diff --git "a/babylm_hop_words4_10M_seed0/runs/checkpoint-1500/trainer_state.json" "b/babylm_hop_words4_10M_seed0/runs/checkpoint-1500/trainer_state.json" new file mode 100644--- /dev/null +++ "b/babylm_hop_words4_10M_seed0/runs/checkpoint-1500/trainer_state.json" @@ -0,0 +1,11733 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.893939393939394, + "eval_steps": 10, + "global_step": 1500, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0012626262626262627, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8571, + "step": 1 + }, + { + "epoch": 0.0025252525252525255, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7758, + "step": 2 + }, + { + "epoch": 0.003787878787878788, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8383, + "step": 3 + }, + { + "epoch": 0.005050505050505051, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.9088, + "step": 4 + }, + { + "epoch": 0.006313131313131313, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8748, + "step": 5 + }, + { + "epoch": 0.007575757575757576, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8575, + "step": 6 + }, + { + "epoch": 0.008838383838383838, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8271, + "step": 7 + }, + { + "epoch": 0.010101010101010102, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8336, + "step": 8 + }, + { + "epoch": 0.011363636363636364, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8037, + "step": 9 + }, + { + "epoch": 0.012626262626262626, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8323, + "step": 10 + }, + { + "epoch": 0.012626262626262626, + "eval_loss": 2.8736937046051025, + "eval_runtime": 37.7802, + "eval_samples_per_second": 26.469, + "eval_steps_per_second": 1.112, + "step": 10 + }, + { + "epoch": 0.013888888888888888, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8653, + "step": 11 + }, + { + "epoch": 0.015151515151515152, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8104, + "step": 12 + }, + { + "epoch": 0.016414141414141416, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8489, + "step": 13 + }, + { + "epoch": 0.017676767676767676, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8442, + "step": 14 + }, + { + "epoch": 0.01893939393939394, + "grad_norm": 5.356770992279053, + "learning_rate": 2.1008403361344538e-08, + "loss": 2.8207, + "step": 15 + }, + { + "epoch": 0.020202020202020204, + "grad_norm": 4.954462051391602, + "learning_rate": 4.2016806722689076e-08, + "loss": 2.8104, + "step": 16 + }, + { + "epoch": 0.021464646464646464, + "grad_norm": 5.248059272766113, + "learning_rate": 6.302521008403361e-08, + "loss": 2.8178, + "step": 17 + }, + { + "epoch": 0.022727272727272728, + "grad_norm": 5.277204513549805, + "learning_rate": 8.403361344537815e-08, + "loss": 2.8255, + "step": 18 + }, + { + "epoch": 0.023989898989898988, + "grad_norm": 5.344228267669678, + "learning_rate": 1.050420168067227e-07, + "loss": 2.8018, + "step": 19 + }, + { + "epoch": 0.025252525252525252, + "grad_norm": 5.004310131072998, + "learning_rate": 1.2605042016806723e-07, + "loss": 2.8337, + "step": 20 + }, + { + "epoch": 0.025252525252525252, + "eval_loss": 2.872413158416748, + "eval_runtime": 37.7426, + "eval_samples_per_second": 26.495, + "eval_steps_per_second": 1.113, + "step": 20 + }, + { + "epoch": 0.026515151515151516, + "grad_norm": 5.209287166595459, + "learning_rate": 1.4705882352941178e-07, + "loss": 2.8246, + "step": 21 + }, + { + "epoch": 0.027777777777777776, + "grad_norm": 5.369327545166016, + "learning_rate": 1.680672268907563e-07, + "loss": 2.8791, + "step": 22 + }, + { + "epoch": 0.02904040404040404, + "grad_norm": 4.646446228027344, + "learning_rate": 1.8907563025210085e-07, + "loss": 2.8425, + "step": 23 + }, + { + "epoch": 0.030303030303030304, + "grad_norm": 4.949256420135498, + "learning_rate": 2.100840336134454e-07, + "loss": 2.8629, + "step": 24 + }, + { + "epoch": 0.03156565656565657, + "grad_norm": 4.709571838378906, + "learning_rate": 2.3109243697478996e-07, + "loss": 2.8264, + "step": 25 + }, + { + "epoch": 0.03282828282828283, + "grad_norm": 4.519557952880859, + "learning_rate": 2.5210084033613445e-07, + "loss": 2.8143, + "step": 26 + }, + { + "epoch": 0.03409090909090909, + "grad_norm": 4.445594310760498, + "learning_rate": 2.73109243697479e-07, + "loss": 2.8031, + "step": 27 + }, + { + "epoch": 0.03535353535353535, + "grad_norm": 4.26489782333374, + "learning_rate": 2.9411764705882356e-07, + "loss": 2.7719, + "step": 28 + }, + { + "epoch": 0.036616161616161616, + "grad_norm": 4.153639793395996, + "learning_rate": 3.151260504201681e-07, + "loss": 2.8251, + "step": 29 + }, + { + "epoch": 0.03787878787878788, + "grad_norm": 3.40531325340271, + "learning_rate": 3.361344537815126e-07, + "loss": 2.849, + "step": 30 + }, + { + "epoch": 0.03787878787878788, + "eval_loss": 2.825382709503174, + "eval_runtime": 37.0715, + "eval_samples_per_second": 26.975, + "eval_steps_per_second": 1.133, + "step": 30 + }, + { + "epoch": 0.039141414141414144, + "grad_norm": 3.4759507179260254, + "learning_rate": 3.5714285714285716e-07, + "loss": 2.7653, + "step": 31 + }, + { + "epoch": 0.04040404040404041, + "grad_norm": 3.4147374629974365, + "learning_rate": 3.781512605042017e-07, + "loss": 2.7916, + "step": 32 + }, + { + "epoch": 0.041666666666666664, + "grad_norm": 3.2775769233703613, + "learning_rate": 3.991596638655462e-07, + "loss": 2.8072, + "step": 33 + }, + { + "epoch": 0.04292929292929293, + "grad_norm": 3.3541715145111084, + "learning_rate": 4.201680672268908e-07, + "loss": 2.7944, + "step": 34 + }, + { + "epoch": 0.04419191919191919, + "grad_norm": 2.991971969604492, + "learning_rate": 4.4117647058823536e-07, + "loss": 2.7827, + "step": 35 + }, + { + "epoch": 0.045454545454545456, + "grad_norm": 2.9127862453460693, + "learning_rate": 4.621848739495799e-07, + "loss": 2.7332, + "step": 36 + }, + { + "epoch": 0.04671717171717172, + "grad_norm": 2.839707136154175, + "learning_rate": 4.831932773109245e-07, + "loss": 2.7568, + "step": 37 + }, + { + "epoch": 0.047979797979797977, + "grad_norm": 2.9395432472229004, + "learning_rate": 5.042016806722689e-07, + "loss": 2.7445, + "step": 38 + }, + { + "epoch": 0.04924242424242424, + "grad_norm": 2.869896650314331, + "learning_rate": 5.252100840336135e-07, + "loss": 2.7157, + "step": 39 + }, + { + "epoch": 0.050505050505050504, + "grad_norm": 2.9064571857452393, + "learning_rate": 5.46218487394958e-07, + "loss": 2.7049, + "step": 40 + }, + { + "epoch": 0.050505050505050504, + "eval_loss": 2.7477543354034424, + "eval_runtime": 37.0669, + "eval_samples_per_second": 26.978, + "eval_steps_per_second": 1.133, + "step": 40 + }, + { + "epoch": 0.05176767676767677, + "grad_norm": 2.6264421939849854, + "learning_rate": 5.672268907563026e-07, + "loss": 2.7651, + "step": 41 + }, + { + "epoch": 0.05303030303030303, + "grad_norm": 2.518627405166626, + "learning_rate": 5.882352941176471e-07, + "loss": 2.7241, + "step": 42 + }, + { + "epoch": 0.054292929292929296, + "grad_norm": 2.521034002304077, + "learning_rate": 6.092436974789917e-07, + "loss": 2.6691, + "step": 43 + }, + { + "epoch": 0.05555555555555555, + "grad_norm": 2.4679362773895264, + "learning_rate": 6.302521008403362e-07, + "loss": 2.743, + "step": 44 + }, + { + "epoch": 0.056818181818181816, + "grad_norm": 2.5817418098449707, + "learning_rate": 6.512605042016808e-07, + "loss": 2.7067, + "step": 45 + }, + { + "epoch": 0.05808080808080808, + "grad_norm": 2.3704075813293457, + "learning_rate": 6.722689075630252e-07, + "loss": 2.6495, + "step": 46 + }, + { + "epoch": 0.059343434343434344, + "grad_norm": 2.2321906089782715, + "learning_rate": 6.932773109243699e-07, + "loss": 2.6606, + "step": 47 + }, + { + "epoch": 0.06060606060606061, + "grad_norm": 1.985407829284668, + "learning_rate": 7.142857142857143e-07, + "loss": 2.662, + "step": 48 + }, + { + "epoch": 0.06186868686868687, + "grad_norm": 1.945688247680664, + "learning_rate": 7.352941176470589e-07, + "loss": 2.6319, + "step": 49 + }, + { + "epoch": 0.06313131313131314, + "grad_norm": 1.9392870664596558, + "learning_rate": 7.563025210084034e-07, + "loss": 2.6094, + "step": 50 + }, + { + "epoch": 0.06313131313131314, + "eval_loss": 2.6554172039031982, + "eval_runtime": 37.0569, + "eval_samples_per_second": 26.986, + "eval_steps_per_second": 1.133, + "step": 50 + }, + { + "epoch": 0.06439393939393939, + "grad_norm": 1.9975577592849731, + "learning_rate": 7.77310924369748e-07, + "loss": 2.6045, + "step": 51 + }, + { + "epoch": 0.06565656565656566, + "grad_norm": 1.9733693599700928, + "learning_rate": 7.983193277310924e-07, + "loss": 2.649, + "step": 52 + }, + { + "epoch": 0.06691919191919192, + "grad_norm": 1.7324917316436768, + "learning_rate": 8.193277310924371e-07, + "loss": 2.5689, + "step": 53 + }, + { + "epoch": 0.06818181818181818, + "grad_norm": 1.7553645372390747, + "learning_rate": 8.403361344537816e-07, + "loss": 2.644, + "step": 54 + }, + { + "epoch": 0.06944444444444445, + "grad_norm": 1.8163801431655884, + "learning_rate": 8.613445378151261e-07, + "loss": 2.4943, + "step": 55 + }, + { + "epoch": 0.0707070707070707, + "grad_norm": 1.7717015743255615, + "learning_rate": 8.823529411764707e-07, + "loss": 2.6093, + "step": 56 + }, + { + "epoch": 0.07196969696969698, + "grad_norm": 1.7013663053512573, + "learning_rate": 9.033613445378152e-07, + "loss": 2.6376, + "step": 57 + }, + { + "epoch": 0.07323232323232323, + "grad_norm": 1.752943992614746, + "learning_rate": 9.243697478991598e-07, + "loss": 2.5172, + "step": 58 + }, + { + "epoch": 0.07449494949494949, + "grad_norm": 1.5362415313720703, + "learning_rate": 9.453781512605043e-07, + "loss": 2.6129, + "step": 59 + }, + { + "epoch": 0.07575757575757576, + "grad_norm": 1.514885425567627, + "learning_rate": 9.66386554621849e-07, + "loss": 2.6064, + "step": 60 + }, + { + "epoch": 0.07575757575757576, + "eval_loss": 2.5934340953826904, + "eval_runtime": 37.0782, + "eval_samples_per_second": 26.97, + "eval_steps_per_second": 1.133, + "step": 60 + }, + { + "epoch": 0.07702020202020202, + "grad_norm": 1.5915579795837402, + "learning_rate": 9.873949579831934e-07, + "loss": 2.5491, + "step": 61 + }, + { + "epoch": 0.07828282828282829, + "grad_norm": 1.601773977279663, + "learning_rate": 1.0084033613445378e-06, + "loss": 2.5685, + "step": 62 + }, + { + "epoch": 0.07954545454545454, + "grad_norm": 1.8327804803848267, + "learning_rate": 1.0294117647058825e-06, + "loss": 2.5763, + "step": 63 + }, + { + "epoch": 0.08080808080808081, + "grad_norm": 1.5276741981506348, + "learning_rate": 1.050420168067227e-06, + "loss": 2.5494, + "step": 64 + }, + { + "epoch": 0.08207070707070707, + "grad_norm": 1.7065422534942627, + "learning_rate": 1.0714285714285714e-06, + "loss": 2.449, + "step": 65 + }, + { + "epoch": 0.08333333333333333, + "grad_norm": 2.0551910400390625, + "learning_rate": 1.092436974789916e-06, + "loss": 2.4371, + "step": 66 + }, + { + "epoch": 0.0845959595959596, + "grad_norm": 2.154855728149414, + "learning_rate": 1.1134453781512607e-06, + "loss": 2.5417, + "step": 67 + }, + { + "epoch": 0.08585858585858586, + "grad_norm": 1.7922390699386597, + "learning_rate": 1.1344537815126051e-06, + "loss": 2.4754, + "step": 68 + }, + { + "epoch": 0.08712121212121213, + "grad_norm": 1.745632290840149, + "learning_rate": 1.1554621848739498e-06, + "loss": 2.4669, + "step": 69 + }, + { + "epoch": 0.08838383838383838, + "grad_norm": 1.8907713890075684, + "learning_rate": 1.1764705882352942e-06, + "loss": 2.4943, + "step": 70 + }, + { + "epoch": 0.08838383838383838, + "eval_loss": 2.518237829208374, + "eval_runtime": 37.074, + "eval_samples_per_second": 26.973, + "eval_steps_per_second": 1.133, + "step": 70 + }, + { + "epoch": 0.08964646464646464, + "grad_norm": 1.8178164958953857, + "learning_rate": 1.1974789915966389e-06, + "loss": 2.4891, + "step": 71 + }, + { + "epoch": 0.09090909090909091, + "grad_norm": 1.970969796180725, + "learning_rate": 1.2184873949579833e-06, + "loss": 2.4797, + "step": 72 + }, + { + "epoch": 0.09217171717171717, + "grad_norm": 2.0211431980133057, + "learning_rate": 1.2394957983193278e-06, + "loss": 2.4716, + "step": 73 + }, + { + "epoch": 0.09343434343434344, + "grad_norm": 2.650787591934204, + "learning_rate": 1.2605042016806724e-06, + "loss": 2.4915, + "step": 74 + }, + { + "epoch": 0.0946969696969697, + "grad_norm": 2.235499143600464, + "learning_rate": 1.2815126050420169e-06, + "loss": 2.3917, + "step": 75 + }, + { + "epoch": 0.09595959595959595, + "grad_norm": 2.245999813079834, + "learning_rate": 1.3025210084033615e-06, + "loss": 2.4942, + "step": 76 + }, + { + "epoch": 0.09722222222222222, + "grad_norm": 2.4414420127868652, + "learning_rate": 1.323529411764706e-06, + "loss": 2.4004, + "step": 77 + }, + { + "epoch": 0.09848484848484848, + "grad_norm": 2.279937982559204, + "learning_rate": 1.3445378151260504e-06, + "loss": 2.4043, + "step": 78 + }, + { + "epoch": 0.09974747474747475, + "grad_norm": 2.3830068111419678, + "learning_rate": 1.3655462184873949e-06, + "loss": 2.4096, + "step": 79 + }, + { + "epoch": 0.10101010101010101, + "grad_norm": 2.0119733810424805, + "learning_rate": 1.3865546218487397e-06, + "loss": 2.4227, + "step": 80 + }, + { + "epoch": 0.10101010101010101, + "eval_loss": 2.4316186904907227, + "eval_runtime": 37.0837, + "eval_samples_per_second": 26.966, + "eval_steps_per_second": 1.133, + "step": 80 + }, + { + "epoch": 0.10227272727272728, + "grad_norm": 2.1273374557495117, + "learning_rate": 1.4075630252100842e-06, + "loss": 2.3541, + "step": 81 + }, + { + "epoch": 0.10353535353535354, + "grad_norm": 2.6852898597717285, + "learning_rate": 1.4285714285714286e-06, + "loss": 2.3321, + "step": 82 + }, + { + "epoch": 0.10479797979797979, + "grad_norm": 2.2622721195220947, + "learning_rate": 1.4495798319327733e-06, + "loss": 2.3166, + "step": 83 + }, + { + "epoch": 0.10606060606060606, + "grad_norm": 2.98258900642395, + "learning_rate": 1.4705882352941177e-06, + "loss": 2.3733, + "step": 84 + }, + { + "epoch": 0.10732323232323232, + "grad_norm": 2.5140459537506104, + "learning_rate": 1.4915966386554624e-06, + "loss": 2.314, + "step": 85 + }, + { + "epoch": 0.10858585858585859, + "grad_norm": 2.637608051300049, + "learning_rate": 1.5126050420168068e-06, + "loss": 2.3479, + "step": 86 + }, + { + "epoch": 0.10984848484848485, + "grad_norm": 2.399508476257324, + "learning_rate": 1.5336134453781515e-06, + "loss": 2.2932, + "step": 87 + }, + { + "epoch": 0.1111111111111111, + "grad_norm": 3.0835793018341064, + "learning_rate": 1.554621848739496e-06, + "loss": 2.2886, + "step": 88 + }, + { + "epoch": 0.11237373737373738, + "grad_norm": 3.288769483566284, + "learning_rate": 1.5756302521008404e-06, + "loss": 2.2914, + "step": 89 + }, + { + "epoch": 0.11363636363636363, + "grad_norm": 2.538361072540283, + "learning_rate": 1.5966386554621848e-06, + "loss": 2.3247, + "step": 90 + }, + { + "epoch": 0.11363636363636363, + "eval_loss": 2.3510072231292725, + "eval_runtime": 37.0847, + "eval_samples_per_second": 26.965, + "eval_steps_per_second": 1.133, + "step": 90 + }, + { + "epoch": 0.1148989898989899, + "grad_norm": 2.8016724586486816, + "learning_rate": 1.6176470588235297e-06, + "loss": 2.281, + "step": 91 + }, + { + "epoch": 0.11616161616161616, + "grad_norm": 2.9251859188079834, + "learning_rate": 1.6386554621848741e-06, + "loss": 2.2715, + "step": 92 + }, + { + "epoch": 0.11742424242424243, + "grad_norm": 2.595750570297241, + "learning_rate": 1.6596638655462186e-06, + "loss": 2.2127, + "step": 93 + }, + { + "epoch": 0.11868686868686869, + "grad_norm": 2.6989378929138184, + "learning_rate": 1.6806722689075632e-06, + "loss": 2.2623, + "step": 94 + }, + { + "epoch": 0.11994949494949494, + "grad_norm": 2.554863929748535, + "learning_rate": 1.7016806722689077e-06, + "loss": 2.2649, + "step": 95 + }, + { + "epoch": 0.12121212121212122, + "grad_norm": 2.7013020515441895, + "learning_rate": 1.7226890756302521e-06, + "loss": 2.2428, + "step": 96 + }, + { + "epoch": 0.12247474747474747, + "grad_norm": 2.7415151596069336, + "learning_rate": 1.7436974789915968e-06, + "loss": 2.1882, + "step": 97 + }, + { + "epoch": 0.12373737373737374, + "grad_norm": 2.672806739807129, + "learning_rate": 1.7647058823529414e-06, + "loss": 2.2302, + "step": 98 + }, + { + "epoch": 0.125, + "grad_norm": 2.9508326053619385, + "learning_rate": 1.7857142857142859e-06, + "loss": 2.1894, + "step": 99 + }, + { + "epoch": 0.12626262626262627, + "grad_norm": 2.4668166637420654, + "learning_rate": 1.8067226890756303e-06, + "loss": 2.2043, + "step": 100 + }, + { + "epoch": 0.12626262626262627, + "eval_loss": 2.2788307666778564, + "eval_runtime": 37.1156, + "eval_samples_per_second": 26.943, + "eval_steps_per_second": 1.132, + "step": 100 + }, + { + "epoch": 0.1275252525252525, + "grad_norm": 2.8241467475891113, + "learning_rate": 1.8277310924369748e-06, + "loss": 2.2109, + "step": 101 + }, + { + "epoch": 0.12878787878787878, + "grad_norm": 2.4179229736328125, + "learning_rate": 1.8487394957983196e-06, + "loss": 2.2193, + "step": 102 + }, + { + "epoch": 0.13005050505050506, + "grad_norm": 2.714315891265869, + "learning_rate": 1.869747899159664e-06, + "loss": 2.2474, + "step": 103 + }, + { + "epoch": 0.13131313131313133, + "grad_norm": 2.8560359477996826, + "learning_rate": 1.8907563025210085e-06, + "loss": 2.2205, + "step": 104 + }, + { + "epoch": 0.13257575757575757, + "grad_norm": 2.7780051231384277, + "learning_rate": 1.9117647058823528e-06, + "loss": 2.1834, + "step": 105 + }, + { + "epoch": 0.13383838383838384, + "grad_norm": 3.3773467540740967, + "learning_rate": 1.932773109243698e-06, + "loss": 2.2063, + "step": 106 + }, + { + "epoch": 0.1351010101010101, + "grad_norm": 2.927551746368408, + "learning_rate": 1.953781512605042e-06, + "loss": 2.1592, + "step": 107 + }, + { + "epoch": 0.13636363636363635, + "grad_norm": 2.8094165325164795, + "learning_rate": 1.9747899159663867e-06, + "loss": 2.2619, + "step": 108 + }, + { + "epoch": 0.13762626262626262, + "grad_norm": 2.5312564373016357, + "learning_rate": 1.9957983193277314e-06, + "loss": 2.2323, + "step": 109 + }, + { + "epoch": 0.1388888888888889, + "grad_norm": 3.175746202468872, + "learning_rate": 2.0168067226890756e-06, + "loss": 2.1776, + "step": 110 + }, + { + "epoch": 0.1388888888888889, + "eval_loss": 2.2208564281463623, + "eval_runtime": 37.3885, + "eval_samples_per_second": 26.746, + "eval_steps_per_second": 1.123, + "step": 110 + }, + { + "epoch": 0.14015151515151514, + "grad_norm": 2.660886287689209, + "learning_rate": 2.0378151260504203e-06, + "loss": 2.1626, + "step": 111 + }, + { + "epoch": 0.1414141414141414, + "grad_norm": 2.6948511600494385, + "learning_rate": 2.058823529411765e-06, + "loss": 2.1731, + "step": 112 + }, + { + "epoch": 0.14267676767676768, + "grad_norm": 2.826976776123047, + "learning_rate": 2.0798319327731096e-06, + "loss": 2.1865, + "step": 113 + }, + { + "epoch": 0.14393939393939395, + "grad_norm": 3.210033655166626, + "learning_rate": 2.100840336134454e-06, + "loss": 2.1325, + "step": 114 + }, + { + "epoch": 0.1452020202020202, + "grad_norm": 2.509060859680176, + "learning_rate": 2.1218487394957985e-06, + "loss": 2.0926, + "step": 115 + }, + { + "epoch": 0.14646464646464646, + "grad_norm": 3.51330304145813, + "learning_rate": 2.1428571428571427e-06, + "loss": 2.1577, + "step": 116 + }, + { + "epoch": 0.14772727272727273, + "grad_norm": 2.468395471572876, + "learning_rate": 2.163865546218488e-06, + "loss": 2.1, + "step": 117 + }, + { + "epoch": 0.14898989898989898, + "grad_norm": 3.081529378890991, + "learning_rate": 2.184873949579832e-06, + "loss": 2.1082, + "step": 118 + }, + { + "epoch": 0.15025252525252525, + "grad_norm": 2.8856358528137207, + "learning_rate": 2.2058823529411767e-06, + "loss": 2.1288, + "step": 119 + }, + { + "epoch": 0.15151515151515152, + "grad_norm": 3.323089838027954, + "learning_rate": 2.2268907563025214e-06, + "loss": 2.0933, + "step": 120 + }, + { + "epoch": 0.15151515151515152, + "eval_loss": 2.1834487915039062, + "eval_runtime": 37.4344, + "eval_samples_per_second": 26.713, + "eval_steps_per_second": 1.122, + "step": 120 + }, + { + "epoch": 0.1527777777777778, + "grad_norm": 2.8871560096740723, + "learning_rate": 2.2478991596638656e-06, + "loss": 2.1519, + "step": 121 + }, + { + "epoch": 0.15404040404040403, + "grad_norm": 3.0718588829040527, + "learning_rate": 2.2689075630252102e-06, + "loss": 2.0321, + "step": 122 + }, + { + "epoch": 0.1553030303030303, + "grad_norm": 2.820739507675171, + "learning_rate": 2.289915966386555e-06, + "loss": 2.0944, + "step": 123 + }, + { + "epoch": 0.15656565656565657, + "grad_norm": 3.1259310245513916, + "learning_rate": 2.3109243697478996e-06, + "loss": 2.0141, + "step": 124 + }, + { + "epoch": 0.15782828282828282, + "grad_norm": 3.303004741668701, + "learning_rate": 2.331932773109244e-06, + "loss": 2.1095, + "step": 125 + }, + { + "epoch": 0.1590909090909091, + "grad_norm": 3.1838674545288086, + "learning_rate": 2.3529411764705885e-06, + "loss": 2.0727, + "step": 126 + }, + { + "epoch": 0.16035353535353536, + "grad_norm": 3.1013786792755127, + "learning_rate": 2.3739495798319327e-06, + "loss": 2.0963, + "step": 127 + }, + { + "epoch": 0.16161616161616163, + "grad_norm": 2.803283214569092, + "learning_rate": 2.3949579831932778e-06, + "loss": 2.0551, + "step": 128 + }, + { + "epoch": 0.16287878787878787, + "grad_norm": 3.1616814136505127, + "learning_rate": 2.415966386554622e-06, + "loss": 2.0485, + "step": 129 + }, + { + "epoch": 0.16414141414141414, + "grad_norm": 3.333782196044922, + "learning_rate": 2.4369747899159667e-06, + "loss": 2.041, + "step": 130 + }, + { + "epoch": 0.16414141414141414, + "eval_loss": 2.150329351425171, + "eval_runtime": 37.494, + "eval_samples_per_second": 26.671, + "eval_steps_per_second": 1.12, + "step": 130 + }, + { + "epoch": 0.16540404040404041, + "grad_norm": 2.9714694023132324, + "learning_rate": 2.457983193277311e-06, + "loss": 2.0686, + "step": 131 + }, + { + "epoch": 0.16666666666666666, + "grad_norm": 3.216895818710327, + "learning_rate": 2.4789915966386555e-06, + "loss": 2.003, + "step": 132 + }, + { + "epoch": 0.16792929292929293, + "grad_norm": 2.9636809825897217, + "learning_rate": 2.5e-06, + "loss": 2.0534, + "step": 133 + }, + { + "epoch": 0.1691919191919192, + "grad_norm": 3.6477177143096924, + "learning_rate": 2.521008403361345e-06, + "loss": 2.077, + "step": 134 + }, + { + "epoch": 0.17045454545454544, + "grad_norm": 2.7437169551849365, + "learning_rate": 2.542016806722689e-06, + "loss": 2.0501, + "step": 135 + }, + { + "epoch": 0.1717171717171717, + "grad_norm": 3.2992663383483887, + "learning_rate": 2.5630252100840338e-06, + "loss": 2.0284, + "step": 136 + }, + { + "epoch": 0.17297979797979798, + "grad_norm": 3.0735950469970703, + "learning_rate": 2.5840336134453784e-06, + "loss": 2.025, + "step": 137 + }, + { + "epoch": 0.17424242424242425, + "grad_norm": 3.588418960571289, + "learning_rate": 2.605042016806723e-06, + "loss": 2.002, + "step": 138 + }, + { + "epoch": 0.1755050505050505, + "grad_norm": 2.8772335052490234, + "learning_rate": 2.6260504201680677e-06, + "loss": 1.9873, + "step": 139 + }, + { + "epoch": 0.17676767676767677, + "grad_norm": 2.9078969955444336, + "learning_rate": 2.647058823529412e-06, + "loss": 2.011, + "step": 140 + }, + { + "epoch": 0.17676767676767677, + "eval_loss": 2.1225531101226807, + "eval_runtime": 37.4531, + "eval_samples_per_second": 26.7, + "eval_steps_per_second": 1.121, + "step": 140 + }, + { + "epoch": 0.17803030303030304, + "grad_norm": 3.556654214859009, + "learning_rate": 2.6680672268907566e-06, + "loss": 2.0401, + "step": 141 + }, + { + "epoch": 0.17929292929292928, + "grad_norm": 2.9375200271606445, + "learning_rate": 2.689075630252101e-06, + "loss": 2.0277, + "step": 142 + }, + { + "epoch": 0.18055555555555555, + "grad_norm": 3.4019014835357666, + "learning_rate": 2.7100840336134455e-06, + "loss": 1.975, + "step": 143 + }, + { + "epoch": 0.18181818181818182, + "grad_norm": 2.7653748989105225, + "learning_rate": 2.7310924369747897e-06, + "loss": 2.0102, + "step": 144 + }, + { + "epoch": 0.1830808080808081, + "grad_norm": 3.4416797161102295, + "learning_rate": 2.752100840336135e-06, + "loss": 2.0093, + "step": 145 + }, + { + "epoch": 0.18434343434343434, + "grad_norm": 3.2531726360321045, + "learning_rate": 2.7731092436974795e-06, + "loss": 1.9852, + "step": 146 + }, + { + "epoch": 0.1856060606060606, + "grad_norm": 3.171502113342285, + "learning_rate": 2.7941176470588237e-06, + "loss": 1.9165, + "step": 147 + }, + { + "epoch": 0.18686868686868688, + "grad_norm": 3.114550828933716, + "learning_rate": 2.8151260504201684e-06, + "loss": 1.9516, + "step": 148 + }, + { + "epoch": 0.18813131313131312, + "grad_norm": 3.3601152896881104, + "learning_rate": 2.8361344537815126e-06, + "loss": 2.0183, + "step": 149 + }, + { + "epoch": 0.1893939393939394, + "grad_norm": 3.059622049331665, + "learning_rate": 2.8571428571428573e-06, + "loss": 1.9682, + "step": 150 + }, + { + "epoch": 0.1893939393939394, + "eval_loss": 2.0962483882904053, + "eval_runtime": 37.2364, + "eval_samples_per_second": 26.855, + "eval_steps_per_second": 1.128, + "step": 150 + }, + { + "epoch": 0.19065656565656566, + "grad_norm": 2.8749427795410156, + "learning_rate": 2.8781512605042015e-06, + "loss": 2.0089, + "step": 151 + }, + { + "epoch": 0.1919191919191919, + "grad_norm": 3.067267894744873, + "learning_rate": 2.8991596638655466e-06, + "loss": 1.9661, + "step": 152 + }, + { + "epoch": 0.19318181818181818, + "grad_norm": 2.7126266956329346, + "learning_rate": 2.9201680672268912e-06, + "loss": 1.982, + "step": 153 + }, + { + "epoch": 0.19444444444444445, + "grad_norm": 3.1330652236938477, + "learning_rate": 2.9411764705882355e-06, + "loss": 1.9753, + "step": 154 + }, + { + "epoch": 0.19570707070707072, + "grad_norm": 3.158313274383545, + "learning_rate": 2.96218487394958e-06, + "loss": 1.9431, + "step": 155 + }, + { + "epoch": 0.19696969696969696, + "grad_norm": 3.30806303024292, + "learning_rate": 2.9831932773109248e-06, + "loss": 1.9812, + "step": 156 + }, + { + "epoch": 0.19823232323232323, + "grad_norm": 3.0344884395599365, + "learning_rate": 3.004201680672269e-06, + "loss": 1.9677, + "step": 157 + }, + { + "epoch": 0.1994949494949495, + "grad_norm": 2.9032909870147705, + "learning_rate": 3.0252100840336137e-06, + "loss": 1.9132, + "step": 158 + }, + { + "epoch": 0.20075757575757575, + "grad_norm": 3.459934711456299, + "learning_rate": 3.046218487394958e-06, + "loss": 1.9581, + "step": 159 + }, + { + "epoch": 0.20202020202020202, + "grad_norm": 2.9455478191375732, + "learning_rate": 3.067226890756303e-06, + "loss": 1.9509, + "step": 160 + }, + { + "epoch": 0.20202020202020202, + "eval_loss": 2.074032783508301, + "eval_runtime": 37.1824, + "eval_samples_per_second": 26.894, + "eval_steps_per_second": 1.13, + "step": 160 + }, + { + "epoch": 0.2032828282828283, + "grad_norm": 3.2347211837768555, + "learning_rate": 3.0882352941176476e-06, + "loss": 1.9743, + "step": 161 + }, + { + "epoch": 0.20454545454545456, + "grad_norm": 3.3152947425842285, + "learning_rate": 3.109243697478992e-06, + "loss": 1.8971, + "step": 162 + }, + { + "epoch": 0.2058080808080808, + "grad_norm": 3.035548210144043, + "learning_rate": 3.1302521008403365e-06, + "loss": 1.8762, + "step": 163 + }, + { + "epoch": 0.20707070707070707, + "grad_norm": 3.1004066467285156, + "learning_rate": 3.1512605042016808e-06, + "loss": 1.9512, + "step": 164 + }, + { + "epoch": 0.20833333333333334, + "grad_norm": 3.2280282974243164, + "learning_rate": 3.1722689075630254e-06, + "loss": 1.8856, + "step": 165 + }, + { + "epoch": 0.20959595959595959, + "grad_norm": 3.021969795227051, + "learning_rate": 3.1932773109243696e-06, + "loss": 1.9127, + "step": 166 + }, + { + "epoch": 0.21085858585858586, + "grad_norm": 3.1647000312805176, + "learning_rate": 3.2142857142857147e-06, + "loss": 2.038, + "step": 167 + }, + { + "epoch": 0.21212121212121213, + "grad_norm": 3.0084872245788574, + "learning_rate": 3.2352941176470594e-06, + "loss": 1.9536, + "step": 168 + }, + { + "epoch": 0.21338383838383837, + "grad_norm": 2.951509952545166, + "learning_rate": 3.2563025210084036e-06, + "loss": 1.933, + "step": 169 + }, + { + "epoch": 0.21464646464646464, + "grad_norm": 3.472856283187866, + "learning_rate": 3.2773109243697483e-06, + "loss": 1.934, + "step": 170 + }, + { + "epoch": 0.21464646464646464, + "eval_loss": 2.0450048446655273, + "eval_runtime": 37.1669, + "eval_samples_per_second": 26.906, + "eval_steps_per_second": 1.13, + "step": 170 + }, + { + "epoch": 0.2159090909090909, + "grad_norm": 3.1104896068573, + "learning_rate": 3.2983193277310925e-06, + "loss": 1.9008, + "step": 171 + }, + { + "epoch": 0.21717171717171718, + "grad_norm": 2.847781181335449, + "learning_rate": 3.319327731092437e-06, + "loss": 1.9592, + "step": 172 + }, + { + "epoch": 0.21843434343434343, + "grad_norm": 3.384554624557495, + "learning_rate": 3.3403361344537814e-06, + "loss": 1.9087, + "step": 173 + }, + { + "epoch": 0.2196969696969697, + "grad_norm": 3.221564292907715, + "learning_rate": 3.3613445378151265e-06, + "loss": 1.9107, + "step": 174 + }, + { + "epoch": 0.22095959595959597, + "grad_norm": 3.30204439163208, + "learning_rate": 3.382352941176471e-06, + "loss": 1.8879, + "step": 175 + }, + { + "epoch": 0.2222222222222222, + "grad_norm": 3.549259662628174, + "learning_rate": 3.4033613445378154e-06, + "loss": 1.9592, + "step": 176 + }, + { + "epoch": 0.22348484848484848, + "grad_norm": 3.498199462890625, + "learning_rate": 3.42436974789916e-06, + "loss": 1.984, + "step": 177 + }, + { + "epoch": 0.22474747474747475, + "grad_norm": 3.576317548751831, + "learning_rate": 3.4453781512605043e-06, + "loss": 1.8897, + "step": 178 + }, + { + "epoch": 0.22601010101010102, + "grad_norm": 3.5337026119232178, + "learning_rate": 3.466386554621849e-06, + "loss": 1.9607, + "step": 179 + }, + { + "epoch": 0.22727272727272727, + "grad_norm": 3.342886209487915, + "learning_rate": 3.4873949579831936e-06, + "loss": 1.8797, + "step": 180 + }, + { + "epoch": 0.22727272727272727, + "eval_loss": 2.02158522605896, + "eval_runtime": 37.1382, + "eval_samples_per_second": 26.926, + "eval_steps_per_second": 1.131, + "step": 180 + }, + { + "epoch": 0.22853535353535354, + "grad_norm": 2.941906213760376, + "learning_rate": 3.508403361344538e-06, + "loss": 1.9, + "step": 181 + }, + { + "epoch": 0.2297979797979798, + "grad_norm": 3.1913063526153564, + "learning_rate": 3.529411764705883e-06, + "loss": 1.8704, + "step": 182 + }, + { + "epoch": 0.23106060606060605, + "grad_norm": 3.0619900226593018, + "learning_rate": 3.550420168067227e-06, + "loss": 1.8851, + "step": 183 + }, + { + "epoch": 0.23232323232323232, + "grad_norm": 3.051288604736328, + "learning_rate": 3.5714285714285718e-06, + "loss": 1.9422, + "step": 184 + }, + { + "epoch": 0.2335858585858586, + "grad_norm": 2.999845266342163, + "learning_rate": 3.5924369747899164e-06, + "loss": 1.8757, + "step": 185 + }, + { + "epoch": 0.23484848484848486, + "grad_norm": 3.262355089187622, + "learning_rate": 3.6134453781512607e-06, + "loss": 1.9114, + "step": 186 + }, + { + "epoch": 0.2361111111111111, + "grad_norm": 3.0354340076446533, + "learning_rate": 3.6344537815126053e-06, + "loss": 1.8547, + "step": 187 + }, + { + "epoch": 0.23737373737373738, + "grad_norm": 2.9401369094848633, + "learning_rate": 3.6554621848739496e-06, + "loss": 1.8974, + "step": 188 + }, + { + "epoch": 0.23863636363636365, + "grad_norm": 3.0544443130493164, + "learning_rate": 3.6764705882352946e-06, + "loss": 1.8662, + "step": 189 + }, + { + "epoch": 0.2398989898989899, + "grad_norm": 2.7257750034332275, + "learning_rate": 3.6974789915966393e-06, + "loss": 1.8737, + "step": 190 + }, + { + "epoch": 0.2398989898989899, + "eval_loss": 2.0065128803253174, + "eval_runtime": 37.1296, + "eval_samples_per_second": 26.933, + "eval_steps_per_second": 1.131, + "step": 190 + }, + { + "epoch": 0.24116161616161616, + "grad_norm": 2.8592522144317627, + "learning_rate": 3.7184873949579835e-06, + "loss": 1.8984, + "step": 191 + }, + { + "epoch": 0.24242424242424243, + "grad_norm": 2.833958864212036, + "learning_rate": 3.739495798319328e-06, + "loss": 1.9084, + "step": 192 + }, + { + "epoch": 0.24368686868686867, + "grad_norm": 2.7842230796813965, + "learning_rate": 3.7605042016806724e-06, + "loss": 1.8843, + "step": 193 + }, + { + "epoch": 0.24494949494949494, + "grad_norm": 3.1309409141540527, + "learning_rate": 3.781512605042017e-06, + "loss": 1.8259, + "step": 194 + }, + { + "epoch": 0.24621212121212122, + "grad_norm": 3.120788335800171, + "learning_rate": 3.8025210084033613e-06, + "loss": 1.8951, + "step": 195 + }, + { + "epoch": 0.2474747474747475, + "grad_norm": 3.2228143215179443, + "learning_rate": 3.8235294117647055e-06, + "loss": 1.8608, + "step": 196 + }, + { + "epoch": 0.24873737373737373, + "grad_norm": 3.3362114429473877, + "learning_rate": 3.844537815126051e-06, + "loss": 1.9163, + "step": 197 + }, + { + "epoch": 0.25, + "grad_norm": 3.0973799228668213, + "learning_rate": 3.865546218487396e-06, + "loss": 1.911, + "step": 198 + }, + { + "epoch": 0.25126262626262624, + "grad_norm": 3.5567545890808105, + "learning_rate": 3.8865546218487395e-06, + "loss": 1.8795, + "step": 199 + }, + { + "epoch": 0.25252525252525254, + "grad_norm": 3.4758830070495605, + "learning_rate": 3.907563025210084e-06, + "loss": 1.8813, + "step": 200 + }, + { + "epoch": 0.25252525252525254, + "eval_loss": 1.9892102479934692, + "eval_runtime": 37.1311, + "eval_samples_per_second": 26.932, + "eval_steps_per_second": 1.131, + "step": 200 + }, + { + "epoch": 0.2537878787878788, + "grad_norm": 3.5564475059509277, + "learning_rate": 3.928571428571429e-06, + "loss": 1.8469, + "step": 201 + }, + { + "epoch": 0.255050505050505, + "grad_norm": 3.7004899978637695, + "learning_rate": 3.9495798319327735e-06, + "loss": 1.873, + "step": 202 + }, + { + "epoch": 0.2563131313131313, + "grad_norm": 3.9285225868225098, + "learning_rate": 3.970588235294118e-06, + "loss": 1.8596, + "step": 203 + }, + { + "epoch": 0.25757575757575757, + "grad_norm": 3.1388907432556152, + "learning_rate": 3.991596638655463e-06, + "loss": 1.7991, + "step": 204 + }, + { + "epoch": 0.2588383838383838, + "grad_norm": 3.3797011375427246, + "learning_rate": 4.0126050420168075e-06, + "loss": 1.8682, + "step": 205 + }, + { + "epoch": 0.2601010101010101, + "grad_norm": 3.32183575630188, + "learning_rate": 4.033613445378151e-06, + "loss": 1.8396, + "step": 206 + }, + { + "epoch": 0.26136363636363635, + "grad_norm": 3.457291603088379, + "learning_rate": 4.054621848739496e-06, + "loss": 1.8314, + "step": 207 + }, + { + "epoch": 0.26262626262626265, + "grad_norm": 3.2906265258789062, + "learning_rate": 4.075630252100841e-06, + "loss": 1.844, + "step": 208 + }, + { + "epoch": 0.2638888888888889, + "grad_norm": 3.081913948059082, + "learning_rate": 4.096638655462185e-06, + "loss": 1.8198, + "step": 209 + }, + { + "epoch": 0.26515151515151514, + "grad_norm": 2.9697105884552, + "learning_rate": 4.11764705882353e-06, + "loss": 1.7885, + "step": 210 + }, + { + "epoch": 0.26515151515151514, + "eval_loss": 1.9749963283538818, + "eval_runtime": 36.962, + "eval_samples_per_second": 27.055, + "eval_steps_per_second": 1.136, + "step": 210 + }, + { + "epoch": 0.26641414141414144, + "grad_norm": 3.0732104778289795, + "learning_rate": 4.138655462184874e-06, + "loss": 1.8463, + "step": 211 + }, + { + "epoch": 0.2676767676767677, + "grad_norm": 3.4714958667755127, + "learning_rate": 4.159663865546219e-06, + "loss": 1.8537, + "step": 212 + }, + { + "epoch": 0.2689393939393939, + "grad_norm": 2.955864429473877, + "learning_rate": 4.180672268907564e-06, + "loss": 1.8566, + "step": 213 + }, + { + "epoch": 0.2702020202020202, + "grad_norm": 3.5314948558807373, + "learning_rate": 4.201680672268908e-06, + "loss": 1.7625, + "step": 214 + }, + { + "epoch": 0.27146464646464646, + "grad_norm": 2.8855810165405273, + "learning_rate": 4.222689075630252e-06, + "loss": 1.7968, + "step": 215 + }, + { + "epoch": 0.2727272727272727, + "grad_norm": 3.7866363525390625, + "learning_rate": 4.243697478991597e-06, + "loss": 1.8341, + "step": 216 + }, + { + "epoch": 0.273989898989899, + "grad_norm": 3.0352272987365723, + "learning_rate": 4.264705882352942e-06, + "loss": 1.8276, + "step": 217 + }, + { + "epoch": 0.27525252525252525, + "grad_norm": 3.4880247116088867, + "learning_rate": 4.2857142857142855e-06, + "loss": 1.8194, + "step": 218 + }, + { + "epoch": 0.2765151515151515, + "grad_norm": 3.6807212829589844, + "learning_rate": 4.306722689075631e-06, + "loss": 1.8495, + "step": 219 + }, + { + "epoch": 0.2777777777777778, + "grad_norm": 3.029017448425293, + "learning_rate": 4.327731092436976e-06, + "loss": 1.8125, + "step": 220 + }, + { + "epoch": 0.2777777777777778, + "eval_loss": 1.9633830785751343, + "eval_runtime": 37.1259, + "eval_samples_per_second": 26.935, + "eval_steps_per_second": 1.131, + "step": 220 + }, + { + "epoch": 0.27904040404040403, + "grad_norm": 3.4820573329925537, + "learning_rate": 4.3487394957983194e-06, + "loss": 1.8159, + "step": 221 + }, + { + "epoch": 0.2803030303030303, + "grad_norm": 3.2057833671569824, + "learning_rate": 4.369747899159664e-06, + "loss": 1.8215, + "step": 222 + }, + { + "epoch": 0.2815656565656566, + "grad_norm": 3.477302312850952, + "learning_rate": 4.390756302521009e-06, + "loss": 1.7837, + "step": 223 + }, + { + "epoch": 0.2828282828282828, + "grad_norm": 2.8349595069885254, + "learning_rate": 4.411764705882353e-06, + "loss": 1.8555, + "step": 224 + }, + { + "epoch": 0.2840909090909091, + "grad_norm": 3.616478443145752, + "learning_rate": 4.432773109243698e-06, + "loss": 1.8067, + "step": 225 + }, + { + "epoch": 0.28535353535353536, + "grad_norm": 3.111969470977783, + "learning_rate": 4.453781512605043e-06, + "loss": 1.8322, + "step": 226 + }, + { + "epoch": 0.2866161616161616, + "grad_norm": 3.248695135116577, + "learning_rate": 4.474789915966387e-06, + "loss": 1.8004, + "step": 227 + }, + { + "epoch": 0.2878787878787879, + "grad_norm": 3.017082691192627, + "learning_rate": 4.495798319327731e-06, + "loss": 1.776, + "step": 228 + }, + { + "epoch": 0.28914141414141414, + "grad_norm": 3.6790099143981934, + "learning_rate": 4.516806722689076e-06, + "loss": 1.8176, + "step": 229 + }, + { + "epoch": 0.2904040404040404, + "grad_norm": 3.1513001918792725, + "learning_rate": 4.5378151260504205e-06, + "loss": 1.8016, + "step": 230 + }, + { + "epoch": 0.2904040404040404, + "eval_loss": 1.949769377708435, + "eval_runtime": 37.2573, + "eval_samples_per_second": 26.84, + "eval_steps_per_second": 1.127, + "step": 230 + }, + { + "epoch": 0.2916666666666667, + "grad_norm": 3.0559096336364746, + "learning_rate": 4.558823529411765e-06, + "loss": 1.7915, + "step": 231 + }, + { + "epoch": 0.29292929292929293, + "grad_norm": 3.0974984169006348, + "learning_rate": 4.57983193277311e-06, + "loss": 1.8277, + "step": 232 + }, + { + "epoch": 0.29419191919191917, + "grad_norm": 2.8514139652252197, + "learning_rate": 4.600840336134454e-06, + "loss": 1.7616, + "step": 233 + }, + { + "epoch": 0.29545454545454547, + "grad_norm": 3.355198383331299, + "learning_rate": 4.621848739495799e-06, + "loss": 1.841, + "step": 234 + }, + { + "epoch": 0.2967171717171717, + "grad_norm": 3.1467435359954834, + "learning_rate": 4.642857142857144e-06, + "loss": 1.7393, + "step": 235 + }, + { + "epoch": 0.29797979797979796, + "grad_norm": 4.065819263458252, + "learning_rate": 4.663865546218488e-06, + "loss": 1.7508, + "step": 236 + }, + { + "epoch": 0.29924242424242425, + "grad_norm": 3.2602460384368896, + "learning_rate": 4.684873949579832e-06, + "loss": 1.7415, + "step": 237 + }, + { + "epoch": 0.3005050505050505, + "grad_norm": 3.6884894371032715, + "learning_rate": 4.705882352941177e-06, + "loss": 1.7608, + "step": 238 + }, + { + "epoch": 0.30176767676767674, + "grad_norm": 3.9843380451202393, + "learning_rate": 4.7268907563025216e-06, + "loss": 1.8519, + "step": 239 + }, + { + "epoch": 0.30303030303030304, + "grad_norm": 3.2823405265808105, + "learning_rate": 4.747899159663865e-06, + "loss": 1.7689, + "step": 240 + }, + { + "epoch": 0.30303030303030304, + "eval_loss": 1.939724326133728, + "eval_runtime": 37.465, + "eval_samples_per_second": 26.692, + "eval_steps_per_second": 1.121, + "step": 240 + }, + { + "epoch": 0.3042929292929293, + "grad_norm": 3.4200189113616943, + "learning_rate": 4.768907563025211e-06, + "loss": 1.7944, + "step": 241 + }, + { + "epoch": 0.3055555555555556, + "grad_norm": 3.7541189193725586, + "learning_rate": 4.7899159663865555e-06, + "loss": 1.8151, + "step": 242 + }, + { + "epoch": 0.3068181818181818, + "grad_norm": 3.9103832244873047, + "learning_rate": 4.810924369747899e-06, + "loss": 1.7921, + "step": 243 + }, + { + "epoch": 0.30808080808080807, + "grad_norm": 3.2769975662231445, + "learning_rate": 4.831932773109244e-06, + "loss": 1.8408, + "step": 244 + }, + { + "epoch": 0.30934343434343436, + "grad_norm": 3.414168357849121, + "learning_rate": 4.852941176470589e-06, + "loss": 1.7937, + "step": 245 + }, + { + "epoch": 0.3106060606060606, + "grad_norm": 3.219625473022461, + "learning_rate": 4.873949579831933e-06, + "loss": 1.796, + "step": 246 + }, + { + "epoch": 0.31186868686868685, + "grad_norm": 3.2573912143707275, + "learning_rate": 4.894957983193277e-06, + "loss": 1.8052, + "step": 247 + }, + { + "epoch": 0.31313131313131315, + "grad_norm": 3.4201040267944336, + "learning_rate": 4.915966386554622e-06, + "loss": 1.8529, + "step": 248 + }, + { + "epoch": 0.3143939393939394, + "grad_norm": 3.7448058128356934, + "learning_rate": 4.936974789915967e-06, + "loss": 1.8183, + "step": 249 + }, + { + "epoch": 0.31565656565656564, + "grad_norm": 3.4108386039733887, + "learning_rate": 4.957983193277311e-06, + "loss": 1.7926, + "step": 250 + }, + { + "epoch": 0.31565656565656564, + "eval_loss": 1.9319047927856445, + "eval_runtime": 37.4745, + "eval_samples_per_second": 26.685, + "eval_steps_per_second": 1.121, + "step": 250 + }, + { + "epoch": 0.31691919191919193, + "grad_norm": 3.50185227394104, + "learning_rate": 4.978991596638656e-06, + "loss": 1.792, + "step": 251 + }, + { + "epoch": 0.3181818181818182, + "grad_norm": 3.225581407546997, + "learning_rate": 5e-06, + "loss": 1.7286, + "step": 252 + }, + { + "epoch": 0.3194444444444444, + "grad_norm": 3.3070669174194336, + "learning_rate": 4.997661365762395e-06, + "loss": 1.7957, + "step": 253 + }, + { + "epoch": 0.3207070707070707, + "grad_norm": 3.498079776763916, + "learning_rate": 4.99532273152479e-06, + "loss": 1.823, + "step": 254 + }, + { + "epoch": 0.32196969696969696, + "grad_norm": 3.316534996032715, + "learning_rate": 4.9929840972871845e-06, + "loss": 1.8249, + "step": 255 + }, + { + "epoch": 0.32323232323232326, + "grad_norm": 3.1641833782196045, + "learning_rate": 4.990645463049579e-06, + "loss": 1.7658, + "step": 256 + }, + { + "epoch": 0.3244949494949495, + "grad_norm": 3.0794270038604736, + "learning_rate": 4.988306828811974e-06, + "loss": 1.8146, + "step": 257 + }, + { + "epoch": 0.32575757575757575, + "grad_norm": 3.2410383224487305, + "learning_rate": 4.985968194574369e-06, + "loss": 1.7894, + "step": 258 + }, + { + "epoch": 0.32702020202020204, + "grad_norm": 3.0899274349212646, + "learning_rate": 4.983629560336764e-06, + "loss": 1.7381, + "step": 259 + }, + { + "epoch": 0.3282828282828283, + "grad_norm": 3.0132803916931152, + "learning_rate": 4.981290926099159e-06, + "loss": 1.7713, + "step": 260 + }, + { + "epoch": 0.3282828282828283, + "eval_loss": 1.9123589992523193, + "eval_runtime": 37.3201, + "eval_samples_per_second": 26.795, + "eval_steps_per_second": 1.125, + "step": 260 + }, + { + "epoch": 0.32954545454545453, + "grad_norm": 2.92564058303833, + "learning_rate": 4.9789522918615536e-06, + "loss": 1.7659, + "step": 261 + }, + { + "epoch": 0.33080808080808083, + "grad_norm": 3.1402878761291504, + "learning_rate": 4.976613657623948e-06, + "loss": 1.8439, + "step": 262 + }, + { + "epoch": 0.33207070707070707, + "grad_norm": 3.170881509780884, + "learning_rate": 4.974275023386343e-06, + "loss": 1.7605, + "step": 263 + }, + { + "epoch": 0.3333333333333333, + "grad_norm": 3.014850378036499, + "learning_rate": 4.971936389148738e-06, + "loss": 1.7899, + "step": 264 + }, + { + "epoch": 0.3345959595959596, + "grad_norm": 3.169818878173828, + "learning_rate": 4.969597754911132e-06, + "loss": 1.8263, + "step": 265 + }, + { + "epoch": 0.33585858585858586, + "grad_norm": 2.8709678649902344, + "learning_rate": 4.967259120673527e-06, + "loss": 1.7834, + "step": 266 + }, + { + "epoch": 0.3371212121212121, + "grad_norm": 2.7527096271514893, + "learning_rate": 4.964920486435922e-06, + "loss": 1.8106, + "step": 267 + }, + { + "epoch": 0.3383838383838384, + "grad_norm": 2.9960227012634277, + "learning_rate": 4.962581852198316e-06, + "loss": 1.7452, + "step": 268 + }, + { + "epoch": 0.33964646464646464, + "grad_norm": 2.979163885116577, + "learning_rate": 4.9602432179607115e-06, + "loss": 1.6943, + "step": 269 + }, + { + "epoch": 0.3409090909090909, + "grad_norm": 3.2396063804626465, + "learning_rate": 4.957904583723106e-06, + "loss": 1.7798, + "step": 270 + }, + { + "epoch": 0.3409090909090909, + "eval_loss": 1.9126806259155273, + "eval_runtime": 37.123, + "eval_samples_per_second": 26.937, + "eval_steps_per_second": 1.131, + "step": 270 + }, + { + "epoch": 0.3421717171717172, + "grad_norm": 2.7868010997772217, + "learning_rate": 4.955565949485501e-06, + "loss": 1.8053, + "step": 271 + }, + { + "epoch": 0.3434343434343434, + "grad_norm": 2.8895349502563477, + "learning_rate": 4.9532273152478956e-06, + "loss": 1.7672, + "step": 272 + }, + { + "epoch": 0.3446969696969697, + "grad_norm": 3.0839710235595703, + "learning_rate": 4.95088868101029e-06, + "loss": 1.7414, + "step": 273 + }, + { + "epoch": 0.34595959595959597, + "grad_norm": 2.81974458694458, + "learning_rate": 4.948550046772685e-06, + "loss": 1.7527, + "step": 274 + }, + { + "epoch": 0.3472222222222222, + "grad_norm": 3.7469775676727295, + "learning_rate": 4.9462114125350805e-06, + "loss": 1.7689, + "step": 275 + }, + { + "epoch": 0.3484848484848485, + "grad_norm": 3.434573173522949, + "learning_rate": 4.943872778297475e-06, + "loss": 1.739, + "step": 276 + }, + { + "epoch": 0.34974747474747475, + "grad_norm": 2.985041379928589, + "learning_rate": 4.941534144059869e-06, + "loss": 1.7894, + "step": 277 + }, + { + "epoch": 0.351010101010101, + "grad_norm": 3.441204309463501, + "learning_rate": 4.939195509822265e-06, + "loss": 1.7804, + "step": 278 + }, + { + "epoch": 0.3522727272727273, + "grad_norm": 2.8371102809906006, + "learning_rate": 4.936856875584659e-06, + "loss": 1.7415, + "step": 279 + }, + { + "epoch": 0.35353535353535354, + "grad_norm": 3.662773847579956, + "learning_rate": 4.9345182413470535e-06, + "loss": 1.7894, + "step": 280 + }, + { + "epoch": 0.35353535353535354, + "eval_loss": 1.915844440460205, + "eval_runtime": 37.0941, + "eval_samples_per_second": 26.958, + "eval_steps_per_second": 1.132, + "step": 280 + }, + { + "epoch": 0.3547979797979798, + "grad_norm": 3.3254036903381348, + "learning_rate": 4.932179607109449e-06, + "loss": 1.7238, + "step": 281 + }, + { + "epoch": 0.3560606060606061, + "grad_norm": 2.9872665405273438, + "learning_rate": 4.929840972871843e-06, + "loss": 1.769, + "step": 282 + }, + { + "epoch": 0.3573232323232323, + "grad_norm": 3.0163521766662598, + "learning_rate": 4.9275023386342376e-06, + "loss": 1.7702, + "step": 283 + }, + { + "epoch": 0.35858585858585856, + "grad_norm": 2.9687387943267822, + "learning_rate": 4.925163704396633e-06, + "loss": 1.7927, + "step": 284 + }, + { + "epoch": 0.35984848484848486, + "grad_norm": 2.7064011096954346, + "learning_rate": 4.922825070159027e-06, + "loss": 1.7638, + "step": 285 + }, + { + "epoch": 0.3611111111111111, + "grad_norm": 3.0043134689331055, + "learning_rate": 4.9204864359214225e-06, + "loss": 1.8041, + "step": 286 + }, + { + "epoch": 0.36237373737373735, + "grad_norm": 3.037100315093994, + "learning_rate": 4.918147801683817e-06, + "loss": 1.7579, + "step": 287 + }, + { + "epoch": 0.36363636363636365, + "grad_norm": 2.7528676986694336, + "learning_rate": 4.915809167446211e-06, + "loss": 1.7629, + "step": 288 + }, + { + "epoch": 0.3648989898989899, + "grad_norm": 3.075815439224243, + "learning_rate": 4.913470533208607e-06, + "loss": 1.7966, + "step": 289 + }, + { + "epoch": 0.3661616161616162, + "grad_norm": 2.964785575866699, + "learning_rate": 4.911131898971001e-06, + "loss": 1.801, + "step": 290 + }, + { + "epoch": 0.3661616161616162, + "eval_loss": 1.8839410543441772, + "eval_runtime": 37.0844, + "eval_samples_per_second": 26.965, + "eval_steps_per_second": 1.133, + "step": 290 + }, + { + "epoch": 0.36742424242424243, + "grad_norm": 2.668747663497925, + "learning_rate": 4.908793264733396e-06, + "loss": 1.7703, + "step": 291 + }, + { + "epoch": 0.3686868686868687, + "grad_norm": 2.814751148223877, + "learning_rate": 4.906454630495791e-06, + "loss": 1.7075, + "step": 292 + }, + { + "epoch": 0.369949494949495, + "grad_norm": 3.005922794342041, + "learning_rate": 4.904115996258186e-06, + "loss": 1.7418, + "step": 293 + }, + { + "epoch": 0.3712121212121212, + "grad_norm": 3.050936460494995, + "learning_rate": 4.90177736202058e-06, + "loss": 1.793, + "step": 294 + }, + { + "epoch": 0.37247474747474746, + "grad_norm": 2.822004556655884, + "learning_rate": 4.899438727782975e-06, + "loss": 1.8303, + "step": 295 + }, + { + "epoch": 0.37373737373737376, + "grad_norm": 2.950254201889038, + "learning_rate": 4.89710009354537e-06, + "loss": 1.808, + "step": 296 + }, + { + "epoch": 0.375, + "grad_norm": 2.8140666484832764, + "learning_rate": 4.8947614593077645e-06, + "loss": 1.7262, + "step": 297 + }, + { + "epoch": 0.37626262626262624, + "grad_norm": 2.899620532989502, + "learning_rate": 4.89242282507016e-06, + "loss": 1.7157, + "step": 298 + }, + { + "epoch": 0.37752525252525254, + "grad_norm": 2.805753469467163, + "learning_rate": 4.890084190832554e-06, + "loss": 1.7294, + "step": 299 + }, + { + "epoch": 0.3787878787878788, + "grad_norm": 3.110246181488037, + "learning_rate": 4.887745556594949e-06, + "loss": 1.7415, + "step": 300 + }, + { + "epoch": 0.3787878787878788, + "eval_loss": 1.892285943031311, + "eval_runtime": 37.0732, + "eval_samples_per_second": 26.974, + "eval_steps_per_second": 1.133, + "step": 300 + }, + { + "epoch": 0.380050505050505, + "grad_norm": 2.9197490215301514, + "learning_rate": 4.885406922357344e-06, + "loss": 1.7132, + "step": 301 + }, + { + "epoch": 0.3813131313131313, + "grad_norm": 3.3165123462677, + "learning_rate": 4.883068288119738e-06, + "loss": 1.777, + "step": 302 + }, + { + "epoch": 0.38257575757575757, + "grad_norm": 2.8409054279327393, + "learning_rate": 4.880729653882133e-06, + "loss": 1.7092, + "step": 303 + }, + { + "epoch": 0.3838383838383838, + "grad_norm": 2.6800901889801025, + "learning_rate": 4.878391019644528e-06, + "loss": 1.7622, + "step": 304 + }, + { + "epoch": 0.3851010101010101, + "grad_norm": 3.2847490310668945, + "learning_rate": 4.876052385406922e-06, + "loss": 1.729, + "step": 305 + }, + { + "epoch": 0.38636363636363635, + "grad_norm": 2.786057949066162, + "learning_rate": 4.873713751169318e-06, + "loss": 1.7168, + "step": 306 + }, + { + "epoch": 0.38762626262626265, + "grad_norm": 3.236050605773926, + "learning_rate": 4.871375116931712e-06, + "loss": 1.6809, + "step": 307 + }, + { + "epoch": 0.3888888888888889, + "grad_norm": 2.851910352706909, + "learning_rate": 4.869036482694107e-06, + "loss": 1.7336, + "step": 308 + }, + { + "epoch": 0.39015151515151514, + "grad_norm": 2.892421007156372, + "learning_rate": 4.866697848456502e-06, + "loss": 1.7451, + "step": 309 + }, + { + "epoch": 0.39141414141414144, + "grad_norm": 2.8175048828125, + "learning_rate": 4.864359214218897e-06, + "loss": 1.7734, + "step": 310 + }, + { + "epoch": 0.39141414141414144, + "eval_loss": 1.8777843713760376, + "eval_runtime": 36.9321, + "eval_samples_per_second": 27.077, + "eval_steps_per_second": 1.137, + "step": 310 + }, + { + "epoch": 0.3926767676767677, + "grad_norm": 3.005329132080078, + "learning_rate": 4.8620205799812915e-06, + "loss": 1.7476, + "step": 311 + }, + { + "epoch": 0.3939393939393939, + "grad_norm": 2.849017858505249, + "learning_rate": 4.859681945743686e-06, + "loss": 1.7854, + "step": 312 + }, + { + "epoch": 0.3952020202020202, + "grad_norm": 2.801048517227173, + "learning_rate": 4.857343311506081e-06, + "loss": 1.7186, + "step": 313 + }, + { + "epoch": 0.39646464646464646, + "grad_norm": 2.5772831439971924, + "learning_rate": 4.855004677268476e-06, + "loss": 1.7279, + "step": 314 + }, + { + "epoch": 0.3977272727272727, + "grad_norm": 2.7969934940338135, + "learning_rate": 4.85266604303087e-06, + "loss": 1.7475, + "step": 315 + }, + { + "epoch": 0.398989898989899, + "grad_norm": 2.7067770957946777, + "learning_rate": 4.850327408793265e-06, + "loss": 1.7055, + "step": 316 + }, + { + "epoch": 0.40025252525252525, + "grad_norm": 2.70928955078125, + "learning_rate": 4.84798877455566e-06, + "loss": 1.7203, + "step": 317 + }, + { + "epoch": 0.4015151515151515, + "grad_norm": 2.7704474925994873, + "learning_rate": 4.845650140318054e-06, + "loss": 1.8058, + "step": 318 + }, + { + "epoch": 0.4027777777777778, + "grad_norm": 2.7680182456970215, + "learning_rate": 4.843311506080449e-06, + "loss": 1.7162, + "step": 319 + }, + { + "epoch": 0.40404040404040403, + "grad_norm": 3.0953330993652344, + "learning_rate": 4.840972871842844e-06, + "loss": 1.7617, + "step": 320 + }, + { + "epoch": 0.40404040404040403, + "eval_loss": 1.8680965900421143, + "eval_runtime": 37.0382, + "eval_samples_per_second": 26.999, + "eval_steps_per_second": 1.134, + "step": 320 + }, + { + "epoch": 0.4053030303030303, + "grad_norm": 2.8340392112731934, + "learning_rate": 4.838634237605239e-06, + "loss": 1.7767, + "step": 321 + }, + { + "epoch": 0.4065656565656566, + "grad_norm": 3.057861804962158, + "learning_rate": 4.836295603367634e-06, + "loss": 1.8043, + "step": 322 + }, + { + "epoch": 0.4078282828282828, + "grad_norm": 3.0123419761657715, + "learning_rate": 4.833956969130029e-06, + "loss": 1.7561, + "step": 323 + }, + { + "epoch": 0.4090909090909091, + "grad_norm": 3.0873124599456787, + "learning_rate": 4.831618334892423e-06, + "loss": 1.7581, + "step": 324 + }, + { + "epoch": 0.41035353535353536, + "grad_norm": 2.6600539684295654, + "learning_rate": 4.8292797006548184e-06, + "loss": 1.7005, + "step": 325 + }, + { + "epoch": 0.4116161616161616, + "grad_norm": 2.905280590057373, + "learning_rate": 4.826941066417213e-06, + "loss": 1.7592, + "step": 326 + }, + { + "epoch": 0.4128787878787879, + "grad_norm": 3.0642189979553223, + "learning_rate": 4.824602432179607e-06, + "loss": 1.6456, + "step": 327 + }, + { + "epoch": 0.41414141414141414, + "grad_norm": 2.606531858444214, + "learning_rate": 4.8222637979420025e-06, + "loss": 1.6731, + "step": 328 + }, + { + "epoch": 0.4154040404040404, + "grad_norm": 2.9918158054351807, + "learning_rate": 4.819925163704397e-06, + "loss": 1.6854, + "step": 329 + }, + { + "epoch": 0.4166666666666667, + "grad_norm": 2.721940040588379, + "learning_rate": 4.817586529466791e-06, + "loss": 1.7346, + "step": 330 + }, + { + "epoch": 0.4166666666666667, + "eval_loss": 1.863736629486084, + "eval_runtime": 37.0887, + "eval_samples_per_second": 26.962, + "eval_steps_per_second": 1.132, + "step": 330 + }, + { + "epoch": 0.41792929292929293, + "grad_norm": 3.1076390743255615, + "learning_rate": 4.815247895229187e-06, + "loss": 1.6286, + "step": 331 + }, + { + "epoch": 0.41919191919191917, + "grad_norm": 3.1235783100128174, + "learning_rate": 4.812909260991581e-06, + "loss": 1.7404, + "step": 332 + }, + { + "epoch": 0.42045454545454547, + "grad_norm": 3.1211330890655518, + "learning_rate": 4.8105706267539755e-06, + "loss": 1.7143, + "step": 333 + }, + { + "epoch": 0.4217171717171717, + "grad_norm": 3.072317600250244, + "learning_rate": 4.808231992516371e-06, + "loss": 1.7514, + "step": 334 + }, + { + "epoch": 0.42297979797979796, + "grad_norm": 2.880671739578247, + "learning_rate": 4.805893358278765e-06, + "loss": 1.7018, + "step": 335 + }, + { + "epoch": 0.42424242424242425, + "grad_norm": 2.700063943862915, + "learning_rate": 4.8035547240411604e-06, + "loss": 1.7089, + "step": 336 + }, + { + "epoch": 0.4255050505050505, + "grad_norm": 2.9038445949554443, + "learning_rate": 4.801216089803556e-06, + "loss": 1.7483, + "step": 337 + }, + { + "epoch": 0.42676767676767674, + "grad_norm": 2.7822766304016113, + "learning_rate": 4.79887745556595e-06, + "loss": 1.6927, + "step": 338 + }, + { + "epoch": 0.42803030303030304, + "grad_norm": 2.957554340362549, + "learning_rate": 4.7965388213283445e-06, + "loss": 1.7162, + "step": 339 + }, + { + "epoch": 0.4292929292929293, + "grad_norm": 3.1333892345428467, + "learning_rate": 4.79420018709074e-06, + "loss": 1.731, + "step": 340 + }, + { + "epoch": 0.4292929292929293, + "eval_loss": 1.852413535118103, + "eval_runtime": 37.1809, + "eval_samples_per_second": 26.896, + "eval_steps_per_second": 1.13, + "step": 340 + }, + { + "epoch": 0.4305555555555556, + "grad_norm": 2.9801084995269775, + "learning_rate": 4.791861552853134e-06, + "loss": 1.6892, + "step": 341 + }, + { + "epoch": 0.4318181818181818, + "grad_norm": 2.5077271461486816, + "learning_rate": 4.789522918615529e-06, + "loss": 1.637, + "step": 342 + }, + { + "epoch": 0.43308080808080807, + "grad_norm": 2.8654391765594482, + "learning_rate": 4.787184284377924e-06, + "loss": 1.6915, + "step": 343 + }, + { + "epoch": 0.43434343434343436, + "grad_norm": 2.6204264163970947, + "learning_rate": 4.784845650140318e-06, + "loss": 1.6722, + "step": 344 + }, + { + "epoch": 0.4356060606060606, + "grad_norm": 2.633991003036499, + "learning_rate": 4.782507015902713e-06, + "loss": 1.6797, + "step": 345 + }, + { + "epoch": 0.43686868686868685, + "grad_norm": 2.7203924655914307, + "learning_rate": 4.780168381665108e-06, + "loss": 1.7018, + "step": 346 + }, + { + "epoch": 0.43813131313131315, + "grad_norm": 2.6789209842681885, + "learning_rate": 4.7778297474275024e-06, + "loss": 1.7153, + "step": 347 + }, + { + "epoch": 0.4393939393939394, + "grad_norm": 2.6285433769226074, + "learning_rate": 4.775491113189898e-06, + "loss": 1.7082, + "step": 348 + }, + { + "epoch": 0.44065656565656564, + "grad_norm": 2.623793840408325, + "learning_rate": 4.773152478952292e-06, + "loss": 1.7094, + "step": 349 + }, + { + "epoch": 0.44191919191919193, + "grad_norm": 2.647160291671753, + "learning_rate": 4.7708138447146865e-06, + "loss": 1.6528, + "step": 350 + }, + { + "epoch": 0.44191919191919193, + "eval_loss": 1.8506718873977661, + "eval_runtime": 37.4232, + "eval_samples_per_second": 26.721, + "eval_steps_per_second": 1.122, + "step": 350 + }, + { + "epoch": 0.4431818181818182, + "grad_norm": 2.664553642272949, + "learning_rate": 4.768475210477082e-06, + "loss": 1.7008, + "step": 351 + }, + { + "epoch": 0.4444444444444444, + "grad_norm": 2.8486251831054688, + "learning_rate": 4.766136576239477e-06, + "loss": 1.6811, + "step": 352 + }, + { + "epoch": 0.4457070707070707, + "grad_norm": 2.7514266967773438, + "learning_rate": 4.7637979420018715e-06, + "loss": 1.6969, + "step": 353 + }, + { + "epoch": 0.44696969696969696, + "grad_norm": 2.848083257675171, + "learning_rate": 4.761459307764266e-06, + "loss": 1.7058, + "step": 354 + }, + { + "epoch": 0.44823232323232326, + "grad_norm": 2.759218454360962, + "learning_rate": 4.759120673526661e-06, + "loss": 1.6109, + "step": 355 + }, + { + "epoch": 0.4494949494949495, + "grad_norm": 2.905318021774292, + "learning_rate": 4.756782039289056e-06, + "loss": 1.7689, + "step": 356 + }, + { + "epoch": 0.45075757575757575, + "grad_norm": 2.845646381378174, + "learning_rate": 4.75444340505145e-06, + "loss": 1.6687, + "step": 357 + }, + { + "epoch": 0.45202020202020204, + "grad_norm": 2.8609306812286377, + "learning_rate": 4.752104770813845e-06, + "loss": 1.6332, + "step": 358 + }, + { + "epoch": 0.4532828282828283, + "grad_norm": 3.016479969024658, + "learning_rate": 4.74976613657624e-06, + "loss": 1.6692, + "step": 359 + }, + { + "epoch": 0.45454545454545453, + "grad_norm": 3.2790887355804443, + "learning_rate": 4.747427502338635e-06, + "loss": 1.704, + "step": 360 + }, + { + "epoch": 0.45454545454545453, + "eval_loss": 1.848663330078125, + "eval_runtime": 37.4817, + "eval_samples_per_second": 26.68, + "eval_steps_per_second": 1.121, + "step": 360 + }, + { + "epoch": 0.45580808080808083, + "grad_norm": 3.1137545108795166, + "learning_rate": 4.745088868101029e-06, + "loss": 1.6797, + "step": 361 + }, + { + "epoch": 0.45707070707070707, + "grad_norm": 2.7598798274993896, + "learning_rate": 4.742750233863424e-06, + "loss": 1.6477, + "step": 362 + }, + { + "epoch": 0.4583333333333333, + "grad_norm": 3.4851579666137695, + "learning_rate": 4.740411599625819e-06, + "loss": 1.6481, + "step": 363 + }, + { + "epoch": 0.4595959595959596, + "grad_norm": 2.9285919666290283, + "learning_rate": 4.7380729653882135e-06, + "loss": 1.7094, + "step": 364 + }, + { + "epoch": 0.46085858585858586, + "grad_norm": 2.901458501815796, + "learning_rate": 4.735734331150608e-06, + "loss": 1.6906, + "step": 365 + }, + { + "epoch": 0.4621212121212121, + "grad_norm": 2.8737902641296387, + "learning_rate": 4.733395696913003e-06, + "loss": 1.6819, + "step": 366 + }, + { + "epoch": 0.4633838383838384, + "grad_norm": 3.2480952739715576, + "learning_rate": 4.7310570626753984e-06, + "loss": 1.6827, + "step": 367 + }, + { + "epoch": 0.46464646464646464, + "grad_norm": 2.8714661598205566, + "learning_rate": 4.728718428437793e-06, + "loss": 1.7349, + "step": 368 + }, + { + "epoch": 0.4659090909090909, + "grad_norm": 2.8410098552703857, + "learning_rate": 4.726379794200187e-06, + "loss": 1.7214, + "step": 369 + }, + { + "epoch": 0.4671717171717172, + "grad_norm": 2.744807720184326, + "learning_rate": 4.7240411599625826e-06, + "loss": 1.6742, + "step": 370 + }, + { + "epoch": 0.4671717171717172, + "eval_loss": 1.842191457748413, + "eval_runtime": 37.454, + "eval_samples_per_second": 26.699, + "eval_steps_per_second": 1.121, + "step": 370 + }, + { + "epoch": 0.4684343434343434, + "grad_norm": 2.8135552406311035, + "learning_rate": 4.721702525724977e-06, + "loss": 1.6957, + "step": 371 + }, + { + "epoch": 0.4696969696969697, + "grad_norm": 2.856764078140259, + "learning_rate": 4.719363891487372e-06, + "loss": 1.6802, + "step": 372 + }, + { + "epoch": 0.47095959595959597, + "grad_norm": 3.0707688331604004, + "learning_rate": 4.717025257249767e-06, + "loss": 1.7383, + "step": 373 + }, + { + "epoch": 0.4722222222222222, + "grad_norm": 2.8832876682281494, + "learning_rate": 4.714686623012161e-06, + "loss": 1.6309, + "step": 374 + }, + { + "epoch": 0.4734848484848485, + "grad_norm": 2.6840622425079346, + "learning_rate": 4.712347988774556e-06, + "loss": 1.6624, + "step": 375 + }, + { + "epoch": 0.47474747474747475, + "grad_norm": 3.2920548915863037, + "learning_rate": 4.710009354536951e-06, + "loss": 1.6887, + "step": 376 + }, + { + "epoch": 0.476010101010101, + "grad_norm": 3.0317189693450928, + "learning_rate": 4.707670720299345e-06, + "loss": 1.6514, + "step": 377 + }, + { + "epoch": 0.4772727272727273, + "grad_norm": 2.7284793853759766, + "learning_rate": 4.7053320860617404e-06, + "loss": 1.6595, + "step": 378 + }, + { + "epoch": 0.47853535353535354, + "grad_norm": 2.86665415763855, + "learning_rate": 4.702993451824135e-06, + "loss": 1.7219, + "step": 379 + }, + { + "epoch": 0.4797979797979798, + "grad_norm": 3.041191577911377, + "learning_rate": 4.700654817586529e-06, + "loss": 1.6959, + "step": 380 + }, + { + "epoch": 0.4797979797979798, + "eval_loss": 1.8414148092269897, + "eval_runtime": 37.2537, + "eval_samples_per_second": 26.843, + "eval_steps_per_second": 1.127, + "step": 380 + }, + { + "epoch": 0.4810606060606061, + "grad_norm": 3.1729421615600586, + "learning_rate": 4.6983161833489246e-06, + "loss": 1.6538, + "step": 381 + }, + { + "epoch": 0.4823232323232323, + "grad_norm": 2.8562991619110107, + "learning_rate": 4.69597754911132e-06, + "loss": 1.6667, + "step": 382 + }, + { + "epoch": 0.48358585858585856, + "grad_norm": 3.0348711013793945, + "learning_rate": 4.693638914873714e-06, + "loss": 1.6888, + "step": 383 + }, + { + "epoch": 0.48484848484848486, + "grad_norm": 2.9156477451324463, + "learning_rate": 4.6913002806361095e-06, + "loss": 1.628, + "step": 384 + }, + { + "epoch": 0.4861111111111111, + "grad_norm": 2.832689046859741, + "learning_rate": 4.688961646398504e-06, + "loss": 1.664, + "step": 385 + }, + { + "epoch": 0.48737373737373735, + "grad_norm": 2.9858946800231934, + "learning_rate": 4.686623012160898e-06, + "loss": 1.6696, + "step": 386 + }, + { + "epoch": 0.48863636363636365, + "grad_norm": 2.732468843460083, + "learning_rate": 4.684284377923294e-06, + "loss": 1.6444, + "step": 387 + }, + { + "epoch": 0.4898989898989899, + "grad_norm": 2.7801132202148438, + "learning_rate": 4.681945743685688e-06, + "loss": 1.7084, + "step": 388 + }, + { + "epoch": 0.4911616161616162, + "grad_norm": 2.646303415298462, + "learning_rate": 4.6796071094480824e-06, + "loss": 1.6963, + "step": 389 + }, + { + "epoch": 0.49242424242424243, + "grad_norm": 2.8750250339508057, + "learning_rate": 4.677268475210478e-06, + "loss": 1.6814, + "step": 390 + }, + { + "epoch": 0.49242424242424243, + "eval_loss": 1.8283894062042236, + "eval_runtime": 37.2058, + "eval_samples_per_second": 26.877, + "eval_steps_per_second": 1.129, + "step": 390 + }, + { + "epoch": 0.4936868686868687, + "grad_norm": 2.5373270511627197, + "learning_rate": 4.674929840972872e-06, + "loss": 1.5956, + "step": 391 + }, + { + "epoch": 0.494949494949495, + "grad_norm": 2.837170124053955, + "learning_rate": 4.6725912067352666e-06, + "loss": 1.6237, + "step": 392 + }, + { + "epoch": 0.4962121212121212, + "grad_norm": 3.199463367462158, + "learning_rate": 4.670252572497662e-06, + "loss": 1.7087, + "step": 393 + }, + { + "epoch": 0.49747474747474746, + "grad_norm": 2.9850947856903076, + "learning_rate": 4.667913938260056e-06, + "loss": 1.5913, + "step": 394 + }, + { + "epoch": 0.49873737373737376, + "grad_norm": 2.954148292541504, + "learning_rate": 4.665575304022451e-06, + "loss": 1.6272, + "step": 395 + }, + { + "epoch": 0.5, + "grad_norm": 2.8625757694244385, + "learning_rate": 4.663236669784846e-06, + "loss": 1.6623, + "step": 396 + }, + { + "epoch": 0.5012626262626263, + "grad_norm": 3.196354627609253, + "learning_rate": 4.66089803554724e-06, + "loss": 1.7249, + "step": 397 + }, + { + "epoch": 0.5025252525252525, + "grad_norm": 2.7257323265075684, + "learning_rate": 4.658559401309636e-06, + "loss": 1.665, + "step": 398 + }, + { + "epoch": 0.5037878787878788, + "grad_norm": 3.0976016521453857, + "learning_rate": 4.656220767072031e-06, + "loss": 1.6763, + "step": 399 + }, + { + "epoch": 0.5050505050505051, + "grad_norm": 3.0398128032684326, + "learning_rate": 4.653882132834425e-06, + "loss": 1.6375, + "step": 400 + }, + { + "epoch": 0.5050505050505051, + "eval_loss": 1.8324880599975586, + "eval_runtime": 37.1754, + "eval_samples_per_second": 26.9, + "eval_steps_per_second": 1.13, + "step": 400 + }, + { + "epoch": 0.5063131313131313, + "grad_norm": 3.314958095550537, + "learning_rate": 4.65154349859682e-06, + "loss": 1.6699, + "step": 401 + }, + { + "epoch": 0.5075757575757576, + "grad_norm": 2.8478870391845703, + "learning_rate": 4.649204864359215e-06, + "loss": 1.6495, + "step": 402 + }, + { + "epoch": 0.5088383838383839, + "grad_norm": 2.8476510047912598, + "learning_rate": 4.646866230121609e-06, + "loss": 1.633, + "step": 403 + }, + { + "epoch": 0.51010101010101, + "grad_norm": 2.916667938232422, + "learning_rate": 4.644527595884004e-06, + "loss": 1.6092, + "step": 404 + }, + { + "epoch": 0.5113636363636364, + "grad_norm": 2.977226495742798, + "learning_rate": 4.642188961646399e-06, + "loss": 1.6588, + "step": 405 + }, + { + "epoch": 0.5126262626262627, + "grad_norm": 3.0540225505828857, + "learning_rate": 4.6398503274087935e-06, + "loss": 1.6529, + "step": 406 + }, + { + "epoch": 0.5138888888888888, + "grad_norm": 2.902035713195801, + "learning_rate": 4.637511693171188e-06, + "loss": 1.5983, + "step": 407 + }, + { + "epoch": 0.5151515151515151, + "grad_norm": 2.9009487628936768, + "learning_rate": 4.635173058933583e-06, + "loss": 1.6389, + "step": 408 + }, + { + "epoch": 0.5164141414141414, + "grad_norm": 2.904982805252075, + "learning_rate": 4.632834424695978e-06, + "loss": 1.6971, + "step": 409 + }, + { + "epoch": 0.5176767676767676, + "grad_norm": 2.8340721130371094, + "learning_rate": 4.630495790458373e-06, + "loss": 1.6336, + "step": 410 + }, + { + "epoch": 0.5176767676767676, + "eval_loss": 1.8218064308166504, + "eval_runtime": 36.9963, + "eval_samples_per_second": 27.03, + "eval_steps_per_second": 1.135, + "step": 410 + }, + { + "epoch": 0.5189393939393939, + "grad_norm": 2.9507815837860107, + "learning_rate": 4.628157156220767e-06, + "loss": 1.6795, + "step": 411 + }, + { + "epoch": 0.5202020202020202, + "grad_norm": 2.868793249130249, + "learning_rate": 4.625818521983162e-06, + "loss": 1.6494, + "step": 412 + }, + { + "epoch": 0.5214646464646465, + "grad_norm": 2.786395788192749, + "learning_rate": 4.623479887745557e-06, + "loss": 1.7156, + "step": 413 + }, + { + "epoch": 0.5227272727272727, + "grad_norm": 2.7456934452056885, + "learning_rate": 4.621141253507952e-06, + "loss": 1.6528, + "step": 414 + }, + { + "epoch": 0.523989898989899, + "grad_norm": 3.1211695671081543, + "learning_rate": 4.618802619270347e-06, + "loss": 1.6891, + "step": 415 + }, + { + "epoch": 0.5252525252525253, + "grad_norm": 3.2319254875183105, + "learning_rate": 4.616463985032741e-06, + "loss": 1.6452, + "step": 416 + }, + { + "epoch": 0.5265151515151515, + "grad_norm": 2.6718053817749023, + "learning_rate": 4.614125350795136e-06, + "loss": 1.6442, + "step": 417 + }, + { + "epoch": 0.5277777777777778, + "grad_norm": 2.985020637512207, + "learning_rate": 4.611786716557531e-06, + "loss": 1.6935, + "step": 418 + }, + { + "epoch": 0.5290404040404041, + "grad_norm": 2.798720598220825, + "learning_rate": 4.609448082319925e-06, + "loss": 1.6303, + "step": 419 + }, + { + "epoch": 0.5303030303030303, + "grad_norm": 2.860403299331665, + "learning_rate": 4.6071094480823205e-06, + "loss": 1.6445, + "step": 420 + }, + { + "epoch": 0.5303030303030303, + "eval_loss": 1.8173401355743408, + "eval_runtime": 37.1334, + "eval_samples_per_second": 26.93, + "eval_steps_per_second": 1.131, + "step": 420 + }, + { + "epoch": 0.5315656565656566, + "grad_norm": 2.883209705352783, + "learning_rate": 4.604770813844715e-06, + "loss": 1.6182, + "step": 421 + }, + { + "epoch": 0.5328282828282829, + "grad_norm": 3.0330183506011963, + "learning_rate": 4.60243217960711e-06, + "loss": 1.6208, + "step": 422 + }, + { + "epoch": 0.5340909090909091, + "grad_norm": 3.224184274673462, + "learning_rate": 4.6000935453695046e-06, + "loss": 1.6687, + "step": 423 + }, + { + "epoch": 0.5353535353535354, + "grad_norm": 3.0345349311828613, + "learning_rate": 4.597754911131899e-06, + "loss": 1.6654, + "step": 424 + }, + { + "epoch": 0.5366161616161617, + "grad_norm": 3.1425750255584717, + "learning_rate": 4.595416276894294e-06, + "loss": 1.6569, + "step": 425 + }, + { + "epoch": 0.5378787878787878, + "grad_norm": 2.6387979984283447, + "learning_rate": 4.593077642656689e-06, + "loss": 1.6868, + "step": 426 + }, + { + "epoch": 0.5391414141414141, + "grad_norm": 2.959446430206299, + "learning_rate": 4.590739008419083e-06, + "loss": 1.621, + "step": 427 + }, + { + "epoch": 0.5404040404040404, + "grad_norm": 3.009673833847046, + "learning_rate": 4.588400374181478e-06, + "loss": 1.6414, + "step": 428 + }, + { + "epoch": 0.5416666666666666, + "grad_norm": 2.827531576156616, + "learning_rate": 4.586061739943874e-06, + "loss": 1.6561, + "step": 429 + }, + { + "epoch": 0.5429292929292929, + "grad_norm": 2.8770735263824463, + "learning_rate": 4.583723105706268e-06, + "loss": 1.6685, + "step": 430 + }, + { + "epoch": 0.5429292929292929, + "eval_loss": 1.8234716653823853, + "eval_runtime": 37.1647, + "eval_samples_per_second": 26.907, + "eval_steps_per_second": 1.13, + "step": 430 + }, + { + "epoch": 0.5441919191919192, + "grad_norm": 3.1215782165527344, + "learning_rate": 4.5813844714686625e-06, + "loss": 1.6548, + "step": 431 + }, + { + "epoch": 0.5454545454545454, + "grad_norm": 2.6587517261505127, + "learning_rate": 4.579045837231058e-06, + "loss": 1.649, + "step": 432 + }, + { + "epoch": 0.5467171717171717, + "grad_norm": 2.7059030532836914, + "learning_rate": 4.576707202993452e-06, + "loss": 1.6293, + "step": 433 + }, + { + "epoch": 0.547979797979798, + "grad_norm": 2.7390925884246826, + "learning_rate": 4.574368568755847e-06, + "loss": 1.6957, + "step": 434 + }, + { + "epoch": 0.5492424242424242, + "grad_norm": 2.9457757472991943, + "learning_rate": 4.572029934518242e-06, + "loss": 1.6714, + "step": 435 + }, + { + "epoch": 0.5505050505050505, + "grad_norm": 2.6871347427368164, + "learning_rate": 4.569691300280636e-06, + "loss": 1.6648, + "step": 436 + }, + { + "epoch": 0.5517676767676768, + "grad_norm": 3.1708056926727295, + "learning_rate": 4.5673526660430315e-06, + "loss": 1.6415, + "step": 437 + }, + { + "epoch": 0.553030303030303, + "grad_norm": 2.8767378330230713, + "learning_rate": 4.565014031805426e-06, + "loss": 1.6475, + "step": 438 + }, + { + "epoch": 0.5542929292929293, + "grad_norm": 3.083566188812256, + "learning_rate": 4.56267539756782e-06, + "loss": 1.632, + "step": 439 + }, + { + "epoch": 0.5555555555555556, + "grad_norm": 2.584460735321045, + "learning_rate": 4.560336763330216e-06, + "loss": 1.6469, + "step": 440 + }, + { + "epoch": 0.5555555555555556, + "eval_loss": 1.8140095472335815, + "eval_runtime": 37.1986, + "eval_samples_per_second": 26.883, + "eval_steps_per_second": 1.129, + "step": 440 + }, + { + "epoch": 0.5568181818181818, + "grad_norm": 2.5885088443756104, + "learning_rate": 4.55799812909261e-06, + "loss": 1.6058, + "step": 441 + }, + { + "epoch": 0.5580808080808081, + "grad_norm": 2.8467836380004883, + "learning_rate": 4.5556594948550045e-06, + "loss": 1.6959, + "step": 442 + }, + { + "epoch": 0.5593434343434344, + "grad_norm": 2.764326333999634, + "learning_rate": 4.5533208606174e-06, + "loss": 1.6565, + "step": 443 + }, + { + "epoch": 0.5606060606060606, + "grad_norm": 2.88089919090271, + "learning_rate": 4.550982226379795e-06, + "loss": 1.6651, + "step": 444 + }, + { + "epoch": 0.5618686868686869, + "grad_norm": 2.6239469051361084, + "learning_rate": 4.548643592142189e-06, + "loss": 1.6255, + "step": 445 + }, + { + "epoch": 0.5631313131313131, + "grad_norm": 2.7748732566833496, + "learning_rate": 4.546304957904585e-06, + "loss": 1.6587, + "step": 446 + }, + { + "epoch": 0.5643939393939394, + "grad_norm": 2.7075202465057373, + "learning_rate": 4.543966323666979e-06, + "loss": 1.6448, + "step": 447 + }, + { + "epoch": 0.5656565656565656, + "grad_norm": 2.9291634559631348, + "learning_rate": 4.5416276894293735e-06, + "loss": 1.6564, + "step": 448 + }, + { + "epoch": 0.5669191919191919, + "grad_norm": 3.0039470195770264, + "learning_rate": 4.539289055191769e-06, + "loss": 1.623, + "step": 449 + }, + { + "epoch": 0.5681818181818182, + "grad_norm": 3.085277557373047, + "learning_rate": 4.536950420954163e-06, + "loss": 1.6176, + "step": 450 + }, + { + "epoch": 0.5681818181818182, + "eval_loss": 1.8022958040237427, + "eval_runtime": 37.2204, + "eval_samples_per_second": 26.867, + "eval_steps_per_second": 1.128, + "step": 450 + }, + { + "epoch": 0.5694444444444444, + "grad_norm": 2.93575119972229, + "learning_rate": 4.534611786716558e-06, + "loss": 1.615, + "step": 451 + }, + { + "epoch": 0.5707070707070707, + "grad_norm": 2.847141742706299, + "learning_rate": 4.532273152478953e-06, + "loss": 1.6474, + "step": 452 + }, + { + "epoch": 0.571969696969697, + "grad_norm": 3.0299603939056396, + "learning_rate": 4.529934518241347e-06, + "loss": 1.6424, + "step": 453 + }, + { + "epoch": 0.5732323232323232, + "grad_norm": 3.0307669639587402, + "learning_rate": 4.527595884003742e-06, + "loss": 1.6473, + "step": 454 + }, + { + "epoch": 0.5744949494949495, + "grad_norm": 2.864269971847534, + "learning_rate": 4.525257249766137e-06, + "loss": 1.6631, + "step": 455 + }, + { + "epoch": 0.5757575757575758, + "grad_norm": 2.606637716293335, + "learning_rate": 4.522918615528531e-06, + "loss": 1.6118, + "step": 456 + }, + { + "epoch": 0.577020202020202, + "grad_norm": 3.127891778945923, + "learning_rate": 4.520579981290926e-06, + "loss": 1.6286, + "step": 457 + }, + { + "epoch": 0.5782828282828283, + "grad_norm": 2.8809893131256104, + "learning_rate": 4.518241347053321e-06, + "loss": 1.6158, + "step": 458 + }, + { + "epoch": 0.5795454545454546, + "grad_norm": 3.0662436485290527, + "learning_rate": 4.515902712815716e-06, + "loss": 1.7336, + "step": 459 + }, + { + "epoch": 0.5808080808080808, + "grad_norm": 3.0795772075653076, + "learning_rate": 4.513564078578111e-06, + "loss": 1.6974, + "step": 460 + }, + { + "epoch": 0.5808080808080808, + "eval_loss": 1.7981404066085815, + "eval_runtime": 37.299, + "eval_samples_per_second": 26.81, + "eval_steps_per_second": 1.126, + "step": 460 + }, + { + "epoch": 0.5820707070707071, + "grad_norm": 3.091688871383667, + "learning_rate": 4.511225444340506e-06, + "loss": 1.6803, + "step": 461 + }, + { + "epoch": 0.5833333333333334, + "grad_norm": 2.5597071647644043, + "learning_rate": 4.5088868101029005e-06, + "loss": 1.6104, + "step": 462 + }, + { + "epoch": 0.5845959595959596, + "grad_norm": 2.849806070327759, + "learning_rate": 4.506548175865295e-06, + "loss": 1.6147, + "step": 463 + }, + { + "epoch": 0.5858585858585859, + "grad_norm": 2.9791204929351807, + "learning_rate": 4.50420954162769e-06, + "loss": 1.6376, + "step": 464 + }, + { + "epoch": 0.5871212121212122, + "grad_norm": 2.8582305908203125, + "learning_rate": 4.501870907390085e-06, + "loss": 1.6831, + "step": 465 + }, + { + "epoch": 0.5883838383838383, + "grad_norm": 2.8445184230804443, + "learning_rate": 4.499532273152479e-06, + "loss": 1.6463, + "step": 466 + }, + { + "epoch": 0.5896464646464646, + "grad_norm": 2.858168125152588, + "learning_rate": 4.497193638914874e-06, + "loss": 1.6662, + "step": 467 + }, + { + "epoch": 0.5909090909090909, + "grad_norm": 2.829702854156494, + "learning_rate": 4.494855004677269e-06, + "loss": 1.6058, + "step": 468 + }, + { + "epoch": 0.5921717171717171, + "grad_norm": 2.9292726516723633, + "learning_rate": 4.492516370439663e-06, + "loss": 1.6341, + "step": 469 + }, + { + "epoch": 0.5934343434343434, + "grad_norm": 2.9006292819976807, + "learning_rate": 4.490177736202058e-06, + "loss": 1.5726, + "step": 470 + }, + { + "epoch": 0.5934343434343434, + "eval_loss": 1.7953916788101196, + "eval_runtime": 37.5058, + "eval_samples_per_second": 26.663, + "eval_steps_per_second": 1.12, + "step": 470 + }, + { + "epoch": 0.5946969696969697, + "grad_norm": 2.8116230964660645, + "learning_rate": 4.487839101964453e-06, + "loss": 1.6083, + "step": 471 + }, + { + "epoch": 0.5959595959595959, + "grad_norm": 2.540607452392578, + "learning_rate": 4.485500467726848e-06, + "loss": 1.6208, + "step": 472 + }, + { + "epoch": 0.5972222222222222, + "grad_norm": 2.7339096069335938, + "learning_rate": 4.4831618334892425e-06, + "loss": 1.6839, + "step": 473 + }, + { + "epoch": 0.5984848484848485, + "grad_norm": 2.9094762802124023, + "learning_rate": 4.480823199251638e-06, + "loss": 1.6962, + "step": 474 + }, + { + "epoch": 0.5997474747474747, + "grad_norm": 2.7733027935028076, + "learning_rate": 4.478484565014032e-06, + "loss": 1.6252, + "step": 475 + }, + { + "epoch": 0.601010101010101, + "grad_norm": 2.5709428787231445, + "learning_rate": 4.4761459307764274e-06, + "loss": 1.5896, + "step": 476 + }, + { + "epoch": 0.6022727272727273, + "grad_norm": 2.8358278274536133, + "learning_rate": 4.473807296538822e-06, + "loss": 1.6247, + "step": 477 + }, + { + "epoch": 0.6035353535353535, + "grad_norm": 2.7504897117614746, + "learning_rate": 4.471468662301216e-06, + "loss": 1.7138, + "step": 478 + }, + { + "epoch": 0.6047979797979798, + "grad_norm": 2.6965620517730713, + "learning_rate": 4.4691300280636115e-06, + "loss": 1.6054, + "step": 479 + }, + { + "epoch": 0.6060606060606061, + "grad_norm": 2.9401891231536865, + "learning_rate": 4.466791393826006e-06, + "loss": 1.6172, + "step": 480 + }, + { + "epoch": 0.6060606060606061, + "eval_loss": 1.7946635484695435, + "eval_runtime": 37.5366, + "eval_samples_per_second": 26.641, + "eval_steps_per_second": 1.119, + "step": 480 + }, + { + "epoch": 0.6073232323232324, + "grad_norm": 2.71535587310791, + "learning_rate": 4.4644527595884e-06, + "loss": 1.61, + "step": 481 + }, + { + "epoch": 0.6085858585858586, + "grad_norm": 2.7910654544830322, + "learning_rate": 4.462114125350796e-06, + "loss": 1.6193, + "step": 482 + }, + { + "epoch": 0.6098484848484849, + "grad_norm": 2.809241533279419, + "learning_rate": 4.45977549111319e-06, + "loss": 1.602, + "step": 483 + }, + { + "epoch": 0.6111111111111112, + "grad_norm": 2.7152137756347656, + "learning_rate": 4.457436856875585e-06, + "loss": 1.6516, + "step": 484 + }, + { + "epoch": 0.6123737373737373, + "grad_norm": 2.7653465270996094, + "learning_rate": 4.45509822263798e-06, + "loss": 1.5904, + "step": 485 + }, + { + "epoch": 0.6136363636363636, + "grad_norm": 2.729876756668091, + "learning_rate": 4.452759588400374e-06, + "loss": 1.6259, + "step": 486 + }, + { + "epoch": 0.61489898989899, + "grad_norm": 2.6290180683135986, + "learning_rate": 4.4504209541627694e-06, + "loss": 1.5796, + "step": 487 + }, + { + "epoch": 0.6161616161616161, + "grad_norm": 2.939884662628174, + "learning_rate": 4.448082319925164e-06, + "loss": 1.6774, + "step": 488 + }, + { + "epoch": 0.6174242424242424, + "grad_norm": 2.7554702758789062, + "learning_rate": 4.445743685687559e-06, + "loss": 1.6226, + "step": 489 + }, + { + "epoch": 0.6186868686868687, + "grad_norm": 2.890458106994629, + "learning_rate": 4.4434050514499535e-06, + "loss": 1.592, + "step": 490 + }, + { + "epoch": 0.6186868686868687, + "eval_loss": 1.7868437767028809, + "eval_runtime": 37.5267, + "eval_samples_per_second": 26.648, + "eval_steps_per_second": 1.119, + "step": 490 + }, + { + "epoch": 0.6199494949494949, + "grad_norm": 2.757178783416748, + "learning_rate": 4.441066417212349e-06, + "loss": 1.6297, + "step": 491 + }, + { + "epoch": 0.6212121212121212, + "grad_norm": 2.7179954051971436, + "learning_rate": 4.438727782974743e-06, + "loss": 1.5685, + "step": 492 + }, + { + "epoch": 0.6224747474747475, + "grad_norm": 2.824514389038086, + "learning_rate": 4.436389148737138e-06, + "loss": 1.6625, + "step": 493 + }, + { + "epoch": 0.6237373737373737, + "grad_norm": 3.0062615871429443, + "learning_rate": 4.434050514499533e-06, + "loss": 1.5889, + "step": 494 + }, + { + "epoch": 0.625, + "grad_norm": 2.962176561355591, + "learning_rate": 4.431711880261927e-06, + "loss": 1.6813, + "step": 495 + }, + { + "epoch": 0.6262626262626263, + "grad_norm": 2.7531204223632812, + "learning_rate": 4.429373246024323e-06, + "loss": 1.6152, + "step": 496 + }, + { + "epoch": 0.6275252525252525, + "grad_norm": 2.764021873474121, + "learning_rate": 4.427034611786717e-06, + "loss": 1.6164, + "step": 497 + }, + { + "epoch": 0.6287878787878788, + "grad_norm": 2.836826801300049, + "learning_rate": 4.4246959775491114e-06, + "loss": 1.6192, + "step": 498 + }, + { + "epoch": 0.6300505050505051, + "grad_norm": 2.871220827102661, + "learning_rate": 4.422357343311507e-06, + "loss": 1.6096, + "step": 499 + }, + { + "epoch": 0.6313131313131313, + "grad_norm": 2.8129167556762695, + "learning_rate": 4.420018709073901e-06, + "loss": 1.6674, + "step": 500 + }, + { + "epoch": 0.6313131313131313, + "eval_loss": 1.785156011581421, + "eval_runtime": 37.2971, + "eval_samples_per_second": 26.812, + "eval_steps_per_second": 1.126, + "step": 500 + }, + { + "epoch": 0.6325757575757576, + "grad_norm": 2.775108814239502, + "learning_rate": 4.4176800748362955e-06, + "loss": 1.5851, + "step": 501 + }, + { + "epoch": 0.6338383838383839, + "grad_norm": 2.7775518894195557, + "learning_rate": 4.415341440598691e-06, + "loss": 1.6839, + "step": 502 + }, + { + "epoch": 0.63510101010101, + "grad_norm": 2.7997918128967285, + "learning_rate": 4.413002806361085e-06, + "loss": 1.5849, + "step": 503 + }, + { + "epoch": 0.6363636363636364, + "grad_norm": 2.91103196144104, + "learning_rate": 4.4106641721234805e-06, + "loss": 1.6369, + "step": 504 + }, + { + "epoch": 0.6376262626262627, + "grad_norm": 2.6800882816314697, + "learning_rate": 4.408325537885875e-06, + "loss": 1.6049, + "step": 505 + }, + { + "epoch": 0.6388888888888888, + "grad_norm": 2.7451021671295166, + "learning_rate": 4.40598690364827e-06, + "loss": 1.658, + "step": 506 + }, + { + "epoch": 0.6401515151515151, + "grad_norm": 2.794813632965088, + "learning_rate": 4.403648269410665e-06, + "loss": 1.6015, + "step": 507 + }, + { + "epoch": 0.6414141414141414, + "grad_norm": 2.769373655319214, + "learning_rate": 4.401309635173059e-06, + "loss": 1.6217, + "step": 508 + }, + { + "epoch": 0.6426767676767676, + "grad_norm": 2.9081523418426514, + "learning_rate": 4.398971000935454e-06, + "loss": 1.6235, + "step": 509 + }, + { + "epoch": 0.6439393939393939, + "grad_norm": 2.707618236541748, + "learning_rate": 4.396632366697849e-06, + "loss": 1.6003, + "step": 510 + }, + { + "epoch": 0.6439393939393939, + "eval_loss": 1.7876425981521606, + "eval_runtime": 36.9832, + "eval_samples_per_second": 27.039, + "eval_steps_per_second": 1.136, + "step": 510 + }, + { + "epoch": 0.6452020202020202, + "grad_norm": 2.7923099994659424, + "learning_rate": 4.394293732460244e-06, + "loss": 1.5647, + "step": 511 + }, + { + "epoch": 0.6464646464646465, + "grad_norm": 2.8449175357818604, + "learning_rate": 4.391955098222638e-06, + "loss": 1.6456, + "step": 512 + }, + { + "epoch": 0.6477272727272727, + "grad_norm": 2.755856990814209, + "learning_rate": 4.389616463985033e-06, + "loss": 1.6962, + "step": 513 + }, + { + "epoch": 0.648989898989899, + "grad_norm": 2.5578861236572266, + "learning_rate": 4.387277829747428e-06, + "loss": 1.6097, + "step": 514 + }, + { + "epoch": 0.6502525252525253, + "grad_norm": 2.8804502487182617, + "learning_rate": 4.3849391955098225e-06, + "loss": 1.6212, + "step": 515 + }, + { + "epoch": 0.6515151515151515, + "grad_norm": 2.684215545654297, + "learning_rate": 4.382600561272217e-06, + "loss": 1.5529, + "step": 516 + }, + { + "epoch": 0.6527777777777778, + "grad_norm": 2.9128026962280273, + "learning_rate": 4.380261927034612e-06, + "loss": 1.6424, + "step": 517 + }, + { + "epoch": 0.6540404040404041, + "grad_norm": 2.733415126800537, + "learning_rate": 4.377923292797007e-06, + "loss": 1.6201, + "step": 518 + }, + { + "epoch": 0.6553030303030303, + "grad_norm": 2.7211062908172607, + "learning_rate": 4.375584658559401e-06, + "loss": 1.5833, + "step": 519 + }, + { + "epoch": 0.6565656565656566, + "grad_norm": 2.7308316230773926, + "learning_rate": 4.373246024321796e-06, + "loss": 1.6288, + "step": 520 + }, + { + "epoch": 0.6565656565656566, + "eval_loss": 1.7787725925445557, + "eval_runtime": 37.1089, + "eval_samples_per_second": 26.948, + "eval_steps_per_second": 1.132, + "step": 520 + }, + { + "epoch": 0.6578282828282829, + "grad_norm": 2.760362148284912, + "learning_rate": 4.3709073900841915e-06, + "loss": 1.5814, + "step": 521 + }, + { + "epoch": 0.6590909090909091, + "grad_norm": 2.7625153064727783, + "learning_rate": 4.368568755846586e-06, + "loss": 1.6239, + "step": 522 + }, + { + "epoch": 0.6603535353535354, + "grad_norm": 2.715710401535034, + "learning_rate": 4.366230121608981e-06, + "loss": 1.6069, + "step": 523 + }, + { + "epoch": 0.6616161616161617, + "grad_norm": 2.7753846645355225, + "learning_rate": 4.363891487371376e-06, + "loss": 1.6178, + "step": 524 + }, + { + "epoch": 0.6628787878787878, + "grad_norm": 2.650716543197632, + "learning_rate": 4.36155285313377e-06, + "loss": 1.6534, + "step": 525 + }, + { + "epoch": 0.6641414141414141, + "grad_norm": 2.573969602584839, + "learning_rate": 4.359214218896165e-06, + "loss": 1.6721, + "step": 526 + }, + { + "epoch": 0.6654040404040404, + "grad_norm": 2.9017715454101562, + "learning_rate": 4.35687558465856e-06, + "loss": 1.6273, + "step": 527 + }, + { + "epoch": 0.6666666666666666, + "grad_norm": 2.691643714904785, + "learning_rate": 4.354536950420954e-06, + "loss": 1.6199, + "step": 528 + }, + { + "epoch": 0.6679292929292929, + "grad_norm": 2.818662166595459, + "learning_rate": 4.3521983161833494e-06, + "loss": 1.5758, + "step": 529 + }, + { + "epoch": 0.6691919191919192, + "grad_norm": 3.017738103866577, + "learning_rate": 4.349859681945744e-06, + "loss": 1.6437, + "step": 530 + }, + { + "epoch": 0.6691919191919192, + "eval_loss": 1.7835458517074585, + "eval_runtime": 37.1359, + "eval_samples_per_second": 26.928, + "eval_steps_per_second": 1.131, + "step": 530 + }, + { + "epoch": 0.6704545454545454, + "grad_norm": 2.8027284145355225, + "learning_rate": 4.347521047708138e-06, + "loss": 1.6688, + "step": 531 + }, + { + "epoch": 0.6717171717171717, + "grad_norm": 2.6642649173736572, + "learning_rate": 4.3451824134705335e-06, + "loss": 1.6327, + "step": 532 + }, + { + "epoch": 0.672979797979798, + "grad_norm": 2.7406396865844727, + "learning_rate": 4.342843779232928e-06, + "loss": 1.6297, + "step": 533 + }, + { + "epoch": 0.6742424242424242, + "grad_norm": 2.758538007736206, + "learning_rate": 4.340505144995323e-06, + "loss": 1.5276, + "step": 534 + }, + { + "epoch": 0.6755050505050505, + "grad_norm": 2.815817356109619, + "learning_rate": 4.3381665107577185e-06, + "loss": 1.6192, + "step": 535 + }, + { + "epoch": 0.6767676767676768, + "grad_norm": 2.8416755199432373, + "learning_rate": 4.335827876520113e-06, + "loss": 1.5966, + "step": 536 + }, + { + "epoch": 0.678030303030303, + "grad_norm": 3.014401435852051, + "learning_rate": 4.333489242282507e-06, + "loss": 1.6104, + "step": 537 + }, + { + "epoch": 0.6792929292929293, + "grad_norm": 2.731356143951416, + "learning_rate": 4.331150608044903e-06, + "loss": 1.6177, + "step": 538 + }, + { + "epoch": 0.6805555555555556, + "grad_norm": 2.633310317993164, + "learning_rate": 4.328811973807297e-06, + "loss": 1.5004, + "step": 539 + }, + { + "epoch": 0.6818181818181818, + "grad_norm": 2.977476119995117, + "learning_rate": 4.3264733395696914e-06, + "loss": 1.6171, + "step": 540 + }, + { + "epoch": 0.6818181818181818, + "eval_loss": 1.7790141105651855, + "eval_runtime": 37.12, + "eval_samples_per_second": 26.94, + "eval_steps_per_second": 1.131, + "step": 540 + }, + { + "epoch": 0.6830808080808081, + "grad_norm": 2.769422769546509, + "learning_rate": 4.324134705332087e-06, + "loss": 1.5591, + "step": 541 + }, + { + "epoch": 0.6843434343434344, + "grad_norm": 3.088426351547241, + "learning_rate": 4.321796071094481e-06, + "loss": 1.5603, + "step": 542 + }, + { + "epoch": 0.6856060606060606, + "grad_norm": 2.6784212589263916, + "learning_rate": 4.3194574368568755e-06, + "loss": 1.6061, + "step": 543 + }, + { + "epoch": 0.6868686868686869, + "grad_norm": 3.209821939468384, + "learning_rate": 4.317118802619271e-06, + "loss": 1.585, + "step": 544 + }, + { + "epoch": 0.6881313131313131, + "grad_norm": 2.940791368484497, + "learning_rate": 4.314780168381665e-06, + "loss": 1.6092, + "step": 545 + }, + { + "epoch": 0.6893939393939394, + "grad_norm": 2.679884672164917, + "learning_rate": 4.3124415341440605e-06, + "loss": 1.6207, + "step": 546 + }, + { + "epoch": 0.6906565656565656, + "grad_norm": 2.853809118270874, + "learning_rate": 4.310102899906455e-06, + "loss": 1.5674, + "step": 547 + }, + { + "epoch": 0.6919191919191919, + "grad_norm": 2.693838119506836, + "learning_rate": 4.307764265668849e-06, + "loss": 1.5886, + "step": 548 + }, + { + "epoch": 0.6931818181818182, + "grad_norm": 2.871680736541748, + "learning_rate": 4.305425631431245e-06, + "loss": 1.6309, + "step": 549 + }, + { + "epoch": 0.6944444444444444, + "grad_norm": 2.9908292293548584, + "learning_rate": 4.30308699719364e-06, + "loss": 1.5577, + "step": 550 + }, + { + "epoch": 0.6944444444444444, + "eval_loss": 1.7790372371673584, + "eval_runtime": 37.114, + "eval_samples_per_second": 26.944, + "eval_steps_per_second": 1.132, + "step": 550 + }, + { + "epoch": 0.6957070707070707, + "grad_norm": 2.883481740951538, + "learning_rate": 4.300748362956034e-06, + "loss": 1.5749, + "step": 551 + }, + { + "epoch": 0.696969696969697, + "grad_norm": 2.7523458003997803, + "learning_rate": 4.298409728718429e-06, + "loss": 1.624, + "step": 552 + }, + { + "epoch": 0.6982323232323232, + "grad_norm": 2.69509220123291, + "learning_rate": 4.296071094480824e-06, + "loss": 1.5856, + "step": 553 + }, + { + "epoch": 0.6994949494949495, + "grad_norm": 2.6340222358703613, + "learning_rate": 4.293732460243218e-06, + "loss": 1.5857, + "step": 554 + }, + { + "epoch": 0.7007575757575758, + "grad_norm": 2.4998607635498047, + "learning_rate": 4.291393826005613e-06, + "loss": 1.6038, + "step": 555 + }, + { + "epoch": 0.702020202020202, + "grad_norm": 2.7913858890533447, + "learning_rate": 4.289055191768008e-06, + "loss": 1.6411, + "step": 556 + }, + { + "epoch": 0.7032828282828283, + "grad_norm": 2.7149479389190674, + "learning_rate": 4.2867165575304025e-06, + "loss": 1.592, + "step": 557 + }, + { + "epoch": 0.7045454545454546, + "grad_norm": 2.7187132835388184, + "learning_rate": 4.284377923292797e-06, + "loss": 1.5413, + "step": 558 + }, + { + "epoch": 0.7058080808080808, + "grad_norm": 2.8913674354553223, + "learning_rate": 4.282039289055192e-06, + "loss": 1.6105, + "step": 559 + }, + { + "epoch": 0.7070707070707071, + "grad_norm": 2.8470897674560547, + "learning_rate": 4.279700654817587e-06, + "loss": 1.6481, + "step": 560 + }, + { + "epoch": 0.7070707070707071, + "eval_loss": 1.7711386680603027, + "eval_runtime": 37.1243, + "eval_samples_per_second": 26.937, + "eval_steps_per_second": 1.131, + "step": 560 + }, + { + "epoch": 0.7083333333333334, + "grad_norm": 2.7525687217712402, + "learning_rate": 4.277362020579982e-06, + "loss": 1.5749, + "step": 561 + }, + { + "epoch": 0.7095959595959596, + "grad_norm": 2.6935207843780518, + "learning_rate": 4.275023386342376e-06, + "loss": 1.637, + "step": 562 + }, + { + "epoch": 0.7108585858585859, + "grad_norm": 2.6935207843780518, + "learning_rate": 4.275023386342376e-06, + "loss": 1.5903, + "step": 563 + }, + { + "epoch": 0.7121212121212122, + "grad_norm": 2.992992401123047, + "learning_rate": 4.272684752104771e-06, + "loss": 1.6654, + "step": 564 + }, + { + "epoch": 0.7133838383838383, + "grad_norm": 2.7314348220825195, + "learning_rate": 4.270346117867166e-06, + "loss": 1.6775, + "step": 565 + }, + { + "epoch": 0.7146464646464646, + "grad_norm": 2.6952013969421387, + "learning_rate": 4.26800748362956e-06, + "loss": 1.5936, + "step": 566 + }, + { + "epoch": 0.7159090909090909, + "grad_norm": 2.8244829177856445, + "learning_rate": 4.265668849391956e-06, + "loss": 1.6228, + "step": 567 + }, + { + "epoch": 0.7171717171717171, + "grad_norm": 2.9110143184661865, + "learning_rate": 4.26333021515435e-06, + "loss": 1.6016, + "step": 568 + }, + { + "epoch": 0.7184343434343434, + "grad_norm": 2.7953436374664307, + "learning_rate": 4.260991580916745e-06, + "loss": 1.611, + "step": 569 + }, + { + "epoch": 0.7196969696969697, + "grad_norm": 2.854447364807129, + "learning_rate": 4.25865294667914e-06, + "loss": 1.6365, + "step": 570 + }, + { + "epoch": 0.7196969696969697, + "eval_loss": 1.7711069583892822, + "eval_runtime": 37.1374, + "eval_samples_per_second": 26.927, + "eval_steps_per_second": 1.131, + "step": 570 + }, + { + "epoch": 0.7209595959595959, + "grad_norm": 2.7885239124298096, + "learning_rate": 4.256314312441534e-06, + "loss": 1.6106, + "step": 571 + }, + { + "epoch": 0.7222222222222222, + "grad_norm": 2.8012993335723877, + "learning_rate": 4.2539756782039295e-06, + "loss": 1.6081, + "step": 572 + }, + { + "epoch": 0.7234848484848485, + "grad_norm": 2.9223906993865967, + "learning_rate": 4.251637043966324e-06, + "loss": 1.5836, + "step": 573 + }, + { + "epoch": 0.7247474747474747, + "grad_norm": 2.885830879211426, + "learning_rate": 4.249298409728719e-06, + "loss": 1.5616, + "step": 574 + }, + { + "epoch": 0.726010101010101, + "grad_norm": 2.687372922897339, + "learning_rate": 4.2469597754911136e-06, + "loss": 1.5873, + "step": 575 + }, + { + "epoch": 0.7272727272727273, + "grad_norm": 3.0156168937683105, + "learning_rate": 4.244621141253508e-06, + "loss": 1.5763, + "step": 576 + }, + { + "epoch": 0.7285353535353535, + "grad_norm": 3.061842918395996, + "learning_rate": 4.242282507015903e-06, + "loss": 1.6148, + "step": 577 + }, + { + "epoch": 0.7297979797979798, + "grad_norm": 2.7467715740203857, + "learning_rate": 4.239943872778298e-06, + "loss": 1.577, + "step": 578 + }, + { + "epoch": 0.7310606060606061, + "grad_norm": 2.669501543045044, + "learning_rate": 4.237605238540692e-06, + "loss": 1.5653, + "step": 579 + }, + { + "epoch": 0.7323232323232324, + "grad_norm": 2.874349355697632, + "learning_rate": 4.235266604303087e-06, + "loss": 1.6523, + "step": 580 + }, + { + "epoch": 0.7323232323232324, + "eval_loss": 1.7680890560150146, + "eval_runtime": 37.2931, + "eval_samples_per_second": 26.815, + "eval_steps_per_second": 1.126, + "step": 580 + }, + { + "epoch": 0.7335858585858586, + "grad_norm": 2.7468504905700684, + "learning_rate": 4.232927970065482e-06, + "loss": 1.5935, + "step": 581 + }, + { + "epoch": 0.7348484848484849, + "grad_norm": 2.936697244644165, + "learning_rate": 4.230589335827877e-06, + "loss": 1.6335, + "step": 582 + }, + { + "epoch": 0.7361111111111112, + "grad_norm": 2.6878974437713623, + "learning_rate": 4.2282507015902715e-06, + "loss": 1.5699, + "step": 583 + }, + { + "epoch": 0.7373737373737373, + "grad_norm": 3.3578853607177734, + "learning_rate": 4.225912067352667e-06, + "loss": 1.5979, + "step": 584 + }, + { + "epoch": 0.7386363636363636, + "grad_norm": 2.8441176414489746, + "learning_rate": 4.223573433115061e-06, + "loss": 1.5736, + "step": 585 + }, + { + "epoch": 0.73989898989899, + "grad_norm": 2.7186129093170166, + "learning_rate": 4.221234798877456e-06, + "loss": 1.636, + "step": 586 + }, + { + "epoch": 0.7411616161616161, + "grad_norm": 2.940002918243408, + "learning_rate": 4.218896164639851e-06, + "loss": 1.5721, + "step": 587 + }, + { + "epoch": 0.7424242424242424, + "grad_norm": 2.676284074783325, + "learning_rate": 4.216557530402245e-06, + "loss": 1.5972, + "step": 588 + }, + { + "epoch": 0.7436868686868687, + "grad_norm": 2.9176571369171143, + "learning_rate": 4.2142188961646405e-06, + "loss": 1.6045, + "step": 589 + }, + { + "epoch": 0.7449494949494949, + "grad_norm": 2.7716288566589355, + "learning_rate": 4.211880261927035e-06, + "loss": 1.5988, + "step": 590 + }, + { + "epoch": 0.7449494949494949, + "eval_loss": 1.775631070137024, + "eval_runtime": 37.4113, + "eval_samples_per_second": 26.73, + "eval_steps_per_second": 1.123, + "step": 590 + }, + { + "epoch": 0.7462121212121212, + "grad_norm": 2.7201945781707764, + "learning_rate": 4.209541627689429e-06, + "loss": 1.5767, + "step": 591 + }, + { + "epoch": 0.7474747474747475, + "grad_norm": 2.838362455368042, + "learning_rate": 4.207202993451825e-06, + "loss": 1.5793, + "step": 592 + }, + { + "epoch": 0.7487373737373737, + "grad_norm": 2.6511716842651367, + "learning_rate": 4.204864359214219e-06, + "loss": 1.5776, + "step": 593 + }, + { + "epoch": 0.75, + "grad_norm": 2.7062957286834717, + "learning_rate": 4.2025257249766135e-06, + "loss": 1.5593, + "step": 594 + }, + { + "epoch": 0.7512626262626263, + "grad_norm": 2.808570384979248, + "learning_rate": 4.200187090739009e-06, + "loss": 1.6249, + "step": 595 + }, + { + "epoch": 0.7525252525252525, + "grad_norm": 2.600278377532959, + "learning_rate": 4.197848456501403e-06, + "loss": 1.6171, + "step": 596 + }, + { + "epoch": 0.7537878787878788, + "grad_norm": 2.9669382572174072, + "learning_rate": 4.195509822263798e-06, + "loss": 1.608, + "step": 597 + }, + { + "epoch": 0.7550505050505051, + "grad_norm": 2.6576340198516846, + "learning_rate": 4.193171188026194e-06, + "loss": 1.5986, + "step": 598 + }, + { + "epoch": 0.7563131313131313, + "grad_norm": 2.787060022354126, + "learning_rate": 4.190832553788588e-06, + "loss": 1.548, + "step": 599 + }, + { + "epoch": 0.7575757575757576, + "grad_norm": 2.854971408843994, + "learning_rate": 4.1884939195509825e-06, + "loss": 1.6421, + "step": 600 + }, + { + "epoch": 0.7575757575757576, + "eval_loss": 1.7721046209335327, + "eval_runtime": 37.4471, + "eval_samples_per_second": 26.704, + "eval_steps_per_second": 1.122, + "step": 600 + }, + { + "epoch": 0.7588383838383839, + "grad_norm": 2.7417967319488525, + "learning_rate": 4.186155285313378e-06, + "loss": 1.6399, + "step": 601 + }, + { + "epoch": 0.76010101010101, + "grad_norm": 3.0036182403564453, + "learning_rate": 4.183816651075772e-06, + "loss": 1.5953, + "step": 602 + }, + { + "epoch": 0.7613636363636364, + "grad_norm": 2.502861261367798, + "learning_rate": 4.181478016838167e-06, + "loss": 1.5284, + "step": 603 + }, + { + "epoch": 0.7626262626262627, + "grad_norm": 2.7262094020843506, + "learning_rate": 4.179139382600562e-06, + "loss": 1.619, + "step": 604 + }, + { + "epoch": 0.7638888888888888, + "grad_norm": 2.8748724460601807, + "learning_rate": 4.176800748362956e-06, + "loss": 1.5625, + "step": 605 + }, + { + "epoch": 0.7651515151515151, + "grad_norm": 2.7272109985351562, + "learning_rate": 4.174462114125351e-06, + "loss": 1.5935, + "step": 606 + }, + { + "epoch": 0.7664141414141414, + "grad_norm": 2.7930479049682617, + "learning_rate": 4.172123479887746e-06, + "loss": 1.61, + "step": 607 + }, + { + "epoch": 0.7676767676767676, + "grad_norm": 2.952056407928467, + "learning_rate": 4.16978484565014e-06, + "loss": 1.6369, + "step": 608 + }, + { + "epoch": 0.7689393939393939, + "grad_norm": 2.979564905166626, + "learning_rate": 4.167446211412535e-06, + "loss": 1.5748, + "step": 609 + }, + { + "epoch": 0.7702020202020202, + "grad_norm": 2.7945773601531982, + "learning_rate": 4.16510757717493e-06, + "loss": 1.6096, + "step": 610 + }, + { + "epoch": 0.7702020202020202, + "eval_loss": 1.7568180561065674, + "eval_runtime": 36.969, + "eval_samples_per_second": 27.05, + "eval_steps_per_second": 1.136, + "step": 610 + }, + { + "epoch": 0.7714646464646465, + "grad_norm": 2.9217135906219482, + "learning_rate": 4.1627689429373245e-06, + "loss": 1.5902, + "step": 611 + }, + { + "epoch": 0.7727272727272727, + "grad_norm": 2.8184452056884766, + "learning_rate": 4.16043030869972e-06, + "loss": 1.5775, + "step": 612 + }, + { + "epoch": 0.773989898989899, + "grad_norm": 2.7121212482452393, + "learning_rate": 4.158091674462115e-06, + "loss": 1.5778, + "step": 613 + }, + { + "epoch": 0.7752525252525253, + "grad_norm": 2.7600955963134766, + "learning_rate": 4.1557530402245095e-06, + "loss": 1.6145, + "step": 614 + }, + { + "epoch": 0.7765151515151515, + "grad_norm": 2.6388845443725586, + "learning_rate": 4.153414405986904e-06, + "loss": 1.6475, + "step": 615 + }, + { + "epoch": 0.7777777777777778, + "grad_norm": 2.684379816055298, + "learning_rate": 4.151075771749299e-06, + "loss": 1.6187, + "step": 616 + }, + { + "epoch": 0.7790404040404041, + "grad_norm": 2.6907055377960205, + "learning_rate": 4.1487371375116936e-06, + "loss": 1.6108, + "step": 617 + }, + { + "epoch": 0.7803030303030303, + "grad_norm": 2.8458995819091797, + "learning_rate": 4.146398503274088e-06, + "loss": 1.5596, + "step": 618 + }, + { + "epoch": 0.7815656565656566, + "grad_norm": 2.68843674659729, + "learning_rate": 4.144059869036483e-06, + "loss": 1.5912, + "step": 619 + }, + { + "epoch": 0.7828282828282829, + "grad_norm": 2.7620413303375244, + "learning_rate": 4.141721234798878e-06, + "loss": 1.627, + "step": 620 + }, + { + "epoch": 0.7828282828282829, + "eval_loss": 1.757076621055603, + "eval_runtime": 37.0489, + "eval_samples_per_second": 26.991, + "eval_steps_per_second": 1.134, + "step": 620 + }, + { + "epoch": 0.7840909090909091, + "grad_norm": 2.7084949016571045, + "learning_rate": 4.139382600561272e-06, + "loss": 1.5249, + "step": 621 + }, + { + "epoch": 0.7853535353535354, + "grad_norm": 3.0410406589508057, + "learning_rate": 4.137043966323667e-06, + "loss": 1.6458, + "step": 622 + }, + { + "epoch": 0.7866161616161617, + "grad_norm": 2.8445630073547363, + "learning_rate": 4.134705332086062e-06, + "loss": 1.6404, + "step": 623 + }, + { + "epoch": 0.7878787878787878, + "grad_norm": 3.0764353275299072, + "learning_rate": 4.132366697848457e-06, + "loss": 1.5041, + "step": 624 + }, + { + "epoch": 0.7891414141414141, + "grad_norm": 2.626544237136841, + "learning_rate": 4.1300280636108515e-06, + "loss": 1.6145, + "step": 625 + }, + { + "epoch": 0.7904040404040404, + "grad_norm": 2.9735796451568604, + "learning_rate": 4.127689429373246e-06, + "loss": 1.5672, + "step": 626 + }, + { + "epoch": 0.7916666666666666, + "grad_norm": 2.7798562049865723, + "learning_rate": 4.125350795135641e-06, + "loss": 1.6272, + "step": 627 + }, + { + "epoch": 0.7929292929292929, + "grad_norm": 2.7438931465148926, + "learning_rate": 4.123012160898036e-06, + "loss": 1.5318, + "step": 628 + }, + { + "epoch": 0.7941919191919192, + "grad_norm": 3.103142499923706, + "learning_rate": 4.120673526660431e-06, + "loss": 1.6, + "step": 629 + }, + { + "epoch": 0.7954545454545454, + "grad_norm": 2.9056239128112793, + "learning_rate": 4.118334892422825e-06, + "loss": 1.6581, + "step": 630 + }, + { + "epoch": 0.7954545454545454, + "eval_loss": 1.7552870512008667, + "eval_runtime": 37.1223, + "eval_samples_per_second": 26.938, + "eval_steps_per_second": 1.131, + "step": 630 + }, + { + "epoch": 0.7967171717171717, + "grad_norm": 2.7320070266723633, + "learning_rate": 4.1159962581852205e-06, + "loss": 1.5808, + "step": 631 + }, + { + "epoch": 0.797979797979798, + "grad_norm": 2.9518637657165527, + "learning_rate": 4.113657623947615e-06, + "loss": 1.6034, + "step": 632 + }, + { + "epoch": 0.7992424242424242, + "grad_norm": 2.7135164737701416, + "learning_rate": 4.111318989710009e-06, + "loss": 1.5684, + "step": 633 + }, + { + "epoch": 0.8005050505050505, + "grad_norm": 2.948230266571045, + "learning_rate": 4.108980355472405e-06, + "loss": 1.6378, + "step": 634 + }, + { + "epoch": 0.8017676767676768, + "grad_norm": 2.816206216812134, + "learning_rate": 4.106641721234799e-06, + "loss": 1.6118, + "step": 635 + }, + { + "epoch": 0.803030303030303, + "grad_norm": 2.8032639026641846, + "learning_rate": 4.104303086997194e-06, + "loss": 1.5627, + "step": 636 + }, + { + "epoch": 0.8042929292929293, + "grad_norm": 3.032348871231079, + "learning_rate": 4.101964452759589e-06, + "loss": 1.6192, + "step": 637 + }, + { + "epoch": 0.8055555555555556, + "grad_norm": 2.9512710571289062, + "learning_rate": 4.099625818521983e-06, + "loss": 1.5655, + "step": 638 + }, + { + "epoch": 0.8068181818181818, + "grad_norm": 2.738973379135132, + "learning_rate": 4.097287184284378e-06, + "loss": 1.5686, + "step": 639 + }, + { + "epoch": 0.8080808080808081, + "grad_norm": 2.6510915756225586, + "learning_rate": 4.094948550046773e-06, + "loss": 1.5889, + "step": 640 + }, + { + "epoch": 0.8080808080808081, + "eval_loss": 1.762879490852356, + "eval_runtime": 37.1457, + "eval_samples_per_second": 26.921, + "eval_steps_per_second": 1.131, + "step": 640 + }, + { + "epoch": 0.8093434343434344, + "grad_norm": 2.6933884620666504, + "learning_rate": 4.092609915809167e-06, + "loss": 1.5436, + "step": 641 + }, + { + "epoch": 0.8106060606060606, + "grad_norm": 2.703784227371216, + "learning_rate": 4.0902712815715625e-06, + "loss": 1.5869, + "step": 642 + }, + { + "epoch": 0.8118686868686869, + "grad_norm": 2.7742321491241455, + "learning_rate": 4.087932647333958e-06, + "loss": 1.5779, + "step": 643 + }, + { + "epoch": 0.8131313131313131, + "grad_norm": 2.836832284927368, + "learning_rate": 4.085594013096352e-06, + "loss": 1.5511, + "step": 644 + }, + { + "epoch": 0.8143939393939394, + "grad_norm": 2.9212028980255127, + "learning_rate": 4.083255378858747e-06, + "loss": 1.5604, + "step": 645 + }, + { + "epoch": 0.8156565656565656, + "grad_norm": 2.8608527183532715, + "learning_rate": 4.080916744621142e-06, + "loss": 1.5975, + "step": 646 + }, + { + "epoch": 0.8169191919191919, + "grad_norm": 2.8246703147888184, + "learning_rate": 4.078578110383536e-06, + "loss": 1.5962, + "step": 647 + }, + { + "epoch": 0.8181818181818182, + "grad_norm": 3.0648651123046875, + "learning_rate": 4.076239476145932e-06, + "loss": 1.6267, + "step": 648 + }, + { + "epoch": 0.8194444444444444, + "grad_norm": 2.849240303039551, + "learning_rate": 4.073900841908326e-06, + "loss": 1.6271, + "step": 649 + }, + { + "epoch": 0.8207070707070707, + "grad_norm": 2.8817555904388428, + "learning_rate": 4.07156220767072e-06, + "loss": 1.5916, + "step": 650 + }, + { + "epoch": 0.8207070707070707, + "eval_loss": 1.7502237558364868, + "eval_runtime": 37.1497, + "eval_samples_per_second": 26.918, + "eval_steps_per_second": 1.131, + "step": 650 + }, + { + "epoch": 0.821969696969697, + "grad_norm": 2.5747952461242676, + "learning_rate": 4.069223573433116e-06, + "loss": 1.5661, + "step": 651 + }, + { + "epoch": 0.8232323232323232, + "grad_norm": 2.7721974849700928, + "learning_rate": 4.06688493919551e-06, + "loss": 1.5903, + "step": 652 + }, + { + "epoch": 0.8244949494949495, + "grad_norm": 2.9606690406799316, + "learning_rate": 4.0645463049579045e-06, + "loss": 1.5431, + "step": 653 + }, + { + "epoch": 0.8257575757575758, + "grad_norm": 2.941169261932373, + "learning_rate": 4.0622076707203e-06, + "loss": 1.5711, + "step": 654 + }, + { + "epoch": 0.827020202020202, + "grad_norm": 2.8408572673797607, + "learning_rate": 4.059869036482694e-06, + "loss": 1.5975, + "step": 655 + }, + { + "epoch": 0.8282828282828283, + "grad_norm": 2.619875431060791, + "learning_rate": 4.057530402245089e-06, + "loss": 1.5194, + "step": 656 + }, + { + "epoch": 0.8295454545454546, + "grad_norm": 2.910982847213745, + "learning_rate": 4.055191768007484e-06, + "loss": 1.6318, + "step": 657 + }, + { + "epoch": 0.8308080808080808, + "grad_norm": 2.747422456741333, + "learning_rate": 4.052853133769879e-06, + "loss": 1.6205, + "step": 658 + }, + { + "epoch": 0.8320707070707071, + "grad_norm": 2.6889796257019043, + "learning_rate": 4.050514499532274e-06, + "loss": 1.5537, + "step": 659 + }, + { + "epoch": 0.8333333333333334, + "grad_norm": 2.9849202632904053, + "learning_rate": 4.048175865294669e-06, + "loss": 1.5438, + "step": 660 + }, + { + "epoch": 0.8333333333333334, + "eval_loss": 1.7484922409057617, + "eval_runtime": 37.161, + "eval_samples_per_second": 26.91, + "eval_steps_per_second": 1.13, + "step": 660 + }, + { + "epoch": 0.8345959595959596, + "grad_norm": 2.816930055618286, + "learning_rate": 4.045837231057063e-06, + "loss": 1.555, + "step": 661 + }, + { + "epoch": 0.8358585858585859, + "grad_norm": 2.990126371383667, + "learning_rate": 4.043498596819458e-06, + "loss": 1.6263, + "step": 662 + }, + { + "epoch": 0.8371212121212122, + "grad_norm": 2.898263692855835, + "learning_rate": 4.041159962581853e-06, + "loss": 1.5522, + "step": 663 + }, + { + "epoch": 0.8383838383838383, + "grad_norm": 2.7824151515960693, + "learning_rate": 4.038821328344247e-06, + "loss": 1.5366, + "step": 664 + }, + { + "epoch": 0.8396464646464646, + "grad_norm": 2.9943954944610596, + "learning_rate": 4.036482694106642e-06, + "loss": 1.6214, + "step": 665 + }, + { + "epoch": 0.8409090909090909, + "grad_norm": 2.628308057785034, + "learning_rate": 4.034144059869037e-06, + "loss": 1.5797, + "step": 666 + }, + { + "epoch": 0.8421717171717171, + "grad_norm": 2.7565176486968994, + "learning_rate": 4.0318054256314315e-06, + "loss": 1.5567, + "step": 667 + }, + { + "epoch": 0.8434343434343434, + "grad_norm": 2.688354969024658, + "learning_rate": 4.029466791393826e-06, + "loss": 1.5516, + "step": 668 + }, + { + "epoch": 0.8446969696969697, + "grad_norm": 2.899620771408081, + "learning_rate": 4.027128157156221e-06, + "loss": 1.6014, + "step": 669 + }, + { + "epoch": 0.8459595959595959, + "grad_norm": 2.7894718647003174, + "learning_rate": 4.024789522918616e-06, + "loss": 1.5688, + "step": 670 + }, + { + "epoch": 0.8459595959595959, + "eval_loss": 1.7486555576324463, + "eval_runtime": 37.1824, + "eval_samples_per_second": 26.894, + "eval_steps_per_second": 1.13, + "step": 670 + }, + { + "epoch": 0.8472222222222222, + "grad_norm": 2.7441444396972656, + "learning_rate": 4.02245088868101e-06, + "loss": 1.5949, + "step": 671 + }, + { + "epoch": 0.8484848484848485, + "grad_norm": 2.7088451385498047, + "learning_rate": 4.020112254443405e-06, + "loss": 1.5878, + "step": 672 + }, + { + "epoch": 0.8497474747474747, + "grad_norm": 2.756920576095581, + "learning_rate": 4.0177736202058005e-06, + "loss": 1.6356, + "step": 673 + }, + { + "epoch": 0.851010101010101, + "grad_norm": 2.5605697631835938, + "learning_rate": 4.015434985968195e-06, + "loss": 1.5586, + "step": 674 + }, + { + "epoch": 0.8522727272727273, + "grad_norm": 2.946573257446289, + "learning_rate": 4.01309635173059e-06, + "loss": 1.5537, + "step": 675 + }, + { + "epoch": 0.8535353535353535, + "grad_norm": 2.65500545501709, + "learning_rate": 4.010757717492985e-06, + "loss": 1.5875, + "step": 676 + }, + { + "epoch": 0.8547979797979798, + "grad_norm": 2.6632118225097656, + "learning_rate": 4.008419083255379e-06, + "loss": 1.5329, + "step": 677 + }, + { + "epoch": 0.8560606060606061, + "grad_norm": 2.8909032344818115, + "learning_rate": 4.006080449017774e-06, + "loss": 1.6396, + "step": 678 + }, + { + "epoch": 0.8573232323232324, + "grad_norm": 2.8780951499938965, + "learning_rate": 4.003741814780169e-06, + "loss": 1.6189, + "step": 679 + }, + { + "epoch": 0.8585858585858586, + "grad_norm": 2.7757010459899902, + "learning_rate": 4.001403180542563e-06, + "loss": 1.5968, + "step": 680 + }, + { + "epoch": 0.8585858585858586, + "eval_loss": 1.7423925399780273, + "eval_runtime": 37.1537, + "eval_samples_per_second": 26.915, + "eval_steps_per_second": 1.13, + "step": 680 + }, + { + "epoch": 0.8598484848484849, + "grad_norm": 2.6610794067382812, + "learning_rate": 3.9990645463049584e-06, + "loss": 1.5258, + "step": 681 + }, + { + "epoch": 0.8611111111111112, + "grad_norm": 2.9292140007019043, + "learning_rate": 3.996725912067353e-06, + "loss": 1.5669, + "step": 682 + }, + { + "epoch": 0.8623737373737373, + "grad_norm": 2.7885360717773438, + "learning_rate": 3.994387277829747e-06, + "loss": 1.5737, + "step": 683 + }, + { + "epoch": 0.8636363636363636, + "grad_norm": 2.7511544227600098, + "learning_rate": 3.9920486435921425e-06, + "loss": 1.5715, + "step": 684 + }, + { + "epoch": 0.86489898989899, + "grad_norm": 2.8561315536499023, + "learning_rate": 3.989710009354537e-06, + "loss": 1.5965, + "step": 685 + }, + { + "epoch": 0.8661616161616161, + "grad_norm": 2.9127237796783447, + "learning_rate": 3.987371375116932e-06, + "loss": 1.5407, + "step": 686 + }, + { + "epoch": 0.8674242424242424, + "grad_norm": 2.735577344894409, + "learning_rate": 3.985032740879327e-06, + "loss": 1.5449, + "step": 687 + }, + { + "epoch": 0.8686868686868687, + "grad_norm": 2.6626689434051514, + "learning_rate": 3.982694106641721e-06, + "loss": 1.5173, + "step": 688 + }, + { + "epoch": 0.8699494949494949, + "grad_norm": 2.901592969894409, + "learning_rate": 3.980355472404116e-06, + "loss": 1.5757, + "step": 689 + }, + { + "epoch": 0.8712121212121212, + "grad_norm": 2.8801231384277344, + "learning_rate": 3.978016838166512e-06, + "loss": 1.5633, + "step": 690 + }, + { + "epoch": 0.8712121212121212, + "eval_loss": 1.7453432083129883, + "eval_runtime": 37.4368, + "eval_samples_per_second": 26.712, + "eval_steps_per_second": 1.122, + "step": 690 + }, + { + "epoch": 0.8724747474747475, + "grad_norm": 2.7463626861572266, + "learning_rate": 3.975678203928906e-06, + "loss": 1.5786, + "step": 691 + }, + { + "epoch": 0.8737373737373737, + "grad_norm": 2.659820318222046, + "learning_rate": 3.9733395696913004e-06, + "loss": 1.5923, + "step": 692 + }, + { + "epoch": 0.875, + "grad_norm": 2.759030342102051, + "learning_rate": 3.971000935453696e-06, + "loss": 1.6217, + "step": 693 + }, + { + "epoch": 0.8762626262626263, + "grad_norm": 2.689967632293701, + "learning_rate": 3.96866230121609e-06, + "loss": 1.5533, + "step": 694 + }, + { + "epoch": 0.8775252525252525, + "grad_norm": 3.1230740547180176, + "learning_rate": 3.9663236669784845e-06, + "loss": 1.6538, + "step": 695 + }, + { + "epoch": 0.8787878787878788, + "grad_norm": 2.657022476196289, + "learning_rate": 3.96398503274088e-06, + "loss": 1.5432, + "step": 696 + }, + { + "epoch": 0.8800505050505051, + "grad_norm": 2.919379711151123, + "learning_rate": 3.961646398503274e-06, + "loss": 1.5867, + "step": 697 + }, + { + "epoch": 0.8813131313131313, + "grad_norm": 2.607135057449341, + "learning_rate": 3.9593077642656695e-06, + "loss": 1.5079, + "step": 698 + }, + { + "epoch": 0.8825757575757576, + "grad_norm": 2.9978976249694824, + "learning_rate": 3.956969130028064e-06, + "loss": 1.5905, + "step": 699 + }, + { + "epoch": 0.8838383838383839, + "grad_norm": 2.8117148876190186, + "learning_rate": 3.954630495790458e-06, + "loss": 1.5588, + "step": 700 + }, + { + "epoch": 0.8838383838383839, + "eval_loss": 1.7376645803451538, + "eval_runtime": 37.5045, + "eval_samples_per_second": 26.663, + "eval_steps_per_second": 1.12, + "step": 700 + }, + { + "epoch": 0.88510101010101, + "grad_norm": 2.7135095596313477, + "learning_rate": 3.952291861552854e-06, + "loss": 1.6013, + "step": 701 + }, + { + "epoch": 0.8863636363636364, + "grad_norm": 2.812061071395874, + "learning_rate": 3.949953227315248e-06, + "loss": 1.5518, + "step": 702 + }, + { + "epoch": 0.8876262626262627, + "grad_norm": 3.0532422065734863, + "learning_rate": 3.9476145930776424e-06, + "loss": 1.5579, + "step": 703 + }, + { + "epoch": 0.8888888888888888, + "grad_norm": 2.8306567668914795, + "learning_rate": 3.945275958840038e-06, + "loss": 1.5478, + "step": 704 + }, + { + "epoch": 0.8901515151515151, + "grad_norm": 2.737921953201294, + "learning_rate": 3.942937324602433e-06, + "loss": 1.4958, + "step": 705 + }, + { + "epoch": 0.8914141414141414, + "grad_norm": 2.902211904525757, + "learning_rate": 3.940598690364827e-06, + "loss": 1.5715, + "step": 706 + }, + { + "epoch": 0.8926767676767676, + "grad_norm": 2.725818395614624, + "learning_rate": 3.938260056127222e-06, + "loss": 1.5502, + "step": 707 + }, + { + "epoch": 0.8939393939393939, + "grad_norm": 2.9530506134033203, + "learning_rate": 3.935921421889617e-06, + "loss": 1.564, + "step": 708 + }, + { + "epoch": 0.8952020202020202, + "grad_norm": 2.798231363296509, + "learning_rate": 3.9335827876520115e-06, + "loss": 1.5445, + "step": 709 + }, + { + "epoch": 0.8964646464646465, + "grad_norm": 3.0233471393585205, + "learning_rate": 3.931244153414407e-06, + "loss": 1.6193, + "step": 710 + }, + { + "epoch": 0.8964646464646465, + "eval_loss": 1.742077350616455, + "eval_runtime": 37.0255, + "eval_samples_per_second": 27.008, + "eval_steps_per_second": 1.134, + "step": 710 + }, + { + "epoch": 0.8977272727272727, + "grad_norm": 2.84309458732605, + "learning_rate": 3.928905519176801e-06, + "loss": 1.5379, + "step": 711 + }, + { + "epoch": 0.898989898989899, + "grad_norm": 2.7412185668945312, + "learning_rate": 3.926566884939196e-06, + "loss": 1.5837, + "step": 712 + }, + { + "epoch": 0.9002525252525253, + "grad_norm": 3.1094822883605957, + "learning_rate": 3.924228250701591e-06, + "loss": 1.5621, + "step": 713 + }, + { + "epoch": 0.9015151515151515, + "grad_norm": 2.776059627532959, + "learning_rate": 3.921889616463985e-06, + "loss": 1.5714, + "step": 714 + }, + { + "epoch": 0.9027777777777778, + "grad_norm": 2.7650489807128906, + "learning_rate": 3.91955098222638e-06, + "loss": 1.5418, + "step": 715 + }, + { + "epoch": 0.9040404040404041, + "grad_norm": 2.8586974143981934, + "learning_rate": 3.917212347988775e-06, + "loss": 1.5891, + "step": 716 + }, + { + "epoch": 0.9053030303030303, + "grad_norm": 2.74332594871521, + "learning_rate": 3.914873713751169e-06, + "loss": 1.5492, + "step": 717 + }, + { + "epoch": 0.9065656565656566, + "grad_norm": 2.931682586669922, + "learning_rate": 3.912535079513564e-06, + "loss": 1.54, + "step": 718 + }, + { + "epoch": 0.9078282828282829, + "grad_norm": 2.678812026977539, + "learning_rate": 3.910196445275959e-06, + "loss": 1.5677, + "step": 719 + }, + { + "epoch": 0.9090909090909091, + "grad_norm": 2.7282893657684326, + "learning_rate": 3.907857811038354e-06, + "loss": 1.5413, + "step": 720 + }, + { + "epoch": 0.9090909090909091, + "eval_loss": 1.7409753799438477, + "eval_runtime": 37.1883, + "eval_samples_per_second": 26.89, + "eval_steps_per_second": 1.129, + "step": 720 + }, + { + "epoch": 0.9103535353535354, + "grad_norm": 3.073997735977173, + "learning_rate": 3.905519176800749e-06, + "loss": 1.5734, + "step": 721 + }, + { + "epoch": 0.9116161616161617, + "grad_norm": 2.6249935626983643, + "learning_rate": 3.903180542563144e-06, + "loss": 1.5063, + "step": 722 + }, + { + "epoch": 0.9128787878787878, + "grad_norm": 2.7758615016937256, + "learning_rate": 3.9008419083255384e-06, + "loss": 1.5112, + "step": 723 + }, + { + "epoch": 0.9141414141414141, + "grad_norm": 2.8954596519470215, + "learning_rate": 3.898503274087933e-06, + "loss": 1.5508, + "step": 724 + }, + { + "epoch": 0.9154040404040404, + "grad_norm": 2.79073166847229, + "learning_rate": 3.896164639850328e-06, + "loss": 1.5773, + "step": 725 + }, + { + "epoch": 0.9166666666666666, + "grad_norm": 2.687027931213379, + "learning_rate": 3.8938260056127226e-06, + "loss": 1.5243, + "step": 726 + }, + { + "epoch": 0.9179292929292929, + "grad_norm": 2.5704851150512695, + "learning_rate": 3.891487371375117e-06, + "loss": 1.5075, + "step": 727 + }, + { + "epoch": 0.9191919191919192, + "grad_norm": 2.8684048652648926, + "learning_rate": 3.889148737137512e-06, + "loss": 1.6471, + "step": 728 + }, + { + "epoch": 0.9204545454545454, + "grad_norm": 2.8995590209960938, + "learning_rate": 3.886810102899907e-06, + "loss": 1.5429, + "step": 729 + }, + { + "epoch": 0.9217171717171717, + "grad_norm": 2.7575254440307617, + "learning_rate": 3.884471468662301e-06, + "loss": 1.5241, + "step": 730 + }, + { + "epoch": 0.9217171717171717, + "eval_loss": 1.7294232845306396, + "eval_runtime": 37.1941, + "eval_samples_per_second": 26.886, + "eval_steps_per_second": 1.129, + "step": 730 + }, + { + "epoch": 0.922979797979798, + "grad_norm": 2.7057321071624756, + "learning_rate": 3.882132834424696e-06, + "loss": 1.5498, + "step": 731 + }, + { + "epoch": 0.9242424242424242, + "grad_norm": 2.7679436206817627, + "learning_rate": 3.879794200187091e-06, + "loss": 1.563, + "step": 732 + }, + { + "epoch": 0.9255050505050505, + "grad_norm": 2.7258687019348145, + "learning_rate": 3.877455565949485e-06, + "loss": 1.5016, + "step": 733 + }, + { + "epoch": 0.9267676767676768, + "grad_norm": 2.8474221229553223, + "learning_rate": 3.8751169317118804e-06, + "loss": 1.5318, + "step": 734 + }, + { + "epoch": 0.928030303030303, + "grad_norm": 2.7718396186828613, + "learning_rate": 3.872778297474276e-06, + "loss": 1.5919, + "step": 735 + }, + { + "epoch": 0.9292929292929293, + "grad_norm": 2.7785022258758545, + "learning_rate": 3.87043966323667e-06, + "loss": 1.5399, + "step": 736 + }, + { + "epoch": 0.9305555555555556, + "grad_norm": 2.742476463317871, + "learning_rate": 3.868101028999065e-06, + "loss": 1.5428, + "step": 737 + }, + { + "epoch": 0.9318181818181818, + "grad_norm": 2.7050726413726807, + "learning_rate": 3.86576239476146e-06, + "loss": 1.5604, + "step": 738 + }, + { + "epoch": 0.9330808080808081, + "grad_norm": 2.6755242347717285, + "learning_rate": 3.863423760523854e-06, + "loss": 1.5409, + "step": 739 + }, + { + "epoch": 0.9343434343434344, + "grad_norm": 2.6947829723358154, + "learning_rate": 3.8610851262862495e-06, + "loss": 1.5537, + "step": 740 + }, + { + "epoch": 0.9343434343434344, + "eval_loss": 1.7396037578582764, + "eval_runtime": 37.1912, + "eval_samples_per_second": 26.888, + "eval_steps_per_second": 1.129, + "step": 740 + }, + { + "epoch": 0.9356060606060606, + "grad_norm": 2.729816436767578, + "learning_rate": 3.858746492048644e-06, + "loss": 1.5581, + "step": 741 + }, + { + "epoch": 0.9368686868686869, + "grad_norm": 2.6951887607574463, + "learning_rate": 3.856407857811038e-06, + "loss": 1.5929, + "step": 742 + }, + { + "epoch": 0.9381313131313131, + "grad_norm": 2.6601064205169678, + "learning_rate": 3.854069223573434e-06, + "loss": 1.4881, + "step": 743 + }, + { + "epoch": 0.9393939393939394, + "grad_norm": 2.623898983001709, + "learning_rate": 3.851730589335828e-06, + "loss": 1.5499, + "step": 744 + }, + { + "epoch": 0.9406565656565656, + "grad_norm": 2.703871250152588, + "learning_rate": 3.8493919550982224e-06, + "loss": 1.5585, + "step": 745 + }, + { + "epoch": 0.9419191919191919, + "grad_norm": 2.674893379211426, + "learning_rate": 3.847053320860618e-06, + "loss": 1.5586, + "step": 746 + }, + { + "epoch": 0.9431818181818182, + "grad_norm": 2.8353004455566406, + "learning_rate": 3.844714686623012e-06, + "loss": 1.5937, + "step": 747 + }, + { + "epoch": 0.9444444444444444, + "grad_norm": 2.68009614944458, + "learning_rate": 3.842376052385407e-06, + "loss": 1.5192, + "step": 748 + }, + { + "epoch": 0.9457070707070707, + "grad_norm": 2.6847407817840576, + "learning_rate": 3.840037418147802e-06, + "loss": 1.512, + "step": 749 + }, + { + "epoch": 0.946969696969697, + "grad_norm": 2.624290943145752, + "learning_rate": 3.837698783910197e-06, + "loss": 1.5348, + "step": 750 + }, + { + "epoch": 0.946969696969697, + "eval_loss": 1.738332748413086, + "eval_runtime": 37.1782, + "eval_samples_per_second": 26.898, + "eval_steps_per_second": 1.13, + "step": 750 + }, + { + "epoch": 0.9482323232323232, + "grad_norm": 2.7467777729034424, + "learning_rate": 3.8353601496725915e-06, + "loss": 1.5603, + "step": 751 + }, + { + "epoch": 0.9494949494949495, + "grad_norm": 2.851987600326538, + "learning_rate": 3.833021515434987e-06, + "loss": 1.5974, + "step": 752 + }, + { + "epoch": 0.9507575757575758, + "grad_norm": 2.6642227172851562, + "learning_rate": 3.830682881197381e-06, + "loss": 1.5138, + "step": 753 + }, + { + "epoch": 0.952020202020202, + "grad_norm": 2.7351491451263428, + "learning_rate": 3.828344246959776e-06, + "loss": 1.5656, + "step": 754 + }, + { + "epoch": 0.9532828282828283, + "grad_norm": 2.593748092651367, + "learning_rate": 3.826005612722171e-06, + "loss": 1.5395, + "step": 755 + }, + { + "epoch": 0.9545454545454546, + "grad_norm": 2.639180898666382, + "learning_rate": 3.823666978484565e-06, + "loss": 1.5349, + "step": 756 + }, + { + "epoch": 0.9558080808080808, + "grad_norm": 2.805148124694824, + "learning_rate": 3.82132834424696e-06, + "loss": 1.5805, + "step": 757 + }, + { + "epoch": 0.9570707070707071, + "grad_norm": 2.5619425773620605, + "learning_rate": 3.818989710009355e-06, + "loss": 1.5389, + "step": 758 + }, + { + "epoch": 0.9583333333333334, + "grad_norm": 2.7224011421203613, + "learning_rate": 3.816651075771749e-06, + "loss": 1.5316, + "step": 759 + }, + { + "epoch": 0.9595959595959596, + "grad_norm": 2.4667656421661377, + "learning_rate": 3.8143124415341442e-06, + "loss": 1.4942, + "step": 760 + }, + { + "epoch": 0.9595959595959596, + "eval_loss": 1.7205606698989868, + "eval_runtime": 37.1654, + "eval_samples_per_second": 26.907, + "eval_steps_per_second": 1.13, + "step": 760 + }, + { + "epoch": 0.9608585858585859, + "grad_norm": 2.456643581390381, + "learning_rate": 3.811973807296539e-06, + "loss": 1.538, + "step": 761 + }, + { + "epoch": 0.9621212121212122, + "grad_norm": 2.696720838546753, + "learning_rate": 3.8096351730589335e-06, + "loss": 1.5462, + "step": 762 + }, + { + "epoch": 0.9633838383838383, + "grad_norm": 2.7314131259918213, + "learning_rate": 3.8072965388213284e-06, + "loss": 1.5931, + "step": 763 + }, + { + "epoch": 0.9646464646464646, + "grad_norm": 2.616532325744629, + "learning_rate": 3.804957904583723e-06, + "loss": 1.5469, + "step": 764 + }, + { + "epoch": 0.9659090909090909, + "grad_norm": 2.7471885681152344, + "learning_rate": 3.8026192703461185e-06, + "loss": 1.5431, + "step": 765 + }, + { + "epoch": 0.9671717171717171, + "grad_norm": 2.7478253841400146, + "learning_rate": 3.8002806361085133e-06, + "loss": 1.5452, + "step": 766 + }, + { + "epoch": 0.9684343434343434, + "grad_norm": 2.543790340423584, + "learning_rate": 3.7979420018709077e-06, + "loss": 1.6091, + "step": 767 + }, + { + "epoch": 0.9696969696969697, + "grad_norm": 2.6169509887695312, + "learning_rate": 3.7956033676333026e-06, + "loss": 1.5557, + "step": 768 + }, + { + "epoch": 0.9709595959595959, + "grad_norm": 2.4768588542938232, + "learning_rate": 3.7932647333956974e-06, + "loss": 1.5247, + "step": 769 + }, + { + "epoch": 0.9722222222222222, + "grad_norm": 2.654648780822754, + "learning_rate": 3.7909260991580923e-06, + "loss": 1.5281, + "step": 770 + }, + { + "epoch": 0.9722222222222222, + "eval_loss": 1.7289395332336426, + "eval_runtime": 37.1374, + "eval_samples_per_second": 26.927, + "eval_steps_per_second": 1.131, + "step": 770 + }, + { + "epoch": 0.9734848484848485, + "grad_norm": 2.893073081970215, + "learning_rate": 3.7885874649204867e-06, + "loss": 1.5749, + "step": 771 + }, + { + "epoch": 0.9747474747474747, + "grad_norm": 2.696341037750244, + "learning_rate": 3.7862488306828815e-06, + "loss": 1.5025, + "step": 772 + }, + { + "epoch": 0.976010101010101, + "grad_norm": 2.678222417831421, + "learning_rate": 3.7839101964452764e-06, + "loss": 1.538, + "step": 773 + }, + { + "epoch": 0.9772727272727273, + "grad_norm": 2.7772746086120605, + "learning_rate": 3.7815715622076708e-06, + "loss": 1.5648, + "step": 774 + }, + { + "epoch": 0.9785353535353535, + "grad_norm": 3.076620101928711, + "learning_rate": 3.7792329279700656e-06, + "loss": 1.6114, + "step": 775 + }, + { + "epoch": 0.9797979797979798, + "grad_norm": 2.700056791305542, + "learning_rate": 3.7768942937324605e-06, + "loss": 1.5881, + "step": 776 + }, + { + "epoch": 0.9810606060606061, + "grad_norm": 2.601419687271118, + "learning_rate": 3.7745556594948553e-06, + "loss": 1.5319, + "step": 777 + }, + { + "epoch": 0.9823232323232324, + "grad_norm": 2.6664950847625732, + "learning_rate": 3.7722170252572497e-06, + "loss": 1.5497, + "step": 778 + }, + { + "epoch": 0.9835858585858586, + "grad_norm": 2.7374589443206787, + "learning_rate": 3.7698783910196446e-06, + "loss": 1.5488, + "step": 779 + }, + { + "epoch": 0.9848484848484849, + "grad_norm": 2.797938823699951, + "learning_rate": 3.76753975678204e-06, + "loss": 1.5376, + "step": 780 + }, + { + "epoch": 0.9848484848484849, + "eval_loss": 1.7334709167480469, + "eval_runtime": 37.1372, + "eval_samples_per_second": 26.927, + "eval_steps_per_second": 1.131, + "step": 780 + }, + { + "epoch": 0.9861111111111112, + "grad_norm": 2.881591796875, + "learning_rate": 3.7652011225444347e-06, + "loss": 1.5405, + "step": 781 + }, + { + "epoch": 0.9873737373737373, + "grad_norm": 2.6965644359588623, + "learning_rate": 3.7628624883068295e-06, + "loss": 1.5066, + "step": 782 + }, + { + "epoch": 0.9886363636363636, + "grad_norm": 2.7286899089813232, + "learning_rate": 3.760523854069224e-06, + "loss": 1.5277, + "step": 783 + }, + { + "epoch": 0.98989898989899, + "grad_norm": 2.795644998550415, + "learning_rate": 3.7581852198316188e-06, + "loss": 1.5335, + "step": 784 + }, + { + "epoch": 0.9911616161616161, + "grad_norm": 2.8160440921783447, + "learning_rate": 3.7558465855940136e-06, + "loss": 1.585, + "step": 785 + }, + { + "epoch": 0.9924242424242424, + "grad_norm": 2.7984936237335205, + "learning_rate": 3.753507951356408e-06, + "loss": 1.5569, + "step": 786 + }, + { + "epoch": 0.9936868686868687, + "grad_norm": 2.7596278190612793, + "learning_rate": 3.751169317118803e-06, + "loss": 1.4838, + "step": 787 + }, + { + "epoch": 0.9949494949494949, + "grad_norm": 2.666383743286133, + "learning_rate": 3.7488306828811977e-06, + "loss": 1.5656, + "step": 788 + }, + { + "epoch": 0.9962121212121212, + "grad_norm": 2.8139331340789795, + "learning_rate": 3.7464920486435926e-06, + "loss": 1.5753, + "step": 789 + }, + { + "epoch": 0.9974747474747475, + "grad_norm": 2.7714269161224365, + "learning_rate": 3.744153414405987e-06, + "loss": 1.5187, + "step": 790 + }, + { + "epoch": 0.9974747474747475, + "eval_loss": 1.7257990837097168, + "eval_runtime": 37.1065, + "eval_samples_per_second": 26.949, + "eval_steps_per_second": 1.132, + "step": 790 + }, + { + "epoch": 0.9987373737373737, + "grad_norm": 2.697606325149536, + "learning_rate": 3.741814780168382e-06, + "loss": 1.56, + "step": 791 + }, + { + "epoch": 1.0, + "grad_norm": 2.7685229778289795, + "learning_rate": 3.7394761459307767e-06, + "loss": 1.5705, + "step": 792 + }, + { + "epoch": 1.0012626262626263, + "grad_norm": 2.778200626373291, + "learning_rate": 3.737137511693171e-06, + "loss": 1.5288, + "step": 793 + }, + { + "epoch": 1.0025252525252526, + "grad_norm": 2.7211337089538574, + "learning_rate": 3.734798877455566e-06, + "loss": 1.5348, + "step": 794 + }, + { + "epoch": 1.003787878787879, + "grad_norm": 2.690739154815674, + "learning_rate": 3.7324602432179608e-06, + "loss": 1.4734, + "step": 795 + }, + { + "epoch": 1.005050505050505, + "grad_norm": 2.6809420585632324, + "learning_rate": 3.730121608980356e-06, + "loss": 1.4852, + "step": 796 + }, + { + "epoch": 1.0063131313131313, + "grad_norm": 2.8550117015838623, + "learning_rate": 3.727782974742751e-06, + "loss": 1.4562, + "step": 797 + }, + { + "epoch": 1.0075757575757576, + "grad_norm": 2.713935613632202, + "learning_rate": 3.7254443405051453e-06, + "loss": 1.5005, + "step": 798 + }, + { + "epoch": 1.0088383838383839, + "grad_norm": 2.668363332748413, + "learning_rate": 3.72310570626754e-06, + "loss": 1.485, + "step": 799 + }, + { + "epoch": 1.0101010101010102, + "grad_norm": 2.840769052505493, + "learning_rate": 3.720767072029935e-06, + "loss": 1.4531, + "step": 800 + }, + { + "epoch": 1.0101010101010102, + "eval_loss": 1.725937008857727, + "eval_runtime": 37.2471, + "eval_samples_per_second": 26.848, + "eval_steps_per_second": 1.128, + "step": 800 + }, + { + "epoch": 1.0113636363636365, + "grad_norm": 2.7345798015594482, + "learning_rate": 3.71842843779233e-06, + "loss": 1.5091, + "step": 801 + }, + { + "epoch": 1.0126262626262625, + "grad_norm": 2.9052624702453613, + "learning_rate": 3.7160898035547243e-06, + "loss": 1.5039, + "step": 802 + }, + { + "epoch": 1.0138888888888888, + "grad_norm": 2.893073558807373, + "learning_rate": 3.713751169317119e-06, + "loss": 1.4613, + "step": 803 + }, + { + "epoch": 1.0151515151515151, + "grad_norm": 2.8505427837371826, + "learning_rate": 3.711412535079514e-06, + "loss": 1.4932, + "step": 804 + }, + { + "epoch": 1.0164141414141414, + "grad_norm": 2.9065170288085938, + "learning_rate": 3.7090739008419084e-06, + "loss": 1.5082, + "step": 805 + }, + { + "epoch": 1.0176767676767677, + "grad_norm": 2.888068675994873, + "learning_rate": 3.706735266604303e-06, + "loss": 1.5675, + "step": 806 + }, + { + "epoch": 1.018939393939394, + "grad_norm": 2.7325260639190674, + "learning_rate": 3.704396632366698e-06, + "loss": 1.4649, + "step": 807 + }, + { + "epoch": 1.02020202020202, + "grad_norm": 2.8861136436462402, + "learning_rate": 3.702057998129093e-06, + "loss": 1.4517, + "step": 808 + }, + { + "epoch": 1.0214646464646464, + "grad_norm": 2.8837692737579346, + "learning_rate": 3.6997193638914873e-06, + "loss": 1.5198, + "step": 809 + }, + { + "epoch": 1.0227272727272727, + "grad_norm": 2.790330410003662, + "learning_rate": 3.697380729653882e-06, + "loss": 1.5066, + "step": 810 + }, + { + "epoch": 1.0227272727272727, + "eval_loss": 1.7269084453582764, + "eval_runtime": 36.89, + "eval_samples_per_second": 27.108, + "eval_steps_per_second": 1.139, + "step": 810 + }, + { + "epoch": 1.023989898989899, + "grad_norm": 3.03643536567688, + "learning_rate": 3.6950420954162774e-06, + "loss": 1.4928, + "step": 811 + }, + { + "epoch": 1.0252525252525253, + "grad_norm": 2.841055154800415, + "learning_rate": 3.6927034611786723e-06, + "loss": 1.5273, + "step": 812 + }, + { + "epoch": 1.0265151515151516, + "grad_norm": 2.84134578704834, + "learning_rate": 3.690364826941067e-06, + "loss": 1.5443, + "step": 813 + }, + { + "epoch": 1.0277777777777777, + "grad_norm": 2.837758779525757, + "learning_rate": 3.6880261927034615e-06, + "loss": 1.5586, + "step": 814 + }, + { + "epoch": 1.029040404040404, + "grad_norm": 2.731412410736084, + "learning_rate": 3.6856875584658564e-06, + "loss": 1.4465, + "step": 815 + }, + { + "epoch": 1.0303030303030303, + "grad_norm": 2.8979599475860596, + "learning_rate": 3.683348924228251e-06, + "loss": 1.5663, + "step": 816 + }, + { + "epoch": 1.0315656565656566, + "grad_norm": 2.7740373611450195, + "learning_rate": 3.6810102899906456e-06, + "loss": 1.4578, + "step": 817 + }, + { + "epoch": 1.0328282828282829, + "grad_norm": 2.8984005451202393, + "learning_rate": 3.6786716557530405e-06, + "loss": 1.493, + "step": 818 + }, + { + "epoch": 1.0340909090909092, + "grad_norm": 2.880680561065674, + "learning_rate": 3.6763330215154353e-06, + "loss": 1.4858, + "step": 819 + }, + { + "epoch": 1.0353535353535352, + "grad_norm": 3.0220961570739746, + "learning_rate": 3.67399438727783e-06, + "loss": 1.5009, + "step": 820 + }, + { + "epoch": 1.0353535353535352, + "eval_loss": 1.7245674133300781, + "eval_runtime": 37.0235, + "eval_samples_per_second": 27.01, + "eval_steps_per_second": 1.134, + "step": 820 + }, + { + "epoch": 1.0366161616161615, + "grad_norm": 2.7905962467193604, + "learning_rate": 3.6716557530402246e-06, + "loss": 1.4728, + "step": 821 + }, + { + "epoch": 1.0378787878787878, + "grad_norm": 2.856935501098633, + "learning_rate": 3.6693171188026194e-06, + "loss": 1.521, + "step": 822 + }, + { + "epoch": 1.0391414141414141, + "grad_norm": 2.956584930419922, + "learning_rate": 3.6669784845650143e-06, + "loss": 1.4725, + "step": 823 + }, + { + "epoch": 1.0404040404040404, + "grad_norm": 3.130960702896118, + "learning_rate": 3.6646398503274087e-06, + "loss": 1.6127, + "step": 824 + }, + { + "epoch": 1.0416666666666667, + "grad_norm": 2.8855528831481934, + "learning_rate": 3.6623012160898035e-06, + "loss": 1.5078, + "step": 825 + }, + { + "epoch": 1.0429292929292928, + "grad_norm": 2.8030972480773926, + "learning_rate": 3.659962581852199e-06, + "loss": 1.4635, + "step": 826 + }, + { + "epoch": 1.0441919191919191, + "grad_norm": 3.061593770980835, + "learning_rate": 3.6576239476145936e-06, + "loss": 1.4944, + "step": 827 + }, + { + "epoch": 1.0454545454545454, + "grad_norm": 2.9983572959899902, + "learning_rate": 3.6552853133769885e-06, + "loss": 1.505, + "step": 828 + }, + { + "epoch": 1.0467171717171717, + "grad_norm": 2.7681095600128174, + "learning_rate": 3.652946679139383e-06, + "loss": 1.4774, + "step": 829 + }, + { + "epoch": 1.047979797979798, + "grad_norm": 3.0530946254730225, + "learning_rate": 3.6506080449017777e-06, + "loss": 1.4676, + "step": 830 + }, + { + "epoch": 1.047979797979798, + "eval_loss": 1.7242811918258667, + "eval_runtime": 37.0752, + "eval_samples_per_second": 26.972, + "eval_steps_per_second": 1.133, + "step": 830 + }, + { + "epoch": 1.0492424242424243, + "grad_norm": 2.8383126258850098, + "learning_rate": 3.6482694106641726e-06, + "loss": 1.486, + "step": 831 + }, + { + "epoch": 1.0505050505050506, + "grad_norm": 3.095928907394409, + "learning_rate": 3.6459307764265674e-06, + "loss": 1.5331, + "step": 832 + }, + { + "epoch": 1.0517676767676767, + "grad_norm": 2.9225456714630127, + "learning_rate": 3.643592142188962e-06, + "loss": 1.556, + "step": 833 + }, + { + "epoch": 1.053030303030303, + "grad_norm": 2.9037210941314697, + "learning_rate": 3.6412535079513567e-06, + "loss": 1.4834, + "step": 834 + }, + { + "epoch": 1.0542929292929293, + "grad_norm": 2.915022373199463, + "learning_rate": 3.6389148737137515e-06, + "loss": 1.5346, + "step": 835 + }, + { + "epoch": 1.0555555555555556, + "grad_norm": 3.250411033630371, + "learning_rate": 3.636576239476146e-06, + "loss": 1.4885, + "step": 836 + }, + { + "epoch": 1.0568181818181819, + "grad_norm": 2.759061336517334, + "learning_rate": 3.634237605238541e-06, + "loss": 1.456, + "step": 837 + }, + { + "epoch": 1.0580808080808082, + "grad_norm": 2.955981731414795, + "learning_rate": 3.6318989710009356e-06, + "loss": 1.4889, + "step": 838 + }, + { + "epoch": 1.0593434343434343, + "grad_norm": 3.243338108062744, + "learning_rate": 3.6295603367633305e-06, + "loss": 1.5207, + "step": 839 + }, + { + "epoch": 1.0606060606060606, + "grad_norm": 3.075840950012207, + "learning_rate": 3.627221702525725e-06, + "loss": 1.4906, + "step": 840 + }, + { + "epoch": 1.0606060606060606, + "eval_loss": 1.7290290594100952, + "eval_runtime": 37.097, + "eval_samples_per_second": 26.956, + "eval_steps_per_second": 1.132, + "step": 840 + }, + { + "epoch": 1.0618686868686869, + "grad_norm": 3.149125099182129, + "learning_rate": 3.62488306828812e-06, + "loss": 1.4755, + "step": 841 + }, + { + "epoch": 1.0631313131313131, + "grad_norm": 3.099318504333496, + "learning_rate": 3.622544434050515e-06, + "loss": 1.534, + "step": 842 + }, + { + "epoch": 1.0643939393939394, + "grad_norm": 2.8942148685455322, + "learning_rate": 3.62020579981291e-06, + "loss": 1.4844, + "step": 843 + }, + { + "epoch": 1.0656565656565657, + "grad_norm": 3.142664670944214, + "learning_rate": 3.6178671655753047e-06, + "loss": 1.5179, + "step": 844 + }, + { + "epoch": 1.0669191919191918, + "grad_norm": 3.106552839279175, + "learning_rate": 3.615528531337699e-06, + "loss": 1.4667, + "step": 845 + }, + { + "epoch": 1.0681818181818181, + "grad_norm": 2.8507936000823975, + "learning_rate": 3.613189897100094e-06, + "loss": 1.4463, + "step": 846 + }, + { + "epoch": 1.0694444444444444, + "grad_norm": 3.2198901176452637, + "learning_rate": 3.610851262862489e-06, + "loss": 1.5146, + "step": 847 + }, + { + "epoch": 1.0707070707070707, + "grad_norm": 3.146388292312622, + "learning_rate": 3.6085126286248832e-06, + "loss": 1.4927, + "step": 848 + }, + { + "epoch": 1.071969696969697, + "grad_norm": 2.8304760456085205, + "learning_rate": 3.606173994387278e-06, + "loss": 1.4971, + "step": 849 + }, + { + "epoch": 1.0732323232323233, + "grad_norm": 3.1714529991149902, + "learning_rate": 3.603835360149673e-06, + "loss": 1.5418, + "step": 850 + }, + { + "epoch": 1.0732323232323233, + "eval_loss": 1.725188970565796, + "eval_runtime": 37.117, + "eval_samples_per_second": 26.942, + "eval_steps_per_second": 1.132, + "step": 850 + }, + { + "epoch": 1.0744949494949494, + "grad_norm": 3.06162428855896, + "learning_rate": 3.6014967259120677e-06, + "loss": 1.4959, + "step": 851 + }, + { + "epoch": 1.0757575757575757, + "grad_norm": 2.812206268310547, + "learning_rate": 3.599158091674462e-06, + "loss": 1.5249, + "step": 852 + }, + { + "epoch": 1.077020202020202, + "grad_norm": 3.236210584640503, + "learning_rate": 3.596819457436857e-06, + "loss": 1.5166, + "step": 853 + }, + { + "epoch": 1.0782828282828283, + "grad_norm": 2.8397765159606934, + "learning_rate": 3.594480823199252e-06, + "loss": 1.5043, + "step": 854 + }, + { + "epoch": 1.0795454545454546, + "grad_norm": 2.9215919971466064, + "learning_rate": 3.5921421889616463e-06, + "loss": 1.5235, + "step": 855 + }, + { + "epoch": 1.0808080808080809, + "grad_norm": 2.8219058513641357, + "learning_rate": 3.589803554724041e-06, + "loss": 1.5091, + "step": 856 + }, + { + "epoch": 1.0820707070707072, + "grad_norm": 2.9344675540924072, + "learning_rate": 3.5874649204864364e-06, + "loss": 1.5355, + "step": 857 + }, + { + "epoch": 1.0833333333333333, + "grad_norm": 2.949094533920288, + "learning_rate": 3.5851262862488312e-06, + "loss": 1.5273, + "step": 858 + }, + { + "epoch": 1.0845959595959596, + "grad_norm": 2.9801411628723145, + "learning_rate": 3.582787652011226e-06, + "loss": 1.5078, + "step": 859 + }, + { + "epoch": 1.0858585858585859, + "grad_norm": 2.802182912826538, + "learning_rate": 3.5804490177736205e-06, + "loss": 1.5015, + "step": 860 + }, + { + "epoch": 1.0858585858585859, + "eval_loss": 1.7209502458572388, + "eval_runtime": 37.1238, + "eval_samples_per_second": 26.937, + "eval_steps_per_second": 1.131, + "step": 860 + }, + { + "epoch": 1.0871212121212122, + "grad_norm": 2.857457160949707, + "learning_rate": 3.5781103835360153e-06, + "loss": 1.5057, + "step": 861 + }, + { + "epoch": 1.0883838383838385, + "grad_norm": 2.794478178024292, + "learning_rate": 3.57577174929841e-06, + "loss": 1.4594, + "step": 862 + }, + { + "epoch": 1.0896464646464645, + "grad_norm": 3.0028598308563232, + "learning_rate": 3.573433115060805e-06, + "loss": 1.4666, + "step": 863 + }, + { + "epoch": 1.0909090909090908, + "grad_norm": 2.7701406478881836, + "learning_rate": 3.5710944808231994e-06, + "loss": 1.4596, + "step": 864 + }, + { + "epoch": 1.0921717171717171, + "grad_norm": 2.9705960750579834, + "learning_rate": 3.5687558465855943e-06, + "loss": 1.5092, + "step": 865 + }, + { + "epoch": 1.0934343434343434, + "grad_norm": 2.9820821285247803, + "learning_rate": 3.566417212347989e-06, + "loss": 1.5246, + "step": 866 + }, + { + "epoch": 1.0946969696969697, + "grad_norm": 2.988356113433838, + "learning_rate": 3.5640785781103835e-06, + "loss": 1.5035, + "step": 867 + }, + { + "epoch": 1.095959595959596, + "grad_norm": 2.901599884033203, + "learning_rate": 3.5617399438727784e-06, + "loss": 1.5097, + "step": 868 + }, + { + "epoch": 1.0972222222222223, + "grad_norm": 2.9128799438476562, + "learning_rate": 3.5594013096351732e-06, + "loss": 1.5424, + "step": 869 + }, + { + "epoch": 1.0984848484848484, + "grad_norm": 2.719289779663086, + "learning_rate": 3.557062675397568e-06, + "loss": 1.4313, + "step": 870 + }, + { + "epoch": 1.0984848484848484, + "eval_loss": 1.7159698009490967, + "eval_runtime": 37.122, + "eval_samples_per_second": 26.938, + "eval_steps_per_second": 1.131, + "step": 870 + }, + { + "epoch": 1.0997474747474747, + "grad_norm": 3.0080976486206055, + "learning_rate": 3.5547240411599625e-06, + "loss": 1.4483, + "step": 871 + }, + { + "epoch": 1.101010101010101, + "grad_norm": 2.856523036956787, + "learning_rate": 3.5523854069223578e-06, + "loss": 1.5463, + "step": 872 + }, + { + "epoch": 1.1022727272727273, + "grad_norm": 2.8585214614868164, + "learning_rate": 3.5500467726847526e-06, + "loss": 1.5391, + "step": 873 + }, + { + "epoch": 1.1035353535353536, + "grad_norm": 2.8440098762512207, + "learning_rate": 3.5477081384471474e-06, + "loss": 1.499, + "step": 874 + }, + { + "epoch": 1.10479797979798, + "grad_norm": 2.8114542961120605, + "learning_rate": 3.545369504209542e-06, + "loss": 1.4691, + "step": 875 + }, + { + "epoch": 1.106060606060606, + "grad_norm": 2.980273962020874, + "learning_rate": 3.5430308699719367e-06, + "loss": 1.4541, + "step": 876 + }, + { + "epoch": 1.1073232323232323, + "grad_norm": 3.0053417682647705, + "learning_rate": 3.5406922357343315e-06, + "loss": 1.4895, + "step": 877 + }, + { + "epoch": 1.1085858585858586, + "grad_norm": 2.924046754837036, + "learning_rate": 3.5383536014967264e-06, + "loss": 1.5186, + "step": 878 + }, + { + "epoch": 1.1098484848484849, + "grad_norm": 2.969461679458618, + "learning_rate": 3.536014967259121e-06, + "loss": 1.5213, + "step": 879 + }, + { + "epoch": 1.1111111111111112, + "grad_norm": 2.8936445713043213, + "learning_rate": 3.5336763330215156e-06, + "loss": 1.5585, + "step": 880 + }, + { + "epoch": 1.1111111111111112, + "eval_loss": 1.7150688171386719, + "eval_runtime": 37.1105, + "eval_samples_per_second": 26.947, + "eval_steps_per_second": 1.132, + "step": 880 + }, + { + "epoch": 1.1123737373737375, + "grad_norm": 2.992034435272217, + "learning_rate": 3.5313376987839105e-06, + "loss": 1.5264, + "step": 881 + }, + { + "epoch": 1.1136363636363635, + "grad_norm": 3.0453898906707764, + "learning_rate": 3.5289990645463053e-06, + "loss": 1.5482, + "step": 882 + }, + { + "epoch": 1.1148989898989898, + "grad_norm": 2.7480897903442383, + "learning_rate": 3.5266604303086998e-06, + "loss": 1.4851, + "step": 883 + }, + { + "epoch": 1.1161616161616161, + "grad_norm": 2.9408013820648193, + "learning_rate": 3.5243217960710946e-06, + "loss": 1.536, + "step": 884 + }, + { + "epoch": 1.1174242424242424, + "grad_norm": 2.854438543319702, + "learning_rate": 3.5219831618334894e-06, + "loss": 1.5814, + "step": 885 + }, + { + "epoch": 1.1186868686868687, + "grad_norm": 2.9060397148132324, + "learning_rate": 3.519644527595884e-06, + "loss": 1.5019, + "step": 886 + }, + { + "epoch": 1.119949494949495, + "grad_norm": 2.8741681575775146, + "learning_rate": 3.517305893358279e-06, + "loss": 1.4713, + "step": 887 + }, + { + "epoch": 1.121212121212121, + "grad_norm": 2.972510576248169, + "learning_rate": 3.514967259120674e-06, + "loss": 1.5104, + "step": 888 + }, + { + "epoch": 1.1224747474747474, + "grad_norm": 2.8113558292388916, + "learning_rate": 3.512628624883069e-06, + "loss": 1.4973, + "step": 889 + }, + { + "epoch": 1.1237373737373737, + "grad_norm": 2.9687416553497314, + "learning_rate": 3.5102899906454637e-06, + "loss": 1.4774, + "step": 890 + }, + { + "epoch": 1.1237373737373737, + "eval_loss": 1.7169743776321411, + "eval_runtime": 37.0911, + "eval_samples_per_second": 26.961, + "eval_steps_per_second": 1.132, + "step": 890 + }, + { + "epoch": 1.125, + "grad_norm": 2.911409854888916, + "learning_rate": 3.507951356407858e-06, + "loss": 1.4536, + "step": 891 + }, + { + "epoch": 1.1262626262626263, + "grad_norm": 2.8209526538848877, + "learning_rate": 3.505612722170253e-06, + "loss": 1.5357, + "step": 892 + }, + { + "epoch": 1.1275252525252526, + "grad_norm": 2.811971426010132, + "learning_rate": 3.5032740879326478e-06, + "loss": 1.4889, + "step": 893 + }, + { + "epoch": 1.128787878787879, + "grad_norm": 2.969878673553467, + "learning_rate": 3.5009354536950426e-06, + "loss": 1.4758, + "step": 894 + }, + { + "epoch": 1.130050505050505, + "grad_norm": 2.858635425567627, + "learning_rate": 3.498596819457437e-06, + "loss": 1.5355, + "step": 895 + }, + { + "epoch": 1.1313131313131313, + "grad_norm": 3.018164873123169, + "learning_rate": 3.496258185219832e-06, + "loss": 1.5421, + "step": 896 + }, + { + "epoch": 1.1325757575757576, + "grad_norm": 2.846277952194214, + "learning_rate": 3.4939195509822267e-06, + "loss": 1.4936, + "step": 897 + }, + { + "epoch": 1.1338383838383839, + "grad_norm": 2.965245246887207, + "learning_rate": 3.491580916744621e-06, + "loss": 1.4903, + "step": 898 + }, + { + "epoch": 1.1351010101010102, + "grad_norm": 2.8500895500183105, + "learning_rate": 3.489242282507016e-06, + "loss": 1.5152, + "step": 899 + }, + { + "epoch": 1.1363636363636362, + "grad_norm": 2.8242156505584717, + "learning_rate": 3.486903648269411e-06, + "loss": 1.5531, + "step": 900 + }, + { + "epoch": 1.1363636363636362, + "eval_loss": 1.7180876731872559, + "eval_runtime": 37.4145, + "eval_samples_per_second": 26.728, + "eval_steps_per_second": 1.123, + "step": 900 + }, + { + "epoch": 1.1376262626262625, + "grad_norm": 2.8846635818481445, + "learning_rate": 3.4845650140318057e-06, + "loss": 1.5314, + "step": 901 + }, + { + "epoch": 1.1388888888888888, + "grad_norm": 2.9443984031677246, + "learning_rate": 3.4822263797942e-06, + "loss": 1.4699, + "step": 902 + }, + { + "epoch": 1.1401515151515151, + "grad_norm": 2.7435739040374756, + "learning_rate": 3.4798877455565953e-06, + "loss": 1.4949, + "step": 903 + }, + { + "epoch": 1.1414141414141414, + "grad_norm": 2.979102373123169, + "learning_rate": 3.47754911131899e-06, + "loss": 1.5689, + "step": 904 + }, + { + "epoch": 1.1426767676767677, + "grad_norm": 2.9814586639404297, + "learning_rate": 3.475210477081385e-06, + "loss": 1.5183, + "step": 905 + }, + { + "epoch": 1.143939393939394, + "grad_norm": 2.8380966186523438, + "learning_rate": 3.4728718428437794e-06, + "loss": 1.5177, + "step": 906 + }, + { + "epoch": 1.14520202020202, + "grad_norm": 2.8478410243988037, + "learning_rate": 3.4705332086061743e-06, + "loss": 1.4787, + "step": 907 + }, + { + "epoch": 1.1464646464646464, + "grad_norm": 2.8247780799865723, + "learning_rate": 3.468194574368569e-06, + "loss": 1.425, + "step": 908 + }, + { + "epoch": 1.1477272727272727, + "grad_norm": 2.869649887084961, + "learning_rate": 3.465855940130964e-06, + "loss": 1.4619, + "step": 909 + }, + { + "epoch": 1.148989898989899, + "grad_norm": 2.9439175128936768, + "learning_rate": 3.4635173058933584e-06, + "loss": 1.5194, + "step": 910 + }, + { + "epoch": 1.148989898989899, + "eval_loss": 1.7227450609207153, + "eval_runtime": 36.9723, + "eval_samples_per_second": 27.047, + "eval_steps_per_second": 1.136, + "step": 910 + }, + { + "epoch": 1.1502525252525253, + "grad_norm": 3.0219292640686035, + "learning_rate": 3.4611786716557532e-06, + "loss": 1.5878, + "step": 911 + }, + { + "epoch": 1.1515151515151516, + "grad_norm": 2.988475799560547, + "learning_rate": 3.458840037418148e-06, + "loss": 1.5349, + "step": 912 + }, + { + "epoch": 1.1527777777777777, + "grad_norm": 2.708740234375, + "learning_rate": 3.456501403180543e-06, + "loss": 1.4637, + "step": 913 + }, + { + "epoch": 1.154040404040404, + "grad_norm": 3.0844545364379883, + "learning_rate": 3.4541627689429373e-06, + "loss": 1.546, + "step": 914 + }, + { + "epoch": 1.1553030303030303, + "grad_norm": 2.800487518310547, + "learning_rate": 3.451824134705332e-06, + "loss": 1.5374, + "step": 915 + }, + { + "epoch": 1.1565656565656566, + "grad_norm": 2.8701722621917725, + "learning_rate": 3.449485500467727e-06, + "loss": 1.4549, + "step": 916 + }, + { + "epoch": 1.1578282828282829, + "grad_norm": 3.036418914794922, + "learning_rate": 3.4471468662301214e-06, + "loss": 1.5253, + "step": 917 + }, + { + "epoch": 1.1590909090909092, + "grad_norm": 3.0300986766815186, + "learning_rate": 3.4448082319925167e-06, + "loss": 1.4727, + "step": 918 + }, + { + "epoch": 1.1603535353535355, + "grad_norm": 2.8652496337890625, + "learning_rate": 3.4424695977549116e-06, + "loss": 1.4842, + "step": 919 + }, + { + "epoch": 1.1616161616161615, + "grad_norm": 2.9140491485595703, + "learning_rate": 3.4401309635173064e-06, + "loss": 1.4968, + "step": 920 + }, + { + "epoch": 1.1616161616161615, + "eval_loss": 1.7184255123138428, + "eval_runtime": 37.1186, + "eval_samples_per_second": 26.941, + "eval_steps_per_second": 1.132, + "step": 920 + }, + { + "epoch": 1.1628787878787878, + "grad_norm": 2.9708127975463867, + "learning_rate": 3.4377923292797012e-06, + "loss": 1.4817, + "step": 921 + }, + { + "epoch": 1.1641414141414141, + "grad_norm": 2.954814910888672, + "learning_rate": 3.4354536950420957e-06, + "loss": 1.4937, + "step": 922 + }, + { + "epoch": 1.1654040404040404, + "grad_norm": 2.803945779800415, + "learning_rate": 3.4331150608044905e-06, + "loss": 1.527, + "step": 923 + }, + { + "epoch": 1.1666666666666667, + "grad_norm": 2.863250255584717, + "learning_rate": 3.4307764265668853e-06, + "loss": 1.4922, + "step": 924 + }, + { + "epoch": 1.1679292929292928, + "grad_norm": 2.8030712604522705, + "learning_rate": 3.4284377923292798e-06, + "loss": 1.4427, + "step": 925 + }, + { + "epoch": 1.1691919191919191, + "grad_norm": 2.944110155105591, + "learning_rate": 3.4260991580916746e-06, + "loss": 1.4435, + "step": 926 + }, + { + "epoch": 1.1704545454545454, + "grad_norm": 2.9672467708587646, + "learning_rate": 3.4237605238540695e-06, + "loss": 1.5005, + "step": 927 + }, + { + "epoch": 1.1717171717171717, + "grad_norm": 2.920205593109131, + "learning_rate": 3.4214218896164643e-06, + "loss": 1.5303, + "step": 928 + }, + { + "epoch": 1.172979797979798, + "grad_norm": 3.1062121391296387, + "learning_rate": 3.4190832553788587e-06, + "loss": 1.5395, + "step": 929 + }, + { + "epoch": 1.1742424242424243, + "grad_norm": 3.1106555461883545, + "learning_rate": 3.4167446211412536e-06, + "loss": 1.5038, + "step": 930 + }, + { + "epoch": 1.1742424242424243, + "eval_loss": 1.7127314805984497, + "eval_runtime": 37.1193, + "eval_samples_per_second": 26.94, + "eval_steps_per_second": 1.131, + "step": 930 + }, + { + "epoch": 1.1755050505050506, + "grad_norm": 2.7898054122924805, + "learning_rate": 3.4144059869036484e-06, + "loss": 1.5096, + "step": 931 + }, + { + "epoch": 1.1767676767676767, + "grad_norm": 2.881694793701172, + "learning_rate": 3.4120673526660432e-06, + "loss": 1.4556, + "step": 932 + }, + { + "epoch": 1.178030303030303, + "grad_norm": 3.131551742553711, + "learning_rate": 3.4097287184284385e-06, + "loss": 1.5487, + "step": 933 + }, + { + "epoch": 1.1792929292929293, + "grad_norm": 2.9734153747558594, + "learning_rate": 3.407390084190833e-06, + "loss": 1.439, + "step": 934 + }, + { + "epoch": 1.1805555555555556, + "grad_norm": 3.009657144546509, + "learning_rate": 3.4050514499532278e-06, + "loss": 1.462, + "step": 935 + }, + { + "epoch": 1.1818181818181819, + "grad_norm": 2.8378758430480957, + "learning_rate": 3.4027128157156226e-06, + "loss": 1.5083, + "step": 936 + }, + { + "epoch": 1.1830808080808082, + "grad_norm": 2.923969268798828, + "learning_rate": 3.400374181478017e-06, + "loss": 1.5844, + "step": 937 + }, + { + "epoch": 1.1843434343434343, + "grad_norm": 3.2384307384490967, + "learning_rate": 3.398035547240412e-06, + "loss": 1.5175, + "step": 938 + }, + { + "epoch": 1.1856060606060606, + "grad_norm": 3.024385929107666, + "learning_rate": 3.3956969130028067e-06, + "loss": 1.4756, + "step": 939 + }, + { + "epoch": 1.1868686868686869, + "grad_norm": 3.088745594024658, + "learning_rate": 3.3933582787652016e-06, + "loss": 1.4914, + "step": 940 + }, + { + "epoch": 1.1868686868686869, + "eval_loss": 1.705490231513977, + "eval_runtime": 37.1291, + "eval_samples_per_second": 26.933, + "eval_steps_per_second": 1.131, + "step": 940 + }, + { + "epoch": 1.1881313131313131, + "grad_norm": 2.898893117904663, + "learning_rate": 3.391019644527596e-06, + "loss": 1.5002, + "step": 941 + }, + { + "epoch": 1.1893939393939394, + "grad_norm": 2.800103187561035, + "learning_rate": 3.388681010289991e-06, + "loss": 1.4842, + "step": 942 + }, + { + "epoch": 1.1906565656565657, + "grad_norm": 3.0287230014801025, + "learning_rate": 3.3863423760523857e-06, + "loss": 1.5454, + "step": 943 + }, + { + "epoch": 1.1919191919191918, + "grad_norm": 2.9379546642303467, + "learning_rate": 3.3840037418147805e-06, + "loss": 1.5427, + "step": 944 + }, + { + "epoch": 1.1931818181818181, + "grad_norm": 2.879699468612671, + "learning_rate": 3.381665107577175e-06, + "loss": 1.4804, + "step": 945 + }, + { + "epoch": 1.1944444444444444, + "grad_norm": 2.8708608150482178, + "learning_rate": 3.3793264733395698e-06, + "loss": 1.479, + "step": 946 + }, + { + "epoch": 1.1957070707070707, + "grad_norm": 2.8513312339782715, + "learning_rate": 3.3769878391019646e-06, + "loss": 1.5343, + "step": 947 + }, + { + "epoch": 1.196969696969697, + "grad_norm": 2.8715100288391113, + "learning_rate": 3.37464920486436e-06, + "loss": 1.5114, + "step": 948 + }, + { + "epoch": 1.1982323232323233, + "grad_norm": 2.819213628768921, + "learning_rate": 3.3723105706267543e-06, + "loss": 1.4873, + "step": 949 + }, + { + "epoch": 1.1994949494949494, + "grad_norm": 2.946578025817871, + "learning_rate": 3.369971936389149e-06, + "loss": 1.5378, + "step": 950 + }, + { + "epoch": 1.1994949494949494, + "eval_loss": 1.7094330787658691, + "eval_runtime": 37.1453, + "eval_samples_per_second": 26.921, + "eval_steps_per_second": 1.131, + "step": 950 + }, + { + "epoch": 1.2007575757575757, + "grad_norm": 2.8170437812805176, + "learning_rate": 3.367633302151544e-06, + "loss": 1.4416, + "step": 951 + }, + { + "epoch": 1.202020202020202, + "grad_norm": 2.9654042720794678, + "learning_rate": 3.365294667913939e-06, + "loss": 1.5115, + "step": 952 + }, + { + "epoch": 1.2032828282828283, + "grad_norm": 2.9138922691345215, + "learning_rate": 3.3629560336763333e-06, + "loss": 1.5106, + "step": 953 + }, + { + "epoch": 1.2045454545454546, + "grad_norm": 2.989961862564087, + "learning_rate": 3.360617399438728e-06, + "loss": 1.5278, + "step": 954 + }, + { + "epoch": 1.2058080808080809, + "grad_norm": 2.9620680809020996, + "learning_rate": 3.358278765201123e-06, + "loss": 1.4949, + "step": 955 + }, + { + "epoch": 1.2070707070707072, + "grad_norm": 2.945505380630493, + "learning_rate": 3.3559401309635174e-06, + "loss": 1.5018, + "step": 956 + }, + { + "epoch": 1.2083333333333333, + "grad_norm": 2.806464433670044, + "learning_rate": 3.353601496725912e-06, + "loss": 1.4951, + "step": 957 + }, + { + "epoch": 1.2095959595959596, + "grad_norm": 2.7847559452056885, + "learning_rate": 3.351262862488307e-06, + "loss": 1.4259, + "step": 958 + }, + { + "epoch": 1.2108585858585859, + "grad_norm": 2.760003089904785, + "learning_rate": 3.348924228250702e-06, + "loss": 1.4757, + "step": 959 + }, + { + "epoch": 1.2121212121212122, + "grad_norm": 3.037522792816162, + "learning_rate": 3.3465855940130963e-06, + "loss": 1.5462, + "step": 960 + }, + { + "epoch": 1.2121212121212122, + "eval_loss": 1.7071956396102905, + "eval_runtime": 37.1351, + "eval_samples_per_second": 26.929, + "eval_steps_per_second": 1.131, + "step": 960 + }, + { + "epoch": 1.2133838383838385, + "grad_norm": 2.7755539417266846, + "learning_rate": 3.344246959775491e-06, + "loss": 1.453, + "step": 961 + }, + { + "epoch": 1.2146464646464645, + "grad_norm": 2.910369634628296, + "learning_rate": 3.341908325537886e-06, + "loss": 1.4914, + "step": 962 + }, + { + "epoch": 1.2159090909090908, + "grad_norm": 2.8999392986297607, + "learning_rate": 3.339569691300281e-06, + "loss": 1.5389, + "step": 963 + }, + { + "epoch": 1.2171717171717171, + "grad_norm": 2.891995906829834, + "learning_rate": 3.337231057062676e-06, + "loss": 1.4915, + "step": 964 + }, + { + "epoch": 1.2184343434343434, + "grad_norm": 2.814624309539795, + "learning_rate": 3.3348924228250705e-06, + "loss": 1.4936, + "step": 965 + }, + { + "epoch": 1.2196969696969697, + "grad_norm": 2.971163034439087, + "learning_rate": 3.3325537885874654e-06, + "loss": 1.4813, + "step": 966 + }, + { + "epoch": 1.220959595959596, + "grad_norm": 2.934155225753784, + "learning_rate": 3.33021515434986e-06, + "loss": 1.5162, + "step": 967 + }, + { + "epoch": 1.2222222222222223, + "grad_norm": 3.007847309112549, + "learning_rate": 3.3278765201122546e-06, + "loss": 1.4741, + "step": 968 + }, + { + "epoch": 1.2234848484848484, + "grad_norm": 2.977207660675049, + "learning_rate": 3.3255378858746495e-06, + "loss": 1.5186, + "step": 969 + }, + { + "epoch": 1.2247474747474747, + "grad_norm": 2.926567316055298, + "learning_rate": 3.3231992516370443e-06, + "loss": 1.5111, + "step": 970 + }, + { + "epoch": 1.2247474747474747, + "eval_loss": 1.7068836688995361, + "eval_runtime": 37.1419, + "eval_samples_per_second": 26.924, + "eval_steps_per_second": 1.131, + "step": 970 + }, + { + "epoch": 1.226010101010101, + "grad_norm": 3.063671350479126, + "learning_rate": 3.320860617399439e-06, + "loss": 1.4376, + "step": 971 + }, + { + "epoch": 1.2272727272727273, + "grad_norm": 2.7263271808624268, + "learning_rate": 3.3185219831618336e-06, + "loss": 1.4555, + "step": 972 + }, + { + "epoch": 1.2285353535353536, + "grad_norm": 2.912658929824829, + "learning_rate": 3.3161833489242284e-06, + "loss": 1.5116, + "step": 973 + }, + { + "epoch": 1.22979797979798, + "grad_norm": 3.2216858863830566, + "learning_rate": 3.3138447146866233e-06, + "loss": 1.5124, + "step": 974 + }, + { + "epoch": 1.231060606060606, + "grad_norm": 2.9138946533203125, + "learning_rate": 3.311506080449018e-06, + "loss": 1.4917, + "step": 975 + }, + { + "epoch": 1.2323232323232323, + "grad_norm": 3.0519120693206787, + "learning_rate": 3.3091674462114125e-06, + "loss": 1.4986, + "step": 976 + }, + { + "epoch": 1.2335858585858586, + "grad_norm": 2.8186402320861816, + "learning_rate": 3.3068288119738074e-06, + "loss": 1.4835, + "step": 977 + }, + { + "epoch": 1.2348484848484849, + "grad_norm": 2.799189329147339, + "learning_rate": 3.304490177736202e-06, + "loss": 1.5069, + "step": 978 + }, + { + "epoch": 1.2361111111111112, + "grad_norm": 3.010010004043579, + "learning_rate": 3.3021515434985975e-06, + "loss": 1.4276, + "step": 979 + }, + { + "epoch": 1.2373737373737375, + "grad_norm": 3.153592109680176, + "learning_rate": 3.299812909260992e-06, + "loss": 1.5459, + "step": 980 + }, + { + "epoch": 1.2373737373737375, + "eval_loss": 1.7070198059082031, + "eval_runtime": 37.1397, + "eval_samples_per_second": 26.925, + "eval_steps_per_second": 1.131, + "step": 980 + }, + { + "epoch": 1.2386363636363638, + "grad_norm": 2.7818424701690674, + "learning_rate": 3.2974742750233867e-06, + "loss": 1.5045, + "step": 981 + }, + { + "epoch": 1.2398989898989898, + "grad_norm": 2.9211390018463135, + "learning_rate": 3.2951356407857816e-06, + "loss": 1.4966, + "step": 982 + }, + { + "epoch": 1.2411616161616161, + "grad_norm": 2.979743242263794, + "learning_rate": 3.2927970065481764e-06, + "loss": 1.4754, + "step": 983 + }, + { + "epoch": 1.2424242424242424, + "grad_norm": 2.8476738929748535, + "learning_rate": 3.290458372310571e-06, + "loss": 1.4619, + "step": 984 + }, + { + "epoch": 1.2436868686868687, + "grad_norm": 2.8304896354675293, + "learning_rate": 3.2881197380729657e-06, + "loss": 1.4192, + "step": 985 + }, + { + "epoch": 1.244949494949495, + "grad_norm": 3.3150880336761475, + "learning_rate": 3.2857811038353605e-06, + "loss": 1.515, + "step": 986 + }, + { + "epoch": 1.246212121212121, + "grad_norm": 3.049398422241211, + "learning_rate": 3.283442469597755e-06, + "loss": 1.5382, + "step": 987 + }, + { + "epoch": 1.2474747474747474, + "grad_norm": 3.0910539627075195, + "learning_rate": 3.2811038353601498e-06, + "loss": 1.5458, + "step": 988 + }, + { + "epoch": 1.2487373737373737, + "grad_norm": 3.002788782119751, + "learning_rate": 3.2787652011225446e-06, + "loss": 1.4668, + "step": 989 + }, + { + "epoch": 1.25, + "grad_norm": 2.9342103004455566, + "learning_rate": 3.2764265668849395e-06, + "loss": 1.5187, + "step": 990 + }, + { + "epoch": 1.25, + "eval_loss": 1.7082287073135376, + "eval_runtime": 37.1399, + "eval_samples_per_second": 26.925, + "eval_steps_per_second": 1.131, + "step": 990 + }, + { + "epoch": 1.2512626262626263, + "grad_norm": 2.7596492767333984, + "learning_rate": 3.274087932647334e-06, + "loss": 1.5191, + "step": 991 + }, + { + "epoch": 1.2525252525252526, + "grad_norm": 2.949368953704834, + "learning_rate": 3.2717492984097287e-06, + "loss": 1.4776, + "step": 992 + }, + { + "epoch": 1.253787878787879, + "grad_norm": 2.7099363803863525, + "learning_rate": 3.2694106641721236e-06, + "loss": 1.455, + "step": 993 + }, + { + "epoch": 1.255050505050505, + "grad_norm": 2.8631885051727295, + "learning_rate": 3.267072029934519e-06, + "loss": 1.4394, + "step": 994 + }, + { + "epoch": 1.2563131313131313, + "grad_norm": 3.09389591217041, + "learning_rate": 3.2647333956969137e-06, + "loss": 1.4958, + "step": 995 + }, + { + "epoch": 1.2575757575757576, + "grad_norm": 3.108271360397339, + "learning_rate": 3.262394761459308e-06, + "loss": 1.5108, + "step": 996 + }, + { + "epoch": 1.2588383838383839, + "grad_norm": 3.079176902770996, + "learning_rate": 3.260056127221703e-06, + "loss": 1.4749, + "step": 997 + }, + { + "epoch": 1.2601010101010102, + "grad_norm": 3.0458056926727295, + "learning_rate": 3.257717492984098e-06, + "loss": 1.4911, + "step": 998 + }, + { + "epoch": 1.2613636363636362, + "grad_norm": 2.952151298522949, + "learning_rate": 3.2553788587464922e-06, + "loss": 1.4842, + "step": 999 + }, + { + "epoch": 1.2626262626262625, + "grad_norm": 2.7918665409088135, + "learning_rate": 3.253040224508887e-06, + "loss": 1.4684, + "step": 1000 + }, + { + "epoch": 1.2626262626262625, + "eval_loss": 1.7100834846496582, + "eval_runtime": 37.3594, + "eval_samples_per_second": 26.767, + "eval_steps_per_second": 1.124, + "step": 1000 + }, + { + "epoch": 1.2638888888888888, + "grad_norm": 2.798064947128296, + "learning_rate": 3.250701590271282e-06, + "loss": 1.4295, + "step": 1001 + }, + { + "epoch": 1.2651515151515151, + "grad_norm": 2.852234363555908, + "learning_rate": 3.2483629560336767e-06, + "loss": 1.5039, + "step": 1002 + }, + { + "epoch": 1.2664141414141414, + "grad_norm": 2.762645721435547, + "learning_rate": 3.246024321796071e-06, + "loss": 1.4365, + "step": 1003 + }, + { + "epoch": 1.2676767676767677, + "grad_norm": 2.944326162338257, + "learning_rate": 3.243685687558466e-06, + "loss": 1.4849, + "step": 1004 + }, + { + "epoch": 1.268939393939394, + "grad_norm": 2.9298877716064453, + "learning_rate": 3.241347053320861e-06, + "loss": 1.4397, + "step": 1005 + }, + { + "epoch": 1.2702020202020203, + "grad_norm": 3.1259615421295166, + "learning_rate": 3.2390084190832553e-06, + "loss": 1.4819, + "step": 1006 + }, + { + "epoch": 1.2714646464646464, + "grad_norm": 3.114792823791504, + "learning_rate": 3.23666978484565e-06, + "loss": 1.4728, + "step": 1007 + }, + { + "epoch": 1.2727272727272727, + "grad_norm": 2.9800360202789307, + "learning_rate": 3.234331150608045e-06, + "loss": 1.4426, + "step": 1008 + }, + { + "epoch": 1.273989898989899, + "grad_norm": 2.8487257957458496, + "learning_rate": 3.2319925163704402e-06, + "loss": 1.461, + "step": 1009 + }, + { + "epoch": 1.2752525252525253, + "grad_norm": 2.800760507583618, + "learning_rate": 3.229653882132835e-06, + "loss": 1.4764, + "step": 1010 + }, + { + "epoch": 1.2752525252525253, + "eval_loss": 1.701302409172058, + "eval_runtime": 36.9359, + "eval_samples_per_second": 27.074, + "eval_steps_per_second": 1.137, + "step": 1010 + }, + { + "epoch": 1.2765151515151514, + "grad_norm": 2.851012706756592, + "learning_rate": 3.2273152478952295e-06, + "loss": 1.5124, + "step": 1011 + }, + { + "epoch": 1.2777777777777777, + "grad_norm": 2.9498190879821777, + "learning_rate": 3.2249766136576243e-06, + "loss": 1.4789, + "step": 1012 + }, + { + "epoch": 1.279040404040404, + "grad_norm": 2.80710768699646, + "learning_rate": 3.222637979420019e-06, + "loss": 1.5592, + "step": 1013 + }, + { + "epoch": 1.2803030303030303, + "grad_norm": 2.9525952339172363, + "learning_rate": 3.220299345182414e-06, + "loss": 1.4941, + "step": 1014 + }, + { + "epoch": 1.2815656565656566, + "grad_norm": 2.9770407676696777, + "learning_rate": 3.2179607109448084e-06, + "loss": 1.4437, + "step": 1015 + }, + { + "epoch": 1.2828282828282829, + "grad_norm": 3.2553279399871826, + "learning_rate": 3.2156220767072033e-06, + "loss": 1.5248, + "step": 1016 + }, + { + "epoch": 1.2840909090909092, + "grad_norm": 3.0591554641723633, + "learning_rate": 3.213283442469598e-06, + "loss": 1.5704, + "step": 1017 + }, + { + "epoch": 1.2853535353535355, + "grad_norm": 2.776968002319336, + "learning_rate": 3.2109448082319925e-06, + "loss": 1.4239, + "step": 1018 + }, + { + "epoch": 1.2866161616161615, + "grad_norm": 2.8728222846984863, + "learning_rate": 3.2086061739943874e-06, + "loss": 1.5115, + "step": 1019 + }, + { + "epoch": 1.2878787878787878, + "grad_norm": 2.907308340072632, + "learning_rate": 3.2062675397567822e-06, + "loss": 1.4764, + "step": 1020 + }, + { + "epoch": 1.2878787878787878, + "eval_loss": 1.7016223669052124, + "eval_runtime": 37.0728, + "eval_samples_per_second": 26.974, + "eval_steps_per_second": 1.133, + "step": 1020 + }, + { + "epoch": 1.2891414141414141, + "grad_norm": 3.0686309337615967, + "learning_rate": 3.203928905519177e-06, + "loss": 1.5175, + "step": 1021 + }, + { + "epoch": 1.2904040404040404, + "grad_norm": 2.8636863231658936, + "learning_rate": 3.2015902712815715e-06, + "loss": 1.5025, + "step": 1022 + }, + { + "epoch": 1.2916666666666667, + "grad_norm": 2.8539068698883057, + "learning_rate": 3.1992516370439663e-06, + "loss": 1.4774, + "step": 1023 + }, + { + "epoch": 1.2929292929292928, + "grad_norm": 2.9830520153045654, + "learning_rate": 3.196913002806361e-06, + "loss": 1.533, + "step": 1024 + }, + { + "epoch": 1.2941919191919191, + "grad_norm": 2.8806374073028564, + "learning_rate": 3.1945743685687564e-06, + "loss": 1.5298, + "step": 1025 + }, + { + "epoch": 1.2954545454545454, + "grad_norm": 2.932870626449585, + "learning_rate": 3.1922357343311513e-06, + "loss": 1.5028, + "step": 1026 + }, + { + "epoch": 1.2967171717171717, + "grad_norm": 2.889601945877075, + "learning_rate": 3.1898971000935457e-06, + "loss": 1.4534, + "step": 1027 + }, + { + "epoch": 1.297979797979798, + "grad_norm": 2.930546522140503, + "learning_rate": 3.1875584658559405e-06, + "loss": 1.5047, + "step": 1028 + }, + { + "epoch": 1.2992424242424243, + "grad_norm": 3.0944340229034424, + "learning_rate": 3.1852198316183354e-06, + "loss": 1.5198, + "step": 1029 + }, + { + "epoch": 1.3005050505050506, + "grad_norm": 2.8930089473724365, + "learning_rate": 3.18288119738073e-06, + "loss": 1.4655, + "step": 1030 + }, + { + "epoch": 1.3005050505050506, + "eval_loss": 1.7008370161056519, + "eval_runtime": 37.1263, + "eval_samples_per_second": 26.935, + "eval_steps_per_second": 1.131, + "step": 1030 + }, + { + "epoch": 1.3017676767676767, + "grad_norm": 2.888303518295288, + "learning_rate": 3.1805425631431246e-06, + "loss": 1.4249, + "step": 1031 + }, + { + "epoch": 1.303030303030303, + "grad_norm": 3.0144314765930176, + "learning_rate": 3.1782039289055195e-06, + "loss": 1.4963, + "step": 1032 + }, + { + "epoch": 1.3042929292929293, + "grad_norm": 3.0118296146392822, + "learning_rate": 3.1758652946679143e-06, + "loss": 1.5119, + "step": 1033 + }, + { + "epoch": 1.3055555555555556, + "grad_norm": 2.8148205280303955, + "learning_rate": 3.1735266604303087e-06, + "loss": 1.4921, + "step": 1034 + }, + { + "epoch": 1.3068181818181819, + "grad_norm": 2.811166286468506, + "learning_rate": 3.1711880261927036e-06, + "loss": 1.4689, + "step": 1035 + }, + { + "epoch": 1.308080808080808, + "grad_norm": 2.924088478088379, + "learning_rate": 3.1688493919550984e-06, + "loss": 1.5682, + "step": 1036 + }, + { + "epoch": 1.3093434343434343, + "grad_norm": 2.905988931655884, + "learning_rate": 3.166510757717493e-06, + "loss": 1.5104, + "step": 1037 + }, + { + "epoch": 1.3106060606060606, + "grad_norm": 2.972963571548462, + "learning_rate": 3.1641721234798877e-06, + "loss": 1.4543, + "step": 1038 + }, + { + "epoch": 1.3118686868686869, + "grad_norm": 2.8034229278564453, + "learning_rate": 3.1618334892422825e-06, + "loss": 1.45, + "step": 1039 + }, + { + "epoch": 1.3131313131313131, + "grad_norm": 2.8977267742156982, + "learning_rate": 3.159494855004678e-06, + "loss": 1.4531, + "step": 1040 + }, + { + "epoch": 1.3131313131313131, + "eval_loss": 1.7027597427368164, + "eval_runtime": 37.0629, + "eval_samples_per_second": 26.981, + "eval_steps_per_second": 1.133, + "step": 1040 + }, + { + "epoch": 1.3143939393939394, + "grad_norm": 3.106973886489868, + "learning_rate": 3.1571562207670726e-06, + "loss": 1.4723, + "step": 1041 + }, + { + "epoch": 1.3156565656565657, + "grad_norm": 2.7628490924835205, + "learning_rate": 3.154817586529467e-06, + "loss": 1.4902, + "step": 1042 + }, + { + "epoch": 1.316919191919192, + "grad_norm": 3.119751453399658, + "learning_rate": 3.152478952291862e-06, + "loss": 1.5326, + "step": 1043 + }, + { + "epoch": 1.3181818181818181, + "grad_norm": 3.1470835208892822, + "learning_rate": 3.1501403180542568e-06, + "loss": 1.4626, + "step": 1044 + }, + { + "epoch": 1.3194444444444444, + "grad_norm": 2.866586208343506, + "learning_rate": 3.1478016838166516e-06, + "loss": 1.5006, + "step": 1045 + }, + { + "epoch": 1.3207070707070707, + "grad_norm": 2.878464698791504, + "learning_rate": 3.145463049579046e-06, + "loss": 1.4727, + "step": 1046 + }, + { + "epoch": 1.321969696969697, + "grad_norm": 2.8567826747894287, + "learning_rate": 3.143124415341441e-06, + "loss": 1.4202, + "step": 1047 + }, + { + "epoch": 1.3232323232323233, + "grad_norm": 2.8718442916870117, + "learning_rate": 3.1407857811038357e-06, + "loss": 1.473, + "step": 1048 + }, + { + "epoch": 1.3244949494949494, + "grad_norm": 3.003040313720703, + "learning_rate": 3.13844714686623e-06, + "loss": 1.4784, + "step": 1049 + }, + { + "epoch": 1.3257575757575757, + "grad_norm": 3.080288887023926, + "learning_rate": 3.136108512628625e-06, + "loss": 1.4514, + "step": 1050 + }, + { + "epoch": 1.3257575757575757, + "eval_loss": 1.704396367073059, + "eval_runtime": 37.0757, + "eval_samples_per_second": 26.972, + "eval_steps_per_second": 1.133, + "step": 1050 + }, + { + "epoch": 1.327020202020202, + "grad_norm": 2.8699166774749756, + "learning_rate": 3.13376987839102e-06, + "loss": 1.4966, + "step": 1051 + }, + { + "epoch": 1.3282828282828283, + "grad_norm": 2.979238271713257, + "learning_rate": 3.1314312441534146e-06, + "loss": 1.4422, + "step": 1052 + }, + { + "epoch": 1.3295454545454546, + "grad_norm": 2.8612024784088135, + "learning_rate": 3.129092609915809e-06, + "loss": 1.4982, + "step": 1053 + }, + { + "epoch": 1.3308080808080809, + "grad_norm": 2.7759032249450684, + "learning_rate": 3.126753975678204e-06, + "loss": 1.413, + "step": 1054 + }, + { + "epoch": 1.3320707070707072, + "grad_norm": 2.9772675037384033, + "learning_rate": 3.124415341440599e-06, + "loss": 1.4789, + "step": 1055 + }, + { + "epoch": 1.3333333333333333, + "grad_norm": 2.9382526874542236, + "learning_rate": 3.122076707202994e-06, + "loss": 1.4547, + "step": 1056 + }, + { + "epoch": 1.3345959595959596, + "grad_norm": 2.85050630569458, + "learning_rate": 3.119738072965389e-06, + "loss": 1.4193, + "step": 1057 + }, + { + "epoch": 1.3358585858585859, + "grad_norm": 3.1148695945739746, + "learning_rate": 3.1173994387277833e-06, + "loss": 1.4527, + "step": 1058 + }, + { + "epoch": 1.3371212121212122, + "grad_norm": 3.3906028270721436, + "learning_rate": 3.115060804490178e-06, + "loss": 1.5273, + "step": 1059 + }, + { + "epoch": 1.3383838383838385, + "grad_norm": 2.9665908813476562, + "learning_rate": 3.112722170252573e-06, + "loss": 1.5237, + "step": 1060 + }, + { + "epoch": 1.3383838383838385, + "eval_loss": 1.7004541158676147, + "eval_runtime": 37.081, + "eval_samples_per_second": 26.968, + "eval_steps_per_second": 1.133, + "step": 1060 + }, + { + "epoch": 1.3396464646464645, + "grad_norm": 2.9809224605560303, + "learning_rate": 3.1103835360149674e-06, + "loss": 1.498, + "step": 1061 + }, + { + "epoch": 1.3409090909090908, + "grad_norm": 2.95265793800354, + "learning_rate": 3.1080449017773622e-06, + "loss": 1.511, + "step": 1062 + }, + { + "epoch": 1.3421717171717171, + "grad_norm": 3.0814359188079834, + "learning_rate": 3.105706267539757e-06, + "loss": 1.4448, + "step": 1063 + }, + { + "epoch": 1.3434343434343434, + "grad_norm": 2.8267405033111572, + "learning_rate": 3.103367633302152e-06, + "loss": 1.4648, + "step": 1064 + }, + { + "epoch": 1.3446969696969697, + "grad_norm": 3.08182954788208, + "learning_rate": 3.1010289990645463e-06, + "loss": 1.5245, + "step": 1065 + }, + { + "epoch": 1.345959595959596, + "grad_norm": 3.183209180831909, + "learning_rate": 3.098690364826941e-06, + "loss": 1.4388, + "step": 1066 + }, + { + "epoch": 1.3472222222222223, + "grad_norm": 3.2121284008026123, + "learning_rate": 3.096351730589336e-06, + "loss": 1.4328, + "step": 1067 + }, + { + "epoch": 1.3484848484848486, + "grad_norm": 3.093921661376953, + "learning_rate": 3.0940130963517304e-06, + "loss": 1.5176, + "step": 1068 + }, + { + "epoch": 1.3497474747474747, + "grad_norm": 2.977710008621216, + "learning_rate": 3.0916744621141253e-06, + "loss": 1.4153, + "step": 1069 + }, + { + "epoch": 1.351010101010101, + "grad_norm": 3.119001865386963, + "learning_rate": 3.08933582787652e-06, + "loss": 1.4879, + "step": 1070 + }, + { + "epoch": 1.351010101010101, + "eval_loss": 1.7006300687789917, + "eval_runtime": 37.0959, + "eval_samples_per_second": 26.957, + "eval_steps_per_second": 1.132, + "step": 1070 + }, + { + "epoch": 1.3522727272727273, + "grad_norm": 3.026405096054077, + "learning_rate": 3.0869971936389154e-06, + "loss": 1.483, + "step": 1071 + }, + { + "epoch": 1.3535353535353536, + "grad_norm": 2.770728826522827, + "learning_rate": 3.0846585594013102e-06, + "loss": 1.5175, + "step": 1072 + }, + { + "epoch": 1.3547979797979797, + "grad_norm": 2.7930657863616943, + "learning_rate": 3.0823199251637047e-06, + "loss": 1.4833, + "step": 1073 + }, + { + "epoch": 1.356060606060606, + "grad_norm": 2.9128804206848145, + "learning_rate": 3.0799812909260995e-06, + "loss": 1.5016, + "step": 1074 + }, + { + "epoch": 1.3573232323232323, + "grad_norm": 3.0432205200195312, + "learning_rate": 3.0776426566884943e-06, + "loss": 1.5211, + "step": 1075 + }, + { + "epoch": 1.3585858585858586, + "grad_norm": 2.9197583198547363, + "learning_rate": 3.075304022450889e-06, + "loss": 1.5065, + "step": 1076 + }, + { + "epoch": 1.3598484848484849, + "grad_norm": 2.944986581802368, + "learning_rate": 3.0729653882132836e-06, + "loss": 1.5015, + "step": 1077 + }, + { + "epoch": 1.3611111111111112, + "grad_norm": 2.7553489208221436, + "learning_rate": 3.0706267539756784e-06, + "loss": 1.4692, + "step": 1078 + }, + { + "epoch": 1.3623737373737375, + "grad_norm": 2.7607858180999756, + "learning_rate": 3.0682881197380733e-06, + "loss": 1.4494, + "step": 1079 + }, + { + "epoch": 1.3636363636363638, + "grad_norm": 2.7896173000335693, + "learning_rate": 3.0659494855004677e-06, + "loss": 1.4625, + "step": 1080 + }, + { + "epoch": 1.3636363636363638, + "eval_loss": 1.6998732089996338, + "eval_runtime": 37.1079, + "eval_samples_per_second": 26.948, + "eval_steps_per_second": 1.132, + "step": 1080 + }, + { + "epoch": 1.3648989898989898, + "grad_norm": 3.020059823989868, + "learning_rate": 3.0636108512628626e-06, + "loss": 1.5331, + "step": 1081 + }, + { + "epoch": 1.3661616161616161, + "grad_norm": 2.805593729019165, + "learning_rate": 3.0612722170252574e-06, + "loss": 1.4867, + "step": 1082 + }, + { + "epoch": 1.3674242424242424, + "grad_norm": 2.965871810913086, + "learning_rate": 3.0589335827876522e-06, + "loss": 1.4494, + "step": 1083 + }, + { + "epoch": 1.3686868686868687, + "grad_norm": 2.8010783195495605, + "learning_rate": 3.0565949485500467e-06, + "loss": 1.4484, + "step": 1084 + }, + { + "epoch": 1.369949494949495, + "grad_norm": 2.985258102416992, + "learning_rate": 3.0542563143124415e-06, + "loss": 1.5029, + "step": 1085 + }, + { + "epoch": 1.371212121212121, + "grad_norm": 2.7528188228607178, + "learning_rate": 3.0519176800748368e-06, + "loss": 1.4695, + "step": 1086 + }, + { + "epoch": 1.3724747474747474, + "grad_norm": 2.9123780727386475, + "learning_rate": 3.0495790458372316e-06, + "loss": 1.5527, + "step": 1087 + }, + { + "epoch": 1.3737373737373737, + "grad_norm": 2.8589160442352295, + "learning_rate": 3.0472404115996265e-06, + "loss": 1.5104, + "step": 1088 + }, + { + "epoch": 1.375, + "grad_norm": 2.853546142578125, + "learning_rate": 3.044901777362021e-06, + "loss": 1.4498, + "step": 1089 + }, + { + "epoch": 1.3762626262626263, + "grad_norm": 2.7989797592163086, + "learning_rate": 3.0425631431244157e-06, + "loss": 1.4494, + "step": 1090 + }, + { + "epoch": 1.3762626262626263, + "eval_loss": 1.6923459768295288, + "eval_runtime": 37.2917, + "eval_samples_per_second": 26.816, + "eval_steps_per_second": 1.126, + "step": 1090 + }, + { + "epoch": 1.3775252525252526, + "grad_norm": 2.8750412464141846, + "learning_rate": 3.0402245088868106e-06, + "loss": 1.5026, + "step": 1091 + }, + { + "epoch": 1.378787878787879, + "grad_norm": 2.8185842037200928, + "learning_rate": 3.037885874649205e-06, + "loss": 1.4859, + "step": 1092 + }, + { + "epoch": 1.380050505050505, + "grad_norm": 2.736332893371582, + "learning_rate": 3.0355472404116e-06, + "loss": 1.4944, + "step": 1093 + }, + { + "epoch": 1.3813131313131313, + "grad_norm": 2.93930983543396, + "learning_rate": 3.0332086061739947e-06, + "loss": 1.4975, + "step": 1094 + }, + { + "epoch": 1.3825757575757576, + "grad_norm": 2.964146852493286, + "learning_rate": 3.0308699719363895e-06, + "loss": 1.4861, + "step": 1095 + }, + { + "epoch": 1.3838383838383839, + "grad_norm": 2.8321545124053955, + "learning_rate": 3.028531337698784e-06, + "loss": 1.4535, + "step": 1096 + }, + { + "epoch": 1.3851010101010102, + "grad_norm": 2.921372413635254, + "learning_rate": 3.0261927034611788e-06, + "loss": 1.5242, + "step": 1097 + }, + { + "epoch": 1.3863636363636362, + "grad_norm": 2.7309911251068115, + "learning_rate": 3.0238540692235736e-06, + "loss": 1.4105, + "step": 1098 + }, + { + "epoch": 1.3876262626262625, + "grad_norm": 2.768899917602539, + "learning_rate": 3.021515434985968e-06, + "loss": 1.4183, + "step": 1099 + }, + { + "epoch": 1.3888888888888888, + "grad_norm": 2.883399248123169, + "learning_rate": 3.019176800748363e-06, + "loss": 1.5307, + "step": 1100 + }, + { + "epoch": 1.3888888888888888, + "eval_loss": 1.6948082447052002, + "eval_runtime": 37.488, + "eval_samples_per_second": 26.675, + "eval_steps_per_second": 1.12, + "step": 1100 + }, + { + "epoch": 1.3901515151515151, + "grad_norm": 2.8610544204711914, + "learning_rate": 3.016838166510758e-06, + "loss": 1.5473, + "step": 1101 + }, + { + "epoch": 1.3914141414141414, + "grad_norm": 2.762694835662842, + "learning_rate": 3.014499532273153e-06, + "loss": 1.4179, + "step": 1102 + }, + { + "epoch": 1.3926767676767677, + "grad_norm": 2.975828170776367, + "learning_rate": 3.012160898035548e-06, + "loss": 1.4959, + "step": 1103 + }, + { + "epoch": 1.393939393939394, + "grad_norm": 3.0633926391601562, + "learning_rate": 3.0098222637979422e-06, + "loss": 1.4979, + "step": 1104 + }, + { + "epoch": 1.3952020202020203, + "grad_norm": 2.972900867462158, + "learning_rate": 3.007483629560337e-06, + "loss": 1.5311, + "step": 1105 + }, + { + "epoch": 1.3964646464646464, + "grad_norm": 2.936033248901367, + "learning_rate": 3.005144995322732e-06, + "loss": 1.4856, + "step": 1106 + }, + { + "epoch": 1.3977272727272727, + "grad_norm": 3.0375452041625977, + "learning_rate": 3.0028063610851268e-06, + "loss": 1.5001, + "step": 1107 + }, + { + "epoch": 1.398989898989899, + "grad_norm": 2.998283624649048, + "learning_rate": 3.000467726847521e-06, + "loss": 1.5219, + "step": 1108 + }, + { + "epoch": 1.4002525252525253, + "grad_norm": 2.831354856491089, + "learning_rate": 2.998129092609916e-06, + "loss": 1.4498, + "step": 1109 + }, + { + "epoch": 1.4015151515151514, + "grad_norm": 2.975390911102295, + "learning_rate": 2.995790458372311e-06, + "loss": 1.4655, + "step": 1110 + }, + { + "epoch": 1.4015151515151514, + "eval_loss": 1.6940916776657104, + "eval_runtime": 36.9469, + "eval_samples_per_second": 27.066, + "eval_steps_per_second": 1.137, + "step": 1110 + }, + { + "epoch": 1.4027777777777777, + "grad_norm": 2.9645442962646484, + "learning_rate": 2.9934518241347053e-06, + "loss": 1.4568, + "step": 1111 + }, + { + "epoch": 1.404040404040404, + "grad_norm": 2.873746633529663, + "learning_rate": 2.9911131898971e-06, + "loss": 1.4483, + "step": 1112 + }, + { + "epoch": 1.4053030303030303, + "grad_norm": 3.1041767597198486, + "learning_rate": 2.988774555659495e-06, + "loss": 1.4914, + "step": 1113 + }, + { + "epoch": 1.4065656565656566, + "grad_norm": 3.0122971534729004, + "learning_rate": 2.98643592142189e-06, + "loss": 1.4494, + "step": 1114 + }, + { + "epoch": 1.4078282828282829, + "grad_norm": 2.842233180999756, + "learning_rate": 2.9840972871842842e-06, + "loss": 1.4274, + "step": 1115 + }, + { + "epoch": 1.4090909090909092, + "grad_norm": 3.058265447616577, + "learning_rate": 2.9817586529466795e-06, + "loss": 1.4637, + "step": 1116 + }, + { + "epoch": 1.4103535353535355, + "grad_norm": 3.06136155128479, + "learning_rate": 2.9794200187090744e-06, + "loss": 1.4791, + "step": 1117 + }, + { + "epoch": 1.4116161616161615, + "grad_norm": 2.8931167125701904, + "learning_rate": 2.977081384471469e-06, + "loss": 1.4476, + "step": 1118 + }, + { + "epoch": 1.4128787878787878, + "grad_norm": 2.8025107383728027, + "learning_rate": 2.974742750233864e-06, + "loss": 1.4891, + "step": 1119 + }, + { + "epoch": 1.4141414141414141, + "grad_norm": 3.1213603019714355, + "learning_rate": 2.9724041159962585e-06, + "loss": 1.5513, + "step": 1120 + }, + { + "epoch": 1.4141414141414141, + "eval_loss": 1.6972154378890991, + "eval_runtime": 37.0863, + "eval_samples_per_second": 26.964, + "eval_steps_per_second": 1.132, + "step": 1120 + }, + { + "epoch": 1.4154040404040404, + "grad_norm": 3.126791477203369, + "learning_rate": 2.9700654817586533e-06, + "loss": 1.4952, + "step": 1121 + }, + { + "epoch": 1.4166666666666667, + "grad_norm": 2.959385395050049, + "learning_rate": 2.967726847521048e-06, + "loss": 1.5098, + "step": 1122 + }, + { + "epoch": 1.4179292929292928, + "grad_norm": 2.907566547393799, + "learning_rate": 2.9653882132834426e-06, + "loss": 1.4959, + "step": 1123 + }, + { + "epoch": 1.4191919191919191, + "grad_norm": 2.9181833267211914, + "learning_rate": 2.9630495790458374e-06, + "loss": 1.4607, + "step": 1124 + }, + { + "epoch": 1.4204545454545454, + "grad_norm": 2.8703668117523193, + "learning_rate": 2.9607109448082323e-06, + "loss": 1.5215, + "step": 1125 + }, + { + "epoch": 1.4217171717171717, + "grad_norm": 3.166255235671997, + "learning_rate": 2.958372310570627e-06, + "loss": 1.4451, + "step": 1126 + }, + { + "epoch": 1.422979797979798, + "grad_norm": 2.863943576812744, + "learning_rate": 2.9560336763330215e-06, + "loss": 1.4481, + "step": 1127 + }, + { + "epoch": 1.4242424242424243, + "grad_norm": 2.975482702255249, + "learning_rate": 2.9536950420954164e-06, + "loss": 1.5118, + "step": 1128 + }, + { + "epoch": 1.4255050505050506, + "grad_norm": 3.077357769012451, + "learning_rate": 2.951356407857811e-06, + "loss": 1.4678, + "step": 1129 + }, + { + "epoch": 1.4267676767676767, + "grad_norm": 3.0375332832336426, + "learning_rate": 2.9490177736202056e-06, + "loss": 1.5036, + "step": 1130 + }, + { + "epoch": 1.4267676767676767, + "eval_loss": 1.6973206996917725, + "eval_runtime": 37.1448, + "eval_samples_per_second": 26.922, + "eval_steps_per_second": 1.131, + "step": 1130 + }, + { + "epoch": 1.428030303030303, + "grad_norm": 3.022068500518799, + "learning_rate": 2.9466791393826005e-06, + "loss": 1.442, + "step": 1131 + }, + { + "epoch": 1.4292929292929293, + "grad_norm": 2.9089183807373047, + "learning_rate": 2.9443405051449957e-06, + "loss": 1.4627, + "step": 1132 + }, + { + "epoch": 1.4305555555555556, + "grad_norm": 2.840149402618408, + "learning_rate": 2.9420018709073906e-06, + "loss": 1.4745, + "step": 1133 + }, + { + "epoch": 1.4318181818181819, + "grad_norm": 2.938650608062744, + "learning_rate": 2.9396632366697854e-06, + "loss": 1.4795, + "step": 1134 + }, + { + "epoch": 1.433080808080808, + "grad_norm": 2.9602437019348145, + "learning_rate": 2.93732460243218e-06, + "loss": 1.4228, + "step": 1135 + }, + { + "epoch": 1.4343434343434343, + "grad_norm": 2.929257869720459, + "learning_rate": 2.9349859681945747e-06, + "loss": 1.5016, + "step": 1136 + }, + { + "epoch": 1.4356060606060606, + "grad_norm": 2.8120505809783936, + "learning_rate": 2.9326473339569695e-06, + "loss": 1.474, + "step": 1137 + }, + { + "epoch": 1.4368686868686869, + "grad_norm": 3.0380513668060303, + "learning_rate": 2.9303086997193644e-06, + "loss": 1.4714, + "step": 1138 + }, + { + "epoch": 1.4381313131313131, + "grad_norm": 2.91106915473938, + "learning_rate": 2.9279700654817588e-06, + "loss": 1.4558, + "step": 1139 + }, + { + "epoch": 1.4393939393939394, + "grad_norm": 2.9970154762268066, + "learning_rate": 2.9256314312441536e-06, + "loss": 1.4895, + "step": 1140 + }, + { + "epoch": 1.4393939393939394, + "eval_loss": 1.6916719675064087, + "eval_runtime": 37.1538, + "eval_samples_per_second": 26.915, + "eval_steps_per_second": 1.13, + "step": 1140 + }, + { + "epoch": 1.4406565656565657, + "grad_norm": 3.041616201400757, + "learning_rate": 2.9232927970065485e-06, + "loss": 1.5177, + "step": 1141 + }, + { + "epoch": 1.441919191919192, + "grad_norm": 2.875385046005249, + "learning_rate": 2.920954162768943e-06, + "loss": 1.4697, + "step": 1142 + }, + { + "epoch": 1.4431818181818181, + "grad_norm": 2.7117881774902344, + "learning_rate": 2.9186155285313377e-06, + "loss": 1.4809, + "step": 1143 + }, + { + "epoch": 1.4444444444444444, + "grad_norm": 2.912123203277588, + "learning_rate": 2.9162768942937326e-06, + "loss": 1.4032, + "step": 1144 + }, + { + "epoch": 1.4457070707070707, + "grad_norm": 2.798440933227539, + "learning_rate": 2.9139382600561274e-06, + "loss": 1.4442, + "step": 1145 + }, + { + "epoch": 1.446969696969697, + "grad_norm": 2.9299135208129883, + "learning_rate": 2.911599625818522e-06, + "loss": 1.4412, + "step": 1146 + }, + { + "epoch": 1.4482323232323233, + "grad_norm": 2.988434076309204, + "learning_rate": 2.909260991580917e-06, + "loss": 1.4473, + "step": 1147 + }, + { + "epoch": 1.4494949494949494, + "grad_norm": 3.0697057247161865, + "learning_rate": 2.906922357343312e-06, + "loss": 1.5151, + "step": 1148 + }, + { + "epoch": 1.4507575757575757, + "grad_norm": 2.9009311199188232, + "learning_rate": 2.9045837231057068e-06, + "loss": 1.4434, + "step": 1149 + }, + { + "epoch": 1.452020202020202, + "grad_norm": 2.9949986934661865, + "learning_rate": 2.902245088868101e-06, + "loss": 1.5436, + "step": 1150 + }, + { + "epoch": 1.452020202020202, + "eval_loss": 1.6880143880844116, + "eval_runtime": 37.1212, + "eval_samples_per_second": 26.939, + "eval_steps_per_second": 1.131, + "step": 1150 + }, + { + "epoch": 1.4532828282828283, + "grad_norm": 2.7827775478363037, + "learning_rate": 2.899906454630496e-06, + "loss": 1.4284, + "step": 1151 + }, + { + "epoch": 1.4545454545454546, + "grad_norm": 2.8879382610321045, + "learning_rate": 2.897567820392891e-06, + "loss": 1.4772, + "step": 1152 + }, + { + "epoch": 1.4558080808080809, + "grad_norm": 3.0119130611419678, + "learning_rate": 2.8952291861552857e-06, + "loss": 1.4546, + "step": 1153 + }, + { + "epoch": 1.4570707070707072, + "grad_norm": 2.9002952575683594, + "learning_rate": 2.89289055191768e-06, + "loss": 1.4898, + "step": 1154 + }, + { + "epoch": 1.4583333333333333, + "grad_norm": 2.8384580612182617, + "learning_rate": 2.890551917680075e-06, + "loss": 1.4685, + "step": 1155 + }, + { + "epoch": 1.4595959595959596, + "grad_norm": 2.9149580001831055, + "learning_rate": 2.88821328344247e-06, + "loss": 1.5133, + "step": 1156 + }, + { + "epoch": 1.4608585858585859, + "grad_norm": 3.041875123977661, + "learning_rate": 2.8858746492048647e-06, + "loss": 1.5038, + "step": 1157 + }, + { + "epoch": 1.4621212121212122, + "grad_norm": 2.9623730182647705, + "learning_rate": 2.883536014967259e-06, + "loss": 1.4776, + "step": 1158 + }, + { + "epoch": 1.4633838383838385, + "grad_norm": 2.935295343399048, + "learning_rate": 2.881197380729654e-06, + "loss": 1.4166, + "step": 1159 + }, + { + "epoch": 1.4646464646464645, + "grad_norm": 2.9556751251220703, + "learning_rate": 2.8788587464920488e-06, + "loss": 1.5132, + "step": 1160 + }, + { + "epoch": 1.4646464646464645, + "eval_loss": 1.69142746925354, + "eval_runtime": 37.1425, + "eval_samples_per_second": 26.923, + "eval_steps_per_second": 1.131, + "step": 1160 + }, + { + "epoch": 1.4659090909090908, + "grad_norm": 2.992587089538574, + "learning_rate": 2.876520112254443e-06, + "loss": 1.4695, + "step": 1161 + }, + { + "epoch": 1.4671717171717171, + "grad_norm": 3.005990982055664, + "learning_rate": 2.8741814780168385e-06, + "loss": 1.5122, + "step": 1162 + }, + { + "epoch": 1.4684343434343434, + "grad_norm": 2.9341115951538086, + "learning_rate": 2.8718428437792333e-06, + "loss": 1.4916, + "step": 1163 + }, + { + "epoch": 1.4696969696969697, + "grad_norm": 2.9665164947509766, + "learning_rate": 2.869504209541628e-06, + "loss": 1.4827, + "step": 1164 + }, + { + "epoch": 1.470959595959596, + "grad_norm": 2.9972221851348877, + "learning_rate": 2.867165575304023e-06, + "loss": 1.4845, + "step": 1165 + }, + { + "epoch": 1.4722222222222223, + "grad_norm": 2.981387138366699, + "learning_rate": 2.8648269410664174e-06, + "loss": 1.4374, + "step": 1166 + }, + { + "epoch": 1.4734848484848486, + "grad_norm": 3.04587459564209, + "learning_rate": 2.8624883068288123e-06, + "loss": 1.431, + "step": 1167 + }, + { + "epoch": 1.4747474747474747, + "grad_norm": 3.194389820098877, + "learning_rate": 2.860149672591207e-06, + "loss": 1.4847, + "step": 1168 + }, + { + "epoch": 1.476010101010101, + "grad_norm": 2.949721574783325, + "learning_rate": 2.857811038353602e-06, + "loss": 1.4596, + "step": 1169 + }, + { + "epoch": 1.4772727272727273, + "grad_norm": 3.0032384395599365, + "learning_rate": 2.8554724041159964e-06, + "loss": 1.4821, + "step": 1170 + }, + { + "epoch": 1.4772727272727273, + "eval_loss": 1.6863130331039429, + "eval_runtime": 37.1386, + "eval_samples_per_second": 26.926, + "eval_steps_per_second": 1.131, + "step": 1170 + }, + { + "epoch": 1.4785353535353536, + "grad_norm": 2.943141460418701, + "learning_rate": 2.853133769878391e-06, + "loss": 1.4998, + "step": 1171 + }, + { + "epoch": 1.4797979797979797, + "grad_norm": 3.0163350105285645, + "learning_rate": 2.850795135640786e-06, + "loss": 1.4565, + "step": 1172 + }, + { + "epoch": 1.481060606060606, + "grad_norm": 2.9713706970214844, + "learning_rate": 2.8484565014031805e-06, + "loss": 1.4663, + "step": 1173 + }, + { + "epoch": 1.4823232323232323, + "grad_norm": 2.9195785522460938, + "learning_rate": 2.8461178671655753e-06, + "loss": 1.4594, + "step": 1174 + }, + { + "epoch": 1.4835858585858586, + "grad_norm": 3.127619504928589, + "learning_rate": 2.84377923292797e-06, + "loss": 1.4742, + "step": 1175 + }, + { + "epoch": 1.4848484848484849, + "grad_norm": 3.0264782905578613, + "learning_rate": 2.841440598690365e-06, + "loss": 1.4366, + "step": 1176 + }, + { + "epoch": 1.4861111111111112, + "grad_norm": 3.0541677474975586, + "learning_rate": 2.8391019644527603e-06, + "loss": 1.4629, + "step": 1177 + }, + { + "epoch": 1.4873737373737375, + "grad_norm": 2.8967537879943848, + "learning_rate": 2.8367633302151547e-06, + "loss": 1.4281, + "step": 1178 + }, + { + "epoch": 1.4886363636363638, + "grad_norm": 2.975773811340332, + "learning_rate": 2.8344246959775495e-06, + "loss": 1.4272, + "step": 1179 + }, + { + "epoch": 1.4898989898989898, + "grad_norm": 2.9768712520599365, + "learning_rate": 2.8320860617399444e-06, + "loss": 1.453, + "step": 1180 + }, + { + "epoch": 1.4898989898989898, + "eval_loss": 1.6810754537582397, + "eval_runtime": 37.1659, + "eval_samples_per_second": 26.906, + "eval_steps_per_second": 1.13, + "step": 1180 + }, + { + "epoch": 1.4911616161616161, + "grad_norm": 2.8897762298583984, + "learning_rate": 2.829747427502339e-06, + "loss": 1.4577, + "step": 1181 + }, + { + "epoch": 1.4924242424242424, + "grad_norm": 2.848259925842285, + "learning_rate": 2.8274087932647336e-06, + "loss": 1.4323, + "step": 1182 + }, + { + "epoch": 1.4936868686868687, + "grad_norm": 2.9275119304656982, + "learning_rate": 2.8250701590271285e-06, + "loss": 1.4712, + "step": 1183 + }, + { + "epoch": 1.494949494949495, + "grad_norm": 2.8643906116485596, + "learning_rate": 2.8227315247895233e-06, + "loss": 1.4375, + "step": 1184 + }, + { + "epoch": 1.496212121212121, + "grad_norm": 3.0250704288482666, + "learning_rate": 2.8203928905519177e-06, + "loss": 1.5482, + "step": 1185 + }, + { + "epoch": 1.4974747474747474, + "grad_norm": 3.1019556522369385, + "learning_rate": 2.8180542563143126e-06, + "loss": 1.4618, + "step": 1186 + }, + { + "epoch": 1.4987373737373737, + "grad_norm": 3.1480300426483154, + "learning_rate": 2.8157156220767074e-06, + "loss": 1.4152, + "step": 1187 + }, + { + "epoch": 1.5, + "grad_norm": 2.82671856880188, + "learning_rate": 2.8133769878391023e-06, + "loss": 1.4692, + "step": 1188 + }, + { + "epoch": 1.5012626262626263, + "grad_norm": 3.2128899097442627, + "learning_rate": 2.8110383536014967e-06, + "loss": 1.4087, + "step": 1189 + }, + { + "epoch": 1.5025252525252526, + "grad_norm": 3.1939210891723633, + "learning_rate": 2.8086997193638915e-06, + "loss": 1.4919, + "step": 1190 + }, + { + "epoch": 1.5025252525252526, + "eval_loss": 1.6861544847488403, + "eval_runtime": 37.3328, + "eval_samples_per_second": 26.786, + "eval_steps_per_second": 1.125, + "step": 1190 + }, + { + "epoch": 1.503787878787879, + "grad_norm": 3.022059679031372, + "learning_rate": 2.8063610851262864e-06, + "loss": 1.4865, + "step": 1191 + }, + { + "epoch": 1.5050505050505052, + "grad_norm": 3.0068860054016113, + "learning_rate": 2.804022450888681e-06, + "loss": 1.478, + "step": 1192 + }, + { + "epoch": 1.5063131313131313, + "grad_norm": 3.2898383140563965, + "learning_rate": 2.801683816651076e-06, + "loss": 1.4773, + "step": 1193 + }, + { + "epoch": 1.5075757575757576, + "grad_norm": 3.090339422225952, + "learning_rate": 2.799345182413471e-06, + "loss": 1.4012, + "step": 1194 + }, + { + "epoch": 1.5088383838383839, + "grad_norm": 2.977215528488159, + "learning_rate": 2.7970065481758657e-06, + "loss": 1.4651, + "step": 1195 + }, + { + "epoch": 1.51010101010101, + "grad_norm": 3.1349570751190186, + "learning_rate": 2.7946679139382606e-06, + "loss": 1.5572, + "step": 1196 + }, + { + "epoch": 1.5113636363636362, + "grad_norm": 2.8715298175811768, + "learning_rate": 2.792329279700655e-06, + "loss": 1.4558, + "step": 1197 + }, + { + "epoch": 1.5126262626262625, + "grad_norm": 2.9848744869232178, + "learning_rate": 2.78999064546305e-06, + "loss": 1.4383, + "step": 1198 + }, + { + "epoch": 1.5138888888888888, + "grad_norm": 3.0774083137512207, + "learning_rate": 2.7876520112254447e-06, + "loss": 1.4607, + "step": 1199 + }, + { + "epoch": 1.5151515151515151, + "grad_norm": 2.9721343517303467, + "learning_rate": 2.785313376987839e-06, + "loss": 1.4341, + "step": 1200 + }, + { + "epoch": 1.5151515151515151, + "eval_loss": 1.6859924793243408, + "eval_runtime": 37.4589, + "eval_samples_per_second": 26.696, + "eval_steps_per_second": 1.121, + "step": 1200 + }, + { + "epoch": 1.5164141414141414, + "grad_norm": 3.041095733642578, + "learning_rate": 2.782974742750234e-06, + "loss": 1.4736, + "step": 1201 + }, + { + "epoch": 1.5176767676767677, + "grad_norm": 2.8400142192840576, + "learning_rate": 2.780636108512629e-06, + "loss": 1.4169, + "step": 1202 + }, + { + "epoch": 1.518939393939394, + "grad_norm": 2.785592555999756, + "learning_rate": 2.7782974742750236e-06, + "loss": 1.4553, + "step": 1203 + }, + { + "epoch": 1.5202020202020203, + "grad_norm": 2.75209641456604, + "learning_rate": 2.775958840037418e-06, + "loss": 1.4267, + "step": 1204 + }, + { + "epoch": 1.5214646464646466, + "grad_norm": 2.927558183670044, + "learning_rate": 2.773620205799813e-06, + "loss": 1.4377, + "step": 1205 + }, + { + "epoch": 1.5227272727272727, + "grad_norm": 2.9335508346557617, + "learning_rate": 2.7712815715622077e-06, + "loss": 1.4295, + "step": 1206 + }, + { + "epoch": 1.523989898989899, + "grad_norm": 2.7812745571136475, + "learning_rate": 2.7689429373246026e-06, + "loss": 1.4029, + "step": 1207 + }, + { + "epoch": 1.5252525252525253, + "grad_norm": 3.230062961578369, + "learning_rate": 2.766604303086998e-06, + "loss": 1.539, + "step": 1208 + }, + { + "epoch": 1.5265151515151514, + "grad_norm": 3.1244332790374756, + "learning_rate": 2.7642656688493923e-06, + "loss": 1.4829, + "step": 1209 + }, + { + "epoch": 1.5277777777777777, + "grad_norm": 3.2391650676727295, + "learning_rate": 2.761927034611787e-06, + "loss": 1.4677, + "step": 1210 + }, + { + "epoch": 1.5277777777777777, + "eval_loss": 1.6849907636642456, + "eval_runtime": 36.8638, + "eval_samples_per_second": 27.127, + "eval_steps_per_second": 1.139, + "step": 1210 + }, + { + "epoch": 1.529040404040404, + "grad_norm": 3.1884987354278564, + "learning_rate": 2.759588400374182e-06, + "loss": 1.4185, + "step": 1211 + }, + { + "epoch": 1.5303030303030303, + "grad_norm": 2.959118366241455, + "learning_rate": 2.7572497661365764e-06, + "loss": 1.4536, + "step": 1212 + }, + { + "epoch": 1.5315656565656566, + "grad_norm": 2.882380247116089, + "learning_rate": 2.7549111318989712e-06, + "loss": 1.4872, + "step": 1213 + }, + { + "epoch": 1.5328282828282829, + "grad_norm": 3.0574429035186768, + "learning_rate": 2.752572497661366e-06, + "loss": 1.4343, + "step": 1214 + }, + { + "epoch": 1.5340909090909092, + "grad_norm": 3.1645233631134033, + "learning_rate": 2.750233863423761e-06, + "loss": 1.5101, + "step": 1215 + }, + { + "epoch": 1.5353535353535355, + "grad_norm": 2.83034086227417, + "learning_rate": 2.7478952291861553e-06, + "loss": 1.45, + "step": 1216 + }, + { + "epoch": 1.5366161616161618, + "grad_norm": 2.964780330657959, + "learning_rate": 2.74555659494855e-06, + "loss": 1.453, + "step": 1217 + }, + { + "epoch": 1.5378787878787878, + "grad_norm": 2.9896774291992188, + "learning_rate": 2.743217960710945e-06, + "loss": 1.4648, + "step": 1218 + }, + { + "epoch": 1.5391414141414141, + "grad_norm": 3.013664484024048, + "learning_rate": 2.74087932647334e-06, + "loss": 1.4828, + "step": 1219 + }, + { + "epoch": 1.5404040404040404, + "grad_norm": 3.0011773109436035, + "learning_rate": 2.7385406922357343e-06, + "loss": 1.4705, + "step": 1220 + }, + { + "epoch": 1.5404040404040404, + "eval_loss": 1.6872731447219849, + "eval_runtime": 36.9864, + "eval_samples_per_second": 27.037, + "eval_steps_per_second": 1.136, + "step": 1220 + }, + { + "epoch": 1.5416666666666665, + "grad_norm": 3.239349842071533, + "learning_rate": 2.736202057998129e-06, + "loss": 1.484, + "step": 1221 + }, + { + "epoch": 1.5429292929292928, + "grad_norm": 2.9267003536224365, + "learning_rate": 2.733863423760524e-06, + "loss": 1.4587, + "step": 1222 + }, + { + "epoch": 1.5441919191919191, + "grad_norm": 2.890094518661499, + "learning_rate": 2.7315247895229192e-06, + "loss": 1.4456, + "step": 1223 + }, + { + "epoch": 1.5454545454545454, + "grad_norm": 3.0287511348724365, + "learning_rate": 2.7291861552853136e-06, + "loss": 1.5059, + "step": 1224 + }, + { + "epoch": 1.5467171717171717, + "grad_norm": 3.041609525680542, + "learning_rate": 2.7268475210477085e-06, + "loss": 1.4341, + "step": 1225 + }, + { + "epoch": 1.547979797979798, + "grad_norm": 2.8771862983703613, + "learning_rate": 2.7245088868101033e-06, + "loss": 1.4308, + "step": 1226 + }, + { + "epoch": 1.5492424242424243, + "grad_norm": 3.1173794269561768, + "learning_rate": 2.722170252572498e-06, + "loss": 1.514, + "step": 1227 + }, + { + "epoch": 1.5505050505050506, + "grad_norm": 2.838275671005249, + "learning_rate": 2.7198316183348926e-06, + "loss": 1.4798, + "step": 1228 + }, + { + "epoch": 1.551767676767677, + "grad_norm": 2.9917166233062744, + "learning_rate": 2.7174929840972874e-06, + "loss": 1.5256, + "step": 1229 + }, + { + "epoch": 1.553030303030303, + "grad_norm": 3.071840763092041, + "learning_rate": 2.7151543498596823e-06, + "loss": 1.5187, + "step": 1230 + }, + { + "epoch": 1.553030303030303, + "eval_loss": 1.6850661039352417, + "eval_runtime": 37.0094, + "eval_samples_per_second": 27.02, + "eval_steps_per_second": 1.135, + "step": 1230 + }, + { + "epoch": 1.5542929292929293, + "grad_norm": 3.253169536590576, + "learning_rate": 2.7128157156220767e-06, + "loss": 1.5522, + "step": 1231 + }, + { + "epoch": 1.5555555555555556, + "grad_norm": 3.037151336669922, + "learning_rate": 2.7104770813844715e-06, + "loss": 1.5125, + "step": 1232 + }, + { + "epoch": 1.5568181818181817, + "grad_norm": 2.8486485481262207, + "learning_rate": 2.7081384471468664e-06, + "loss": 1.4169, + "step": 1233 + }, + { + "epoch": 1.558080808080808, + "grad_norm": 2.953824996948242, + "learning_rate": 2.7057998129092612e-06, + "loss": 1.4505, + "step": 1234 + }, + { + "epoch": 1.5593434343434343, + "grad_norm": 3.012303590774536, + "learning_rate": 2.7034611786716557e-06, + "loss": 1.4816, + "step": 1235 + }, + { + "epoch": 1.5606060606060606, + "grad_norm": 3.1798768043518066, + "learning_rate": 2.7011225444340505e-06, + "loss": 1.5116, + "step": 1236 + }, + { + "epoch": 1.5618686868686869, + "grad_norm": 3.11370587348938, + "learning_rate": 2.6987839101964453e-06, + "loss": 1.5035, + "step": 1237 + }, + { + "epoch": 1.5631313131313131, + "grad_norm": 3.0563342571258545, + "learning_rate": 2.69644527595884e-06, + "loss": 1.4387, + "step": 1238 + }, + { + "epoch": 1.5643939393939394, + "grad_norm": 2.9884026050567627, + "learning_rate": 2.6941066417212354e-06, + "loss": 1.428, + "step": 1239 + }, + { + "epoch": 1.5656565656565657, + "grad_norm": 2.955773115158081, + "learning_rate": 2.69176800748363e-06, + "loss": 1.4208, + "step": 1240 + }, + { + "epoch": 1.5656565656565657, + "eval_loss": 1.685523509979248, + "eval_runtime": 37.0248, + "eval_samples_per_second": 27.009, + "eval_steps_per_second": 1.134, + "step": 1240 + }, + { + "epoch": 1.566919191919192, + "grad_norm": 2.9858005046844482, + "learning_rate": 2.6894293732460247e-06, + "loss": 1.4706, + "step": 1241 + }, + { + "epoch": 1.5681818181818183, + "grad_norm": 3.122340440750122, + "learning_rate": 2.6870907390084196e-06, + "loss": 1.4892, + "step": 1242 + }, + { + "epoch": 1.5694444444444444, + "grad_norm": 2.9828262329101562, + "learning_rate": 2.684752104770814e-06, + "loss": 1.4255, + "step": 1243 + }, + { + "epoch": 1.5707070707070707, + "grad_norm": 2.916653633117676, + "learning_rate": 2.682413470533209e-06, + "loss": 1.4241, + "step": 1244 + }, + { + "epoch": 1.571969696969697, + "grad_norm": 3.1615540981292725, + "learning_rate": 2.6800748362956037e-06, + "loss": 1.4719, + "step": 1245 + }, + { + "epoch": 1.573232323232323, + "grad_norm": 3.154130697250366, + "learning_rate": 2.6777362020579985e-06, + "loss": 1.4731, + "step": 1246 + }, + { + "epoch": 1.5744949494949494, + "grad_norm": 3.040848731994629, + "learning_rate": 2.675397567820393e-06, + "loss": 1.5149, + "step": 1247 + }, + { + "epoch": 1.5757575757575757, + "grad_norm": 2.9808318614959717, + "learning_rate": 2.6730589335827878e-06, + "loss": 1.431, + "step": 1248 + }, + { + "epoch": 1.577020202020202, + "grad_norm": 2.8538382053375244, + "learning_rate": 2.6707202993451826e-06, + "loss": 1.4458, + "step": 1249 + }, + { + "epoch": 1.5782828282828283, + "grad_norm": 3.113945960998535, + "learning_rate": 2.6683816651075774e-06, + "loss": 1.4873, + "step": 1250 + }, + { + "epoch": 1.5782828282828283, + "eval_loss": 1.6823841333389282, + "eval_runtime": 37.0384, + "eval_samples_per_second": 26.999, + "eval_steps_per_second": 1.134, + "step": 1250 + }, + { + "epoch": 1.5795454545454546, + "grad_norm": 3.0044517517089844, + "learning_rate": 2.666043030869972e-06, + "loss": 1.4904, + "step": 1251 + }, + { + "epoch": 1.5808080808080809, + "grad_norm": 2.845820665359497, + "learning_rate": 2.6637043966323667e-06, + "loss": 1.4174, + "step": 1252 + }, + { + "epoch": 1.5820707070707072, + "grad_norm": 2.9123542308807373, + "learning_rate": 2.6613657623947616e-06, + "loss": 1.4896, + "step": 1253 + }, + { + "epoch": 1.5833333333333335, + "grad_norm": 3.1593222618103027, + "learning_rate": 2.659027128157157e-06, + "loss": 1.5208, + "step": 1254 + }, + { + "epoch": 1.5845959595959596, + "grad_norm": 3.0782854557037354, + "learning_rate": 2.6566884939195512e-06, + "loss": 1.4911, + "step": 1255 + }, + { + "epoch": 1.5858585858585859, + "grad_norm": 2.77309513092041, + "learning_rate": 2.654349859681946e-06, + "loss": 1.4475, + "step": 1256 + }, + { + "epoch": 1.5871212121212122, + "grad_norm": 2.9154622554779053, + "learning_rate": 2.652011225444341e-06, + "loss": 1.4324, + "step": 1257 + }, + { + "epoch": 1.5883838383838382, + "grad_norm": 3.107395887374878, + "learning_rate": 2.6496725912067358e-06, + "loss": 1.5126, + "step": 1258 + }, + { + "epoch": 1.5896464646464645, + "grad_norm": 2.825025796890259, + "learning_rate": 2.64733395696913e-06, + "loss": 1.4723, + "step": 1259 + }, + { + "epoch": 1.5909090909090908, + "grad_norm": 2.833296060562134, + "learning_rate": 2.644995322731525e-06, + "loss": 1.4516, + "step": 1260 + }, + { + "epoch": 1.5909090909090908, + "eval_loss": 1.6950432062149048, + "eval_runtime": 37.0343, + "eval_samples_per_second": 27.002, + "eval_steps_per_second": 1.134, + "step": 1260 + }, + { + "epoch": 1.5921717171717171, + "grad_norm": 2.9508421421051025, + "learning_rate": 2.64265668849392e-06, + "loss": 1.4002, + "step": 1261 + }, + { + "epoch": 1.5934343434343434, + "grad_norm": 2.8778491020202637, + "learning_rate": 2.6403180542563143e-06, + "loss": 1.4258, + "step": 1262 + }, + { + "epoch": 1.5946969696969697, + "grad_norm": 2.81261944770813, + "learning_rate": 2.637979420018709e-06, + "loss": 1.435, + "step": 1263 + }, + { + "epoch": 1.595959595959596, + "grad_norm": 3.1811511516571045, + "learning_rate": 2.635640785781104e-06, + "loss": 1.4792, + "step": 1264 + }, + { + "epoch": 1.5972222222222223, + "grad_norm": 3.0168304443359375, + "learning_rate": 2.633302151543499e-06, + "loss": 1.5394, + "step": 1265 + }, + { + "epoch": 1.5984848484848486, + "grad_norm": 2.9531311988830566, + "learning_rate": 2.6309635173058932e-06, + "loss": 1.4775, + "step": 1266 + }, + { + "epoch": 1.5997474747474747, + "grad_norm": 2.835839033126831, + "learning_rate": 2.628624883068288e-06, + "loss": 1.478, + "step": 1267 + }, + { + "epoch": 1.601010101010101, + "grad_norm": 2.9014039039611816, + "learning_rate": 2.626286248830683e-06, + "loss": 1.4926, + "step": 1268 + }, + { + "epoch": 1.6022727272727273, + "grad_norm": 2.89461088180542, + "learning_rate": 2.623947614593078e-06, + "loss": 1.448, + "step": 1269 + }, + { + "epoch": 1.6035353535353534, + "grad_norm": 2.8005120754241943, + "learning_rate": 2.621608980355473e-06, + "loss": 1.4733, + "step": 1270 + }, + { + "epoch": 1.6035353535353534, + "eval_loss": 1.6889415979385376, + "eval_runtime": 37.0345, + "eval_samples_per_second": 27.002, + "eval_steps_per_second": 1.134, + "step": 1270 + }, + { + "epoch": 1.6047979797979797, + "grad_norm": 2.9576175212860107, + "learning_rate": 2.6192703461178675e-06, + "loss": 1.5116, + "step": 1271 + }, + { + "epoch": 1.606060606060606, + "grad_norm": 2.926638603210449, + "learning_rate": 2.6169317118802623e-06, + "loss": 1.4744, + "step": 1272 + }, + { + "epoch": 1.6073232323232323, + "grad_norm": 2.911587715148926, + "learning_rate": 2.614593077642657e-06, + "loss": 1.4695, + "step": 1273 + }, + { + "epoch": 1.6085858585858586, + "grad_norm": 2.8878750801086426, + "learning_rate": 2.6122544434050516e-06, + "loss": 1.4574, + "step": 1274 + }, + { + "epoch": 1.6098484848484849, + "grad_norm": 2.8489067554473877, + "learning_rate": 2.6099158091674464e-06, + "loss": 1.4599, + "step": 1275 + }, + { + "epoch": 1.6111111111111112, + "grad_norm": 2.8223719596862793, + "learning_rate": 2.6075771749298412e-06, + "loss": 1.4431, + "step": 1276 + }, + { + "epoch": 1.6123737373737375, + "grad_norm": 2.869910955429077, + "learning_rate": 2.605238540692236e-06, + "loss": 1.48, + "step": 1277 + }, + { + "epoch": 1.6136363636363638, + "grad_norm": 2.8435168266296387, + "learning_rate": 2.6028999064546305e-06, + "loss": 1.4471, + "step": 1278 + }, + { + "epoch": 1.61489898989899, + "grad_norm": 2.9384665489196777, + "learning_rate": 2.6005612722170253e-06, + "loss": 1.5176, + "step": 1279 + }, + { + "epoch": 1.6161616161616161, + "grad_norm": 3.021785020828247, + "learning_rate": 2.59822263797942e-06, + "loss": 1.4789, + "step": 1280 + }, + { + "epoch": 1.6161616161616161, + "eval_loss": 1.6864516735076904, + "eval_runtime": 37.0249, + "eval_samples_per_second": 27.009, + "eval_steps_per_second": 1.134, + "step": 1280 + }, + { + "epoch": 1.6174242424242424, + "grad_norm": 3.007183313369751, + "learning_rate": 2.5958840037418146e-06, + "loss": 1.4686, + "step": 1281 + }, + { + "epoch": 1.6186868686868687, + "grad_norm": 3.115910053253174, + "learning_rate": 2.5935453695042095e-06, + "loss": 1.5382, + "step": 1282 + }, + { + "epoch": 1.6199494949494948, + "grad_norm": 2.98108172416687, + "learning_rate": 2.5912067352666043e-06, + "loss": 1.4569, + "step": 1283 + }, + { + "epoch": 1.621212121212121, + "grad_norm": 2.926832437515259, + "learning_rate": 2.5888681010289996e-06, + "loss": 1.4515, + "step": 1284 + }, + { + "epoch": 1.6224747474747474, + "grad_norm": 2.9045534133911133, + "learning_rate": 2.5865294667913944e-06, + "loss": 1.4259, + "step": 1285 + }, + { + "epoch": 1.6237373737373737, + "grad_norm": 3.0757768154144287, + "learning_rate": 2.584190832553789e-06, + "loss": 1.5436, + "step": 1286 + }, + { + "epoch": 1.625, + "grad_norm": 3.045257806777954, + "learning_rate": 2.5818521983161837e-06, + "loss": 1.4378, + "step": 1287 + }, + { + "epoch": 1.6262626262626263, + "grad_norm": 2.8131601810455322, + "learning_rate": 2.5795135640785785e-06, + "loss": 1.4224, + "step": 1288 + }, + { + "epoch": 1.6275252525252526, + "grad_norm": 2.953326940536499, + "learning_rate": 2.5771749298409734e-06, + "loss": 1.4137, + "step": 1289 + }, + { + "epoch": 1.628787878787879, + "grad_norm": 2.946232318878174, + "learning_rate": 2.5748362956033678e-06, + "loss": 1.4575, + "step": 1290 + }, + { + "epoch": 1.628787878787879, + "eval_loss": 1.6896334886550903, + "eval_runtime": 37.317, + "eval_samples_per_second": 26.797, + "eval_steps_per_second": 1.125, + "step": 1290 + }, + { + "epoch": 1.6300505050505052, + "grad_norm": 2.860867738723755, + "learning_rate": 2.5724976613657626e-06, + "loss": 1.4412, + "step": 1291 + }, + { + "epoch": 1.6313131313131313, + "grad_norm": 3.0417261123657227, + "learning_rate": 2.5701590271281575e-06, + "loss": 1.4919, + "step": 1292 + }, + { + "epoch": 1.6325757575757576, + "grad_norm": 3.21425199508667, + "learning_rate": 2.567820392890552e-06, + "loss": 1.5243, + "step": 1293 + }, + { + "epoch": 1.6338383838383839, + "grad_norm": 3.000302314758301, + "learning_rate": 2.5654817586529467e-06, + "loss": 1.4571, + "step": 1294 + }, + { + "epoch": 1.63510101010101, + "grad_norm": 2.88415789604187, + "learning_rate": 2.5631431244153416e-06, + "loss": 1.4527, + "step": 1295 + }, + { + "epoch": 1.6363636363636362, + "grad_norm": 2.7747156620025635, + "learning_rate": 2.5608044901777364e-06, + "loss": 1.3984, + "step": 1296 + }, + { + "epoch": 1.6376262626262625, + "grad_norm": 2.9298319816589355, + "learning_rate": 2.558465855940131e-06, + "loss": 1.4587, + "step": 1297 + }, + { + "epoch": 1.6388888888888888, + "grad_norm": 3.0199968814849854, + "learning_rate": 2.5561272217025257e-06, + "loss": 1.4539, + "step": 1298 + }, + { + "epoch": 1.6401515151515151, + "grad_norm": 3.09503436088562, + "learning_rate": 2.5537885874649205e-06, + "loss": 1.4551, + "step": 1299 + }, + { + "epoch": 1.6414141414141414, + "grad_norm": 2.972161293029785, + "learning_rate": 2.5514499532273158e-06, + "loss": 1.4123, + "step": 1300 + }, + { + "epoch": 1.6414141414141414, + "eval_loss": 1.6877377033233643, + "eval_runtime": 37.4088, + "eval_samples_per_second": 26.732, + "eval_steps_per_second": 1.123, + "step": 1300 + }, + { + "epoch": 1.6426767676767677, + "grad_norm": 3.081058979034424, + "learning_rate": 2.5491113189897106e-06, + "loss": 1.457, + "step": 1301 + }, + { + "epoch": 1.643939393939394, + "grad_norm": 3.040106773376465, + "learning_rate": 2.546772684752105e-06, + "loss": 1.4465, + "step": 1302 + }, + { + "epoch": 1.6452020202020203, + "grad_norm": 3.133812427520752, + "learning_rate": 2.5444340505145e-06, + "loss": 1.4382, + "step": 1303 + }, + { + "epoch": 1.6464646464646466, + "grad_norm": 2.9701125621795654, + "learning_rate": 2.5420954162768947e-06, + "loss": 1.457, + "step": 1304 + }, + { + "epoch": 1.6477272727272727, + "grad_norm": 2.9272050857543945, + "learning_rate": 2.539756782039289e-06, + "loss": 1.464, + "step": 1305 + }, + { + "epoch": 1.648989898989899, + "grad_norm": 3.048401117324829, + "learning_rate": 2.537418147801684e-06, + "loss": 1.4688, + "step": 1306 + }, + { + "epoch": 1.6502525252525253, + "grad_norm": 2.9103102684020996, + "learning_rate": 2.535079513564079e-06, + "loss": 1.5025, + "step": 1307 + }, + { + "epoch": 1.6515151515151514, + "grad_norm": 3.0841548442840576, + "learning_rate": 2.5327408793264737e-06, + "loss": 1.4938, + "step": 1308 + }, + { + "epoch": 1.6527777777777777, + "grad_norm": 3.022184371948242, + "learning_rate": 2.530402245088868e-06, + "loss": 1.4332, + "step": 1309 + }, + { + "epoch": 1.654040404040404, + "grad_norm": 2.881479024887085, + "learning_rate": 2.528063610851263e-06, + "loss": 1.4541, + "step": 1310 + }, + { + "epoch": 1.654040404040404, + "eval_loss": 1.6814675331115723, + "eval_runtime": 36.9446, + "eval_samples_per_second": 27.068, + "eval_steps_per_second": 1.137, + "step": 1310 + }, + { + "epoch": 1.6553030303030303, + "grad_norm": 2.906719446182251, + "learning_rate": 2.5257249766136578e-06, + "loss": 1.4968, + "step": 1311 + }, + { + "epoch": 1.6565656565656566, + "grad_norm": 3.0566461086273193, + "learning_rate": 2.523386342376052e-06, + "loss": 1.4513, + "step": 1312 + }, + { + "epoch": 1.6578282828282829, + "grad_norm": 2.895026206970215, + "learning_rate": 2.521047708138447e-06, + "loss": 1.4217, + "step": 1313 + }, + { + "epoch": 1.6590909090909092, + "grad_norm": 3.0006465911865234, + "learning_rate": 2.518709073900842e-06, + "loss": 1.4793, + "step": 1314 + }, + { + "epoch": 1.6603535353535355, + "grad_norm": 2.9853575229644775, + "learning_rate": 2.516370439663237e-06, + "loss": 1.4642, + "step": 1315 + }, + { + "epoch": 1.6616161616161618, + "grad_norm": 2.9729392528533936, + "learning_rate": 2.514031805425632e-06, + "loss": 1.4644, + "step": 1316 + }, + { + "epoch": 1.6628787878787878, + "grad_norm": 3.0997586250305176, + "learning_rate": 2.5116931711880264e-06, + "loss": 1.4392, + "step": 1317 + }, + { + "epoch": 1.6641414141414141, + "grad_norm": 2.9621379375457764, + "learning_rate": 2.5093545369504213e-06, + "loss": 1.4387, + "step": 1318 + }, + { + "epoch": 1.6654040404040404, + "grad_norm": 3.0571911334991455, + "learning_rate": 2.507015902712816e-06, + "loss": 1.4638, + "step": 1319 + }, + { + "epoch": 1.6666666666666665, + "grad_norm": 2.9580321311950684, + "learning_rate": 2.504677268475211e-06, + "loss": 1.4985, + "step": 1320 + }, + { + "epoch": 1.6666666666666665, + "eval_loss": 1.6790977716445923, + "eval_runtime": 37.0242, + "eval_samples_per_second": 27.009, + "eval_steps_per_second": 1.134, + "step": 1320 + }, + { + "epoch": 1.6679292929292928, + "grad_norm": 2.87432599067688, + "learning_rate": 2.5023386342376054e-06, + "loss": 1.4058, + "step": 1321 + }, + { + "epoch": 1.6691919191919191, + "grad_norm": 2.925546407699585, + "learning_rate": 2.5e-06, + "loss": 1.4986, + "step": 1322 + }, + { + "epoch": 1.6704545454545454, + "grad_norm": 2.837743043899536, + "learning_rate": 2.497661365762395e-06, + "loss": 1.455, + "step": 1323 + }, + { + "epoch": 1.6717171717171717, + "grad_norm": 3.111581563949585, + "learning_rate": 2.4953227315247895e-06, + "loss": 1.5005, + "step": 1324 + }, + { + "epoch": 1.672979797979798, + "grad_norm": 2.851076126098633, + "learning_rate": 2.4929840972871843e-06, + "loss": 1.4365, + "step": 1325 + }, + { + "epoch": 1.6742424242424243, + "grad_norm": 2.966965675354004, + "learning_rate": 2.4906454630495796e-06, + "loss": 1.4697, + "step": 1326 + }, + { + "epoch": 1.6755050505050506, + "grad_norm": 3.015193223953247, + "learning_rate": 2.488306828811974e-06, + "loss": 1.4486, + "step": 1327 + }, + { + "epoch": 1.676767676767677, + "grad_norm": 3.017681121826172, + "learning_rate": 2.485968194574369e-06, + "loss": 1.4017, + "step": 1328 + }, + { + "epoch": 1.678030303030303, + "grad_norm": 3.071981906890869, + "learning_rate": 2.4836295603367637e-06, + "loss": 1.4572, + "step": 1329 + }, + { + "epoch": 1.6792929292929293, + "grad_norm": 3.051741123199463, + "learning_rate": 2.481290926099158e-06, + "loss": 1.4498, + "step": 1330 + }, + { + "epoch": 1.6792929292929293, + "eval_loss": 1.6771777868270874, + "eval_runtime": 37.0747, + "eval_samples_per_second": 26.973, + "eval_steps_per_second": 1.133, + "step": 1330 + }, + { + "epoch": 1.6805555555555556, + "grad_norm": 2.9561071395874023, + "learning_rate": 2.478952291861553e-06, + "loss": 1.4428, + "step": 1331 + }, + { + "epoch": 1.6818181818181817, + "grad_norm": 2.992584466934204, + "learning_rate": 2.4766136576239478e-06, + "loss": 1.461, + "step": 1332 + }, + { + "epoch": 1.683080808080808, + "grad_norm": 3.0473921298980713, + "learning_rate": 2.4742750233863426e-06, + "loss": 1.489, + "step": 1333 + }, + { + "epoch": 1.6843434343434343, + "grad_norm": 2.8721725940704346, + "learning_rate": 2.4719363891487375e-06, + "loss": 1.487, + "step": 1334 + }, + { + "epoch": 1.6856060606060606, + "grad_norm": 2.8650765419006348, + "learning_rate": 2.4695977549111323e-06, + "loss": 1.4504, + "step": 1335 + }, + { + "epoch": 1.6868686868686869, + "grad_norm": 2.998887538909912, + "learning_rate": 2.4672591206735267e-06, + "loss": 1.4586, + "step": 1336 + }, + { + "epoch": 1.6881313131313131, + "grad_norm": 2.915188789367676, + "learning_rate": 2.4649204864359216e-06, + "loss": 1.4333, + "step": 1337 + }, + { + "epoch": 1.6893939393939394, + "grad_norm": 2.9660091400146484, + "learning_rate": 2.4625818521983164e-06, + "loss": 1.4711, + "step": 1338 + }, + { + "epoch": 1.6906565656565657, + "grad_norm": 2.916274070739746, + "learning_rate": 2.4602432179607113e-06, + "loss": 1.4642, + "step": 1339 + }, + { + "epoch": 1.691919191919192, + "grad_norm": 3.04154896736145, + "learning_rate": 2.4579045837231057e-06, + "loss": 1.4233, + "step": 1340 + }, + { + "epoch": 1.691919191919192, + "eval_loss": 1.6764302253723145, + "eval_runtime": 37.0686, + "eval_samples_per_second": 26.977, + "eval_steps_per_second": 1.133, + "step": 1340 + }, + { + "epoch": 1.6931818181818183, + "grad_norm": 2.9976465702056885, + "learning_rate": 2.4555659494855005e-06, + "loss": 1.4775, + "step": 1341 + }, + { + "epoch": 1.6944444444444444, + "grad_norm": 2.9140090942382812, + "learning_rate": 2.4532273152478954e-06, + "loss": 1.4532, + "step": 1342 + }, + { + "epoch": 1.6957070707070707, + "grad_norm": 3.0224454402923584, + "learning_rate": 2.45088868101029e-06, + "loss": 1.5091, + "step": 1343 + }, + { + "epoch": 1.696969696969697, + "grad_norm": 2.8979439735412598, + "learning_rate": 2.448550046772685e-06, + "loss": 1.4634, + "step": 1344 + }, + { + "epoch": 1.698232323232323, + "grad_norm": 2.913433313369751, + "learning_rate": 2.44621141253508e-06, + "loss": 1.4483, + "step": 1345 + }, + { + "epoch": 1.6994949494949494, + "grad_norm": 2.813455820083618, + "learning_rate": 2.4438727782974743e-06, + "loss": 1.446, + "step": 1346 + }, + { + "epoch": 1.7007575757575757, + "grad_norm": 2.988722324371338, + "learning_rate": 2.441534144059869e-06, + "loss": 1.4908, + "step": 1347 + }, + { + "epoch": 1.702020202020202, + "grad_norm": 2.850446939468384, + "learning_rate": 2.439195509822264e-06, + "loss": 1.4221, + "step": 1348 + }, + { + "epoch": 1.7032828282828283, + "grad_norm": 2.8405508995056152, + "learning_rate": 2.436856875584659e-06, + "loss": 1.4496, + "step": 1349 + }, + { + "epoch": 1.7045454545454546, + "grad_norm": 3.0188651084899902, + "learning_rate": 2.4345182413470537e-06, + "loss": 1.489, + "step": 1350 + }, + { + "epoch": 1.7045454545454546, + "eval_loss": 1.6780083179473877, + "eval_runtime": 37.0724, + "eval_samples_per_second": 26.974, + "eval_steps_per_second": 1.133, + "step": 1350 + }, + { + "epoch": 1.7058080808080809, + "grad_norm": 2.905088424682617, + "learning_rate": 2.4321796071094485e-06, + "loss": 1.4283, + "step": 1351 + }, + { + "epoch": 1.7070707070707072, + "grad_norm": 2.878998279571533, + "learning_rate": 2.429840972871843e-06, + "loss": 1.4357, + "step": 1352 + }, + { + "epoch": 1.7083333333333335, + "grad_norm": 3.1707355976104736, + "learning_rate": 2.427502338634238e-06, + "loss": 1.496, + "step": 1353 + }, + { + "epoch": 1.7095959595959596, + "grad_norm": 2.9610493183135986, + "learning_rate": 2.4251637043966326e-06, + "loss": 1.4905, + "step": 1354 + }, + { + "epoch": 1.7108585858585859, + "grad_norm": 2.8372201919555664, + "learning_rate": 2.422825070159027e-06, + "loss": 1.4563, + "step": 1355 + }, + { + "epoch": 1.7121212121212122, + "grad_norm": 2.8741753101348877, + "learning_rate": 2.420486435921422e-06, + "loss": 1.4525, + "step": 1356 + }, + { + "epoch": 1.7133838383838382, + "grad_norm": 2.973726749420166, + "learning_rate": 2.418147801683817e-06, + "loss": 1.4878, + "step": 1357 + }, + { + "epoch": 1.7146464646464645, + "grad_norm": 3.05717396736145, + "learning_rate": 2.4158091674462116e-06, + "loss": 1.5402, + "step": 1358 + }, + { + "epoch": 1.7159090909090908, + "grad_norm": 2.883772373199463, + "learning_rate": 2.4134705332086064e-06, + "loss": 1.4329, + "step": 1359 + }, + { + "epoch": 1.7171717171717171, + "grad_norm": 2.8246653079986572, + "learning_rate": 2.4111318989710013e-06, + "loss": 1.4821, + "step": 1360 + }, + { + "epoch": 1.7171717171717171, + "eval_loss": 1.6737746000289917, + "eval_runtime": 37.0666, + "eval_samples_per_second": 26.978, + "eval_steps_per_second": 1.133, + "step": 1360 + }, + { + "epoch": 1.7184343434343434, + "grad_norm": 2.905977964401245, + "learning_rate": 2.4087932647333957e-06, + "loss": 1.4378, + "step": 1361 + }, + { + "epoch": 1.7196969696969697, + "grad_norm": 2.9268195629119873, + "learning_rate": 2.4064546304957905e-06, + "loss": 1.4765, + "step": 1362 + }, + { + "epoch": 1.720959595959596, + "grad_norm": 3.1369476318359375, + "learning_rate": 2.4041159962581854e-06, + "loss": 1.4532, + "step": 1363 + }, + { + "epoch": 1.7222222222222223, + "grad_norm": 3.04546856880188, + "learning_rate": 2.4017773620205802e-06, + "loss": 1.4921, + "step": 1364 + }, + { + "epoch": 1.7234848484848486, + "grad_norm": 2.872852325439453, + "learning_rate": 2.399438727782975e-06, + "loss": 1.4323, + "step": 1365 + }, + { + "epoch": 1.7247474747474747, + "grad_norm": 2.961681604385376, + "learning_rate": 2.39710009354537e-06, + "loss": 1.5181, + "step": 1366 + }, + { + "epoch": 1.726010101010101, + "grad_norm": 2.997729778289795, + "learning_rate": 2.3947614593077643e-06, + "loss": 1.4817, + "step": 1367 + }, + { + "epoch": 1.7272727272727273, + "grad_norm": 2.9097466468811035, + "learning_rate": 2.392422825070159e-06, + "loss": 1.4478, + "step": 1368 + }, + { + "epoch": 1.7285353535353534, + "grad_norm": 2.82433819770813, + "learning_rate": 2.390084190832554e-06, + "loss": 1.4065, + "step": 1369 + }, + { + "epoch": 1.7297979797979797, + "grad_norm": 2.940910577774048, + "learning_rate": 2.387745556594949e-06, + "loss": 1.4735, + "step": 1370 + }, + { + "epoch": 1.7297979797979797, + "eval_loss": 1.6717814207077026, + "eval_runtime": 37.0937, + "eval_samples_per_second": 26.959, + "eval_steps_per_second": 1.132, + "step": 1370 + }, + { + "epoch": 1.731060606060606, + "grad_norm": 2.79101300239563, + "learning_rate": 2.3854069223573433e-06, + "loss": 1.4545, + "step": 1371 + }, + { + "epoch": 1.7323232323232323, + "grad_norm": 2.916954517364502, + "learning_rate": 2.3830682881197385e-06, + "loss": 1.4293, + "step": 1372 + }, + { + "epoch": 1.7335858585858586, + "grad_norm": 2.9609909057617188, + "learning_rate": 2.380729653882133e-06, + "loss": 1.4578, + "step": 1373 + }, + { + "epoch": 1.7348484848484849, + "grad_norm": 2.935925245285034, + "learning_rate": 2.378391019644528e-06, + "loss": 1.4577, + "step": 1374 + }, + { + "epoch": 1.7361111111111112, + "grad_norm": 2.9803526401519775, + "learning_rate": 2.3760523854069226e-06, + "loss": 1.4449, + "step": 1375 + }, + { + "epoch": 1.7373737373737375, + "grad_norm": 3.0292913913726807, + "learning_rate": 2.3737137511693175e-06, + "loss": 1.4308, + "step": 1376 + }, + { + "epoch": 1.7386363636363638, + "grad_norm": 3.0247879028320312, + "learning_rate": 2.371375116931712e-06, + "loss": 1.5406, + "step": 1377 + }, + { + "epoch": 1.73989898989899, + "grad_norm": 2.86775541305542, + "learning_rate": 2.3690364826941067e-06, + "loss": 1.4324, + "step": 1378 + }, + { + "epoch": 1.7411616161616161, + "grad_norm": 2.8593552112579346, + "learning_rate": 2.3666978484565016e-06, + "loss": 1.4526, + "step": 1379 + }, + { + "epoch": 1.7424242424242424, + "grad_norm": 2.8735320568084717, + "learning_rate": 2.3643592142188964e-06, + "loss": 1.4097, + "step": 1380 + }, + { + "epoch": 1.7424242424242424, + "eval_loss": 1.675827145576477, + "eval_runtime": 37.1206, + "eval_samples_per_second": 26.939, + "eval_steps_per_second": 1.131, + "step": 1380 + }, + { + "epoch": 1.7436868686868687, + "grad_norm": 2.948153495788574, + "learning_rate": 2.3620205799812913e-06, + "loss": 1.4234, + "step": 1381 + }, + { + "epoch": 1.7449494949494948, + "grad_norm": 2.9344282150268555, + "learning_rate": 2.359681945743686e-06, + "loss": 1.4653, + "step": 1382 + }, + { + "epoch": 1.746212121212121, + "grad_norm": 2.9309439659118652, + "learning_rate": 2.3573433115060805e-06, + "loss": 1.4676, + "step": 1383 + }, + { + "epoch": 1.7474747474747474, + "grad_norm": 3.0762970447540283, + "learning_rate": 2.3550046772684754e-06, + "loss": 1.4407, + "step": 1384 + }, + { + "epoch": 1.7487373737373737, + "grad_norm": 3.149824857711792, + "learning_rate": 2.3526660430308702e-06, + "loss": 1.4629, + "step": 1385 + }, + { + "epoch": 1.75, + "grad_norm": 3.109571695327759, + "learning_rate": 2.3503274087932646e-06, + "loss": 1.494, + "step": 1386 + }, + { + "epoch": 1.7512626262626263, + "grad_norm": 2.8359758853912354, + "learning_rate": 2.34798877455566e-06, + "loss": 1.4056, + "step": 1387 + }, + { + "epoch": 1.7525252525252526, + "grad_norm": 2.8523781299591064, + "learning_rate": 2.3456501403180548e-06, + "loss": 1.4389, + "step": 1388 + }, + { + "epoch": 1.753787878787879, + "grad_norm": 3.0218093395233154, + "learning_rate": 2.343311506080449e-06, + "loss": 1.4587, + "step": 1389 + }, + { + "epoch": 1.7550505050505052, + "grad_norm": 3.0975496768951416, + "learning_rate": 2.340972871842844e-06, + "loss": 1.5133, + "step": 1390 + }, + { + "epoch": 1.7550505050505052, + "eval_loss": 1.671931266784668, + "eval_runtime": 37.2749, + "eval_samples_per_second": 26.828, + "eval_steps_per_second": 1.127, + "step": 1390 + }, + { + "epoch": 1.7563131313131313, + "grad_norm": 3.0177836418151855, + "learning_rate": 2.338634237605239e-06, + "loss": 1.4716, + "step": 1391 + }, + { + "epoch": 1.7575757575757576, + "grad_norm": 3.049105167388916, + "learning_rate": 2.3362956033676333e-06, + "loss": 1.4858, + "step": 1392 + }, + { + "epoch": 1.7588383838383839, + "grad_norm": 3.0062026977539062, + "learning_rate": 2.333956969130028e-06, + "loss": 1.4239, + "step": 1393 + }, + { + "epoch": 1.76010101010101, + "grad_norm": 3.071772813796997, + "learning_rate": 2.331618334892423e-06, + "loss": 1.4654, + "step": 1394 + }, + { + "epoch": 1.7613636363636362, + "grad_norm": 3.140510320663452, + "learning_rate": 2.329279700654818e-06, + "loss": 1.4854, + "step": 1395 + }, + { + "epoch": 1.7626262626262625, + "grad_norm": 2.9903907775878906, + "learning_rate": 2.3269410664172126e-06, + "loss": 1.4406, + "step": 1396 + }, + { + "epoch": 1.7638888888888888, + "grad_norm": 2.839891195297241, + "learning_rate": 2.3246024321796075e-06, + "loss": 1.4498, + "step": 1397 + }, + { + "epoch": 1.7651515151515151, + "grad_norm": 3.0252437591552734, + "learning_rate": 2.322263797942002e-06, + "loss": 1.472, + "step": 1398 + }, + { + "epoch": 1.7664141414141414, + "grad_norm": 3.0214762687683105, + "learning_rate": 2.3199251637043968e-06, + "loss": 1.4031, + "step": 1399 + }, + { + "epoch": 1.7676767676767677, + "grad_norm": 3.190088987350464, + "learning_rate": 2.3175865294667916e-06, + "loss": 1.4497, + "step": 1400 + }, + { + "epoch": 1.7676767676767677, + "eval_loss": 1.673120379447937, + "eval_runtime": 37.4414, + "eval_samples_per_second": 26.708, + "eval_steps_per_second": 1.122, + "step": 1400 + }, + { + "epoch": 1.768939393939394, + "grad_norm": 2.9620201587677, + "learning_rate": 2.3152478952291864e-06, + "loss": 1.4149, + "step": 1401 + }, + { + "epoch": 1.7702020202020203, + "grad_norm": 3.112372875213623, + "learning_rate": 2.312909260991581e-06, + "loss": 1.4569, + "step": 1402 + }, + { + "epoch": 1.7714646464646466, + "grad_norm": 3.138857364654541, + "learning_rate": 2.310570626753976e-06, + "loss": 1.4819, + "step": 1403 + }, + { + "epoch": 1.7727272727272727, + "grad_norm": 3.0957937240600586, + "learning_rate": 2.3082319925163705e-06, + "loss": 1.4374, + "step": 1404 + }, + { + "epoch": 1.773989898989899, + "grad_norm": 3.014039993286133, + "learning_rate": 2.3058933582787654e-06, + "loss": 1.4262, + "step": 1405 + }, + { + "epoch": 1.7752525252525253, + "grad_norm": 2.8127005100250244, + "learning_rate": 2.3035547240411602e-06, + "loss": 1.4468, + "step": 1406 + }, + { + "epoch": 1.7765151515151514, + "grad_norm": 2.9180614948272705, + "learning_rate": 2.301216089803555e-06, + "loss": 1.4259, + "step": 1407 + }, + { + "epoch": 1.7777777777777777, + "grad_norm": 3.0376124382019043, + "learning_rate": 2.2988774555659495e-06, + "loss": 1.4628, + "step": 1408 + }, + { + "epoch": 1.779040404040404, + "grad_norm": 3.031740665435791, + "learning_rate": 2.2965388213283443e-06, + "loss": 1.5058, + "step": 1409 + }, + { + "epoch": 1.7803030303030303, + "grad_norm": 2.9876646995544434, + "learning_rate": 2.294200187090739e-06, + "loss": 1.4605, + "step": 1410 + }, + { + "epoch": 1.7803030303030303, + "eval_loss": 1.668077826499939, + "eval_runtime": 36.9734, + "eval_samples_per_second": 27.046, + "eval_steps_per_second": 1.136, + "step": 1410 + }, + { + "epoch": 1.7815656565656566, + "grad_norm": 2.953744888305664, + "learning_rate": 2.291861552853134e-06, + "loss": 1.4257, + "step": 1411 + }, + { + "epoch": 1.7828282828282829, + "grad_norm": 3.049036741256714, + "learning_rate": 2.289522918615529e-06, + "loss": 1.4476, + "step": 1412 + }, + { + "epoch": 1.7840909090909092, + "grad_norm": 3.102116584777832, + "learning_rate": 2.2871842843779237e-06, + "loss": 1.4701, + "step": 1413 + }, + { + "epoch": 1.7853535353535355, + "grad_norm": 2.842970132827759, + "learning_rate": 2.284845650140318e-06, + "loss": 1.415, + "step": 1414 + }, + { + "epoch": 1.7866161616161618, + "grad_norm": 2.9386725425720215, + "learning_rate": 2.282507015902713e-06, + "loss": 1.4561, + "step": 1415 + }, + { + "epoch": 1.7878787878787878, + "grad_norm": 2.942713499069214, + "learning_rate": 2.280168381665108e-06, + "loss": 1.435, + "step": 1416 + }, + { + "epoch": 1.7891414141414141, + "grad_norm": 2.943105936050415, + "learning_rate": 2.2778297474275022e-06, + "loss": 1.4923, + "step": 1417 + }, + { + "epoch": 1.7904040404040404, + "grad_norm": 2.7094717025756836, + "learning_rate": 2.2754911131898975e-06, + "loss": 1.4457, + "step": 1418 + }, + { + "epoch": 1.7916666666666665, + "grad_norm": 2.7859625816345215, + "learning_rate": 2.2731524789522923e-06, + "loss": 1.3993, + "step": 1419 + }, + { + "epoch": 1.7929292929292928, + "grad_norm": 3.0923988819122314, + "learning_rate": 2.2708138447146868e-06, + "loss": 1.5054, + "step": 1420 + }, + { + "epoch": 1.7929292929292928, + "eval_loss": 1.6687692403793335, + "eval_runtime": 37.0567, + "eval_samples_per_second": 26.986, + "eval_steps_per_second": 1.133, + "step": 1420 + }, + { + "epoch": 1.7941919191919191, + "grad_norm": 2.849726676940918, + "learning_rate": 2.2684752104770816e-06, + "loss": 1.3978, + "step": 1421 + }, + { + "epoch": 1.7954545454545454, + "grad_norm": 3.028932809829712, + "learning_rate": 2.2661365762394764e-06, + "loss": 1.4786, + "step": 1422 + }, + { + "epoch": 1.7967171717171717, + "grad_norm": 2.973188638687134, + "learning_rate": 2.263797942001871e-06, + "loss": 1.4393, + "step": 1423 + }, + { + "epoch": 1.797979797979798, + "grad_norm": 2.884032964706421, + "learning_rate": 2.2614593077642657e-06, + "loss": 1.4197, + "step": 1424 + }, + { + "epoch": 1.7992424242424243, + "grad_norm": 2.9058549404144287, + "learning_rate": 2.2591206735266606e-06, + "loss": 1.4313, + "step": 1425 + }, + { + "epoch": 1.8005050505050506, + "grad_norm": 2.905841827392578, + "learning_rate": 2.2567820392890554e-06, + "loss": 1.439, + "step": 1426 + }, + { + "epoch": 1.801767676767677, + "grad_norm": 2.927028179168701, + "learning_rate": 2.2544434050514502e-06, + "loss": 1.4296, + "step": 1427 + }, + { + "epoch": 1.803030303030303, + "grad_norm": 2.965529203414917, + "learning_rate": 2.252104770813845e-06, + "loss": 1.4185, + "step": 1428 + }, + { + "epoch": 1.8042929292929293, + "grad_norm": 2.9494433403015137, + "learning_rate": 2.2497661365762395e-06, + "loss": 1.5087, + "step": 1429 + }, + { + "epoch": 1.8055555555555556, + "grad_norm": 3.0137765407562256, + "learning_rate": 2.2474275023386343e-06, + "loss": 1.4383, + "step": 1430 + }, + { + "epoch": 1.8055555555555556, + "eval_loss": 1.6728461980819702, + "eval_runtime": 37.1033, + "eval_samples_per_second": 26.952, + "eval_steps_per_second": 1.132, + "step": 1430 + }, + { + "epoch": 1.8068181818181817, + "grad_norm": 2.980062484741211, + "learning_rate": 2.245088868101029e-06, + "loss": 1.4333, + "step": 1431 + }, + { + "epoch": 1.808080808080808, + "grad_norm": 3.143759250640869, + "learning_rate": 2.242750233863424e-06, + "loss": 1.4676, + "step": 1432 + }, + { + "epoch": 1.8093434343434343, + "grad_norm": 3.0134518146514893, + "learning_rate": 2.240411599625819e-06, + "loss": 1.4223, + "step": 1433 + }, + { + "epoch": 1.8106060606060606, + "grad_norm": 3.000018358230591, + "learning_rate": 2.2380729653882137e-06, + "loss": 1.4537, + "step": 1434 + }, + { + "epoch": 1.8118686868686869, + "grad_norm": 2.9828288555145264, + "learning_rate": 2.235734331150608e-06, + "loss": 1.4379, + "step": 1435 + }, + { + "epoch": 1.8131313131313131, + "grad_norm": 2.965486764907837, + "learning_rate": 2.233395696913003e-06, + "loss": 1.4244, + "step": 1436 + }, + { + "epoch": 1.8143939393939394, + "grad_norm": 2.9787070751190186, + "learning_rate": 2.231057062675398e-06, + "loss": 1.4117, + "step": 1437 + }, + { + "epoch": 1.8156565656565657, + "grad_norm": 3.0893537998199463, + "learning_rate": 2.2287184284377927e-06, + "loss": 1.4809, + "step": 1438 + }, + { + "epoch": 1.816919191919192, + "grad_norm": 2.8568801879882812, + "learning_rate": 2.226379794200187e-06, + "loss": 1.4232, + "step": 1439 + }, + { + "epoch": 1.8181818181818183, + "grad_norm": 3.1005046367645264, + "learning_rate": 2.224041159962582e-06, + "loss": 1.482, + "step": 1440 + }, + { + "epoch": 1.8181818181818183, + "eval_loss": 1.6738661527633667, + "eval_runtime": 37.1286, + "eval_samples_per_second": 26.933, + "eval_steps_per_second": 1.131, + "step": 1440 + }, + { + "epoch": 1.8194444444444444, + "grad_norm": 2.9064700603485107, + "learning_rate": 2.2217025257249768e-06, + "loss": 1.4179, + "step": 1441 + }, + { + "epoch": 1.8207070707070707, + "grad_norm": 3.0111782550811768, + "learning_rate": 2.2193638914873716e-06, + "loss": 1.4084, + "step": 1442 + }, + { + "epoch": 1.821969696969697, + "grad_norm": 2.9269816875457764, + "learning_rate": 2.2170252572497665e-06, + "loss": 1.4361, + "step": 1443 + }, + { + "epoch": 1.823232323232323, + "grad_norm": 2.9985902309417725, + "learning_rate": 2.2146866230121613e-06, + "loss": 1.4345, + "step": 1444 + }, + { + "epoch": 1.8244949494949494, + "grad_norm": 3.1306066513061523, + "learning_rate": 2.2123479887745557e-06, + "loss": 1.4728, + "step": 1445 + }, + { + "epoch": 1.8257575757575757, + "grad_norm": 2.9876086711883545, + "learning_rate": 2.2100093545369506e-06, + "loss": 1.378, + "step": 1446 + }, + { + "epoch": 1.827020202020202, + "grad_norm": 3.018763780593872, + "learning_rate": 2.2076707202993454e-06, + "loss": 1.4506, + "step": 1447 + }, + { + "epoch": 1.8282828282828283, + "grad_norm": 2.960573673248291, + "learning_rate": 2.2053320860617402e-06, + "loss": 1.4674, + "step": 1448 + }, + { + "epoch": 1.8295454545454546, + "grad_norm": 2.849881887435913, + "learning_rate": 2.202993451824135e-06, + "loss": 1.4687, + "step": 1449 + }, + { + "epoch": 1.8308080808080809, + "grad_norm": 2.8798882961273193, + "learning_rate": 2.2006548175865295e-06, + "loss": 1.3868, + "step": 1450 + }, + { + "epoch": 1.8308080808080809, + "eval_loss": 1.6657519340515137, + "eval_runtime": 37.0993, + "eval_samples_per_second": 26.955, + "eval_steps_per_second": 1.132, + "step": 1450 + }, + { + "epoch": 1.8320707070707072, + "grad_norm": 2.937471628189087, + "learning_rate": 2.1983161833489243e-06, + "loss": 1.4085, + "step": 1451 + }, + { + "epoch": 1.8333333333333335, + "grad_norm": 3.0815699100494385, + "learning_rate": 2.195977549111319e-06, + "loss": 1.4055, + "step": 1452 + }, + { + "epoch": 1.8345959595959596, + "grad_norm": 2.927917718887329, + "learning_rate": 2.193638914873714e-06, + "loss": 1.4014, + "step": 1453 + }, + { + "epoch": 1.8358585858585859, + "grad_norm": 2.9576570987701416, + "learning_rate": 2.1913002806361085e-06, + "loss": 1.4394, + "step": 1454 + }, + { + "epoch": 1.8371212121212122, + "grad_norm": 2.9505863189697266, + "learning_rate": 2.1889616463985033e-06, + "loss": 1.4018, + "step": 1455 + }, + { + "epoch": 1.8383838383838382, + "grad_norm": 3.014965772628784, + "learning_rate": 2.186623012160898e-06, + "loss": 1.3735, + "step": 1456 + }, + { + "epoch": 1.8396464646464645, + "grad_norm": 3.121122121810913, + "learning_rate": 2.184284377923293e-06, + "loss": 1.4289, + "step": 1457 + }, + { + "epoch": 1.8409090909090908, + "grad_norm": 2.9710142612457275, + "learning_rate": 2.181945743685688e-06, + "loss": 1.4494, + "step": 1458 + }, + { + "epoch": 1.8421717171717171, + "grad_norm": 3.028554916381836, + "learning_rate": 2.1796071094480827e-06, + "loss": 1.5151, + "step": 1459 + }, + { + "epoch": 1.8434343434343434, + "grad_norm": 2.8973751068115234, + "learning_rate": 2.177268475210477e-06, + "loss": 1.4171, + "step": 1460 + }, + { + "epoch": 1.8434343434343434, + "eval_loss": 1.6670986413955688, + "eval_runtime": 37.0937, + "eval_samples_per_second": 26.959, + "eval_steps_per_second": 1.132, + "step": 1460 + }, + { + "epoch": 1.8446969696969697, + "grad_norm": 3.046865701675415, + "learning_rate": 2.174929840972872e-06, + "loss": 1.4016, + "step": 1461 + }, + { + "epoch": 1.845959595959596, + "grad_norm": 3.0509603023529053, + "learning_rate": 2.1725912067352668e-06, + "loss": 1.4196, + "step": 1462 + }, + { + "epoch": 1.8472222222222223, + "grad_norm": 2.951892137527466, + "learning_rate": 2.1702525724976616e-06, + "loss": 1.4063, + "step": 1463 + }, + { + "epoch": 1.8484848484848486, + "grad_norm": 3.2126667499542236, + "learning_rate": 2.1679139382600565e-06, + "loss": 1.5106, + "step": 1464 + }, + { + "epoch": 1.8497474747474747, + "grad_norm": 2.965576648712158, + "learning_rate": 2.1655753040224513e-06, + "loss": 1.4096, + "step": 1465 + }, + { + "epoch": 1.851010101010101, + "grad_norm": 2.8640592098236084, + "learning_rate": 2.1632366697848457e-06, + "loss": 1.389, + "step": 1466 + }, + { + "epoch": 1.8522727272727273, + "grad_norm": 3.1476943492889404, + "learning_rate": 2.1608980355472406e-06, + "loss": 1.4793, + "step": 1467 + }, + { + "epoch": 1.8535353535353534, + "grad_norm": 2.814695119857788, + "learning_rate": 2.1585594013096354e-06, + "loss": 1.4265, + "step": 1468 + }, + { + "epoch": 1.8547979797979797, + "grad_norm": 2.9728431701660156, + "learning_rate": 2.1562207670720302e-06, + "loss": 1.386, + "step": 1469 + }, + { + "epoch": 1.856060606060606, + "grad_norm": 2.9062652587890625, + "learning_rate": 2.1538821328344247e-06, + "loss": 1.3986, + "step": 1470 + }, + { + "epoch": 1.856060606060606, + "eval_loss": 1.6698763370513916, + "eval_runtime": 37.1129, + "eval_samples_per_second": 26.945, + "eval_steps_per_second": 1.132, + "step": 1470 + }, + { + "epoch": 1.8573232323232323, + "grad_norm": 2.950857162475586, + "learning_rate": 2.15154349859682e-06, + "loss": 1.4193, + "step": 1471 + }, + { + "epoch": 1.8585858585858586, + "grad_norm": 3.017594337463379, + "learning_rate": 2.1492048643592144e-06, + "loss": 1.4318, + "step": 1472 + }, + { + "epoch": 1.8598484848484849, + "grad_norm": 2.991514205932617, + "learning_rate": 2.146866230121609e-06, + "loss": 1.4462, + "step": 1473 + }, + { + "epoch": 1.8611111111111112, + "grad_norm": 3.149475336074829, + "learning_rate": 2.144527595884004e-06, + "loss": 1.4486, + "step": 1474 + }, + { + "epoch": 1.8623737373737375, + "grad_norm": 2.948828935623169, + "learning_rate": 2.1421889616463985e-06, + "loss": 1.4522, + "step": 1475 + }, + { + "epoch": 1.8636363636363638, + "grad_norm": 3.0383694171905518, + "learning_rate": 2.1398503274087933e-06, + "loss": 1.45, + "step": 1476 + }, + { + "epoch": 1.86489898989899, + "grad_norm": 3.093449831008911, + "learning_rate": 2.137511693171188e-06, + "loss": 1.4266, + "step": 1477 + }, + { + "epoch": 1.8661616161616161, + "grad_norm": 3.017549753189087, + "learning_rate": 2.135173058933583e-06, + "loss": 1.4067, + "step": 1478 + }, + { + "epoch": 1.8674242424242424, + "grad_norm": 2.985044002532959, + "learning_rate": 2.132834424695978e-06, + "loss": 1.4904, + "step": 1479 + }, + { + "epoch": 1.8686868686868687, + "grad_norm": 2.9600346088409424, + "learning_rate": 2.1304957904583727e-06, + "loss": 1.4006, + "step": 1480 + }, + { + "epoch": 1.8686868686868687, + "eval_loss": 1.6648480892181396, + "eval_runtime": 37.1225, + "eval_samples_per_second": 26.938, + "eval_steps_per_second": 1.131, + "step": 1480 + }, + { + "epoch": 1.8699494949494948, + "grad_norm": 2.8905625343322754, + "learning_rate": 2.128157156220767e-06, + "loss": 1.4096, + "step": 1481 + }, + { + "epoch": 1.871212121212121, + "grad_norm": 2.95342755317688, + "learning_rate": 2.125818521983162e-06, + "loss": 1.3888, + "step": 1482 + }, + { + "epoch": 1.8724747474747474, + "grad_norm": 2.9438536167144775, + "learning_rate": 2.1234798877455568e-06, + "loss": 1.413, + "step": 1483 + }, + { + "epoch": 1.8737373737373737, + "grad_norm": 3.1257364749908447, + "learning_rate": 2.1211412535079516e-06, + "loss": 1.4902, + "step": 1484 + }, + { + "epoch": 1.875, + "grad_norm": 3.1377382278442383, + "learning_rate": 2.118802619270346e-06, + "loss": 1.4815, + "step": 1485 + }, + { + "epoch": 1.8762626262626263, + "grad_norm": 3.122253656387329, + "learning_rate": 2.116463985032741e-06, + "loss": 1.4695, + "step": 1486 + }, + { + "epoch": 1.8775252525252526, + "grad_norm": 3.0128729343414307, + "learning_rate": 2.1141253507951357e-06, + "loss": 1.4673, + "step": 1487 + }, + { + "epoch": 1.878787878787879, + "grad_norm": 2.921835422515869, + "learning_rate": 2.1117867165575306e-06, + "loss": 1.4325, + "step": 1488 + }, + { + "epoch": 1.8800505050505052, + "grad_norm": 3.19289231300354, + "learning_rate": 2.1094480823199254e-06, + "loss": 1.4504, + "step": 1489 + }, + { + "epoch": 1.8813131313131313, + "grad_norm": 2.8579299449920654, + "learning_rate": 2.1071094480823203e-06, + "loss": 1.4042, + "step": 1490 + }, + { + "epoch": 1.8813131313131313, + "eval_loss": 1.6603271961212158, + "eval_runtime": 37.2997, + "eval_samples_per_second": 26.81, + "eval_steps_per_second": 1.126, + "step": 1490 + }, + { + "epoch": 1.8825757575757576, + "grad_norm": 2.9113969802856445, + "learning_rate": 2.1047708138447147e-06, + "loss": 1.4128, + "step": 1491 + }, + { + "epoch": 1.8838383838383839, + "grad_norm": 3.058133363723755, + "learning_rate": 2.1024321796071095e-06, + "loss": 1.4574, + "step": 1492 + }, + { + "epoch": 1.88510101010101, + "grad_norm": 3.0066583156585693, + "learning_rate": 2.1000935453695044e-06, + "loss": 1.4521, + "step": 1493 + }, + { + "epoch": 1.8863636363636362, + "grad_norm": 2.9154908657073975, + "learning_rate": 2.097754911131899e-06, + "loss": 1.4103, + "step": 1494 + }, + { + "epoch": 1.8876262626262625, + "grad_norm": 2.977062702178955, + "learning_rate": 2.095416276894294e-06, + "loss": 1.4003, + "step": 1495 + }, + { + "epoch": 1.8888888888888888, + "grad_norm": 2.9048168659210205, + "learning_rate": 2.093077642656689e-06, + "loss": 1.4216, + "step": 1496 + }, + { + "epoch": 1.8901515151515151, + "grad_norm": 3.008733034133911, + "learning_rate": 2.0907390084190833e-06, + "loss": 1.4516, + "step": 1497 + }, + { + "epoch": 1.8914141414141414, + "grad_norm": 3.0832529067993164, + "learning_rate": 2.088400374181478e-06, + "loss": 1.4248, + "step": 1498 + }, + { + "epoch": 1.8926767676767677, + "grad_norm": 2.848066568374634, + "learning_rate": 2.086061739943873e-06, + "loss": 1.4042, + "step": 1499 + }, + { + "epoch": 1.893939393939394, + "grad_norm": 2.846782684326172, + "learning_rate": 2.0837231057062674e-06, + "loss": 1.409, + "step": 1500 + }, + { + "epoch": 1.893939393939394, + "eval_loss": 1.6672260761260986, + "eval_runtime": 37.4695, + "eval_samples_per_second": 26.688, + "eval_steps_per_second": 1.121, + "step": 1500 + } + ], + "logging_steps": 1, + "max_steps": 2376, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 100, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 4.67596411404288e+17, + "train_batch_size": 3, + "trial_name": null, + "trial_params": null +}