05, + "loss": 0.3297, + "step": 2280 + }, + { + "epoch": 0.27, + "grad_norm": 7.798212276898864, + "learning_rate": 1.998371675866344e-05, + "loss": 0.3389, + "step": 2300 + }, + { + "epoch": 0.27, + "grad_norm": 7.811483576458649, + "learning_rate": 1.9982594424923566e-05, + "loss": 0.332, + "step": 2320 + }, + { + "epoch": 0.27, + "grad_norm": 6.967736687994363, + "learning_rate": 1.9981434724208126e-05, + "loss": 0.3099, + "step": 2340 + }, + { + "epoch": 0.28, + "grad_norm": 9.635115819134949, + "learning_rate": 1.9980237660858565e-05, + "loss": 0.3285, + "step": 2360 + }, + { + "epoch": 0.28, + "grad_norm": 6.720069081541748, + "learning_rate": 1.997900323935619e-05, + "loss": 0.343, + "step": 2380 + }, + { + "epoch": 0.28, + "grad_norm": 10.300889386045979, + "learning_rate": 1.9977731464322162e-05, + "loss": 0.3667, + "step": 2400 + }, + { + "epoch": 0.28, + "grad_norm": 7.888854177455323, + "learning_rate": 1.9976422340517495e-05, + "loss": 0.3397, + "step": 2420 + }, + { + "epoch": 0.29, + "grad_norm": 9.643051064776678, + "learning_rate": 1.9975075872843e-05, + "loss": 0.3371, + "step": 2440 + }, + { + "epoch": 0.29, + "grad_norm": 6.166262467342929, + "learning_rate": 1.9973692066339298e-05, + "loss": 0.3424, + "step": 2460 + }, + { + "epoch": 0.29, + "grad_norm": 7.520601841401043, + "learning_rate": 1.9972270926186794e-05, + "loss": 0.3083, + "step": 2480 + }, + { + "epoch": 0.29, + "grad_norm": 6.095472200047568, + "learning_rate": 1.9970812457705642e-05, + "loss": 0.3512, + "step": 2500 + }, + { + "epoch": 0.29, + "grad_norm": 7.259492119293066, + "learning_rate": 1.9969316666355757e-05, + "loss": 0.3309, + "step": 2520 + }, + { + "epoch": 0.3, + "grad_norm": 8.355978245335697, + "learning_rate": 1.996778355773676e-05, + "loss": 0.3296, + "step": 2540 + }, + { + "epoch": 0.3, + "grad_norm": 9.612502269967832, + "learning_rate": 1.996621313758798e-05, + "loss": 0.3391, + "step": 2560 + }, + { + "epoch": 0.3, + "grad_norm": 8.148285817739712, + "learning_rate": 1.9964605411788424e-05, + "loss": 0.3416, + "step": 2580 + }, + { + "epoch": 0.3, + "grad_norm": 7.645206131186397, + "learning_rate": 1.9962960386356748e-05, + "loss": 0.352, + "step": 2600 + }, + { + "epoch": 0.31, + "grad_norm": 5.406075006199623, + "learning_rate": 1.9961278067451258e-05, + "loss": 0.3194, + "step": 2620 + }, + { + "epoch": 0.31, + "grad_norm": 7.23674984010716, + "learning_rate": 1.995955846136986e-05, + "loss": 0.3216, + "step": 2640 + }, + { + "epoch": 0.31, + "grad_norm": 8.044071872202302, + "learning_rate": 1.9957801574550044e-05, + "loss": 0.321, + "step": 2660 + }, + { + "epoch": 0.31, + "grad_norm": 7.2624583573058485, + "learning_rate": 1.9956007413568874e-05, + "loss": 0.3352, + "step": 2680 + }, + { + "epoch": 0.32, + "grad_norm": 9.277277549105909, + "learning_rate": 1.9954175985142955e-05, + "loss": 0.3391, + "step": 2700 + }, + { + "epoch": 0.32, + "grad_norm": 10.975275864316648, + "learning_rate": 1.9952307296128395e-05, + "loss": 0.3298, + "step": 2720 + }, + { + "epoch": 0.32, + "grad_norm": 11.416959735607907, + "learning_rate": 1.9950401353520795e-05, + "loss": 0.3382, + "step": 2740 + }, + { + "epoch": 0.32, + "grad_norm": 7.384993942351236, + "learning_rate": 1.994845816445522e-05, + "loss": 0.3422, + "step": 2760 + }, + { + "epoch": 0.33, + "grad_norm": 6.775839902047113, + "learning_rate": 1.9946477736206167e-05, + "loss": 0.3694, + "step": 2780 + }, + { + "epoch": 0.33, + "grad_norm": 8.246414330131564, + "learning_rate": 1.9944460076187546e-05, + "loss": 0.3458, + "step": 2800 + }, + { + "epoch": 0.33, + "grad_norm": 7.152660285783071, + "learning_rate": 1.9942405191952646e-05, + "loss": 0.3426, + "step": 2820 + }, + { + "epoch": 0.33, + "grad_norm": 6.9408668390055315, + "learning_rate": 1.9940313091194096e-05, + "loss": 0.3097, + "step": 2840 + }, + { + "epoch": 0.33, + "grad_norm": 5.797513942002627, + "learning_rate": 1.9938183781743873e-05, + "loss": 0.3102, + "step": 2860 + }, + { + "epoch": 0.34, + "grad_norm": 7.361935180345544, + "learning_rate": 1.9936017271573223e-05, + "loss": 0.3102, + "step": 2880 + }, + { + "epoch": 0.34, + "grad_norm": 7.4796257480282, + "learning_rate": 1.9933813568792678e-05, + "loss": 0.3256, + "step": 2900 + }, + { + "epoch": 0.34, + "grad_norm": 5.500192812227557, + "learning_rate": 1.9931572681651984e-05, + "loss": 0.3107, + "step": 2920 + }, + { + "epoch": 0.34, + "grad_norm": 13.84776864376109, + "learning_rate": 1.9929294618540103e-05, + "loss": 0.3062, + "step": 2940 + }, + { + "epoch": 0.35, + "grad_norm": 7.937410128828084, + "learning_rate": 1.992697938798516e-05, + "loss": 0.3293, + "step": 2960 + }, + { + "epoch": 0.35, + "grad_norm": 5.447364190742796, + "learning_rate": 1.9924626998654433e-05, + "loss": 0.3318, + "step": 2980 + }, + { + "epoch": 0.35, + "grad_norm": 7.825239369201605, + "learning_rate": 1.9922237459354287e-05, + "loss": 0.3238, + "step": 3000 + }, + { + "epoch": 0.35, + "grad_norm": 6.466926354506395, + "learning_rate": 1.9919810779030177e-05, + "loss": 0.3136, + "step": 3020 + }, + { + "epoch": 0.36, + "grad_norm": 10.115420483963225, + "learning_rate": 1.9917346966766596e-05, + "loss": 0.3329, + "step": 3040 + }, + { + "epoch": 0.36, + "grad_norm": 6.08100630613984, + "learning_rate": 1.991484603178704e-05, + "loss": 0.2999, + "step": 3060 + }, + { + "epoch": 0.36, + "grad_norm": 6.153474908414458, + "learning_rate": 1.991230798345398e-05, + "loss": 0.3344, + "step": 3080 + }, + { + "epoch": 0.36, + "grad_norm": 6.956842076140817, + "learning_rate": 1.990973283126882e-05, + "loss": 0.3356, + "step": 3100 + }, + { + "epoch": 0.37, + "grad_norm": 7.977473013654998, + "learning_rate": 1.990712058487187e-05, + "loss": 0.2984, + "step": 3120 + }, + { + "epoch": 0.37, + "grad_norm": 7.472067890046691, + "learning_rate": 1.9904471254042304e-05, + "loss": 0.308, + "step": 3140 + }, + { + "epoch": 0.37, + "grad_norm": 6.98090482864238, + "learning_rate": 1.9901784848698125e-05, + "loss": 0.3036, + "step": 3160 + }, + { + "epoch": 0.37, + "grad_norm": 7.229553593404789, + "learning_rate": 1.989906137889613e-05, + "loss": 0.3027, + "step": 3180 + }, + { + "epoch": 0.37, + "grad_norm": 5.9700766482137295, + "learning_rate": 1.9896300854831862e-05, + "loss": 0.3329, + "step": 3200 + }, + { + "epoch": 0.38, + "grad_norm": 7.864131148182889, + "learning_rate": 1.989350328683959e-05, + "loss": 0.308, + "step": 3220 + }, + { + "epoch": 0.38, + "grad_norm": 20.04577205534678, + "learning_rate": 1.989066868539225e-05, + "loss": 0.3217, + "step": 3240 + }, + { + "epoch": 0.38, + "grad_norm": 8.46751929768339, + "learning_rate": 1.9887797061101424e-05, + "loss": 0.3036, + "step": 3260 + }, + { + "epoch": 0.38, + "grad_norm": 7.667111535917501, + "learning_rate": 1.9884888424717296e-05, + "loss": 0.3084, + "step": 3280 + }, + { + "epoch": 0.39, + "grad_norm": 6.615069450077278, + "learning_rate": 1.9881942787128594e-05, + "loss": 0.3177, + "step": 3300 + }, + { + "epoch": 0.39, + "grad_norm": 9.324876177990959, + "learning_rate": 1.9878960159362572e-05, + "loss": 0.3496, + "step": 3320 + }, + { + "epoch": 0.39, + "grad_norm": 7.1979554134343635, + "learning_rate": 1.987594055258496e-05, + "loss": 0.3167, + "step": 3340 + }, + { + "epoch": 0.39, + "grad_norm": 29.076306154493622, + "learning_rate": 1.987288397809993e-05, + "loss": 0.34, + "step": 3360 + }, + { + "epoch": 0.4, + "grad_norm": 7.861933206527291, + "learning_rate": 1.986979044735002e-05, + "loss": 0.3424, + "step": 3380 + }, + { + "epoch": 0.4, + "grad_norm": 5.7878723156599445, + "learning_rate": 1.9866659971916146e-05, + "loss": 0.3157, + "step": 3400 + }, + { + "epoch": 0.4, + "grad_norm": 8.470531768859948, + "learning_rate": 1.9863492563517513e-05, + "loss": 0.3201, + "step": 3420 + }, + { + "epoch": 0.4, + "grad_norm": 8.104846079245535, + "learning_rate": 1.98602882340116e-05, + "loss": 0.322, + "step": 3440 + }, + { + "epoch": 0.4, + "grad_norm": 6.180289951670451, + "learning_rate": 1.9857046995394084e-05, + "loss": 0.2991, + "step": 3460 + }, + { + "epoch": 0.41, + "grad_norm": 4.962160167060266, + "learning_rate": 1.985376885979884e-05, + "loss": 0.3097, + "step": 3480 + }, + { + "epoch": 0.41, + "grad_norm": 6.813966925636774, + "learning_rate": 1.9850453839497844e-05, + "loss": 0.3294, + "step": 3500 + }, + { + "epoch": 0.41, + "grad_norm": 5.902763508299491, + "learning_rate": 1.9847101946901172e-05, + "loss": 0.2967, + "step": 3520 + }, + { + "epoch": 0.41, + "grad_norm": 7.359798475906595, + "learning_rate": 1.984371319455693e-05, + "loss": 0.3268, + "step": 3540 + }, + { + "epoch": 0.42, + "grad_norm": 7.377881796357637, + "learning_rate": 1.9840287595151205e-05, + "loss": 0.3251, + "step": 3560 + }, + { + "epoch": 0.42, + "grad_norm": 7.261601866242624, + "learning_rate": 1.9836825161508032e-05, + "loss": 0.284, + "step": 3580 + }, + { + "epoch": 0.42, + "grad_norm": 6.123609423918951, + "learning_rate": 1.983332590658934e-05, + "loss": 0.3137, + "step": 3600 + }, + { + "epoch": 0.42, + "grad_norm": 7.205290164862431, + "learning_rate": 1.9829789843494885e-05, + "loss": 0.3, + "step": 3620 + }, + { + "epoch": 0.43, + "grad_norm": 6.0471563544848035, + "learning_rate": 1.9826216985462245e-05, + "loss": 0.2991, + "step": 3640 + }, + { + "epoch": 0.43, + "grad_norm": 7.436589545934819, + "learning_rate": 1.9822607345866715e-05, + "loss": 0.3299, + "step": 3660 + }, + { + "epoch": 0.43, + "grad_norm": 8.240853065109379, + "learning_rate": 1.9818960938221308e-05, + "loss": 0.2886, + "step": 3680 + }, + { + "epoch": 0.43, + "grad_norm": 5.451034188795461, + "learning_rate": 1.9815277776176663e-05, + "loss": 0.3138, + "step": 3700 + }, + { + "epoch": 0.44, + "grad_norm": 7.874196329434049, + "learning_rate": 1.9811557873521027e-05, + "loss": 0.2946, + "step": 3720 + }, + { + "epoch": 0.44, + "grad_norm": 6.547528436867746, + "learning_rate": 1.9807801244180174e-05, + "loss": 0.2978, + "step": 3740 + }, + { + "epoch": 0.44, + "grad_norm": 5.816214082411462, + "learning_rate": 1.980400790221739e-05, + "loss": 0.3067, + "step": 3760 + }, + { + "epoch": 0.44, + "grad_norm": 6.245697982302123, + "learning_rate": 1.9800177861833372e-05, + "loss": 0.3191, + "step": 3780 + }, + { + "epoch": 0.44, + "grad_norm": 7.932045331416635, + "learning_rate": 1.9796311137366212e-05, + "loss": 0.3439, + "step": 3800 + }, + { + "epoch": 0.45, + "grad_norm": 6.088006858952151, + "learning_rate": 1.9792407743291343e-05, + "loss": 0.33, + "step": 3820 + }, + { + "epoch": 0.45, + "grad_norm": 5.770262468473867, + "learning_rate": 1.9788467694221453e-05, + "loss": 0.2897, + "step": 3840 + }, + { + "epoch": 0.45, + "grad_norm": 7.978864884873927, + "learning_rate": 1.9784491004906468e-05, + "loss": 0.3118, + "step": 3860 + }, + { + "epoch": 0.45, + "grad_norm": 6.631368778757732, + "learning_rate": 1.9780477690233476e-05, + "loss": 0.3165, + "step": 3880 + }, + { + "epoch": 0.46, + "grad_norm": 6.541290304773085, + "learning_rate": 1.9776427765226664e-05, + "loss": 0.3379, + "step": 3900 + }, + { + "epoch": 0.46, + "grad_norm": 6.42886929503141, + "learning_rate": 1.9772341245047285e-05, + "loss": 0.3136, + "step": 3920 + }, + { + "epoch": 0.46, + "grad_norm": 9.004651033459043, + "learning_rate": 1.9768218144993593e-05, + "loss": 0.311, + "step": 3940 + }, + { + "epoch": 0.46, + "grad_norm": 6.241942743461084, + "learning_rate": 1.976405848050077e-05, + "loss": 0.3224, + "step": 3960 + }, + { + "epoch": 0.47, + "grad_norm": 7.030346892956473, + "learning_rate": 1.975986226714089e-05, + "loss": 0.3513, + "step": 3980 + }, + { + "epoch": 0.47, + "grad_norm": 6.984184655106913, + "learning_rate": 1.9755629520622835e-05, + "loss": 0.3031, + "step": 4000 + }, + { + "epoch": 0.47, + "grad_norm": 3.6594620639728253, + "learning_rate": 1.9751360256792274e-05, + "loss": 0.2819, + "step": 4020 + }, + { + "epoch": 0.47, + "grad_norm": 3.9453364305431418, + "learning_rate": 1.9747054491631567e-05, + "loss": 0.2806, + "step": 4040 + }, + { + "epoch": 0.48, + "grad_norm": 3.8004852838340475, + "learning_rate": 1.9742712241259726e-05, + "loss": 0.2712, + "step": 4060 + }, + { + "epoch": 0.48, + "grad_norm": 4.671864793440983, + "learning_rate": 1.9738333521932348e-05, + "loss": 0.264, + "step": 4080 + }, + { + "epoch": 0.48, + "grad_norm": 3.9158215253486452, + "learning_rate": 1.9733918350041546e-05, + "loss": 0.2466, + "step": 4100 + }, + { + "epoch": 0.48, + "grad_norm": 3.232730170815344, + "learning_rate": 1.972946674211591e-05, + "loss": 0.28, + "step": 4120 + }, + { + "epoch": 0.48, + "grad_norm": 5.26102706338969, + "learning_rate": 1.972497871482043e-05, + "loss": 0.2793, + "step": 4140 + }, + { + "epoch": 0.49, + "grad_norm": 6.036896745386087, + "learning_rate": 1.972045428495642e-05, + "loss": 0.2551, + "step": 4160 + }, + { + "epoch": 0.49, + "grad_norm": 3.4899809349872823, + "learning_rate": 1.971589346946149e-05, + "loss": 0.2746, + "step": 4180 + }, + { + "epoch": 0.49, + "grad_norm": 4.003065603442425, + "learning_rate": 1.9711296285409443e-05, + "loss": 0.2639, + "step": 4200 + }, + { + "epoch": 0.49, + "grad_norm": 2.8477299266017626, + "learning_rate": 1.9706662750010255e-05, + "loss": 0.2689, + "step": 4220 + }, + { + "epoch": 0.5, + "grad_norm": 3.5491058459283837, + "learning_rate": 1.970199288060996e-05, + "loss": 0.2642, + "step": 4240 + }, + { + "epoch": 0.5, + "grad_norm": 3.658049963633355, + "learning_rate": 1.969728669469063e-05, + "loss": 0.2699, + "step": 4260 + }, + { + "epoch": 0.5, + "grad_norm": 3.7093939532444127, + "learning_rate": 1.9692544209870287e-05, + "loss": 0.2697, + "step": 4280 + }, + { + "epoch": 0.5, + "grad_norm": 3.0342042036143444, + "learning_rate": 1.968776544390283e-05, + "loss": 0.2654, + "step": 4300 + }, + { + "epoch": 0.51, + "grad_norm": 3.8174199112021174, + "learning_rate": 1.9682950414678004e-05, + "loss": 0.2525, + "step": 4320 + }, + { + "epoch": 0.51, + "grad_norm": 5.35961771156863, + "learning_rate": 1.9678099140221284e-05, + "loss": 0.2464, + "step": 4340 + }, + { + "epoch": 0.51, + "grad_norm": 5.063501894675061, + "learning_rate": 1.9673211638693844e-05, + "loss": 0.2788, + "step": 4360 + }, + { + "epoch": 0.51, + "grad_norm": 3.8581072620927723, + "learning_rate": 1.9668287928392478e-05, + "loss": 0.2743, + "step": 4380 + }, + { + "epoch": 0.51, + "grad_norm": 3.386524586463125, + "learning_rate": 1.9663328027749523e-05, + "loss": 0.2547, + "step": 4400 + }, + { + "epoch": 0.52, + "grad_norm": 4.049063661443696, + "learning_rate": 1.9658331955332807e-05, + "loss": 0.2787, + "step": 4420 + }, + { + "epoch": 0.52, + "grad_norm": 5.12761026294579, + "learning_rate": 1.9653299729845564e-05, + "loss": 0.2577, + "step": 4440 + }, + { + "epoch": 0.52, + "grad_norm": 3.808933278135121, + "learning_rate": 1.9648231370126375e-05, + "loss": 0.2654, + "step": 4460 + }, + { + "epoch": 0.52, + "grad_norm": 4.402659150192465, + "learning_rate": 1.9643126895149087e-05, + "loss": 0.2603, + "step": 4480 + }, + { + "epoch": 0.53, + "grad_norm": 8.492295307503369, + "learning_rate": 1.963798632402275e-05, + "loss": 0.2691, + "step": 4500 + }, + { + "epoch": 0.53, + "grad_norm": 3.628014593555578, + "learning_rate": 1.9632809675991544e-05, + "loss": 0.2711, + "step": 4520 + }, + { + "epoch": 0.53, + "grad_norm": 5.207256154885955, + "learning_rate": 1.9627596970434708e-05, + "loss": 0.2604, + "step": 4540 + }, + { + "epoch": 0.53, + "grad_norm": 3.892539047603965, + "learning_rate": 1.962234822686646e-05, + "loss": 0.2693, + "step": 4560 + }, + { + "epoch": 0.54, + "grad_norm": 4.005334872796194, + "learning_rate": 1.961706346493594e-05, + "loss": 0.2774, + "step": 4580 + }, + { + "epoch": 0.54, + "grad_norm": 4.456881895914548, + "learning_rate": 1.9611742704427106e-05, + "loss": 0.2531, + "step": 4600 + }, + { + "epoch": 0.54, + "grad_norm": 3.925512807597538, + "learning_rate": 1.9606385965258697e-05, + "loss": 0.2507, + "step": 4620 + }, + { + "epoch": 0.54, + "grad_norm": 4.401211069070843, + "learning_rate": 1.9600993267484147e-05, + "loss": 0.262, + "step": 4640 + }, + { + "epoch": 0.55, + "grad_norm": 5.3437799820979075, + "learning_rate": 1.9595564631291478e-05, + "loss": 0.2516, + "step": 4660 + }, + { + "epoch": 0.55, + "grad_norm": 4.553539308066039, + "learning_rate": 1.9590100077003278e-05, + "loss": 0.2603, + "step": 4680 + }, + { + "epoch": 0.55, + "grad_norm": 3.169670353545694, + "learning_rate": 1.958459962507658e-05, + "loss": 0.2875, + "step": 4700 + }, + { + "epoch": 0.55, + "grad_norm": 3.572973282405317, + "learning_rate": 1.9579063296102812e-05, + "loss": 0.2671, + "step": 4720 + }, + { + "epoch": 0.55, + "grad_norm": 3.59782258517998, + "learning_rate": 1.957349111080771e-05, + "loss": 0.2837, + "step": 4740 + }, + { + "epoch": 0.56, + "grad_norm": 5.199297439358037, + "learning_rate": 1.956788309005123e-05, + "loss": 0.2753, + "step": 4760 + }, + { + "epoch": 0.56, + "grad_norm": 4.993257936146954, + "learning_rate": 1.95622392548275e-05, + "loss": 0.2714, + "step": 4780 + }, + { + "epoch": 0.56, + "grad_norm": 3.7288744650364642, + "learning_rate": 1.955655962626471e-05, + "loss": 0.2609, + "step": 4800 + }, + { + "epoch": 0.56, + "grad_norm": 5.377784540016259, + "learning_rate": 1.9550844225625043e-05, + "loss": 0.2605, + "step": 4820 + }, + { + "epoch": 0.57, + "grad_norm": 3.1237811954279353, + "learning_rate": 1.954509307430461e-05, + "loss": 0.2569, + "step": 4840 + }, + { + "epoch": 0.57, + "grad_norm": 4.101705029145583, + "learning_rate": 1.953930619383335e-05, + "loss": 0.2719, + "step": 4860 + }, + { + "epoch": 0.57, + "grad_norm": 2.734092828632754, + "learning_rate": 1.953348360587496e-05, + "loss": 0.2503, + "step": 4880 + }, + { + "epoch": 0.57, + "grad_norm": 3.4516150854343604, + "learning_rate": 1.9527625332226804e-05, + "loss": 0.2588, + "step": 4900 + }, + { + "epoch": 0.58, + "grad_norm": 3.0561153265178085, + "learning_rate": 1.952173139481985e-05, + "loss": 0.2779, + "step": 4920 + }, + { + "epoch": 0.58, + "grad_norm": 3.5096519082422986, + "learning_rate": 1.9515801815718566e-05, + "loss": 0.2567, + "step": 4940 + }, + { + "epoch": 0.58, + "grad_norm": 3.1690240083159993, + "learning_rate": 1.9509836617120856e-05, + "loss": 0.2466, + "step": 4960 + }, + { + "epoch": 0.58, + "grad_norm": 3.2970371296868484, + "learning_rate": 1.950383582135796e-05, + "loss": 0.2762, + "step": 4980 + }, + { + "epoch": 0.59, + "grad_norm": 4.326207366177782, + "learning_rate": 1.9497799450894387e-05, + "loss": 0.2681, + "step": 5000 + }, + { + "epoch": 0.59, + "grad_norm": 3.977255410491104, + "learning_rate": 1.949172752832782e-05, + "loss": 0.235, + "step": 5020 + }, + { + "epoch": 0.59, + "grad_norm": 3.5634772315616665, + "learning_rate": 1.9485620076389036e-05, + "loss": 0.254, + "step": 5040 + }, + { + "epoch": 0.59, + "grad_norm": 2.9667663371741377, + "learning_rate": 1.9479477117941815e-05, + "loss": 0.2582, + "step": 5060 + }, + { + "epoch": 0.59, + "grad_norm": 3.628580495546253, + "learning_rate": 1.9473298675982863e-05, + "loss": 0.2526, + "step": 5080 + }, + { + "epoch": 0.6, + "grad_norm": 7.393689026504584, + "learning_rate": 1.946708477364172e-05, + "loss": 0.2587, + "step": 5100 + }, + { + "epoch": 0.6, + "grad_norm": 3.1975448011133407, + "learning_rate": 1.9460835434180677e-05, + "loss": 0.2679, + "step": 5120 + }, + { + "epoch": 0.6, + "grad_norm": 3.231683615655966, + "learning_rate": 1.9454550680994684e-05, + "loss": 0.2589, + "step": 5140 + }, + { + "epoch": 0.6, + "grad_norm": 2.956736460495959, + "learning_rate": 1.9448230537611266e-05, + "loss": 0.2594, + "step": 5160 + }, + { + "epoch": 0.61, + "grad_norm": 3.326543297808526, + "learning_rate": 1.9441875027690435e-05, + "loss": 0.2556, + "step": 5180 + }, + { + "epoch": 0.61, + "grad_norm": 3.635188025395725, + "learning_rate": 1.94354841750246e-05, + "loss": 0.2391, + "step": 5200 + }, + { + "epoch": 0.61, + "grad_norm": 4.934642109517512, + "learning_rate": 1.942905800353848e-05, + "loss": 0.2576, + "step": 5220 + }, + { + "epoch": 0.61, + "grad_norm": 3.8685357103793767, + "learning_rate": 1.9422596537289016e-05, + "loss": 0.2735, + "step": 5240 + }, + { + "epoch": 0.62, + "grad_norm": 4.387694986472, + "learning_rate": 1.941609980046527e-05, + "loss": 0.2584, + "step": 5260 + }, + { + "epoch": 0.62, + "grad_norm": 2.8046325848231977, + "learning_rate": 1.9409567817388347e-05, + "loss": 0.2527, + "step": 5280 + }, + { + "epoch": 0.62, + "grad_norm": 3.4012361074638466, + "learning_rate": 1.94030006125113e-05, + "loss": 0.2566, + "step": 5300 + }, + { + "epoch": 0.62, + "grad_norm": 5.450089536158873, + "learning_rate": 1.9396398210419038e-05, + "loss": 0.2667, + "step": 5320 + }, + { + "epoch": 0.62, + "grad_norm": 3.7967573233998677, + "learning_rate": 1.938976063582824e-05, + "loss": 0.2536, + "step": 5340 + }, + { + "epoch": 0.63, + "grad_norm": 3.0487888683110866, + "learning_rate": 1.938308791358723e-05, + "loss": 0.2673, + "step": 5360 + }, + { + "epoch": 0.63, + "grad_norm": 3.2736066169116285, + "learning_rate": 1.9376380068675956e-05, + "loss": 0.2592, + "step": 5380 + }, + { + "epoch": 0.63, + "grad_norm": 9.727952335441667, + "learning_rate": 1.93696371262058e-05, + "loss": 0.273, + "step": 5400 + }, + { + "epoch": 0.63, + "grad_norm": 3.3824664284674686, + "learning_rate": 1.936285911141958e-05, + "loss": 0.2787, + "step": 5420 + }, + { + "epoch": 0.64, + "grad_norm": 6.879080376487243, + "learning_rate": 1.935604604969138e-05, + "loss": 0.2698, + "step": 5440 + }, + { + "epoch": 0.64, + "grad_norm": 4.579168728014292, + "learning_rate": 1.934919796652649e-05, + "loss": 0.2589, + "step": 5460 + }, + { + "epoch": 0.64, + "grad_norm": 4.6776324362587784, + "learning_rate": 1.9342314887561323e-05, + "loss": 0.2544, + "step": 5480 + }, + { + "epoch": 0.64, + "grad_norm": 3.693997622998789, + "learning_rate": 1.9335396838563282e-05, + "loss": 0.2446, + "step": 5500 + }, + { + "epoch": 0.65, + "grad_norm": 5.06873813162095, + "learning_rate": 1.932844384543069e-05, + "loss": 0.2618, + "step": 5520 + }, + { + "epoch": 0.65, + "grad_norm": 6.816411566527297, + "learning_rate": 1.9321455934192693e-05, + "loss": 0.2614, + "step": 5540 + }, + { + "epoch": 0.65, + "grad_norm": 3.5475232040366853, + "learning_rate": 1.9314433131009146e-05, + "loss": 0.2942, + "step": 5560 + }, + { + "epoch": 0.65, + "grad_norm": 5.287550198652632, + "learning_rate": 1.9307375462170527e-05, + "loss": 0.2482, + "step": 5580 + }, + { + "epoch": 0.66, + "grad_norm": 11.389854038766874, + "learning_rate": 1.9300282954097847e-05, + "loss": 0.245, + "step": 5600 + }, + { + "epoch": 0.66, + "grad_norm": 4.178435806663215, + "learning_rate": 1.9293155633342534e-05, + "loss": 0.2534, + "step": 5620 + }, + { + "epoch": 0.66, + "grad_norm": 3.1710922703595035, + "learning_rate": 1.928599352658633e-05, + "loss": 0.2622, + "step": 5640 + }, + { + "epoch": 0.66, + "grad_norm": 2.6946627953356996, + "learning_rate": 1.927879666064122e-05, + "loss": 0.2626, + "step": 5660 + }, + { + "epoch": 0.66, + "grad_norm": 4.8581812333670955, + "learning_rate": 1.92715650624493e-05, + "loss": 0.2346, + "step": 5680 + }, + { + "epoch": 0.67, + "grad_norm": 3.092927765677092, + "learning_rate": 1.926429875908269e-05, + "loss": 0.2585, + "step": 5700 + }, + { + "epoch": 0.67, + "grad_norm": 4.029661826813332, + "learning_rate": 1.9256997777743437e-05, + "loss": 0.2467, + "step": 5720 + }, + { + "epoch": 0.67, + "grad_norm": 4.642691104141709, + "learning_rate": 1.924966214576341e-05, + "loss": 0.2681, + "step": 5740 + }, + { + "epoch": 0.67, + "grad_norm": 7.315047122076067, + "learning_rate": 1.9242291890604186e-05, + "loss": 0.2508, + "step": 5760 + }, + { + "epoch": 0.68, + "grad_norm": 3.540549397482083, + "learning_rate": 1.923488703985696e-05, + "loss": 0.2592, + "step": 5780 + }, + { + "epoch": 0.68, + "grad_norm": 3.475018859373003, + "learning_rate": 1.9227447621242458e-05, + "loss": 0.2422, + "step": 5800 + }, + { + "epoch": 0.68, + "grad_norm": 3.8676644079356355, + "learning_rate": 1.921997366261078e-05, + "loss": 0.2603, + "step": 5820 + }, + { + "epoch": 0.68, + "grad_norm": 3.5885966628483, + "learning_rate": 1.9212465191941346e-05, + "loss": 0.2542, + "step": 5840 + }, + { + "epoch": 0.69, + "grad_norm": 3.6495890137770886, + "learning_rate": 1.920492223734279e-05, + "loss": 0.2374, + "step": 5860 + }, + { + "epoch": 0.69, + "grad_norm": 4.323488251305847, + "learning_rate": 1.9197344827052813e-05, + "loss": 0.2687, + "step": 5880 + }, + { + "epoch": 0.69, + "grad_norm": 5.051497760408599, + "learning_rate": 1.9189732989438123e-05, + "loss": 0.2678, + "step": 5900 + }, + { + "epoch": 0.69, + "grad_norm": 3.5559342765441495, + "learning_rate": 1.9182086752994295e-05, + "loss": 0.2726, + "step": 5920 + }, + { + "epoch": 0.7, + "grad_norm": 3.029184640641301, + "learning_rate": 1.9174406146345694e-05, + "loss": 0.246, + "step": 5940 + }, + { + "epoch": 0.7, + "grad_norm": 3.4792605785250994, + "learning_rate": 1.9166691198245343e-05, + "loss": 0.2843, + "step": 5960 + }, + { + "epoch": 0.7, + "grad_norm": 3.0247490770922, + "learning_rate": 1.915894193757483e-05, + "loss": 0.2528, + "step": 5980 + }, + { + "epoch": 0.7, + "grad_norm": 3.7727567007247855, + "learning_rate": 1.915115839334419e-05, + "loss": 0.2442, + "step": 6000 + }, + { + "epoch": 0.7, + "grad_norm": 2.31881357193246, + "learning_rate": 1.9143340594691806e-05, + "loss": 0.2561, + "step": 6020 + }, + { + "epoch": 0.71, + "grad_norm": 3.138315746753083, + "learning_rate": 1.913548857088429e-05, + "loss": 0.2636, + "step": 6040 + }, + { + "epoch": 0.71, + "grad_norm": 8.010145906221986, + "learning_rate": 1.9127602351316393e-05, + "loss": 0.2876, + "step": 6060 + }, + { + "epoch": 0.71, + "grad_norm": 3.867596475572881, + "learning_rate": 1.9119681965510858e-05, + "loss": 0.2659, + "step": 6080 + }, + { + "epoch": 0.71, + "grad_norm": 3.3587166074477777, + "learning_rate": 1.9111727443118357e-05, + "loss": 0.2634, + "step": 6100 + }, + { + "epoch": 0.72, + "grad_norm": 2.89895288044486, + "learning_rate": 1.9103738813917336e-05, + "loss": 0.2533, + "step": 6120 + }, + { + "epoch": 0.72, + "grad_norm": 4.037752097559083, + "learning_rate": 1.909571610781393e-05, + "loss": 0.2485, + "step": 6140 + }, + { + "epoch": 0.72, + "grad_norm": 3.763608653829743, + "learning_rate": 1.908765935484185e-05, + "loss": 0.259, + "step": 6160 + }, + { + "epoch": 0.72, + "grad_norm": 3.161229351032073, + "learning_rate": 1.9079568585162248e-05, + "loss": 0.2508, + "step": 6180 + }, + { + "epoch": 0.73, + "grad_norm": 2.725004696623228, + "learning_rate": 1.9071443829063644e-05, + "loss": 0.2524, + "step": 6200 + }, + { + "epoch": 0.73, + "grad_norm": 3.7116897623627985, + "learning_rate": 1.906328511696177e-05, + "loss": 0.2598, + "step": 6220 + }, + { + "epoch": 0.73, + "grad_norm": 2.9004110110815544, + "learning_rate": 1.9055092479399474e-05, + "loss": 0.2553, + "step": 6240 + }, + { + "epoch": 0.73, + "grad_norm": 4.311030925472568, + "learning_rate": 1.9046865947046627e-05, + "loss": 0.2678, + "step": 6260 + }, + { + "epoch": 0.73, + "grad_norm": 3.5377476420175977, + "learning_rate": 1.903860555069997e-05, + "loss": 0.2379, + "step": 6280 + }, + { + "epoch": 0.74, + "grad_norm": 3.58934533910836, + "learning_rate": 1.9030311321283022e-05, + "loss": 0.2608, + "step": 6300 + }, + { + "epoch": 0.74, + "grad_norm": 3.4612535457241624, + "learning_rate": 1.9021983289845956e-05, + "loss": 0.246, + "step": 6320 + }, + { + "epoch": 0.74, + "grad_norm": 3.9763337015158258, + "learning_rate": 1.9013621487565498e-05, + "loss": 0.238, + "step": 6340 + }, + { + "epoch": 0.74, + "grad_norm": 3.05488683337997, + "learning_rate": 1.9005225945744782e-05, + "loss": 0.2472, + "step": 6360 + }, + { + "epoch": 0.75, + "grad_norm": 3.2350295137123313, + "learning_rate": 1.8996796695813256e-05, + "loss": 0.261, + "step": 6380 + }, + { + "epoch": 0.75, + "grad_norm": 3.2500906711548123, + "learning_rate": 1.8988333769326564e-05, + "loss": 0.2641, + "step": 6400 + }, + { + "epoch": 0.75, + "grad_norm": 2.815087548041305, + "learning_rate": 1.8979837197966415e-05, + "loss": 0.2583, + "step": 6420 + }, + { + "epoch": 0.75, + "grad_norm": 3.7397965536014257, + "learning_rate": 1.8971307013540464e-05, + "loss": 0.2485, + "step": 6440 + }, + { + "epoch": 0.76, + "grad_norm": 2.7592373806083423, + "learning_rate": 1.8962743247982205e-05, + "loss": 0.2641, + "step": 6460 + }, + { + "epoch": 0.76, + "grad_norm": 2.818080452105616, + "learning_rate": 1.895414593335085e-05, + "loss": 0.2554, + "step": 6480 + }, + { + "epoch": 0.76, + "grad_norm": 3.627385505536174, + "learning_rate": 1.8945515101831203e-05, + "loss": 0.2693, + "step": 6500 + }, + { + "epoch": 0.76, + "grad_norm": 3.255509767944512, + "learning_rate": 1.893685078573354e-05, + "loss": 0.2459, + "step": 6520 + }, + { + "epoch": 0.77, + "grad_norm": 4.135182845698409, + "learning_rate": 1.8928153017493488e-05, + "loss": 0.2499, + "step": 6540 + }, + { + "epoch": 0.77, + "grad_norm": 5.892816677202276, + "learning_rate": 1.8919421829671904e-05, + "loss": 0.2786, + "step": 6560 + }, + { + "epoch": 0.77, + "grad_norm": 3.479892875732629, + "learning_rate": 1.891065725495476e-05, + "loss": 0.265, + "step": 6580 + }, + { + "epoch": 0.77, + "grad_norm": 3.8343809741781216, + "learning_rate": 1.890185932615301e-05, + "loss": 0.2527, + "step": 6600 + }, + { + "epoch": 0.77, + "grad_norm": 3.1643033159708107, + "learning_rate": 1.8893028076202472e-05, + "loss": 0.2741, + "step": 6620 + }, + { + "epoch": 0.78, + "grad_norm": 4.477352585232187, + "learning_rate": 1.8884163538163708e-05, + "loss": 0.2525, + "step": 6640 + }, + { + "epoch": 0.78, + "grad_norm": 3.529760400999529, + "learning_rate": 1.8875265745221894e-05, + "loss": 0.2643, + "step": 6660 + }, + { + "epoch": 0.78, + "grad_norm": 2.3110189076619525, + "learning_rate": 1.8866334730686703e-05, + "loss": 0.2721, + "step": 6680 + }, + { + "epoch": 0.78, + "grad_norm": 3.685226774238388, + "learning_rate": 1.8857370527992165e-05, + "loss": 0.2549, + "step": 6700 + }, + { + "epoch": 0.79, + "grad_norm": 3.8074891978039864, + "learning_rate": 1.8848373170696567e-05, + "loss": 0.2418, + "step": 6720 + }, + { + "epoch": 0.79, + "grad_norm": 7.507877790185021, + "learning_rate": 1.8839342692482302e-05, + "loss": 0.2451, + "step": 6740 + }, + { + "epoch": 0.79, + "grad_norm": 2.7359733051983186, + "learning_rate": 1.8830279127155764e-05, + "loss": 0.2484, + "step": 6760 + }, + { + "epoch": 0.79, + "grad_norm": 2.4635290638854657, + "learning_rate": 1.88211825086472e-05, + "loss": 0.2407, + "step": 6780 + }, + { + "epoch": 0.8, + "grad_norm": 3.040130192327906, + "learning_rate": 1.8812052871010605e-05, + "loss": 0.2544, + "step": 6800 + }, + { + "epoch": 0.8, + "grad_norm": 5.610778632978326, + "learning_rate": 1.8802890248423574e-05, + "loss": 0.2397, + "step": 6820 + }, + { + "epoch": 0.8, + "grad_norm": 4.156547822535594, + "learning_rate": 1.8793694675187197e-05, + "loss": 0.2635, + "step": 6840 + }, + { + "epoch": 0.8, + "grad_norm": 4.057314806159278, + "learning_rate": 1.8784466185725902e-05, + "loss": 0.2598, + "step": 6860 + }, + { + "epoch": 0.81, + "grad_norm": 8.314874066232434, + "learning_rate": 1.877520481458736e-05, + "loss": 0.2706, + "step": 6880 + }, + { + "epoch": 0.81, + "grad_norm": 3.7767818313058195, + "learning_rate": 1.8765910596442316e-05, + "loss": 0.2483, + "step": 6900 + }, + { + "epoch": 0.81, + "grad_norm": 3.53905010431742, + "learning_rate": 1.8756583566084502e-05, + "loss": 0.2683, + "step": 6920 + }, + { + "epoch": 0.81, + "grad_norm": 3.2053331160944625, + "learning_rate": 1.8747223758430467e-05, + "loss": 0.2567, + "step": 6940 + }, + { + "epoch": 0.81, + "grad_norm": 3.655977418308093, + "learning_rate": 1.8737831208519478e-05, + "loss": 0.2433, + "step": 6960 + }, + { + "epoch": 0.82, + "grad_norm": 3.334646935711793, + "learning_rate": 1.8728405951513376e-05, + "loss": 0.2442, + "step": 6980 + }, + { + "epoch": 0.82, + "grad_norm": 3.3678493336120505, + "learning_rate": 1.8718948022696427e-05, + "loss": 0.2634, + "step": 7000 + }, + { + "epoch": 0.82, + "grad_norm": 2.9941614875638956, + "learning_rate": 1.8709457457475224e-05, + "loss": 0.2719, + "step": 7020 + }, + { + "epoch": 0.82, + "grad_norm": 3.3212764283242495, + "learning_rate": 1.8699934291378534e-05, + "loss": 0.2616, + "step": 7040 + }, + { + "epoch": 0.83, + "grad_norm": 3.7449113015848963, + "learning_rate": 1.8690378560057165e-05, + "loss": 0.251, + "step": 7060 + }, + { + "epoch": 0.83, + "grad_norm": 2.588981565454773, + "learning_rate": 1.868079029928383e-05, + "loss": 0.2345, + "step": 7080 + }, + { + "epoch": 0.83, + "grad_norm": 4.188807496871825, + "learning_rate": 1.8671169544953033e-05, + "loss": 0.2537, + "step": 7100 + }, + { + "epoch": 0.83, + "grad_norm": 4.66518430781191, + "learning_rate": 1.866151633308091e-05, + "loss": 0.2397, + "step": 7120 + }, + { + "epoch": 0.84, + "grad_norm": 4.197676860816234, + "learning_rate": 1.8651830699805108e-05, + "loss": 0.2446, + "step": 7140 + }, + { + "epoch": 0.84, + "grad_norm": 3.6434672643458543, + "learning_rate": 1.8642112681384645e-05, + "loss": 0.2402, + "step": 7160 + }, + { + "epoch": 0.84, + "grad_norm": 3.950387723218738, + "learning_rate": 1.8632362314199778e-05, + "loss": 0.2536, + "step": 7180 + }, + { + "epoch": 0.84, + "grad_norm": 9.658211616844094, + "learning_rate": 1.8622579634751864e-05, + "loss": 0.2541, + "step": 7200 + }, + { + "epoch": 0.84, + "grad_norm": 3.799337836345952, + "learning_rate": 1.8612764679663222e-05, + "loss": 0.2586, + "step": 7220 + }, + { + "epoch": 0.85, + "grad_norm": 3.737906856782671, + "learning_rate": 1.8602917485677e-05, + "loss": 0.2539, + "step": 7240 + }, + { + "epoch": 0.85, + "grad_norm": 3.2570157254536016, + "learning_rate": 1.859303808965703e-05, + "loss": 0.2533, + "step": 7260 + }, + { + "epoch": 0.85, + "grad_norm": 2.8736525985585226, + "learning_rate": 1.85831265285877e-05, + "loss": 0.2532, + "step": 7280 + }, + { + "epoch": 0.85, + "grad_norm": 3.5447389329516503, + "learning_rate": 1.857318283957381e-05, + "loss": 0.2387, + "step": 7300 + }, + { + "epoch": 0.86, + "grad_norm": 2.7468496831393403, + "learning_rate": 1.8563207059840436e-05, + "loss": 0.2455, + "step": 7320 + }, + { + "epoch": 0.86, + "grad_norm": 3.483830293921577, + "learning_rate": 1.8553199226732784e-05, + "loss": 0.2456, + "step": 7340 + }, + { + "epoch": 0.86, + "grad_norm": 3.3132796899404635, + "learning_rate": 1.8543159377716054e-05, + "loss": 0.2564, + "step": 7360 + }, + { + "epoch": 0.86, + "grad_norm": 3.3795508326649575, + "learning_rate": 1.8533087550375302e-05, + "loss": 0.2527, + "step": 7380 + }, + { + "epoch": 0.87, + "grad_norm": 4.001941448174156, + "learning_rate": 1.8522983782415306e-05, + "loss": 0.2433, + "step": 7400 + }, + { + "epoch": 0.87, + "grad_norm": 3.4205382361927117, + "learning_rate": 1.85128481116604e-05, + "loss": 0.2538, + "step": 7420 + }, + { + "epoch": 0.87, + "grad_norm": 3.896330422632185, + "learning_rate": 1.8502680576054365e-05, + "loss": 0.2478, + "step": 7440 + }, + { + "epoch": 0.87, + "grad_norm": 5.020228094868969, + "learning_rate": 1.849248121366026e-05, + "loss": 0.2427, + "step": 7460 + }, + { + "epoch": 0.88, + "grad_norm": 3.574276782956127, + "learning_rate": 1.8482250062660293e-05, + "loss": 0.2389, + "step": 7480 + }, + { + "epoch": 0.88, + "grad_norm": 3.5309579103911184, + "learning_rate": 1.847198716135568e-05, + "loss": 0.2567, + "step": 7500 + }, + { + "epoch": 0.88, + "grad_norm": 3.1760414458004, + "learning_rate": 1.846169254816649e-05, + "loss": 0.2291, + "step": 7520 + }, + { + "epoch": 0.88, + "grad_norm": 4.214334621561301, + "learning_rate": 1.8451366261631514e-05, + "loss": 0.2624, + "step": 7540 + }, + { + "epoch": 0.88, + "grad_norm": 2.977221959069712, + "learning_rate": 1.8441008340408115e-05, + "loss": 0.2425, + "step": 7560 + }, + { + "epoch": 0.89, + "grad_norm": 3.78383188925913, + "learning_rate": 1.8430618823272076e-05, + "loss": 0.2437, + "step": 7580 + }, + { + "epoch": 0.89, + "grad_norm": 3.2432934448936224, + "learning_rate": 1.842019774911747e-05, + "loss": 0.2628, + "step": 7600 + }, + { + "epoch": 0.89, + "grad_norm": 3.05382414419798, + "learning_rate": 1.8409745156956503e-05, + "loss": 0.2332, + "step": 7620 + }, + { + "epoch": 0.89, + "grad_norm": 3.098744070302099, + "learning_rate": 1.839926108591937e-05, + "loss": 0.2359, + "step": 7640 + }, + { + "epoch": 0.9, + "grad_norm": 3.427098715766002, + "learning_rate": 1.8388745575254117e-05, + "loss": 0.2429, + "step": 7660 + }, + { + "epoch": 0.9, + "grad_norm": 5.630711008711337, + "learning_rate": 1.8378198664326474e-05, + "loss": 0.2416, + "step": 7680 + }, + { + "epoch": 0.9, + "grad_norm": 3.1144723181358622, + "learning_rate": 1.836762039261974e-05, + "loss": 0.2406, + "step": 7700 + }, + { + "epoch": 0.9, + "grad_norm": 4.070301722184627, + "learning_rate": 1.8357010799734588e-05, + "loss": 0.234, + "step": 7720 + }, + { + "epoch": 0.91, + "grad_norm": 3.255428182680105, + "learning_rate": 1.834636992538898e-05, + "loss": 0.2365, + "step": 7740 + }, + { + "epoch": 0.91, + "grad_norm": 3.4823775104086576, + "learning_rate": 1.8335697809417952e-05, + "loss": 0.2376, + "step": 7760 + }, + { + "epoch": 0.91, + "grad_norm": 5.528269163918348, + "learning_rate": 1.8324994491773507e-05, + "loss": 0.2679, + "step": 7780 + }, + { + "epoch": 0.91, + "grad_norm": 2.7221100312640467, + "learning_rate": 1.8314260012524462e-05, + "loss": 0.2544, + "step": 7800 + }, + { + "epoch": 0.92, + "grad_norm": 3.224515621271091, + "learning_rate": 1.8303494411856278e-05, + "loss": 0.2586, + "step": 7820 + }, + { + "epoch": 0.92, + "grad_norm": 2.798714485709818, + "learning_rate": 1.829269773007092e-05, + "loss": 0.2277, + "step": 7840 + }, + { + "epoch": 0.92, + "grad_norm": 3.890849401904209, + "learning_rate": 1.8281870007586735e-05, + "loss": 0.2565, + "step": 7860 + }, + { + "epoch": 0.92, + "grad_norm": 3.701194549732654, + "learning_rate": 1.8271011284938233e-05, + "loss": 0.2307, + "step": 7880 + }, + { + "epoch": 0.92, + "grad_norm": 3.9057568048025875, + "learning_rate": 1.8260121602776003e-05, + "loss": 0.2464, + "step": 7900 + }, + { + "epoch": 0.93, + "grad_norm": 3.397196786583742, + "learning_rate": 1.8249201001866533e-05, + "loss": 0.25, + "step": 7920 + }, + { + "epoch": 0.93, + "grad_norm": 3.6136942859367496, + "learning_rate": 1.8238249523092046e-05, + "loss": 0.2398, + "step": 7940 + }, + { + "epoch": 0.93, + "grad_norm": 3.6009293204428765, + "learning_rate": 1.8227267207450363e-05, + "loss": 0.2356, + "step": 7960 + }, + { + "epoch": 0.93, + "grad_norm": 3.1462270393488585, + "learning_rate": 1.8216254096054747e-05, + "loss": 0.2292, + "step": 7980 + }, + { + "epoch": 0.94, + "grad_norm": 3.453766085807792, + "learning_rate": 1.8205210230133756e-05, + "loss": 0.2532, + "step": 8000 + }, + { + "epoch": 0.94, + "grad_norm": 4.718419953986993, + "learning_rate": 1.8194135651031063e-05, + "loss": 0.2379, + "step": 8020 + }, + { + "epoch": 0.94, + "grad_norm": 4.074122309800583, + "learning_rate": 1.8183030400205335e-05, + "loss": 0.247, + "step": 8040 + }, + { + "epoch": 0.94, + "grad_norm": 3.597629476607082, + "learning_rate": 1.817189451923005e-05, + "loss": 0.2563, + "step": 8060 + }, + { + "epoch": 0.95, + "grad_norm": 3.2550443969136436, + "learning_rate": 1.8160728049793357e-05, + "loss": 0.2425, + "step": 8080 + }, + { + "epoch": 0.95, + "grad_norm": 2.6027515890006505, + "learning_rate": 1.8149531033697913e-05, + "loss": 0.2329, + "step": 8100 + }, + { + "epoch": 0.95, + "grad_norm": 3.705349797095471, + "learning_rate": 1.8138303512860737e-05, + "loss": 0.2521, + "step": 8120 + }, + { + "epoch": 0.95, + "grad_norm": 3.510055266811937, + "learning_rate": 1.812704552931303e-05, + "loss": 0.2413, + "step": 8140 + }, + { + "epoch": 0.95, + "grad_norm": 3.2863076604207997, + "learning_rate": 1.8115757125200054e-05, + "loss": 0.2471, + "step": 8160 + }, + { + "epoch": 0.96, + "grad_norm": 3.174018473859266, + "learning_rate": 1.810443834278094e-05, + "loss": 0.2429, + "step": 8180 + }, + { + "epoch": 0.96, + "grad_norm": 4.78684161965386, + "learning_rate": 1.8093089224428533e-05, + "loss": 0.21, + "step": 8200 + }, + { + "epoch": 0.96, + "grad_norm": 3.778736256263061, + "learning_rate": 1.8081709812629265e-05, + "loss": 0.2438, + "step": 8220 + }, + { + "epoch": 0.96, + "grad_norm": 2.011964890319697, + "learning_rate": 1.807030014998296e-05, + "loss": 0.2235, + "step": 8240 + }, + { + "epoch": 0.97, + "grad_norm": 2.469231474076661, + "learning_rate": 1.8058860279202698e-05, + "loss": 0.2361, + "step": 8260 + }, + { + "epoch": 0.97, + "grad_norm": 4.194473608952758, + "learning_rate": 1.804739024311463e-05, + "loss": 0.2485, + "step": 8280 + }, + { + "epoch": 0.97, + "grad_norm": 3.648802255045255, + "learning_rate": 1.8035890084657854e-05, + "loss": 0.2448, + "step": 8300 + }, + { + "epoch": 0.97, + "grad_norm": 2.8256525011924443, + "learning_rate": 1.8024359846884213e-05, + "loss": 0.2688, + "step": 8320 + }, + { + "epoch": 0.98, + "grad_norm": 4.409589867929299, + "learning_rate": 1.8012799572958173e-05, + "loss": 0.2405, + "step": 8340 + }, + { + "epoch": 0.98, + "grad_norm": 5.024619525530542, + "learning_rate": 1.8001209306156634e-05, + "loss": 0.2284, + "step": 8360 + }, + { + "epoch": 0.98, + "grad_norm": 3.5087106485202026, + "learning_rate": 1.7989589089868774e-05, + "loss": 0.2298, + "step": 8380 + }, + { + "epoch": 0.98, + "grad_norm": 4.40165070000877, + "learning_rate": 1.7977938967595897e-05, + "loss": 0.2507, + "step": 8400 + }, + { + "epoch": 0.99, + "grad_norm": 3.9660729217673705, + "learning_rate": 1.7966258982951255e-05, + "loss": 0.2459, + "step": 8420 + }, + { + "epoch": 0.99, + "grad_norm": 4.665706172372416, + "learning_rate": 1.7954549179659903e-05, + "loss": 0.2458, + "step": 8440 + }, + { + "epoch": 0.99, + "grad_norm": 3.2908069759526333, + "learning_rate": 1.794280960155851e-05, + "loss": 0.2251, + "step": 8460 + }, + { + "epoch": 0.99, + "grad_norm": 6.7504752662905725, + "learning_rate": 1.7931040292595227e-05, + "loss": 0.254, + "step": 8480 + }, + { + "epoch": 0.99, + "grad_norm": 2.2047073417435787, + "learning_rate": 1.791924129682949e-05, + "loss": 0.2456, + "step": 8500 + }, + { + "epoch": 1.0, + "grad_norm": 3.9048911210632236, + "learning_rate": 1.7907412658431877e-05, + "loss": 0.2307, + "step": 8520 + }, + { + "epoch": 1.0, + "grad_norm": 3.33290909610624, + "learning_rate": 1.7895554421683938e-05, + "loss": 0.2514, + "step": 8540 + }, + { + "epoch": 1.0, + "grad_norm": 3.7515500997507605, + "learning_rate": 1.7883666630978024e-05, + "loss": 0.2189, + "step": 8560 + }, + { + "epoch": 1.0, + "grad_norm": 2.495653761409699, + "learning_rate": 1.7871749330817114e-05, + "loss": 0.216, + "step": 8580 + }, + { + "epoch": 1.01, + "grad_norm": 3.1012959017387556, + "learning_rate": 1.785980256581468e-05, + "loss": 0.2183, + "step": 8600 + }, + { + "epoch": 1.01, + "grad_norm": 2.7851575340492745, + "learning_rate": 1.7847826380694482e-05, + "loss": 0.2086, + "step": 8620 + }, + { + "epoch": 1.01, + "grad_norm": 5.673400169424141, + "learning_rate": 1.783582082029042e-05, + "loss": 0.2216, + "step": 8640 + }, + { + "epoch": 1.01, + "grad_norm": 3.9347827557672113, + "learning_rate": 1.7823785929546363e-05, + "loss": 0.2351, + "step": 8660 + }, + { + "epoch": 1.02, + "grad_norm": 4.263640336790709, + "learning_rate": 1.781172175351599e-05, + "loss": 0.2237, + "step": 8680 + }, + { + "epoch": 1.02, + "grad_norm": 2.8181611623874034, + "learning_rate": 1.779962833736259e-05, + "loss": 0.2078, + "step": 8700 + }, + { + "epoch": 1.02, + "grad_norm": 3.826255497546828, + "learning_rate": 1.7787505726358937e-05, + "loss": 0.2224, + "step": 8720 + }, + { + "epoch": 1.02, + "grad_norm": 3.2239814391820603, + "learning_rate": 1.777535396588709e-05, + "loss": 0.2196, + "step": 8740 + }, + { + "epoch": 1.03, + "grad_norm": 3.4517898394902753, + "learning_rate": 1.7763173101438232e-05, + "loss": 0.2056, + "step": 8760 + }, + { + "epoch": 1.03, + "grad_norm": 4.604513417070624, + "learning_rate": 1.7750963178612498e-05, + "loss": 0.2434, + "step": 8780 + }, + { + "epoch": 1.03, + "grad_norm": 3.835720949606093, + "learning_rate": 1.7738724243118807e-05, + "loss": 0.2282, + "step": 8800 + }, + { + "epoch": 1.03, + "grad_norm": 3.437975625165332, + "learning_rate": 1.7726456340774684e-05, + "loss": 0.25, + "step": 8820 + }, + { + "epoch": 1.03, + "grad_norm": 3.362299298390976, + "learning_rate": 1.771415951750611e-05, + "loss": 0.2301, + "step": 8840 + }, + { + "epoch": 1.04, + "grad_norm": 3.469988077896261, + "learning_rate": 1.7701833819347317e-05, + "loss": 0.2268, + "step": 8860 + }, + { + "epoch": 1.04, + "grad_norm": 3.3257530965726834, + "learning_rate": 1.7689479292440635e-05, + "loss": 0.2362, + "step": 8880 + }, + { + "epoch": 1.04, + "grad_norm": 4.357144130197379, + "learning_rate": 1.767709598303633e-05, + "loss": 0.2179, + "step": 8900 + }, + { + "epoch": 1.04, + "grad_norm": 2.833662313173985, + "learning_rate": 1.7664683937492394e-05, + "loss": 0.2082, + "step": 8920 + }, + { + "epoch": 1.05, + "grad_norm": 3.6248674039596347, + "learning_rate": 1.7652243202274427e-05, + "loss": 0.2399, + "step": 8940 + }, + { + "epoch": 1.05, + "grad_norm": 3.374713885577454, + "learning_rate": 1.7639773823955405e-05, + "loss": 0.2303, + "step": 8960 + }, + { + "epoch": 1.05, + "grad_norm": 4.623143210009699, + "learning_rate": 1.7627275849215543e-05, + "loss": 0.2197, + "step": 8980 + }, + { + "epoch": 1.05, + "grad_norm": 2.7733450860758713, + "learning_rate": 1.7614749324842106e-05, + "loss": 0.2419, + "step": 9000 + }, + { + "epoch": 1.06, + "grad_norm": 4.136130142481253, + "learning_rate": 1.7602194297729238e-05, + "loss": 0.2315, + "step": 9020 + }, + { + "epoch": 1.06, + "grad_norm": 2.5713300459015476, + "learning_rate": 1.758961081487779e-05, + "loss": 0.2285, + "step": 9040 + }, + { + "epoch": 1.06, + "grad_norm": 2.90938161486602, + "learning_rate": 1.7576998923395128e-05, + "loss": 0.2357, + "step": 9060 + }, + { + "epoch": 1.06, + "grad_norm": 3.2699691862001896, + "learning_rate": 1.756435867049498e-05, + "loss": 0.2161, + "step": 9080 + }, + { + "epoch": 1.06, + "grad_norm": 3.353066842604925, + "learning_rate": 1.7551690103497248e-05, + "loss": 0.2379, + "step": 9100 + }, + { + "epoch": 1.07, + "grad_norm": 3.1865972100619815, + "learning_rate": 1.7538993269827817e-05, + "loss": 0.2324, + "step": 9120 + }, + { + "epoch": 1.07, + "grad_norm": 4.115179219937697, + "learning_rate": 1.7526268217018404e-05, + "loss": 0.2165, + "step": 9140 + }, + { + "epoch": 1.07, + "grad_norm": 3.0706453021663553, + "learning_rate": 1.751351499270636e-05, + "loss": 0.2128, + "step": 9160 + }, + { + "epoch": 1.07, + "grad_norm": 3.0307089725880774, + "learning_rate": 1.7500733644634503e-05, + "loss": 0.2153, + "step": 9180 + }, + { + "epoch": 1.08, + "grad_norm": 3.0414096427563435, + "learning_rate": 1.748792422065093e-05, + "loss": 0.2159, + "step": 9200 + }, + { + "epoch": 1.08, + "grad_norm": 3.581867909009255, + "learning_rate": 1.7475086768708844e-05, + "loss": 0.2108, + "step": 9220 + }, + { + "epoch": 1.08, + "grad_norm": 3.3902710057413707, + "learning_rate": 1.7462221336866373e-05, + "loss": 0.2171, + "step": 9240 + }, + { + "epoch": 1.08, + "grad_norm": 4.186442227351193, + "learning_rate": 1.74493279732864e-05, + "loss": 0.213, + "step": 9260 + }, + { + "epoch": 1.09, + "grad_norm": 2.111492196482593, + "learning_rate": 1.743640672623636e-05, + "loss": 0.231, + "step": 9280 + }, + { + "epoch": 1.09, + "grad_norm": 3.3151227536495744, + "learning_rate": 1.7423457644088067e-05, + "loss": 0.2255, + "step": 9300 + }, + { + "epoch": 1.09, + "grad_norm": 2.979480504575997, + "learning_rate": 1.741048077531756e-05, + "loss": 0.2193, + "step": 9320 + }, + { + "epoch": 1.09, + "grad_norm": 8.144688622580304, + "learning_rate": 1.7397476168504877e-05, + "loss": 0.2213, + "step": 9340 + }, + { + "epoch": 1.1, + "grad_norm": 2.3573272104543395, + "learning_rate": 1.738444387233392e-05, + "loss": 0.23, + "step": 9360 + }, + { + "epoch": 1.1, + "grad_norm": 5.626318713423072, + "learning_rate": 1.737138393559222e-05, + "loss": 0.2231, + "step": 9380 + }, + { + "epoch": 1.1, + "grad_norm": 3.659177993436024, + "learning_rate": 1.7358296407170803e-05, + "loss": 0.2201, + "step": 9400 + }, + { + "epoch": 1.1, + "grad_norm": 4.1924943889109505, + "learning_rate": 1.734518133606398e-05, + "loss": 0.2211, + "step": 9420 + }, + { + "epoch": 1.1, + "grad_norm": 3.1828991465696626, + "learning_rate": 1.7332038771369178e-05, + "loss": 0.2178, + "step": 9440 + }, + { + "epoch": 1.11, + "grad_norm": 3.210581516107296, + "learning_rate": 1.7318868762286735e-05, + "loss": 0.2252, + "step": 9460 + }, + { + "epoch": 1.11, + "grad_norm": 3.1129128729970272, + "learning_rate": 1.7305671358119743e-05, + "loss": 0.2218, + "step": 9480 + }, + { + "epoch": 1.11, + "grad_norm": 3.041995792784782, + "learning_rate": 1.7292446608273838e-05, + "loss": 0.2089, + "step": 9500 + }, + { + "epoch": 1.11, + "grad_norm": 4.604556704379658, + "learning_rate": 1.7279194562257034e-05, + "loss": 0.2104, + "step": 9520 + }, + { + "epoch": 1.12, + "grad_norm": 2.4749100524278944, + "learning_rate": 1.7265915269679533e-05, + "loss": 0.2077, + "step": 9540 + }, + { + "epoch": 1.12, + "grad_norm": 3.256783082638953, + "learning_rate": 1.7252608780253527e-05, + "loss": 0.2222, + "step": 9560 + }, + { + "epoch": 1.12, + "grad_norm": 3.250990203251271, + "learning_rate": 1.723927514379304e-05, + "loss": 0.238, + "step": 9580 + }, + { + "epoch": 1.12, + "grad_norm": 4.17869256557804, + "learning_rate": 1.7225914410213695e-05, + "loss": 0.2264, + "step": 9600 + }, + { + "epoch": 1.13, + "grad_norm": 3.001794596192333, + "learning_rate": 1.7212526629532585e-05, + "loss": 0.2156, + "step": 9620 + }, + { + "epoch": 1.13, + "grad_norm": 4.51471973043563, + "learning_rate": 1.7199111851868038e-05, + "loss": 0.2212, + "step": 9640 + }, + { + "epoch": 1.13, + "grad_norm": 3.343622223025876, + "learning_rate": 1.7185670127439453e-05, + "loss": 0.2292, + "step": 9660 + }, + { + "epoch": 1.13, + "grad_norm": 5.351871498264467, + "learning_rate": 1.7172201506567103e-05, + "loss": 0.233, + "step": 9680 + }, + { + "epoch": 1.14, + "grad_norm": 2.480488358067184, + "learning_rate": 1.7158706039671962e-05, + "loss": 0.2253, + "step": 9700 + }, + { + "epoch": 1.14, + "grad_norm": 9.126953333979776, + "learning_rate": 1.7145183777275484e-05, + "loss": 0.2216, + "step": 9720 + }, + { + "epoch": 1.14, + "grad_norm": 3.0992364496831994, + "learning_rate": 1.7131634769999458e-05, + "loss": 0.2339, + "step": 9740 + }, + { + "epoch": 1.14, + "grad_norm": 3.536084536918467, + "learning_rate": 1.7118059068565775e-05, + "loss": 0.2244, + "step": 9760 + }, + { + "epoch": 1.14, + "grad_norm": 2.48499093727593, + "learning_rate": 1.710445672379627e-05, + "loss": 0.221, + "step": 9780 + }, + { + "epoch": 1.15, + "grad_norm": 3.7310914483561097, + "learning_rate": 1.7090827786612515e-05, + "loss": 0.2356, + "step": 9800 + }, + { + "epoch": 1.15, + "grad_norm": 3.0512655853035975, + "learning_rate": 1.7077172308035627e-05, + "loss": 0.2157, + "step": 9820 + }, + { + "epoch": 1.15, + "grad_norm": 2.6139447976363868, + "learning_rate": 1.70634903391861e-05, + "loss": 0.2251, + "step": 9840 + }, + { + "epoch": 1.15, + "grad_norm": 2.7800692505753135, + "learning_rate": 1.7049781931283576e-05, + "loss": 0.2385, + "step": 9860 + }, + { + "epoch": 1.16, + "grad_norm": 3.8229638009279614, + "learning_rate": 1.7036047135646695e-05, + "loss": 0.2163, + "step": 9880 + }, + { + "epoch": 1.16, + "grad_norm": 3.4670030473465205, + "learning_rate": 1.7022286003692863e-05, + "loss": 0.2172, + "step": 9900 + }, + { + "epoch": 1.16, + "grad_norm": 3.1067639606886908, + "learning_rate": 1.700849858693809e-05, + "loss": 0.228, + "step": 9920 + }, + { + "epoch": 1.16, + "grad_norm": 2.775258672531786, + "learning_rate": 1.6994684936996782e-05, + "loss": 0.2174, + "step": 9940 + }, + { + "epoch": 1.17, + "grad_norm": 3.0648920487627045, + "learning_rate": 1.698084510558155e-05, + "loss": 0.2148, + "step": 9960 + }, + { + "epoch": 1.17, + "grad_norm": 3.2600668007495064, + "learning_rate": 1.696697914450302e-05, + "loss": 0.2133, + "step": 9980 + }, + { + "epoch": 1.17, + "grad_norm": 5.51464299005287, + "learning_rate": 1.6953087105669627e-05, + "loss": 0.2297, + "step": 10000 + }, + { + "epoch": 1.17, + "grad_norm": 3.872665563483291, + "learning_rate": 1.693916904108745e-05, + "loss": 0.2277, + "step": 10020 + }, + { + "epoch": 1.17, + "grad_norm": 3.0527595230989917, + "learning_rate": 1.692522500285998e-05, + "loss": 0.2084, + "step": 10040 + }, + { + "epoch": 1.18, + "grad_norm": 3.009517988727613, + "learning_rate": 1.6911255043187952e-05, + "loss": 0.2105, + "step": 10060 + }, + { + "epoch": 1.18, + "grad_norm": 4.461421265567992, + "learning_rate": 1.6897259214369137e-05, + "loss": 0.2156, + "step": 10080 + }, + { + "epoch": 1.18, + "grad_norm": 3.2651682265027757, + "learning_rate": 1.6883237568798148e-05, + "loss": 0.2223, + "step": 10100 + }, + { + "epoch": 1.18, + "grad_norm": 4.560070279232337, + "learning_rate": 1.6869190158966248e-05, + "loss": 0.2032, + "step": 10120 + }, + { + "epoch": 1.19, + "grad_norm": 2.17671049536392, + "learning_rate": 1.6855117037461147e-05, + "loss": 0.2086, + "step": 10140 + }, + { + "epoch": 1.19, + "grad_norm": 2.835463807449662, + "learning_rate": 1.6841018256966817e-05, + "loss": 0.2285, + "step": 10160 + }, + { + "epoch": 1.19, + "grad_norm": 2.9456662236305045, + "learning_rate": 1.682689387026328e-05, + "loss": 0.2288, + "step": 10180 + }, + { + "epoch": 1.19, + "grad_norm": 2.6149578560812046, + "learning_rate": 1.681274393022641e-05, + "loss": 0.2196, + "step": 10200 + }, + { + "epoch": 1.2, + "grad_norm": 2.9578551501440025, + "learning_rate": 1.6798568489827764e-05, + "loss": 0.2233, + "step": 10220 + }, + { + "epoch": 1.2, + "grad_norm": 3.191017752898972, + "learning_rate": 1.6784367602134337e-05, + "loss": 0.2093, + "step": 10240 + }, + { + "epoch": 1.2, + "grad_norm": 3.353192695439166, + "learning_rate": 1.6770141320308405e-05, + "loss": 0.233, + "step": 10260 + }, + { + "epoch": 1.2, + "grad_norm": 3.8431182590288397, + "learning_rate": 1.6755889697607306e-05, + "loss": 0.2206, + "step": 10280 + }, + { + "epoch": 1.21, + "grad_norm": 3.479325951377326, + "learning_rate": 1.6741612787383236e-05, + "loss": 0.2205, + "step": 10300 + }, + { + "epoch": 1.21, + "grad_norm": 2.5880779155399316, + "learning_rate": 1.6727310643083066e-05, + "loss": 0.2103, + "step": 10320 + }, + { + "epoch": 1.21, + "grad_norm": 7.419706714519934, + "learning_rate": 1.671298331824813e-05, + "loss": 0.207, + "step": 10340 + }, + { + "epoch": 1.21, + "grad_norm": 4.622811908173908, + "learning_rate": 1.6698630866514022e-05, + "loss": 0.2205, + "step": 10360 + }, + { + "epoch": 1.21, + "grad_norm": 4.038977026265583, + "learning_rate": 1.6684253341610414e-05, + "loss": 0.2046, + "step": 10380 + }, + { + "epoch": 1.22, + "grad_norm": 3.2506861329229912, + "learning_rate": 1.6669850797360825e-05, + "loss": 0.2161, + "step": 10400 + }, + { + "epoch": 1.22, + "grad_norm": 3.502316934183813, + "learning_rate": 1.665542328768245e-05, + "loss": 0.2202, + "step": 10420 + }, + { + "epoch": 1.22, + "grad_norm": 2.774170254901442, + "learning_rate": 1.6640970866585937e-05, + "loss": 0.2313, + "step": 10440 + }, + { + "epoch": 1.22, + "grad_norm": 3.3567765943400345, + "learning_rate": 1.6626493588175194e-05, + "loss": 0.2118, + "step": 10460 + }, + { + "epoch": 1.23, + "grad_norm": 3.147468966936314, + "learning_rate": 1.6611991506647186e-05, + "loss": 0.2015, + "step": 10480 + }, + { + "epoch": 1.23, + "grad_norm": 2.794186200047937, + "learning_rate": 1.659746467629173e-05, + "loss": 0.2267, + "step": 10500 + }, + { + "epoch": 1.23, + "grad_norm": 3.208243100437357, + "learning_rate": 1.6582913151491286e-05, + "loss": 0.2272, + "step": 10520 + }, + { + "epoch": 1.23, + "grad_norm": 2.8524752057641996, + "learning_rate": 1.6568336986720773e-05, + "loss": 0.2169, + "step": 10540 + }, + { + "epoch": 1.24, + "grad_norm": 3.3114099688426513, + "learning_rate": 1.6553736236547336e-05, + "loss": 0.2161, + "step": 10560 + }, + { + "epoch": 1.24, + "grad_norm": 3.763983657663605, + "learning_rate": 1.653911095563017e-05, + "loss": 0.2252, + "step": 10580 + }, + { + "epoch": 1.24, + "grad_norm": 3.266499087518214, + "learning_rate": 1.6524461198720303e-05, + "loss": 0.1967, + "step": 10600 + }, + { + "epoch": 1.24, + "grad_norm": 3.9946972745895017, + "learning_rate": 1.650978702066038e-05, + "loss": 0.2174, + "step": 10620 + }, + { + "epoch": 1.25, + "grad_norm": 2.8854543268016055, + "learning_rate": 1.649508847638448e-05, + "loss": 0.2136, + "step": 10640 + }, + { + "epoch": 1.25, + "grad_norm": 3.5610681969299653, + "learning_rate": 1.648036562091789e-05, + "loss": 0.225, + "step": 10660 + }, + { + "epoch": 1.25, + "grad_norm": 8.713288377423561, + "learning_rate": 1.6465618509376923e-05, + "loss": 0.2172, + "step": 10680 + }, + { + "epoch": 1.25, + "grad_norm": 3.2387951119923932, + "learning_rate": 1.6450847196968676e-05, + "loss": 0.2131, + "step": 10700 + }, + { + "epoch": 1.25, + "grad_norm": 3.8639660913863234, + "learning_rate": 1.643605173899086e-05, + "loss": 0.2126, + "step": 10720 + }, + { + "epoch": 1.26, + "grad_norm": 4.013150770232071, + "learning_rate": 1.642123219083156e-05, + "loss": 0.2253, + "step": 10740 + }, + { + "epoch": 1.26, + "grad_norm": 3.448212871140601, + "learning_rate": 1.640638860796908e-05, + "loss": 0.2122, + "step": 10760 + }, + { + "epoch": 1.26, + "grad_norm": 4.008865307397649, + "learning_rate": 1.6391521045971654e-05, + "loss": 0.2198, + "step": 10780 + }, + { + "epoch": 1.26, + "grad_norm": 2.668310542854165, + "learning_rate": 1.6376629560497318e-05, + "loss": 0.2303, + "step": 10800 + }, + { + "epoch": 1.27, + "grad_norm": 4.253939653866208, + "learning_rate": 1.636171420729365e-05, + "loss": 0.2165, + "step": 10820 + }, + { + "epoch": 1.27, + "grad_norm": 3.662446989645766, + "learning_rate": 1.634677504219759e-05, + "loss": 0.2093, + "step": 10840 + }, + { + "epoch": 1.27, + "grad_norm": 3.7845535349608115, + "learning_rate": 1.6331812121135216e-05, + "loss": 0.1997, + "step": 10860 + }, + { + "epoch": 1.27, + "grad_norm": 4.21364980338036, + "learning_rate": 1.6316825500121528e-05, + "loss": 0.2186, + "step": 10880 + }, + { + "epoch": 1.28, + "grad_norm": 3.4198924553007153, + "learning_rate": 1.630181523526027e-05, + "loss": 0.2303, + "step": 10900 + }, + { + "epoch": 1.28, + "grad_norm": 3.569247097261353, + "learning_rate": 1.6286781382743678e-05, + "loss": 0.2084, + "step": 10920 + }, + { + "epoch": 1.28, + "grad_norm": 3.2057387689879966, + "learning_rate": 1.6271723998852305e-05, + "loss": 0.2223, + "step": 10940 + }, + { + "epoch": 1.28, + "grad_norm": 3.438084084566052, + "learning_rate": 1.6256643139954787e-05, + "loss": 0.2087, + "step": 10960 + }, + { + "epoch": 1.28, + "grad_norm": 3.1629151324387075, + "learning_rate": 1.6241538862507647e-05, + "loss": 0.2225, + "step": 10980 + }, + { + "epoch": 1.29, + "grad_norm": 3.1696025055249097, + "learning_rate": 1.622641122305507e-05, + "loss": 0.2315, + "step": 11000 + }, + { + "epoch": 1.29, + "grad_norm": 3.053371916892243, + "learning_rate": 1.621126027822871e-05, + "loss": 0.2116, + "step": 11020 + }, + { + "epoch": 1.29, + "grad_norm": 2.9543774083504766, + "learning_rate": 1.619608608474745e-05, + "loss": 0.2254, + "step": 11040 + }, + { + "epoch": 1.29, + "grad_norm": 3.3693833415742453, + "learning_rate": 1.618088869941723e-05, + "loss": 0.2169, + "step": 11060 + }, + { + "epoch": 1.3, + "grad_norm": 3.9995183654737376, + "learning_rate": 1.6165668179130784e-05, + "loss": 0.2365, + "step": 11080 + }, + { + "epoch": 1.3, + "grad_norm": 13.880528903525311, + "learning_rate": 1.6150424580867476e-05, + "loss": 0.2333, + "step": 11100 + }, + { + "epoch": 1.3, + "grad_norm": 3.1836707357447778, + "learning_rate": 1.613515796169305e-05, + "loss": 0.2237, + "step": 11120 + }, + { + "epoch": 1.3, + "grad_norm": 3.8509352600789506, + "learning_rate": 1.6119868378759436e-05, + "loss": 0.2053, + "step": 11140 + }, + { + "epoch": 1.31, + "grad_norm": 3.2583341112322906, + "learning_rate": 1.610455588930453e-05, + "loss": 0.2139, + "step": 11160 + }, + { + "epoch": 1.31, + "grad_norm": 5.386703878469319, + "learning_rate": 1.6089220550651985e-05, + "loss": 0.2104, + "step": 11180 + }, + { + "epoch": 1.31, + "grad_norm": 3.1002668081338114, + "learning_rate": 1.607386242021099e-05, + "loss": 0.2048, + "step": 11200 + }, + { + "epoch": 1.31, + "grad_norm": 4.67566499111419, + "learning_rate": 1.605848155547605e-05, + "loss": 0.2182, + "step": 11220 + }, + { + "epoch": 1.32, + "grad_norm": 1.8252132735330482, + "learning_rate": 1.6043078014026786e-05, + "loss": 0.2211, + "step": 11240 + }, + { + "epoch": 1.32, + "grad_norm": 3.5104873755754635, + "learning_rate": 1.6027651853527706e-05, + "loss": 0.2082, + "step": 11260 + }, + { + "epoch": 1.32, + "grad_norm": 2.596015767829866, + "learning_rate": 1.6012203131728e-05, + "loss": 0.2053, + "step": 11280 + }, + { + "epoch": 1.32, + "grad_norm": 3.9320961337285247, + "learning_rate": 1.5996731906461314e-05, + "loss": 0.2124, + "step": 11300 + }, + { + "epoch": 1.32, + "grad_norm": 3.239895886290778, + "learning_rate": 1.598123823564554e-05, + "loss": 0.2126, + "step": 11320 + }, + { + "epoch": 1.33, + "grad_norm": 7.536836064136331, + "learning_rate": 1.5965722177282595e-05, + "loss": 0.2199, + "step": 11340 + }, + { + "epoch": 1.33, + "grad_norm": 5.562084911160559, + "learning_rate": 1.5950183789458206e-05, + "loss": 0.2246, + "step": 11360 + }, + { + "epoch": 1.33, + "grad_norm": 2.742122377671127, + "learning_rate": 1.5934623130341695e-05, + "loss": 0.2178, + "step": 11380 + }, + { + "epoch": 1.33, + "grad_norm": 4.143294024804004, + "learning_rate": 1.5919040258185753e-05, + "loss": 0.2189, + "step": 11400 + }, + { + "epoch": 1.34, + "grad_norm": 3.6948127005341975, + "learning_rate": 1.5903435231326233e-05, + "loss": 0.2311, + "step": 11420 + }, + { + "epoch": 1.34, + "grad_norm": 2.8456773200400005, + "learning_rate": 1.5887808108181922e-05, + "loss": 0.2047, + "step": 11440 + }, + { + "epoch": 1.34, + "grad_norm": 4.524975240228682, + "learning_rate": 1.5872158947254328e-05, + "loss": 0.216, + "step": 11460 + }, + { + "epoch": 1.34, + "grad_norm": 3.22887435738582, + "learning_rate": 1.585648780712746e-05, + "loss": 0.2139, + "step": 11480 + }, + { + "epoch": 1.35, + "grad_norm": 2.40402306888702, + "learning_rate": 1.5840794746467615e-05, + "loss": 0.2066, + "step": 11500 + }, + { + "epoch": 1.35, + "grad_norm": 3.4971814386926967, + "learning_rate": 1.5825079824023132e-05, + "loss": 0.2131, + "step": 11520 + }, + { + "epoch": 1.35, + "grad_norm": 3.210588645083174, + "learning_rate": 1.5809343098624212e-05, + "loss": 0.2114, + "step": 11540 + }, + { + "epoch": 1.35, + "grad_norm": 2.632446897574305, + "learning_rate": 1.5793584629182667e-05, + "loss": 0.2151, + "step": 11560 + }, + { + "epoch": 1.36, + "grad_norm": 4.5614133612079755, + "learning_rate": 1.577780447469171e-05, + "loss": 0.2151, + "step": 11580 + }, + { + "epoch": 1.36, + "grad_norm": 2.394202553621421, + "learning_rate": 1.5762002694225735e-05, + "loss": 0.2079, + "step": 11600 + }, + { + "epoch": 1.36, + "grad_norm": 2.6251453177782604, + "learning_rate": 1.5746179346940102e-05, + "loss": 0.2159, + "step": 11620 + }, + { + "epoch": 1.36, + "grad_norm": 3.3437649735445256, + "learning_rate": 1.5730334492070893e-05, + "loss": 0.22, + "step": 11640 + }, + { + "epoch": 1.36, + "grad_norm": 2.63771302867986, + "learning_rate": 1.5714468188934716e-05, + "loss": 0.2136, + "step": 11660 + }, + { + "epoch": 1.37, + "grad_norm": 2.655161017296502, + "learning_rate": 1.5698580496928475e-05, + "loss": 0.2067, + "step": 11680 + }, + { + "epoch": 1.37, + "grad_norm": 2.948812575507089, + "learning_rate": 1.5682671475529137e-05, + "loss": 0.2008, + "step": 11700 + }, + { + "epoch": 1.37, + "grad_norm": 3.469779471829569, + "learning_rate": 1.566674118429352e-05, + "loss": 0.2228, + "step": 11720 + }, + { + "epoch": 1.37, + "grad_norm": 3.2002593829411747, + "learning_rate": 1.5650789682858073e-05, + "loss": 0.2149, + "step": 11740 + }, + { + "epoch": 1.38, + "grad_norm": 3.0178239938264864, + "learning_rate": 1.563481703093864e-05, + "loss": 0.2173, + "step": 11760 + }, + { + "epoch": 1.38, + "grad_norm": 3.359858420431831, + "learning_rate": 1.5618823288330245e-05, + "loss": 0.2109, + "step": 11780 + }, + { + "epoch": 1.38, + "grad_norm": 4.244464186067146, + "learning_rate": 1.5602808514906874e-05, + "loss": 0.2172, + "step": 11800 + }, + { + "epoch": 1.38, + "grad_norm": 3.02678280463322, + "learning_rate": 1.5586772770621236e-05, + "loss": 0.2055, + "step": 11820 + }, + { + "epoch": 1.39, + "grad_norm": 3.2941320498135602, + "learning_rate": 1.5570716115504544e-05, + "loss": 0.2212, + "step": 11840 + }, + { + "epoch": 1.39, + "grad_norm": 3.2434629574232474, + "learning_rate": 1.5554638609666298e-05, + "loss": 0.2183, + "step": 11860 + }, + { + "epoch": 1.39, + "grad_norm": 2.7694471424497342, + "learning_rate": 1.5538540313294058e-05, + "loss": 0.2254, + "step": 11880 + }, + { + "epoch": 1.39, + "grad_norm": 3.5269104616476983, + "learning_rate": 1.5522421286653208e-05, + "loss": 0.2201, + "step": 11900 + }, + { + "epoch": 1.39, + "grad_norm": 3.8299784007845363, + "learning_rate": 1.550628159008674e-05, + "loss": 0.2133, + "step": 11920 + }, + { + "epoch": 1.4, + "grad_norm": 5.026447159875988, + "learning_rate": 1.5490121284015026e-05, + "loss": 0.2177, + "step": 11940 + }, + { + "epoch": 1.4, + "grad_norm": 3.55665309344499, + "learning_rate": 1.5473940428935594e-05, + "loss": 0.2269, + "step": 11960 + }, + { + "epoch": 1.4, + "grad_norm": 3.8149873717812732, + "learning_rate": 1.545773908542289e-05, + "loss": 0.2135, + "step": 11980 + }, + { + "epoch": 1.4, + "grad_norm": 4.453465498906573, + "learning_rate": 1.5441517314128075e-05, + "loss": 0.2066, + "step": 12000 + }, + { + "epoch": 1.41, + "grad_norm": 3.4558864961105167, + "learning_rate": 1.5425275175778765e-05, + "loss": 0.2221, + "step": 12020 + }, + { + "epoch": 1.41, + "grad_norm": 3.0249571615921726, + "learning_rate": 1.5409012731178844e-05, + "loss": 0.2176, + "step": 12040 + }, + { + "epoch": 1.41, + "grad_norm": 3.378968625237132, + "learning_rate": 1.539273004120819e-05, + "loss": 0.2129, + "step": 12060 + }, + { + "epoch": 1.41, + "grad_norm": 3.599643022004301, + "learning_rate": 1.5376427166822493e-05, + "loss": 0.2005, + "step": 12080 + }, + { + "epoch": 1.42, + "grad_norm": 4.369469771352546, + "learning_rate": 1.5360104169052983e-05, + "loss": 0.2072, + "step": 12100 + }, + { + "epoch": 1.42, + "grad_norm": 2.9804449117964618, + "learning_rate": 1.534376110900625e-05, + "loss": 0.2235, + "step": 12120 + }, + { + "epoch": 1.42, + "grad_norm": 3.477739662012998, + "learning_rate": 1.532739804786397e-05, + "loss": 0.2044, + "step": 12140 + }, + { + "epoch": 1.42, + "grad_norm": 4.365666269531915, + "learning_rate": 1.5311015046882695e-05, + "loss": 0.2068, + "step": 12160 + }, + { + "epoch": 1.43, + "grad_norm": 3.097981651118829, + "learning_rate": 1.5294612167393633e-05, + "loss": 0.2034, + "step": 12180 + }, + { + "epoch": 1.43, + "grad_norm": 3.1761710102538485, + "learning_rate": 1.5278189470802402e-05, + "loss": 0.2148, + "step": 12200 + }, + { + "epoch": 1.43, + "grad_norm": 6.183507999444912, + "learning_rate": 1.526174701858881e-05, + "loss": 0.2252, + "step": 12220 + }, + { + "epoch": 1.43, + "grad_norm": 3.794441982049424, + "learning_rate": 1.5245284872306627e-05, + "loss": 0.2194, + "step": 12240 + }, + { + "epoch": 1.43, + "grad_norm": 3.061187462874714, + "learning_rate": 1.5228803093583333e-05, + "loss": 0.2255, + "step": 12260 + }, + { + "epoch": 1.44, + "grad_norm": 3.26228124827874, + "learning_rate": 1.5212301744119922e-05, + "loss": 0.2297, + "step": 12280 + }, + { + "epoch": 1.44, + "grad_norm": 2.8942060403720355, + "learning_rate": 1.5195780885690641e-05, + "loss": 0.2013, + "step": 12300 + }, + { + "epoch": 1.44, + "grad_norm": 2.6600186708698534, + "learning_rate": 1.5179240580142782e-05, + "loss": 0.2047, + "step": 12320 + }, + { + "epoch": 1.44, + "grad_norm": 3.969215305715658, + "learning_rate": 1.5162680889396424e-05, + "loss": 0.2173, + "step": 12340 + }, + { + "epoch": 1.45, + "grad_norm": 2.6829571142305904, + "learning_rate": 1.5146101875444226e-05, + "loss": 0.2023, + "step": 12360 + }, + { + "epoch": 1.45, + "grad_norm": 3.6657811020233684, + "learning_rate": 1.5129503600351186e-05, + "loss": 0.2147, + "step": 12380 + }, + { + "epoch": 1.45, + "grad_norm": 3.90125498508867, + "learning_rate": 1.5112886126254405e-05, + "loss": 0.2137, + "step": 12400 + }, + { + "epoch": 1.45, + "grad_norm": 3.147105349280156, + "learning_rate": 1.5096249515362853e-05, + "loss": 0.2118, + "step": 12420 + }, + { + "epoch": 1.46, + "grad_norm": 2.7729387721425645, + "learning_rate": 1.5079593829957144e-05, + "loss": 0.2149, + "step": 12440 + }, + { + "epoch": 1.46, + "grad_norm": 5.342410092487611, + "learning_rate": 1.5062919132389305e-05, + "loss": 0.2169, + "step": 12460 + }, + { + "epoch": 1.46, + "grad_norm": 3.9559478933684638, + "learning_rate": 1.5046225485082526e-05, + "loss": 0.2205, + "step": 12480 + }, + { + "epoch": 1.46, + "grad_norm": 2.726490183748406, + "learning_rate": 1.5029512950530943e-05, + "loss": 0.224, + "step": 12500 + }, + { + "epoch": 1.47, + "grad_norm": 2.990487116043663, + "learning_rate": 1.5012781591299399e-05, + "loss": 0.2287, + "step": 12520 + }, + { + "epoch": 1.47, + "grad_norm": 3.8406463746608215, + "learning_rate": 1.4996031470023205e-05, + "loss": 0.2198, + "step": 12540 + }, + { + "epoch": 1.47, + "grad_norm": 3.29662607708646, + "learning_rate": 1.4979262649407914e-05, + "loss": 0.2198, + "step": 12560 + }, + { + "epoch": 1.47, + "grad_norm": 2.7040482195837834, + "learning_rate": 1.496247519222908e-05, + "loss": 0.2079, + "step": 12580 + }, + { + "epoch": 1.47, + "grad_norm": 2.6175719021573918, + "learning_rate": 1.4945669161332021e-05, + "loss": 0.2198, + "step": 12600 + }, + { + "epoch": 1.48, + "grad_norm": 2.9074738909532285, + "learning_rate": 1.4928844619631588e-05, + "loss": 0.2084, + "step": 12620 + }, + { + "epoch": 1.48, + "grad_norm": 3.1468612473612563, + "learning_rate": 1.4912001630111936e-05, + "loss": 0.2013, + "step": 12640 + }, + { + "epoch": 1.48, + "grad_norm": 3.779361095424968, + "learning_rate": 1.4895140255826273e-05, + "loss": 0.1918, + "step": 12660 + }, + { + "epoch": 1.48, + "grad_norm": 2.696019176777178, + "learning_rate": 1.4878260559896634e-05, + "loss": 0.2126, + "step": 12680 + }, + { + "epoch": 1.49, + "grad_norm": 3.7345031971654103, + "learning_rate": 1.4861362605513643e-05, + "loss": 0.2212, + "step": 12700 + }, + { + "epoch": 1.49, + "grad_norm": 4.465836435468654, + "learning_rate": 1.4844446455936277e-05, + "loss": 0.2211, + "step": 12720 + }, + { + "epoch": 1.49, + "grad_norm": 2.9078487797789645, + "learning_rate": 1.4827512174491631e-05, + "loss": 0.2267, + "step": 12740 + }, + { + "epoch": 1.49, + "grad_norm": 3.296252006498326, + "learning_rate": 1.4810559824574675e-05, + "loss": 0.2216, + "step": 12760 + }, + { + "epoch": 1.5, + "grad_norm": 3.227367236112649, + "learning_rate": 1.4793589469648017e-05, + "loss": 0.2233, + "step": 12780 + }, + { + "epoch": 1.5, + "grad_norm": 3.4962503238332774, + "learning_rate": 1.4776601173241671e-05, + "loss": 0.2269, + "step": 12800 + }, + { + "epoch": 1.5, + "grad_norm": 2.878724712305482, + "learning_rate": 1.4759594998952821e-05, + "loss": 0.2119, + "step": 12820 + }, + { + "epoch": 1.5, + "grad_norm": 3.682489644312493, + "learning_rate": 1.4742571010445572e-05, + "loss": 0.2011, + "step": 12840 + }, + { + "epoch": 1.5, + "grad_norm": 5.266174287756561, + "learning_rate": 1.472552927145072e-05, + "loss": 0.2258, + "step": 12860 + }, + { + "epoch": 1.51, + "grad_norm": 4.045996375421248, + "learning_rate": 1.4708469845765512e-05, + "loss": 0.1922, + "step": 12880 + }, + { + "epoch": 1.51, + "grad_norm": 3.266689729284762, + "learning_rate": 1.4691392797253405e-05, + "loss": 0.2075, + "step": 12900 + }, + { + "epoch": 1.51, + "grad_norm": 2.242467882655721, + "learning_rate": 1.4674298189843837e-05, + "loss": 0.212, + "step": 12920 + }, + { + "epoch": 1.51, + "grad_norm": 2.807279276540352, + "learning_rate": 1.4657186087531962e-05, + "loss": 0.1924, + "step": 12940 + }, + { + "epoch": 1.52, + "grad_norm": 6.9144252531074875, + "learning_rate": 1.4640056554378448e-05, + "loss": 0.224, + "step": 12960 + }, + { + "epoch": 1.52, + "grad_norm": 3.5667764527961645, + "learning_rate": 1.4622909654509204e-05, + "loss": 0.2175, + "step": 12980 + }, + { + "epoch": 1.52, + "grad_norm": 2.431697492432426, + "learning_rate": 1.4605745452115155e-05, + "loss": 0.2172, + "step": 13000 + }, + { + "epoch": 1.52, + "grad_norm": 3.336010970081616, + "learning_rate": 1.4588564011451998e-05, + "loss": 0.2219, + "step": 13020 + }, + { + "epoch": 1.53, + "grad_norm": 2.9985996792871656, + "learning_rate": 1.4571365396839969e-05, + "loss": 0.2226, + "step": 13040 + }, + { + "epoch": 1.53, + "grad_norm": 2.7817912914440677, + "learning_rate": 1.4554149672663591e-05, + "loss": 0.2166, + "step": 13060 + }, + { + "epoch": 1.53, + "grad_norm": 3.5654978937601007, + "learning_rate": 1.4536916903371443e-05, + "loss": 0.209, + "step": 13080 + }, + { + "epoch": 1.53, + "grad_norm": 3.140872774741023, + "learning_rate": 1.4519667153475908e-05, + "loss": 0.2059, + "step": 13100 + }, + { + "epoch": 1.54, + "grad_norm": 3.7735270929856304, + "learning_rate": 1.4502400487552937e-05, + "loss": 0.2078, + "step": 13120 + }, + { + "epoch": 1.54, + "grad_norm": 3.3704008625810826, + "learning_rate": 1.4485116970241814e-05, + "loss": 0.2288, + "step": 13140 + }, + { + "epoch": 1.54, + "grad_norm": 3.058405104235901, + "learning_rate": 1.4467816666244904e-05, + "loss": 0.2145, + "step": 13160 + }, + { + "epoch": 1.54, + "grad_norm": 3.2628541717630295, + "learning_rate": 1.4450499640327418e-05, + "loss": 0.2096, + "step": 13180 + }, + { + "epoch": 1.54, + "grad_norm": 2.8358576319353475, + "learning_rate": 1.443316595731716e-05, + "loss": 0.2162, + "step": 13200 + }, + { + "epoch": 1.55, + "grad_norm": 2.130034709343158, + "learning_rate": 1.4415815682104295e-05, + "loss": 0.2038, + "step": 13220 + }, + { + "epoch": 1.55, + "grad_norm": 3.679976984449645, + "learning_rate": 1.4398448879641102e-05, + "loss": 0.216, + "step": 13240 + }, + { + "epoch": 1.55, + "grad_norm": 3.5668380826834647, + "learning_rate": 1.4381065614941733e-05, + "loss": 0.2262, + "step": 13260 + }, + { + "epoch": 1.55, + "grad_norm": 3.0243831604612765, + "learning_rate": 1.4363665953081969e-05, + "loss": 0.201, + "step": 13280 + }, + { + "epoch": 1.56, + "grad_norm": 3.2212853677485933, + "learning_rate": 1.4346249959198965e-05, + "loss": 0.2163, + "step": 13300 + }, + { + "epoch": 1.56, + "grad_norm": 2.596122759445438, + "learning_rate": 1.4328817698491031e-05, + "loss": 0.22, + "step": 13320 + }, + { + "epoch": 1.56, + "grad_norm": 3.0666724605781996, + "learning_rate": 1.4311369236217367e-05, + "loss": 0.2139, + "step": 13340 + }, + { + "epoch": 1.56, + "grad_norm": 3.657252769223528, + "learning_rate": 1.4293904637697817e-05, + "loss": 0.215, + "step": 13360 + }, + { + "epoch": 1.57, + "grad_norm": 3.4144546667640383, + "learning_rate": 1.4276423968312645e-05, + "loss": 0.227, + "step": 13380 + }, + { + "epoch": 1.57, + "grad_norm": 3.2413250982644675, + "learning_rate": 1.4258927293502273e-05, + "loss": 0.2077, + "step": 13400 + }, + { + "epoch": 1.57, + "grad_norm": 3.5063106320828488, + "learning_rate": 1.4241414678767039e-05, + "loss": 0.2047, + "step": 13420 + }, + { + "epoch": 1.57, + "grad_norm": 2.1504143515238514, + "learning_rate": 1.4223886189666951e-05, + "loss": 0.1982, + "step": 13440 + }, + { + "epoch": 1.58, + "grad_norm": 2.9525594825793005, + "learning_rate": 1.4206341891821456e-05, + "loss": 0.2066, + "step": 13460 + }, + { + "epoch": 1.58, + "grad_norm": 5.213240221920547, + "learning_rate": 1.4188781850909165e-05, + "loss": 0.2246, + "step": 13480 + }, + { + "epoch": 1.58, + "grad_norm": 4.305790200898134, + "learning_rate": 1.417120613266764e-05, + "loss": 0.1994, + "step": 13500 + }, + { + "epoch": 1.58, + "grad_norm": 3.370920187857142, + "learning_rate": 1.4153614802893125e-05, + "loss": 0.2085, + "step": 13520 + }, + { + "epoch": 1.58, + "grad_norm": 3.425048053710641, + "learning_rate": 1.4136007927440307e-05, + "loss": 0.207, + "step": 13540 + }, + { + "epoch": 1.59, + "grad_norm": 3.24691919878633, + "learning_rate": 1.4118385572222072e-05, + "loss": 0.2194, + "step": 13560 + }, + { + "epoch": 1.59, + "grad_norm": 2.8908834805780064, + "learning_rate": 1.4100747803209252e-05, + "loss": 0.2024, + "step": 13580 + }, + { + "epoch": 1.59, + "grad_norm": 3.5087368773081473, + "learning_rate": 1.4083094686430387e-05, + "loss": 0.2102, + "step": 13600 + }, + { + "epoch": 1.59, + "grad_norm": 2.5718331983652485, + "learning_rate": 1.4065426287971473e-05, + "loss": 0.2205, + "step": 13620 + }, + { + "epoch": 1.6, + "grad_norm": 3.0529680412811375, + "learning_rate": 1.4047742673975706e-05, + "loss": 0.214, + "step": 13640 + }, + { + "epoch": 1.6, + "grad_norm": 3.6927881631261505, + "learning_rate": 1.4030043910643254e-05, + "loss": 0.2124, + "step": 13660 + }, + { + "epoch": 1.6, + "grad_norm": 3.3041534353009743, + "learning_rate": 1.401233006423099e-05, + "loss": 0.2033, + "step": 13680 + }, + { + "epoch": 1.6, + "grad_norm": 3.539906112805195, + "learning_rate": 1.3994601201052253e-05, + "loss": 0.2022, + "step": 13700 + }, + { + "epoch": 1.61, + "grad_norm": 3.981704115906401, + "learning_rate": 1.39768573874766e-05, + "loss": 0.213, + "step": 13720 + }, + { + "epoch": 1.61, + "grad_norm": 2.5831456167458375, + "learning_rate": 1.3959098689929555e-05, + "loss": 0.1958, + "step": 13740 + }, + { + "epoch": 1.61, + "grad_norm": 2.4960100759889725, + "learning_rate": 1.3941325174892371e-05, + "loss": 0.2243, + "step": 13760 + }, + { + "epoch": 1.61, + "grad_norm": 2.9399908340726424, + "learning_rate": 1.392353690890175e-05, + "loss": 0.2052, + "step": 13780 + }, + { + "epoch": 1.61, + "grad_norm": 3.161745178851194, + "learning_rate": 1.3905733958549632e-05, + "loss": 0.1905, + "step": 13800 + }, + { + "epoch": 1.62, + "grad_norm": 3.6763150256287305, + "learning_rate": 1.3887916390482927e-05, + "loss": 0.2097, + "step": 13820 + }, + { + "epoch": 1.62, + "grad_norm": 2.589478710103983, + "learning_rate": 1.3870084271403269e-05, + "loss": 0.2108, + "step": 13840 + }, + { + "epoch": 1.62, + "grad_norm": 2.9786347312251205, + "learning_rate": 1.3852237668066763e-05, + "loss": 0.2021, + "step": 13860 + }, + { + "epoch": 1.62, + "grad_norm": 2.918896140906093, + "learning_rate": 1.3834376647283729e-05, + "loss": 0.2184, + "step": 13880 + }, + { + "epoch": 1.63, + "grad_norm": 3.4250636463795074, + "learning_rate": 1.3816501275918475e-05, + "loss": 0.1989, + "step": 13900 + }, + { + "epoch": 1.63, + "grad_norm": 2.7399597682922923, + "learning_rate": 1.3798611620889017e-05, + "loss": 0.2089, + "step": 13920 + }, + { + "epoch": 1.63, + "grad_norm": 3.678704067595457, + "learning_rate": 1.378070774916686e-05, + "loss": 0.2192, + "step": 13940 + }, + { + "epoch": 1.63, + "grad_norm": 3.1242276572434133, + "learning_rate": 1.3762789727776708e-05, + "loss": 0.2157, + "step": 13960 + }, + { + "epoch": 1.64, + "grad_norm": 2.885737395564972, + "learning_rate": 1.3744857623796255e-05, + "loss": 0.2133, + "step": 13980 + }, + { + "epoch": 1.64, + "grad_norm": 2.998690955346801, + "learning_rate": 1.3726911504355908e-05, + "loss": 0.2129, + "step": 14000 + }, + { + "epoch": 1.64, + "grad_norm": 4.013414301345931, + "learning_rate": 1.3708951436638542e-05, + "loss": 0.2031, + "step": 14020 + }, + { + "epoch": 1.64, + "grad_norm": 3.6060060573956503, + "learning_rate": 1.3690977487879247e-05, + "loss": 0.2149, + "step": 14040 + }, + { + "epoch": 1.65, + "grad_norm": 2.8295749719869105, + "learning_rate": 1.3672989725365076e-05, + "loss": 0.2113, + "step": 14060 + }, + { + "epoch": 1.65, + "grad_norm": 4.634259826093287, + "learning_rate": 1.3654988216434802e-05, + "loss": 0.2013, + "step": 14080 + }, + { + "epoch": 1.65, + "grad_norm": 4.019301812951925, + "learning_rate": 1.3636973028478655e-05, + "loss": 0.2079, + "step": 14100 + }, + { + "epoch": 1.65, + "grad_norm": 3.3314202699929387, + "learning_rate": 1.3618944228938072e-05, + "loss": 0.1968, + "step": 14120 + }, + { + "epoch": 1.65, + "grad_norm": 4.011000289395802, + "learning_rate": 1.3600901885305446e-05, + "loss": 0.2075, + "step": 14140 + }, + { + "epoch": 1.66, + "grad_norm": 3.1307277830268183, + "learning_rate": 1.358284606512388e-05, + "loss": 0.2004, + "step": 14160 + }, + { + "epoch": 1.66, + "grad_norm": 2.6402421481891047, + "learning_rate": 1.3564776835986921e-05, + "loss": 0.2102, + "step": 14180 + }, + { + "epoch": 1.66, + "grad_norm": 3.1118142226597745, + "learning_rate": 1.3546694265538316e-05, + "loss": 0.2124, + "step": 14200 + }, + { + "epoch": 1.66, + "grad_norm": 3.4864022554268614, + "learning_rate": 1.3528598421471755e-05, + "loss": 0.2067, + "step": 14220 + }, + { + "epoch": 1.67, + "grad_norm": 2.8051441214080457, + "learning_rate": 1.3510489371530615e-05, + "loss": 0.2163, + "step": 14240 + }, + { + "epoch": 1.67, + "grad_norm": 3.1532788786376638, + "learning_rate": 1.3492367183507726e-05, + "loss": 0.203, + "step": 14260 + }, + { + "epoch": 1.67, + "grad_norm": 3.6383798904923412, + "learning_rate": 1.3474231925245087e-05, + "loss": 0.2117, + "step": 14280 + }, + { + "epoch": 1.67, + "grad_norm": 3.2573585834385925, + "learning_rate": 1.3456083664633626e-05, + "loss": 0.2097, + "step": 14300 + }, + { + "epoch": 1.68, + "grad_norm": 4.797438398942751, + "learning_rate": 1.3437922469612951e-05, + "loss": 0.2116, + "step": 14320 + }, + { + "epoch": 1.68, + "grad_norm": 2.5155576140103646, + "learning_rate": 1.3419748408171094e-05, + "loss": 0.2225, + "step": 14340 + }, + { + "epoch": 1.68, + "grad_norm": 5.78637039662545, + "learning_rate": 1.3401561548344253e-05, + "loss": 0.2074, + "step": 14360 + }, + { + "epoch": 1.68, + "grad_norm": 5.171051974083877, + "learning_rate": 1.3383361958216531e-05, + "loss": 0.1966, + "step": 14380 + }, + { + "epoch": 1.68, + "grad_norm": 3.6388434115246278, + "learning_rate": 1.3365149705919693e-05, + "loss": 0.2202, + "step": 14400 + }, + { + "epoch": 1.69, + "grad_norm": 3.6368767430520426, + "learning_rate": 1.3346924859632908e-05, + "loss": 0.2156, + "step": 14420 + }, + { + "epoch": 1.69, + "grad_norm": 2.6252892652713293, + "learning_rate": 1.3328687487582486e-05, + "loss": 0.2047, + "step": 14440 + }, + { + "epoch": 1.69, + "grad_norm": 3.2310005313042853, + "learning_rate": 1.3310437658041631e-05, + "loss": 0.1995, + "step": 14460 + }, + { + "epoch": 1.69, + "grad_norm": 2.591667902252132, + "learning_rate": 1.3292175439330182e-05, + "loss": 0.1936, + "step": 14480 + }, + { + "epoch": 1.7, + "grad_norm": 6.920226165515457, + "learning_rate": 1.3273900899814365e-05, + "loss": 0.2008, + "step": 14500 + }, + { + "epoch": 1.7, + "grad_norm": 3.051444827700318, + "learning_rate": 1.3255614107906521e-05, + "loss": 0.1976, + "step": 14520 + }, + { + "epoch": 1.7, + "grad_norm": 2.9835362079482435, + "learning_rate": 1.3237315132064863e-05, + "loss": 0.2148, + "step": 14540 + }, + { + "epoch": 1.7, + "grad_norm": 2.8495541541728024, + "learning_rate": 1.321900404079321e-05, + "loss": 0.1977, + "step": 14560 + }, + { + "epoch": 1.71, + "grad_norm": 2.7688725257095443, + "learning_rate": 1.3200680902640744e-05, + "loss": 0.2178, + "step": 14580 + }, + { + "epoch": 1.71, + "grad_norm": 2.7073866045339576, + "learning_rate": 1.318234578620174e-05, + "loss": 0.2146, + "step": 14600 + }, + { + "epoch": 1.71, + "grad_norm": 3.161990242988256, + "learning_rate": 1.3163998760115325e-05, + "loss": 0.2134, + "step": 14620 + }, + { + "epoch": 1.71, + "grad_norm": 2.7928417256953812, + "learning_rate": 1.314563989306519e-05, + "loss": 0.2065, + "step": 14640 + }, + { + "epoch": 1.72, + "grad_norm": 2.356147630324507, + "learning_rate": 1.3127269253779376e-05, + "loss": 0.2105, + "step": 14660 + }, + { + "epoch": 1.72, + "grad_norm": 4.346477722398746, + "learning_rate": 1.310888691102998e-05, + "loss": 0.2134, + "step": 14680 + }, + { + "epoch": 1.72, + "grad_norm": 2.7518253337355763, + "learning_rate": 1.3090492933632926e-05, + "loss": 0.2107, + "step": 14700 + }, + { + "epoch": 1.72, + "grad_norm": 3.241061463203708, + "learning_rate": 1.307208739044767e-05, + "loss": 0.2094, + "step": 14720 + }, + { + "epoch": 1.72, + "grad_norm": 2.666615356507985, + "learning_rate": 1.3053670350376993e-05, + "loss": 0.1989, + "step": 14740 + }, + { + "epoch": 1.73, + "grad_norm": 3.0536397322406663, + "learning_rate": 1.3035241882366692e-05, + "loss": 0.2156, + "step": 14760 + }, + { + "epoch": 1.73, + "grad_norm": 2.3893756469119007, + "learning_rate": 1.3016802055405363e-05, + "loss": 0.2091, + "step": 14780 + }, + { + "epoch": 1.73, + "grad_norm": 3.252027759325605, + "learning_rate": 1.2998350938524114e-05, + "loss": 0.1993, + "step": 14800 + }, + { + "epoch": 1.73, + "grad_norm": 3.9445454371124105, + "learning_rate": 1.2979888600796322e-05, + "loss": 0.2021, + "step": 14820 + }, + { + "epoch": 1.74, + "grad_norm": 3.811107068357804, + "learning_rate": 1.2961415111337374e-05, + "loss": 0.2257, + "step": 14840 + }, + { + "epoch": 1.74, + "grad_norm": 2.437893371140398, + "learning_rate": 1.2942930539304397e-05, + "loss": 0.1892, + "step": 14860 + }, + { + "epoch": 1.74, + "grad_norm": 2.5823821090570105, + "learning_rate": 1.2924434953896014e-05, + "loss": 0.1954, + "step": 14880 + }, + { + "epoch": 1.74, + "grad_norm": 2.817062999748064, + "learning_rate": 1.2905928424352068e-05, + "loss": 0.2016, + "step": 14900 + }, + { + "epoch": 1.75, + "grad_norm": 5.869940669265181, + "learning_rate": 1.288741101995338e-05, + "loss": 0.211, + "step": 14920 + }, + { + "epoch": 1.75, + "grad_norm": 3.047154496896096, + "learning_rate": 1.286888281002148e-05, + "loss": 0.1963, + "step": 14940 + }, + { + "epoch": 1.75, + "grad_norm": 3.7639733962315756, + "learning_rate": 1.2850343863918358e-05, + "loss": 0.2057, + "step": 14960 + }, + { + "epoch": 1.75, + "grad_norm": 2.7118107601240107, + "learning_rate": 1.2831794251046173e-05, + "loss": 0.2064, + "step": 14980 + }, + { + "epoch": 1.76, + "grad_norm": 4.293885901020362, + "learning_rate": 1.281323404084704e-05, + "loss": 0.189, + "step": 15000 + }, + { + "epoch": 1.76, + "grad_norm": 3.136084771606493, + "learning_rate": 1.2794663302802732e-05, + "loss": 0.222, + "step": 15020 + }, + { + "epoch": 1.76, + "grad_norm": 2.33967520945844, + "learning_rate": 1.2776082106434442e-05, + "loss": 0.21, + "step": 15040 + }, + { + "epoch": 1.76, + "grad_norm": 3.058105585694254, + "learning_rate": 1.2757490521302506e-05, + "loss": 0.2066, + "step": 15060 + }, + { + "epoch": 1.76, + "grad_norm": 3.783134432254202, + "learning_rate": 1.2738888617006158e-05, + "loss": 0.204, + "step": 15080 + }, + { + "epoch": 1.77, + "grad_norm": 3.3088612273817577, + "learning_rate": 1.2720276463183264e-05, + "loss": 0.2088, + "step": 15100 + }, + { + "epoch": 1.77, + "grad_norm": 2.5164740415038063, + "learning_rate": 1.2701654129510053e-05, + "loss": 0.2083, + "step": 15120 + }, + { + "epoch": 1.77, + "grad_norm": 2.680468920777465, + "learning_rate": 1.2683021685700867e-05, + "loss": 0.2085, + "step": 15140 + }, + { + "epoch": 1.77, + "grad_norm": 2.7932332800591206, + "learning_rate": 1.26643792015079e-05, + "loss": 0.218, + "step": 15160 + }, + { + "epoch": 1.78, + "grad_norm": 4.372169232777801, + "learning_rate": 1.2645726746720925e-05, + "loss": 0.1995, + "step": 15180 + }, + { + "epoch": 1.78, + "grad_norm": 3.1330033622307245, + "learning_rate": 1.2627064391167045e-05, + "loss": 0.2057, + "step": 15200 + }, + { + "epoch": 1.78, + "grad_norm": 3.6755304848957446, + "learning_rate": 1.2608392204710436e-05, + "loss": 0.2154, + "step": 15220 + }, + { + "epoch": 1.78, + "grad_norm": 2.54686994493603, + "learning_rate": 1.2589710257252058e-05, + "loss": 0.2056, + "step": 15240 + }, + { + "epoch": 1.79, + "grad_norm": 3.1076386937733926, + "learning_rate": 1.2571018618729422e-05, + "loss": 0.2006, + "step": 15260 + }, + { + "epoch": 1.79, + "grad_norm": 4.9484787578205855, + "learning_rate": 1.2552317359116326e-05, + "loss": 0.2089, + "step": 15280 + }, + { + "epoch": 1.79, + "grad_norm": 2.4893284006088634, + "learning_rate": 1.2533606548422577e-05, + "loss": 0.2024, + "step": 15300 + }, + { + "epoch": 1.79, + "grad_norm": 3.1084832681147594, + "learning_rate": 1.2514886256693732e-05, + "loss": 0.2106, + "step": 15320 + }, + { + "epoch": 1.79, + "grad_norm": 3.0389123403917804, + "learning_rate": 1.249615655401085e-05, + "loss": 0.2096, + "step": 15340 + }, + { + "epoch": 1.8, + "grad_norm": 4.392113049940431, + "learning_rate": 1.2477417510490223e-05, + "loss": 0.21, + "step": 15360 + }, + { + "epoch": 1.8, + "grad_norm": 2.4290921871050077, + "learning_rate": 1.245866919628311e-05, + "loss": 0.2101, + "step": 15380 + }, + { + "epoch": 1.8, + "grad_norm": 2.53729167515282, + "learning_rate": 1.2439911681575459e-05, + "loss": 0.2085, + "step": 15400 + }, + { + "epoch": 1.8, + "grad_norm": 0.955267140537957, + "learning_rate": 1.2421145036587681e-05, + "loss": 0.1946, + "step": 15420 + }, + { + "epoch": 1.81, + "grad_norm": 2.3147872747972924, + "learning_rate": 1.2402369331574363e-05, + "loss": 0.2133, + "step": 15440 + }, + { + "epoch": 1.81, + "grad_norm": 2.6069168116002013, + "learning_rate": 1.2383584636824002e-05, + "loss": 0.2151, + "step": 15460 + }, + { + "epoch": 1.81, + "grad_norm": 3.6384307404668057, + "learning_rate": 1.2364791022658756e-05, + "loss": 0.2055, + "step": 15480 + }, + { + "epoch": 1.81, + "grad_norm": 2.7467900528862796, + "learning_rate": 1.2345988559434163e-05, + "loss": 0.1968, + "step": 15500 + }, + { + "epoch": 1.82, + "grad_norm": 3.712915796699261, + "learning_rate": 1.2327177317538903e-05, + "loss": 0.2022, + "step": 15520 + }, + { + "epoch": 1.82, + "grad_norm": 3.0273517830034553, + "learning_rate": 1.2308357367394515e-05, + "loss": 0.2059, + "step": 15540 + }, + { + "epoch": 1.82, + "grad_norm": 2.9422286911490256, + "learning_rate": 1.2289528779455129e-05, + "loss": 0.2109, + "step": 15560 + }, + { + "epoch": 1.82, + "grad_norm": 2.7461702382273097, + "learning_rate": 1.227069162420722e-05, + "loss": 0.2008, + "step": 15580 + }, + { + "epoch": 1.83, + "grad_norm": 2.9259269484541868, + "learning_rate": 1.2251845972169336e-05, + "loss": 0.2087, + "step": 15600 + }, + { + "epoch": 1.83, + "grad_norm": 2.351163202901331, + "learning_rate": 1.2232991893891826e-05, + "loss": 0.2162, + "step": 15620 + }, + { + "epoch": 1.83, + "grad_norm": 3.3864085617483486, + "learning_rate": 1.2214129459956595e-05, + "loss": 0.2109, + "step": 15640 + }, + { + "epoch": 1.83, + "grad_norm": 3.1532912786097556, + "learning_rate": 1.2195258740976816e-05, + "loss": 0.1927, + "step": 15660 + }, + { + "epoch": 1.83, + "grad_norm": 2.32467980179587, + "learning_rate": 1.2176379807596684e-05, + "loss": 0.2083, + "step": 15680 + }, + { + "epoch": 1.84, + "grad_norm": 3.692322632410928, + "learning_rate": 1.2157492730491142e-05, + "loss": 0.1952, + "step": 15700 + }, + { + "epoch": 1.84, + "grad_norm": 3.5592767958117704, + "learning_rate": 1.2138597580365635e-05, + "loss": 0.2259, + "step": 15720 + }, + { + "epoch": 1.84, + "grad_norm": 3.312387140618671, + "learning_rate": 1.21196944279558e-05, + "loss": 0.1996, + "step": 15740 + }, + { + "epoch": 1.84, + "grad_norm": 2.4404373077236974, + "learning_rate": 1.2100783344027257e-05, + "loss": 0.2061, + "step": 15760 + }, + { + "epoch": 1.85, + "grad_norm": 3.0929231405689337, + "learning_rate": 1.2081864399375306e-05, + "loss": 0.1989, + "step": 15780 + }, + { + "epoch": 1.85, + "grad_norm": 3.6918547181282237, + "learning_rate": 1.2062937664824692e-05, + "loss": 0.2039, + "step": 15800 + }, + { + "epoch": 1.85, + "grad_norm": 25.439570612132734, + "learning_rate": 1.2044003211229292e-05, + "loss": 0.2023, + "step": 15820 + }, + { + "epoch": 1.85, + "grad_norm": 3.238254008621471, + "learning_rate": 1.2025061109471903e-05, + "loss": 0.195, + "step": 15840 + }, + { + "epoch": 1.86, + "grad_norm": 2.717237819317576, + "learning_rate": 1.2006111430463945e-05, + "loss": 0.2159, + "step": 15860 + }, + { + "epoch": 1.86, + "grad_norm": 3.3138924317200207, + "learning_rate": 1.198715424514521e-05, + "loss": 0.2037, + "step": 15880 + }, + { + "epoch": 1.86, + "grad_norm": 2.477738254765999, + "learning_rate": 1.1968189624483583e-05, + "loss": 0.2081, + "step": 15900 + }, + { + "epoch": 1.86, + "grad_norm": 5.595275329960663, + "learning_rate": 1.194921763947479e-05, + "loss": 0.2048, + "step": 15920 + }, + { + "epoch": 1.87, + "grad_norm": 6.622925343547304, + "learning_rate": 1.1930238361142116e-05, + "loss": 0.2149, + "step": 15940 + }, + { + "epoch": 1.87, + "grad_norm": 3.576100405484642, + "learning_rate": 1.1911251860536169e-05, + "loss": 0.2103, + "step": 15960 + }, + { + "epoch": 1.87, + "grad_norm": 2.5035188705559004, + "learning_rate": 1.1892258208734573e-05, + "loss": 0.2108, + "step": 15980 + }, + { + "epoch": 1.87, + "grad_norm": 2.6938599141436494, + "learning_rate": 1.1873257476841734e-05, + "loss": 0.2029, + "step": 16000 + }, + { + "epoch": 1.87, + "grad_norm": 2.7076420610940914, + "learning_rate": 1.1854249735988559e-05, + "loss": 0.203, + "step": 16020 + }, + { + "epoch": 1.88, + "grad_norm": 2.53155657006711, + "learning_rate": 1.18352350573322e-05, + "loss": 0.2073, + "step": 16040 + }, + { + "epoch": 1.88, + "grad_norm": 2.8351358589034037, + "learning_rate": 1.1816213512055778e-05, + "loss": 0.1972, + "step": 16060 + }, + { + "epoch": 1.88, + "grad_norm": 1.994239617859996, + "learning_rate": 1.1797185171368113e-05, + "loss": 0.2157, + "step": 16080 + }, + { + "epoch": 1.88, + "grad_norm": 8.626140574352826, + "learning_rate": 1.1778150106503472e-05, + "loss": 0.2219, + "step": 16100 + }, + { + "epoch": 1.89, + "grad_norm": 3.0368386887010326, + "learning_rate": 1.1759108388721294e-05, + "loss": 0.212, + "step": 16120 + }, + { + "epoch": 1.89, + "grad_norm": 3.720694025784477, + "learning_rate": 1.1740060089305923e-05, + "loss": 0.2188, + "step": 16140 + }, + { + "epoch": 1.89, + "grad_norm": 4.140749601972492, + "learning_rate": 1.1721005279566335e-05, + "loss": 0.2205, + "step": 16160 + }, + { + "epoch": 1.89, + "grad_norm": 3.696009886666488, + "learning_rate": 1.170194403083589e-05, + "loss": 0.202, + "step": 16180 + }, + { + "epoch": 1.9, + "grad_norm": 2.913336533021661, + "learning_rate": 1.1682876414472041e-05, + "loss": 0.2122, + "step": 16200 + }, + { + "epoch": 1.9, + "grad_norm": 3.7716338149283706, + "learning_rate": 1.1663802501856097e-05, + "loss": 0.2256, + "step": 16220 + }, + { + "epoch": 1.9, + "grad_norm": 4.6668357591421215, + "learning_rate": 1.1644722364392916e-05, + "loss": 0.1985, + "step": 16240 + }, + { + "epoch": 1.9, + "grad_norm": 2.246832949864308, + "learning_rate": 1.1625636073510669e-05, + "loss": 0.2221, + "step": 16260 + }, + { + "epoch": 1.9, + "grad_norm": 3.962744426709658, + "learning_rate": 1.1606543700660566e-05, + "loss": 0.2083, + "step": 16280 + }, + { + "epoch": 1.91, + "grad_norm": 2.801003078896173, + "learning_rate": 1.1587445317316579e-05, + "loss": 0.2005, + "step": 16300 + }, + { + "epoch": 1.91, + "grad_norm": 2.1091323147793117, + "learning_rate": 1.1568340994975185e-05, + "loss": 0.2045, + "step": 16320 + }, + { + "epoch": 1.91, + "grad_norm": 3.198361161909217, + "learning_rate": 1.1549230805155091e-05, + "loss": 0.206, + "step": 16340 + }, + { + "epoch": 1.91, + "grad_norm": 10.406894789872744, + "learning_rate": 1.1530114819396976e-05, + "loss": 0.2123, + "step": 16360 + }, + { + "epoch": 1.92, + "grad_norm": 3.2619393462547737, + "learning_rate": 1.1510993109263206e-05, + "loss": 0.2041, + "step": 16380 + }, + { + "epoch": 1.92, + "grad_norm": 3.409171994718834, + "learning_rate": 1.149186574633759e-05, + "loss": 0.2056, + "step": 16400 + }, + { + "epoch": 1.92, + "grad_norm": 2.5059832499949737, + "learning_rate": 1.1472732802225081e-05, + "loss": 0.2034, + "step": 16420 + }, + { + "epoch": 1.92, + "grad_norm": 3.267940886449616, + "learning_rate": 1.1453594348551547e-05, + "loss": 0.2012, + "step": 16440 + }, + { + "epoch": 1.93, + "grad_norm": 3.2066882155626875, + "learning_rate": 1.1434450456963462e-05, + "loss": 0.1971, + "step": 16460 + }, + { + "epoch": 1.93, + "grad_norm": 3.1243325092793883, + "learning_rate": 1.1415301199127671e-05, + "loss": 0.2041, + "step": 16480 + }, + { + "epoch": 1.93, + "grad_norm": 2.6001065965955448, + "learning_rate": 1.1396146646731104e-05, + "loss": 0.1996, + "step": 16500 + }, + { + "epoch": 1.93, + "grad_norm": 2.424714226942306, + "learning_rate": 1.1376986871480505e-05, + "loss": 0.1946, + "step": 16520 + }, + { + "epoch": 1.94, + "grad_norm": 2.4162929070497, + "learning_rate": 1.1357821945102177e-05, + "loss": 0.1907, + "step": 16540 + }, + { + "epoch": 1.94, + "grad_norm": 3.868349706003454, + "learning_rate": 1.133865193934171e-05, + "loss": 0.2202, + "step": 16560 + }, + { + "epoch": 1.94, + "grad_norm": 5.899437583367488, + "learning_rate": 1.1319476925963702e-05, + "loss": 0.203, + "step": 16580 + }, + { + "epoch": 1.94, + "grad_norm": 4.061783477977705, + "learning_rate": 1.1300296976751496e-05, + "loss": 0.2078, + "step": 16600 + }, + { + "epoch": 1.94, + "grad_norm": 3.279344277780531, + "learning_rate": 1.1281112163506923e-05, + "loss": 0.2004, + "step": 16620 + }, + { + "epoch": 1.95, + "grad_norm": 3.012593064512875, + "learning_rate": 1.1261922558050012e-05, + "loss": 0.1992, + "step": 16640 + }, + { + "epoch": 1.95, + "grad_norm": 3.108910147265754, + "learning_rate": 1.1242728232218743e-05, + "loss": 0.2006, + "step": 16660 + }, + { + "epoch": 1.95, + "grad_norm": 7.0034021556776525, + "learning_rate": 1.122352925786875e-05, + "loss": 0.2147, + "step": 16680 + }, + { + "epoch": 1.95, + "grad_norm": 1.8028644419788564, + "learning_rate": 1.1204325706873091e-05, + "loss": 0.2024, + "step": 16700 + }, + { + "epoch": 1.96, + "grad_norm": 2.515029836822745, + "learning_rate": 1.1185117651121942e-05, + "loss": 0.2128, + "step": 16720 + }, + { + "epoch": 1.96, + "grad_norm": 5.397378511249806, + "learning_rate": 1.1165905162522347e-05, + "loss": 0.2158, + "step": 16740 + }, + { + "epoch": 1.96, + "grad_norm": 3.52178009365664, + "learning_rate": 1.1146688312997946e-05, + "loss": 0.2038, + "step": 16760 + }, + { + "epoch": 1.96, + "grad_norm": 4.3841535451132305, + "learning_rate": 1.1127467174488697e-05, + "loss": 0.2101, + "step": 16780 + }, + { + "epoch": 1.97, + "grad_norm": 3.869824799113014, + "learning_rate": 1.1108241818950627e-05, + "loss": 0.1925, + "step": 16800 + }, + { + "epoch": 1.97, + "grad_norm": 4.329393772389082, + "learning_rate": 1.1089012318355544e-05, + "loss": 0.2014, + "step": 16820 + }, + { + "epoch": 1.97, + "grad_norm": 3.781154410675153, + "learning_rate": 1.1069778744690771e-05, + "loss": 0.1978, + "step": 16840 + }, + { + "epoch": 1.97, + "grad_norm": 2.96129756942421, + "learning_rate": 1.1050541169958878e-05, + "loss": 0.1964, + "step": 16860 + }, + { + "epoch": 1.98, + "grad_norm": 3.730945241165853, + "learning_rate": 1.1031299666177414e-05, + "loss": 0.1997, + "step": 16880 + }, + { + "epoch": 1.98, + "grad_norm": 3.6034480060136094, + "learning_rate": 1.1012054305378647e-05, + "loss": 0.2064, + "step": 16900 + }, + { + "epoch": 1.98, + "grad_norm": 3.2852161959917856, + "learning_rate": 1.0992805159609274e-05, + "loss": 0.2013, + "step": 16920 + }, + { + "epoch": 1.98, + "grad_norm": 3.2808070292511253, + "learning_rate": 1.0973552300930157e-05, + "loss": 0.2114, + "step": 16940 + }, + { + "epoch": 1.98, + "grad_norm": 2.8603265565226517, + "learning_rate": 1.0954295801416068e-05, + "loss": 0.2045, + "step": 16960 + }, + { + "epoch": 1.99, + "grad_norm": 4.057061413690102, + "learning_rate": 1.0935035733155402e-05, + "loss": 0.1918, + "step": 16980 + }, + { + "epoch": 1.99, + "grad_norm": 3.9395052935238755, + "learning_rate": 1.0915772168249923e-05, + "loss": 0.213, + "step": 17000 + }, + { + "epoch": 1.99, + "grad_norm": 9.969773195187368, + "learning_rate": 1.0896505178814471e-05, + "loss": 0.2098, + "step": 17020 + }, + { + "epoch": 1.99, + "grad_norm": 3.0741477095309926, + "learning_rate": 1.0877234836976714e-05, + "loss": 0.2178, + "step": 17040 + }, + { + "epoch": 2.0, + "grad_norm": 3.0763547812686323, + "learning_rate": 1.0857961214876875e-05, + "loss": 0.2009, + "step": 17060 + }, + { + "epoch": 2.0, + "grad_norm": 2.581952300141561, + "learning_rate": 1.0838684384667446e-05, + "loss": 0.2033, + "step": 17080 + } + ], + "logging_steps": 20, + "max_steps": 34184, + "num_input_tokens_seen": 0, + "num_train_epochs": 4, + "save_steps": 500, + "total_flos": 0.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +}