diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,7101 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 5.0, + "eval_steps": 10000, + "global_step": 100000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.01, + "grad_norm": 17.069135665893555, + "learning_rate": 3.96e-06, + "loss": 1.9267, + "step": 100 + }, + { + "epoch": 0.01, + "grad_norm": 7.49233341217041, + "learning_rate": 7.960000000000002e-06, + "loss": 1.7814, + "step": 200 + }, + { + "epoch": 0.01, + "grad_norm": 10.066174507141113, + "learning_rate": 1.196e-05, + "loss": 1.7612, + "step": 300 + }, + { + "epoch": 0.02, + "grad_norm": 8.212127685546875, + "learning_rate": 1.5960000000000003e-05, + "loss": 1.7866, + "step": 400 + }, + { + "epoch": 0.03, + "grad_norm": 5.494411945343018, + "learning_rate": 1.9960000000000002e-05, + "loss": 1.7809, + "step": 500 + }, + { + "epoch": 0.03, + "grad_norm": 6.4578728675842285, + "learning_rate": 1.9980100502512564e-05, + "loss": 1.7893, + "step": 600 + }, + { + "epoch": 0.04, + "grad_norm": 4.508376121520996, + "learning_rate": 1.996020100502513e-05, + "loss": 1.7851, + "step": 700 + }, + { + "epoch": 0.04, + "grad_norm": 6.78291654586792, + "learning_rate": 1.9940100502512564e-05, + "loss": 1.758, + "step": 800 + }, + { + "epoch": 0.04, + "grad_norm": 3.534212589263916, + "learning_rate": 1.9920000000000002e-05, + "loss": 1.7659, + "step": 900 + }, + { + "epoch": 0.05, + "grad_norm": 6.295835018157959, + "learning_rate": 1.9899899497487437e-05, + "loss": 1.7201, + "step": 1000 + }, + { + "epoch": 0.06, + "grad_norm": 5.502695083618164, + "learning_rate": 1.987979899497488e-05, + "loss": 1.7634, + "step": 1100 + }, + { + "epoch": 0.06, + "grad_norm": 6.535002708435059, + "learning_rate": 1.9859698492462313e-05, + "loss": 1.7322, + "step": 1200 + }, + { + "epoch": 0.07, + "grad_norm": 5.8399834632873535, + "learning_rate": 1.983959798994975e-05, + "loss": 1.7306, + "step": 1300 + }, + { + "epoch": 0.07, + "grad_norm": 3.5027928352355957, + "learning_rate": 1.9819497487437185e-05, + "loss": 1.6763, + "step": 1400 + }, + { + "epoch": 0.07, + "grad_norm": 6.470935821533203, + "learning_rate": 1.9799396984924623e-05, + "loss": 1.687, + "step": 1500 + }, + { + "epoch": 0.08, + "grad_norm": 6.546024799346924, + "learning_rate": 1.977929648241206e-05, + "loss": 1.6944, + "step": 1600 + }, + { + "epoch": 0.09, + "grad_norm": 6.186180591583252, + "learning_rate": 1.97591959798995e-05, + "loss": 1.7085, + "step": 1700 + }, + { + "epoch": 0.09, + "grad_norm": 3.819445848464966, + "learning_rate": 1.9739095477386937e-05, + "loss": 1.6765, + "step": 1800 + }, + { + "epoch": 0.1, + "grad_norm": 4.2868876457214355, + "learning_rate": 1.9718994974874372e-05, + "loss": 1.6888, + "step": 1900 + }, + { + "epoch": 0.1, + "grad_norm": 6.690129280090332, + "learning_rate": 1.969889447236181e-05, + "loss": 1.675, + "step": 2000 + }, + { + "epoch": 0.1, + "grad_norm": 7.435989856719971, + "learning_rate": 1.9678793969849248e-05, + "loss": 1.6913, + "step": 2100 + }, + { + "epoch": 0.11, + "grad_norm": 6.209521293640137, + "learning_rate": 1.9658693467336686e-05, + "loss": 1.7021, + "step": 2200 + }, + { + "epoch": 0.12, + "grad_norm": 7.704258441925049, + "learning_rate": 1.963859296482412e-05, + "loss": 1.6973, + "step": 2300 + }, + { + "epoch": 0.12, + "grad_norm": 6.3551025390625, + "learning_rate": 1.9618492462311562e-05, + "loss": 1.6951, + "step": 2400 + }, + { + "epoch": 0.12, + "grad_norm": 4.278153419494629, + "learning_rate": 1.9598391959798996e-05, + "loss": 1.6749, + "step": 2500 + }, + { + "epoch": 0.13, + "grad_norm": 5.053964614868164, + "learning_rate": 1.9578291457286434e-05, + "loss": 1.6839, + "step": 2600 + }, + { + "epoch": 0.14, + "grad_norm": 5.125670909881592, + "learning_rate": 1.955819095477387e-05, + "loss": 1.6734, + "step": 2700 + }, + { + "epoch": 0.14, + "grad_norm": 3.6489973068237305, + "learning_rate": 1.953809045226131e-05, + "loss": 1.6436, + "step": 2800 + }, + { + "epoch": 0.14, + "grad_norm": 6.31420373916626, + "learning_rate": 1.9517989949748745e-05, + "loss": 1.6781, + "step": 2900 + }, + { + "epoch": 0.15, + "grad_norm": 4.799983501434326, + "learning_rate": 1.9497889447236183e-05, + "loss": 1.6766, + "step": 3000 + }, + { + "epoch": 0.15, + "grad_norm": 3.4388930797576904, + "learning_rate": 1.9477788944723618e-05, + "loss": 1.6613, + "step": 3100 + }, + { + "epoch": 0.16, + "grad_norm": 6.100885391235352, + "learning_rate": 1.945768844221106e-05, + "loss": 1.6885, + "step": 3200 + }, + { + "epoch": 0.17, + "grad_norm": 5.916701793670654, + "learning_rate": 1.9437587939698493e-05, + "loss": 1.6705, + "step": 3300 + }, + { + "epoch": 0.17, + "grad_norm": 4.234365463256836, + "learning_rate": 1.941748743718593e-05, + "loss": 1.6803, + "step": 3400 + }, + { + "epoch": 0.17, + "grad_norm": 5.592808723449707, + "learning_rate": 1.939738693467337e-05, + "loss": 1.6222, + "step": 3500 + }, + { + "epoch": 0.18, + "grad_norm": 6.33900260925293, + "learning_rate": 1.9377286432160804e-05, + "loss": 1.6275, + "step": 3600 + }, + { + "epoch": 0.18, + "grad_norm": 5.385664939880371, + "learning_rate": 1.9357185929648242e-05, + "loss": 1.6234, + "step": 3700 + }, + { + "epoch": 0.19, + "grad_norm": 6.868696689605713, + "learning_rate": 1.933708542713568e-05, + "loss": 1.6584, + "step": 3800 + }, + { + "epoch": 0.2, + "grad_norm": 5.236993789672852, + "learning_rate": 1.9316984924623118e-05, + "loss": 1.6342, + "step": 3900 + }, + { + "epoch": 0.2, + "grad_norm": 5.226047515869141, + "learning_rate": 1.9296884422110552e-05, + "loss": 1.6109, + "step": 4000 + }, + { + "epoch": 0.2, + "grad_norm": 6.2877655029296875, + "learning_rate": 1.9276783919597994e-05, + "loss": 1.632, + "step": 4100 + }, + { + "epoch": 0.21, + "grad_norm": 4.26241397857666, + "learning_rate": 1.925668341708543e-05, + "loss": 1.5813, + "step": 4200 + }, + { + "epoch": 0.21, + "grad_norm": 4.126381874084473, + "learning_rate": 1.9236582914572866e-05, + "loss": 1.6013, + "step": 4300 + }, + { + "epoch": 0.22, + "grad_norm": 4.520874977111816, + "learning_rate": 1.92164824120603e-05, + "loss": 1.6499, + "step": 4400 + }, + { + "epoch": 0.23, + "grad_norm": 4.288824558258057, + "learning_rate": 1.9196381909547742e-05, + "loss": 1.6151, + "step": 4500 + }, + { + "epoch": 0.23, + "grad_norm": 5.139670372009277, + "learning_rate": 1.9176281407035177e-05, + "loss": 1.61, + "step": 4600 + }, + { + "epoch": 0.23, + "grad_norm": 5.32205867767334, + "learning_rate": 1.9156180904522615e-05, + "loss": 1.6125, + "step": 4700 + }, + { + "epoch": 0.24, + "grad_norm": 3.354684829711914, + "learning_rate": 1.913608040201005e-05, + "loss": 1.6219, + "step": 4800 + }, + { + "epoch": 0.24, + "grad_norm": 4.567023754119873, + "learning_rate": 1.911597989949749e-05, + "loss": 1.6326, + "step": 4900 + }, + { + "epoch": 0.25, + "grad_norm": 5.753734588623047, + "learning_rate": 1.9095879396984925e-05, + "loss": 1.5992, + "step": 5000 + }, + { + "epoch": 0.26, + "grad_norm": 4.663640975952148, + "learning_rate": 1.9075778894472363e-05, + "loss": 1.5792, + "step": 5100 + }, + { + "epoch": 0.26, + "grad_norm": 5.779416561126709, + "learning_rate": 1.90556783919598e-05, + "loss": 1.6072, + "step": 5200 + }, + { + "epoch": 0.27, + "grad_norm": 5.422142505645752, + "learning_rate": 1.903557788944724e-05, + "loss": 1.6244, + "step": 5300 + }, + { + "epoch": 0.27, + "grad_norm": 4.999263286590576, + "learning_rate": 1.9015477386934674e-05, + "loss": 1.6056, + "step": 5400 + }, + { + "epoch": 0.28, + "grad_norm": 7.253634452819824, + "learning_rate": 1.8995376884422112e-05, + "loss": 1.5862, + "step": 5500 + }, + { + "epoch": 0.28, + "grad_norm": 5.289574146270752, + "learning_rate": 1.897527638190955e-05, + "loss": 1.6188, + "step": 5600 + }, + { + "epoch": 0.28, + "grad_norm": 4.021580219268799, + "learning_rate": 1.8955175879396988e-05, + "loss": 1.5506, + "step": 5700 + }, + { + "epoch": 0.29, + "grad_norm": 4.235142707824707, + "learning_rate": 1.893527638190955e-05, + "loss": 1.5975, + "step": 5800 + }, + { + "epoch": 0.29, + "grad_norm": 5.086207866668701, + "learning_rate": 1.8915175879396988e-05, + "loss": 1.5796, + "step": 5900 + }, + { + "epoch": 0.3, + "grad_norm": 5.470453262329102, + "learning_rate": 1.8895075376884423e-05, + "loss": 1.5871, + "step": 6000 + }, + { + "epoch": 0.3, + "grad_norm": 6.826024532318115, + "learning_rate": 1.887497487437186e-05, + "loss": 1.5398, + "step": 6100 + }, + { + "epoch": 0.31, + "grad_norm": 4.6210761070251465, + "learning_rate": 1.88548743718593e-05, + "loss": 1.6186, + "step": 6200 + }, + { + "epoch": 0.32, + "grad_norm": 5.72627592086792, + "learning_rate": 1.8834773869346733e-05, + "loss": 1.5377, + "step": 6300 + }, + { + "epoch": 0.32, + "grad_norm": 6.252579212188721, + "learning_rate": 1.881467336683417e-05, + "loss": 1.6209, + "step": 6400 + }, + { + "epoch": 0.33, + "grad_norm": 5.615140914916992, + "learning_rate": 1.879457286432161e-05, + "loss": 1.5482, + "step": 6500 + }, + { + "epoch": 0.33, + "grad_norm": 3.57523512840271, + "learning_rate": 1.8774472361809047e-05, + "loss": 1.5903, + "step": 6600 + }, + { + "epoch": 0.34, + "grad_norm": 7.865254878997803, + "learning_rate": 1.8754371859296482e-05, + "loss": 1.596, + "step": 6700 + }, + { + "epoch": 0.34, + "grad_norm": 5.929252624511719, + "learning_rate": 1.8734271356783923e-05, + "loss": 1.538, + "step": 6800 + }, + { + "epoch": 0.34, + "grad_norm": 3.798943281173706, + "learning_rate": 1.8714170854271358e-05, + "loss": 1.5793, + "step": 6900 + }, + { + "epoch": 0.35, + "grad_norm": 6.028155326843262, + "learning_rate": 1.8694070351758796e-05, + "loss": 1.5486, + "step": 7000 + }, + { + "epoch": 0.35, + "grad_norm": 4.279233932495117, + "learning_rate": 1.867396984924623e-05, + "loss": 1.5538, + "step": 7100 + }, + { + "epoch": 0.36, + "grad_norm": 3.5549476146698, + "learning_rate": 1.8653869346733672e-05, + "loss": 1.571, + "step": 7200 + }, + { + "epoch": 0.36, + "grad_norm": 6.669614315032959, + "learning_rate": 1.8633768844221106e-05, + "loss": 1.5943, + "step": 7300 + }, + { + "epoch": 0.37, + "grad_norm": 5.953083515167236, + "learning_rate": 1.8613668341708544e-05, + "loss": 1.5622, + "step": 7400 + }, + { + "epoch": 0.38, + "grad_norm": 4.570889472961426, + "learning_rate": 1.8593567839195982e-05, + "loss": 1.5548, + "step": 7500 + }, + { + "epoch": 0.38, + "grad_norm": 6.883716583251953, + "learning_rate": 1.857346733668342e-05, + "loss": 1.5523, + "step": 7600 + }, + { + "epoch": 0.39, + "grad_norm": 5.618870258331299, + "learning_rate": 1.8553366834170855e-05, + "loss": 1.573, + "step": 7700 + }, + { + "epoch": 0.39, + "grad_norm": 6.305469989776611, + "learning_rate": 1.8533266331658293e-05, + "loss": 1.5568, + "step": 7800 + }, + { + "epoch": 0.4, + "grad_norm": 5.411215782165527, + "learning_rate": 1.851316582914573e-05, + "loss": 1.5381, + "step": 7900 + }, + { + "epoch": 0.4, + "grad_norm": 4.859926700592041, + "learning_rate": 1.849306532663317e-05, + "loss": 1.5554, + "step": 8000 + }, + { + "epoch": 0.41, + "grad_norm": 3.7912240028381348, + "learning_rate": 1.8473165829145728e-05, + "loss": 1.4991, + "step": 8100 + }, + { + "epoch": 0.41, + "grad_norm": 5.7156901359558105, + "learning_rate": 1.845306532663317e-05, + "loss": 1.567, + "step": 8200 + }, + { + "epoch": 0.41, + "grad_norm": 5.494976043701172, + "learning_rate": 1.8432964824120604e-05, + "loss": 1.5879, + "step": 8300 + }, + { + "epoch": 0.42, + "grad_norm": 4.401707172393799, + "learning_rate": 1.8412864321608042e-05, + "loss": 1.5464, + "step": 8400 + }, + { + "epoch": 0.42, + "grad_norm": 6.136995315551758, + "learning_rate": 1.839276381909548e-05, + "loss": 1.5808, + "step": 8500 + }, + { + "epoch": 0.43, + "grad_norm": 5.482088565826416, + "learning_rate": 1.8372663316582918e-05, + "loss": 1.5014, + "step": 8600 + }, + { + "epoch": 0.43, + "grad_norm": 5.293674945831299, + "learning_rate": 1.8352562814070352e-05, + "loss": 1.5409, + "step": 8700 + }, + { + "epoch": 0.44, + "grad_norm": 6.073010444641113, + "learning_rate": 1.833246231155779e-05, + "loss": 1.5292, + "step": 8800 + }, + { + "epoch": 0.45, + "grad_norm": 5.509156703948975, + "learning_rate": 1.8312361809045228e-05, + "loss": 1.5243, + "step": 8900 + }, + { + "epoch": 0.45, + "grad_norm": 7.400144577026367, + "learning_rate": 1.8292261306532663e-05, + "loss": 1.5577, + "step": 9000 + }, + { + "epoch": 0.46, + "grad_norm": 6.070242881774902, + "learning_rate": 1.8272160804020104e-05, + "loss": 1.5394, + "step": 9100 + }, + { + "epoch": 0.46, + "grad_norm": 5.7604804039001465, + "learning_rate": 1.825206030150754e-05, + "loss": 1.5335, + "step": 9200 + }, + { + "epoch": 0.47, + "grad_norm": 4.665848731994629, + "learning_rate": 1.8231959798994977e-05, + "loss": 1.5484, + "step": 9300 + }, + { + "epoch": 0.47, + "grad_norm": 3.539947271347046, + "learning_rate": 1.821185929648241e-05, + "loss": 1.5424, + "step": 9400 + }, + { + "epoch": 0.47, + "grad_norm": 6.976469039916992, + "learning_rate": 1.8191758793969853e-05, + "loss": 1.5128, + "step": 9500 + }, + { + "epoch": 0.48, + "grad_norm": 4.112757682800293, + "learning_rate": 1.8171658291457287e-05, + "loss": 1.5085, + "step": 9600 + }, + { + "epoch": 0.48, + "grad_norm": 4.922628402709961, + "learning_rate": 1.8151557788944725e-05, + "loss": 1.5107, + "step": 9700 + }, + { + "epoch": 0.49, + "grad_norm": 6.697315692901611, + "learning_rate": 1.813145728643216e-05, + "loss": 1.5258, + "step": 9800 + }, + { + "epoch": 0.49, + "grad_norm": 7.6928253173828125, + "learning_rate": 1.81113567839196e-05, + "loss": 1.525, + "step": 9900 + }, + { + "epoch": 0.5, + "grad_norm": 3.736105442047119, + "learning_rate": 1.8091256281407036e-05, + "loss": 1.5156, + "step": 10000 + }, + { + "epoch": 0.5, + "eval_loss": 1.5033862590789795, + "eval_runtime": 21.8773, + "eval_samples_per_second": 45.709, + "eval_steps_per_second": 5.714, + "step": 10000 + }, + { + "epoch": 0.51, + "grad_norm": 6.009729385375977, + "learning_rate": 1.8071155778894474e-05, + "loss": 1.564, + "step": 10100 + }, + { + "epoch": 0.51, + "grad_norm": 5.207152366638184, + "learning_rate": 1.8051256281407036e-05, + "loss": 1.5506, + "step": 10200 + }, + { + "epoch": 0.52, + "grad_norm": 4.1969685554504395, + "learning_rate": 1.8031155778894474e-05, + "loss": 1.508, + "step": 10300 + }, + { + "epoch": 0.52, + "grad_norm": 4.609541416168213, + "learning_rate": 1.801105527638191e-05, + "loss": 1.5556, + "step": 10400 + }, + { + "epoch": 0.53, + "grad_norm": 4.768685340881348, + "learning_rate": 1.799095477386935e-05, + "loss": 1.5357, + "step": 10500 + }, + { + "epoch": 0.53, + "grad_norm": 6.3910651206970215, + "learning_rate": 1.7970854271356785e-05, + "loss": 1.5377, + "step": 10600 + }, + { + "epoch": 0.54, + "grad_norm": 4.312323570251465, + "learning_rate": 1.7950753768844223e-05, + "loss": 1.5454, + "step": 10700 + }, + { + "epoch": 0.54, + "grad_norm": 3.9479427337646484, + "learning_rate": 1.793065326633166e-05, + "loss": 1.4863, + "step": 10800 + }, + { + "epoch": 0.55, + "grad_norm": 5.744295120239258, + "learning_rate": 1.79105527638191e-05, + "loss": 1.5507, + "step": 10900 + }, + { + "epoch": 0.55, + "grad_norm": 4.2211833000183105, + "learning_rate": 1.7890452261306533e-05, + "loss": 1.5182, + "step": 11000 + }, + { + "epoch": 0.56, + "grad_norm": 4.841630458831787, + "learning_rate": 1.787035175879397e-05, + "loss": 1.4814, + "step": 11100 + }, + { + "epoch": 0.56, + "grad_norm": 6.714913845062256, + "learning_rate": 1.785025125628141e-05, + "loss": 1.4904, + "step": 11200 + }, + { + "epoch": 0.56, + "grad_norm": 6.587597846984863, + "learning_rate": 1.7830150753768847e-05, + "loss": 1.5045, + "step": 11300 + }, + { + "epoch": 0.57, + "grad_norm": 4.343375205993652, + "learning_rate": 1.7810050251256285e-05, + "loss": 1.5143, + "step": 11400 + }, + { + "epoch": 0.57, + "grad_norm": 7.786270618438721, + "learning_rate": 1.778994974874372e-05, + "loss": 1.496, + "step": 11500 + }, + { + "epoch": 0.58, + "grad_norm": 7.0261054039001465, + "learning_rate": 1.7769849246231158e-05, + "loss": 1.5113, + "step": 11600 + }, + { + "epoch": 0.58, + "grad_norm": 5.448154449462891, + "learning_rate": 1.7749748743718592e-05, + "loss": 1.4699, + "step": 11700 + }, + { + "epoch": 0.59, + "grad_norm": 5.4564361572265625, + "learning_rate": 1.7729648241206034e-05, + "loss": 1.4909, + "step": 11800 + }, + { + "epoch": 0.59, + "grad_norm": 5.704242706298828, + "learning_rate": 1.7709547738693468e-05, + "loss": 1.4931, + "step": 11900 + }, + { + "epoch": 0.6, + "grad_norm": 4.819602966308594, + "learning_rate": 1.7689447236180906e-05, + "loss": 1.4981, + "step": 12000 + }, + { + "epoch": 0.6, + "grad_norm": 2.80843186378479, + "learning_rate": 1.766934673366834e-05, + "loss": 1.4753, + "step": 12100 + }, + { + "epoch": 0.61, + "grad_norm": 4.010366439819336, + "learning_rate": 1.7649246231155782e-05, + "loss": 1.4899, + "step": 12200 + }, + { + "epoch": 0.61, + "grad_norm": 6.8596391677856445, + "learning_rate": 1.7629145728643217e-05, + "loss": 1.4951, + "step": 12300 + }, + { + "epoch": 0.62, + "grad_norm": 5.7791643142700195, + "learning_rate": 1.7609246231155782e-05, + "loss": 1.4841, + "step": 12400 + }, + { + "epoch": 0.62, + "grad_norm": 4.629549503326416, + "learning_rate": 1.7589145728643217e-05, + "loss": 1.4918, + "step": 12500 + }, + { + "epoch": 0.63, + "grad_norm": 4.676841735839844, + "learning_rate": 1.7569045226130655e-05, + "loss": 1.4934, + "step": 12600 + }, + { + "epoch": 0.64, + "grad_norm": 5.469869613647461, + "learning_rate": 1.754894472361809e-05, + "loss": 1.4767, + "step": 12700 + }, + { + "epoch": 0.64, + "grad_norm": 4.605990886688232, + "learning_rate": 1.752884422110553e-05, + "loss": 1.4635, + "step": 12800 + }, + { + "epoch": 0.65, + "grad_norm": 5.055588722229004, + "learning_rate": 1.7508743718592966e-05, + "loss": 1.4925, + "step": 12900 + }, + { + "epoch": 0.65, + "grad_norm": 4.2058916091918945, + "learning_rate": 1.7488643216080404e-05, + "loss": 1.4988, + "step": 13000 + }, + { + "epoch": 0.66, + "grad_norm": 13.90904426574707, + "learning_rate": 1.7468542713567838e-05, + "loss": 1.4883, + "step": 13100 + }, + { + "epoch": 0.66, + "grad_norm": 5.441417694091797, + "learning_rate": 1.744844221105528e-05, + "loss": 1.4817, + "step": 13200 + }, + { + "epoch": 0.67, + "grad_norm": 4.476733684539795, + "learning_rate": 1.7428341708542714e-05, + "loss": 1.5099, + "step": 13300 + }, + { + "epoch": 0.67, + "grad_norm": 4.831094264984131, + "learning_rate": 1.7408241206030152e-05, + "loss": 1.4817, + "step": 13400 + }, + { + "epoch": 0.68, + "grad_norm": 6.846999645233154, + "learning_rate": 1.738814070351759e-05, + "loss": 1.473, + "step": 13500 + }, + { + "epoch": 0.68, + "grad_norm": 4.799276351928711, + "learning_rate": 1.7368040201005028e-05, + "loss": 1.4969, + "step": 13600 + }, + { + "epoch": 0.69, + "grad_norm": 7.644506931304932, + "learning_rate": 1.7347939698492463e-05, + "loss": 1.4836, + "step": 13700 + }, + { + "epoch": 0.69, + "grad_norm": 4.794766426086426, + "learning_rate": 1.73278391959799e-05, + "loss": 1.4993, + "step": 13800 + }, + { + "epoch": 0.69, + "grad_norm": 5.148614406585693, + "learning_rate": 1.730773869346734e-05, + "loss": 1.4697, + "step": 13900 + }, + { + "epoch": 0.7, + "grad_norm": 3.6471970081329346, + "learning_rate": 1.7287638190954777e-05, + "loss": 1.4811, + "step": 14000 + }, + { + "epoch": 0.7, + "grad_norm": 5.793773174285889, + "learning_rate": 1.7267537688442214e-05, + "loss": 1.4988, + "step": 14100 + }, + { + "epoch": 0.71, + "grad_norm": 4.239154815673828, + "learning_rate": 1.724743718592965e-05, + "loss": 1.4673, + "step": 14200 + }, + { + "epoch": 0.71, + "grad_norm": 5.415383815765381, + "learning_rate": 1.7227336683417087e-05, + "loss": 1.4966, + "step": 14300 + }, + { + "epoch": 0.72, + "grad_norm": 4.704416275024414, + "learning_rate": 1.720723618090452e-05, + "loss": 1.4996, + "step": 14400 + }, + { + "epoch": 0.72, + "grad_norm": 3.5393893718719482, + "learning_rate": 1.7187336683417087e-05, + "loss": 1.4677, + "step": 14500 + }, + { + "epoch": 0.73, + "grad_norm": 3.5352883338928223, + "learning_rate": 1.7167236180904522e-05, + "loss": 1.4739, + "step": 14600 + }, + { + "epoch": 0.73, + "grad_norm": 5.679812431335449, + "learning_rate": 1.7147135678391963e-05, + "loss": 1.4556, + "step": 14700 + }, + { + "epoch": 0.74, + "grad_norm": Infinity, + "learning_rate": 1.7127236180904526e-05, + "loss": 1.4665, + "step": 14800 + }, + { + "epoch": 0.74, + "grad_norm": 5.901428699493408, + "learning_rate": 1.710713567839196e-05, + "loss": 1.464, + "step": 14900 + }, + { + "epoch": 0.75, + "grad_norm": 4.5120744705200195, + "learning_rate": 1.70870351758794e-05, + "loss": 1.4752, + "step": 15000 + }, + { + "epoch": 0.76, + "grad_norm": 4.715979099273682, + "learning_rate": 1.7066934673366836e-05, + "loss": 1.4632, + "step": 15100 + }, + { + "epoch": 0.76, + "grad_norm": 5.309842109680176, + "learning_rate": 1.704683417085427e-05, + "loss": 1.4474, + "step": 15200 + }, + { + "epoch": 0.77, + "grad_norm": 4.106339454650879, + "learning_rate": 1.7026733668341712e-05, + "loss": 1.4706, + "step": 15300 + }, + { + "epoch": 0.77, + "grad_norm": 9.052672386169434, + "learning_rate": 1.7006633165829147e-05, + "loss": 1.4662, + "step": 15400 + }, + { + "epoch": 0.78, + "grad_norm": 5.6153059005737305, + "learning_rate": 1.6986532663316585e-05, + "loss": 1.485, + "step": 15500 + }, + { + "epoch": 0.78, + "grad_norm": 4.049362659454346, + "learning_rate": 1.696643216080402e-05, + "loss": 1.4847, + "step": 15600 + }, + { + "epoch": 0.79, + "grad_norm": 4.916749477386475, + "learning_rate": 1.694633165829146e-05, + "loss": 1.4487, + "step": 15700 + }, + { + "epoch": 0.79, + "grad_norm": 4.682246685028076, + "learning_rate": 1.6926231155778895e-05, + "loss": 1.4227, + "step": 15800 + }, + { + "epoch": 0.8, + "grad_norm": 3.3342747688293457, + "learning_rate": 1.6906130653266333e-05, + "loss": 1.4564, + "step": 15900 + }, + { + "epoch": 0.8, + "grad_norm": 5.837874412536621, + "learning_rate": 1.6886030150753768e-05, + "loss": 1.4759, + "step": 16000 + }, + { + "epoch": 0.81, + "grad_norm": 4.549025058746338, + "learning_rate": 1.686592964824121e-05, + "loss": 1.4381, + "step": 16100 + }, + { + "epoch": 0.81, + "grad_norm": 6.344630718231201, + "learning_rate": 1.6845829145728644e-05, + "loss": 1.4824, + "step": 16200 + }, + { + "epoch": 0.81, + "grad_norm": 4.8240485191345215, + "learning_rate": 1.6825728643216082e-05, + "loss": 1.4631, + "step": 16300 + }, + { + "epoch": 0.82, + "grad_norm": 4.205628395080566, + "learning_rate": 1.680562814070352e-05, + "loss": 1.4683, + "step": 16400 + }, + { + "epoch": 0.82, + "grad_norm": 4.132819652557373, + "learning_rate": 1.6785527638190958e-05, + "loss": 1.4209, + "step": 16500 + }, + { + "epoch": 0.83, + "grad_norm": 4.3151140213012695, + "learning_rate": 1.6765427135678392e-05, + "loss": 1.4468, + "step": 16600 + }, + { + "epoch": 0.83, + "grad_norm": 5.677152633666992, + "learning_rate": 1.674532663316583e-05, + "loss": 1.4198, + "step": 16700 + }, + { + "epoch": 0.84, + "grad_norm": 4.871775150299072, + "learning_rate": 1.672522613065327e-05, + "loss": 1.4904, + "step": 16800 + }, + { + "epoch": 0.84, + "grad_norm": 5.693517208099365, + "learning_rate": 1.6705125628140706e-05, + "loss": 1.4592, + "step": 16900 + }, + { + "epoch": 0.85, + "grad_norm": 4.093634605407715, + "learning_rate": 1.668502512562814e-05, + "loss": 1.4561, + "step": 17000 + }, + { + "epoch": 0.85, + "grad_norm": 6.508328437805176, + "learning_rate": 1.666492462311558e-05, + "loss": 1.456, + "step": 17100 + }, + { + "epoch": 0.86, + "grad_norm": 8.67950439453125, + "learning_rate": 1.6644824120603017e-05, + "loss": 1.4432, + "step": 17200 + }, + { + "epoch": 0.86, + "grad_norm": 6.435894012451172, + "learning_rate": 1.662472361809045e-05, + "loss": 1.4807, + "step": 17300 + }, + { + "epoch": 0.87, + "grad_norm": 4.387815952301025, + "learning_rate": 1.6604623115577893e-05, + "loss": 1.4171, + "step": 17400 + }, + { + "epoch": 0.88, + "grad_norm": 5.042853832244873, + "learning_rate": 1.6584522613065327e-05, + "loss": 1.4361, + "step": 17500 + }, + { + "epoch": 0.88, + "grad_norm": 4.579937934875488, + "learning_rate": 1.6564422110552765e-05, + "loss": 1.4752, + "step": 17600 + }, + { + "epoch": 0.89, + "grad_norm": 8.229300498962402, + "learning_rate": 1.65443216080402e-05, + "loss": 1.4058, + "step": 17700 + }, + { + "epoch": 0.89, + "grad_norm": 5.82681131362915, + "learning_rate": 1.652422110552764e-05, + "loss": 1.4353, + "step": 17800 + }, + { + "epoch": 0.9, + "grad_norm": 3.8094637393951416, + "learning_rate": 1.6504120603015076e-05, + "loss": 1.4043, + "step": 17900 + }, + { + "epoch": 0.9, + "grad_norm": 5.825170993804932, + "learning_rate": 1.6484020100502514e-05, + "loss": 1.458, + "step": 18000 + }, + { + "epoch": 0.91, + "grad_norm": 5.906398773193359, + "learning_rate": 1.6463919597989952e-05, + "loss": 1.4373, + "step": 18100 + }, + { + "epoch": 0.91, + "grad_norm": 4.367284774780273, + "learning_rate": 1.644381909547739e-05, + "loss": 1.4119, + "step": 18200 + }, + { + "epoch": 0.92, + "grad_norm": 4.767496585845947, + "learning_rate": 1.6423718592964824e-05, + "loss": 1.4378, + "step": 18300 + }, + { + "epoch": 0.92, + "grad_norm": 2.3912670612335205, + "learning_rate": 1.6403618090452262e-05, + "loss": 1.4452, + "step": 18400 + }, + { + "epoch": 0.93, + "grad_norm": 4.836172580718994, + "learning_rate": 1.63835175879397e-05, + "loss": 1.4054, + "step": 18500 + }, + { + "epoch": 0.93, + "grad_norm": 7.216467380523682, + "learning_rate": 1.636341708542714e-05, + "loss": 1.4604, + "step": 18600 + }, + { + "epoch": 0.94, + "grad_norm": 4.356799602508545, + "learning_rate": 1.6343316582914573e-05, + "loss": 1.4552, + "step": 18700 + }, + { + "epoch": 0.94, + "grad_norm": 3.337068557739258, + "learning_rate": 1.632321608040201e-05, + "loss": 1.4597, + "step": 18800 + }, + { + "epoch": 0.94, + "grad_norm": 4.059195518493652, + "learning_rate": 1.630311557788945e-05, + "loss": 1.467, + "step": 18900 + }, + { + "epoch": 0.95, + "grad_norm": 3.697249412536621, + "learning_rate": 1.6283015075376887e-05, + "loss": 1.445, + "step": 19000 + }, + { + "epoch": 0.95, + "grad_norm": 6.429022789001465, + "learning_rate": 1.6262914572864325e-05, + "loss": 1.4536, + "step": 19100 + }, + { + "epoch": 0.96, + "grad_norm": 5.085973739624023, + "learning_rate": 1.624281407035176e-05, + "loss": 1.4507, + "step": 19200 + }, + { + "epoch": 0.96, + "grad_norm": 4.309168815612793, + "learning_rate": 1.6222713567839197e-05, + "loss": 1.4319, + "step": 19300 + }, + { + "epoch": 0.97, + "grad_norm": 5.077241897583008, + "learning_rate": 1.6202613065326635e-05, + "loss": 1.4103, + "step": 19400 + }, + { + "epoch": 0.97, + "grad_norm": 3.984090566635132, + "learning_rate": 1.6182512562814073e-05, + "loss": 1.4104, + "step": 19500 + }, + { + "epoch": 0.98, + "grad_norm": 4.95877742767334, + "learning_rate": 1.6162412060301508e-05, + "loss": 1.4656, + "step": 19600 + }, + { + "epoch": 0.98, + "grad_norm": 4.357282638549805, + "learning_rate": 1.6142311557788946e-05, + "loss": 1.4453, + "step": 19700 + }, + { + "epoch": 0.99, + "grad_norm": 5.499750137329102, + "learning_rate": 1.6122211055276384e-05, + "loss": 1.409, + "step": 19800 + }, + { + "epoch": 0.99, + "grad_norm": 4.081977367401123, + "learning_rate": 1.6102110552763822e-05, + "loss": 1.41, + "step": 19900 + }, + { + "epoch": 1.0, + "grad_norm": 5.961399078369141, + "learning_rate": 1.6082010050251256e-05, + "loss": 1.393, + "step": 20000 + }, + { + "epoch": 1.0, + "eval_loss": 1.431087613105774, + "eval_runtime": 21.8687, + "eval_samples_per_second": 45.727, + "eval_steps_per_second": 5.716, + "step": 20000 + }, + { + "epoch": 1.0, + "grad_norm": 6.539051532745361, + "learning_rate": 1.6062110552763822e-05, + "loss": 1.4339, + "step": 20100 + }, + { + "epoch": 1.01, + "grad_norm": 7.362614631652832, + "learning_rate": 1.6042010050251257e-05, + "loss": 1.4047, + "step": 20200 + }, + { + "epoch": 1.01, + "grad_norm": 4.155520439147949, + "learning_rate": 1.6021909547738695e-05, + "loss": 1.3943, + "step": 20300 + }, + { + "epoch": 1.02, + "grad_norm": 4.347718715667725, + "learning_rate": 1.600180904522613e-05, + "loss": 1.432, + "step": 20400 + }, + { + "epoch": 1.02, + "grad_norm": 4.478184700012207, + "learning_rate": 1.598170854271357e-05, + "loss": 1.4201, + "step": 20500 + }, + { + "epoch": 1.03, + "grad_norm": 8.239706993103027, + "learning_rate": 1.5961608040201005e-05, + "loss": 1.3944, + "step": 20600 + }, + { + "epoch": 1.03, + "grad_norm": 6.581277370452881, + "learning_rate": 1.5941507537688443e-05, + "loss": 1.359, + "step": 20700 + }, + { + "epoch": 1.04, + "grad_norm": 4.083044528961182, + "learning_rate": 1.592140703517588e-05, + "loss": 1.4004, + "step": 20800 + }, + { + "epoch": 1.04, + "grad_norm": 5.052839756011963, + "learning_rate": 1.590130653266332e-05, + "loss": 1.4157, + "step": 20900 + }, + { + "epoch": 1.05, + "grad_norm": 3.8107857704162598, + "learning_rate": 1.5881206030150754e-05, + "loss": 1.3823, + "step": 21000 + }, + { + "epoch": 1.05, + "grad_norm": 6.1900954246521, + "learning_rate": 1.5861105527638192e-05, + "loss": 1.4194, + "step": 21100 + }, + { + "epoch": 1.06, + "grad_norm": 4.510327339172363, + "learning_rate": 1.584100502512563e-05, + "loss": 1.4173, + "step": 21200 + }, + { + "epoch": 1.06, + "grad_norm": 6.412552833557129, + "learning_rate": 1.5820904522613068e-05, + "loss": 1.3996, + "step": 21300 + }, + { + "epoch": 1.07, + "grad_norm": 5.167262077331543, + "learning_rate": 1.5800804020100506e-05, + "loss": 1.4007, + "step": 21400 + }, + { + "epoch": 1.07, + "grad_norm": 4.012689590454102, + "learning_rate": 1.578070351758794e-05, + "loss": 1.3875, + "step": 21500 + }, + { + "epoch": 1.08, + "grad_norm": 6.7843017578125, + "learning_rate": 1.5760603015075378e-05, + "loss": 1.4192, + "step": 21600 + }, + { + "epoch": 1.08, + "grad_norm": 5.4605207443237305, + "learning_rate": 1.5740502512562816e-05, + "loss": 1.421, + "step": 21700 + }, + { + "epoch": 1.09, + "grad_norm": 8.303611755371094, + "learning_rate": 1.5720402010050254e-05, + "loss": 1.45, + "step": 21800 + }, + { + "epoch": 1.09, + "grad_norm": 4.898472309112549, + "learning_rate": 1.570030150753769e-05, + "loss": 1.3982, + "step": 21900 + }, + { + "epoch": 1.1, + "grad_norm": 6.471447467803955, + "learning_rate": 1.5680201005025127e-05, + "loss": 1.4272, + "step": 22000 + }, + { + "epoch": 1.1, + "grad_norm": 7.5459885597229, + "learning_rate": 1.5660100502512565e-05, + "loss": 1.3861, + "step": 22100 + }, + { + "epoch": 1.11, + "grad_norm": 7.108932971954346, + "learning_rate": 1.5640000000000003e-05, + "loss": 1.3946, + "step": 22200 + }, + { + "epoch": 1.11, + "grad_norm": 5.083498954772949, + "learning_rate": 1.5619899497487437e-05, + "loss": 1.4006, + "step": 22300 + }, + { + "epoch": 1.12, + "grad_norm": 6.796627998352051, + "learning_rate": 1.5599798994974875e-05, + "loss": 1.4266, + "step": 22400 + }, + { + "epoch": 1.12, + "grad_norm": 5.1619367599487305, + "learning_rate": 1.5579698492462313e-05, + "loss": 1.3567, + "step": 22500 + }, + { + "epoch": 1.13, + "grad_norm": 5.548572063446045, + "learning_rate": 1.555959798994975e-05, + "loss": 1.4193, + "step": 22600 + }, + { + "epoch": 1.14, + "grad_norm": 4.019492149353027, + "learning_rate": 1.5539497487437186e-05, + "loss": 1.3988, + "step": 22700 + }, + { + "epoch": 1.14, + "grad_norm": 5.359696865081787, + "learning_rate": 1.5519396984924624e-05, + "loss": 1.4046, + "step": 22800 + }, + { + "epoch": 1.15, + "grad_norm": 3.4442272186279297, + "learning_rate": 1.5499296482412062e-05, + "loss": 1.3924, + "step": 22900 + }, + { + "epoch": 1.15, + "grad_norm": 5.838873863220215, + "learning_rate": 1.5479396984924624e-05, + "loss": 1.4141, + "step": 23000 + }, + { + "epoch": 1.16, + "grad_norm": 5.01621150970459, + "learning_rate": 1.5459296482412062e-05, + "loss": 1.3658, + "step": 23100 + }, + { + "epoch": 1.16, + "grad_norm": 5.7665205001831055, + "learning_rate": 1.54391959798995e-05, + "loss": 1.4245, + "step": 23200 + }, + { + "epoch": 1.17, + "grad_norm": 3.1527726650238037, + "learning_rate": 1.5419095477386935e-05, + "loss": 1.3695, + "step": 23300 + }, + { + "epoch": 1.17, + "grad_norm": 6.3304924964904785, + "learning_rate": 1.5398994974874373e-05, + "loss": 1.3978, + "step": 23400 + }, + { + "epoch": 1.18, + "grad_norm": 7.042291164398193, + "learning_rate": 1.537889447236181e-05, + "loss": 1.385, + "step": 23500 + }, + { + "epoch": 1.18, + "grad_norm": 4.397637844085693, + "learning_rate": 1.535879396984925e-05, + "loss": 1.3726, + "step": 23600 + }, + { + "epoch": 1.19, + "grad_norm": 3.7688262462615967, + "learning_rate": 1.5338693467336687e-05, + "loss": 1.3896, + "step": 23700 + }, + { + "epoch": 1.19, + "grad_norm": 4.961839199066162, + "learning_rate": 1.531859296482412e-05, + "loss": 1.3792, + "step": 23800 + }, + { + "epoch": 1.2, + "grad_norm": 4.08626127243042, + "learning_rate": 1.529849246231156e-05, + "loss": 1.3507, + "step": 23900 + }, + { + "epoch": 1.2, + "grad_norm": 4.055938243865967, + "learning_rate": 1.5278391959798997e-05, + "loss": 1.3873, + "step": 24000 + }, + { + "epoch": 1.21, + "grad_norm": 5.093524932861328, + "learning_rate": 1.5258291457286433e-05, + "loss": 1.3763, + "step": 24100 + }, + { + "epoch": 1.21, + "grad_norm": 5.755058288574219, + "learning_rate": 1.523819095477387e-05, + "loss": 1.4164, + "step": 24200 + }, + { + "epoch": 1.22, + "grad_norm": 4.845275402069092, + "learning_rate": 1.5218090452261308e-05, + "loss": 1.3103, + "step": 24300 + }, + { + "epoch": 1.22, + "grad_norm": 5.180044174194336, + "learning_rate": 1.5197989949748746e-05, + "loss": 1.3739, + "step": 24400 + }, + { + "epoch": 1.23, + "grad_norm": 5.913352012634277, + "learning_rate": 1.5178090452261306e-05, + "loss": 1.3946, + "step": 24500 + }, + { + "epoch": 1.23, + "grad_norm": 6.644520282745361, + "learning_rate": 1.5157989949748746e-05, + "loss": 1.3683, + "step": 24600 + }, + { + "epoch": 1.23, + "grad_norm": 4.617815017700195, + "learning_rate": 1.5137889447236182e-05, + "loss": 1.3588, + "step": 24700 + }, + { + "epoch": 1.24, + "grad_norm": 5.690709114074707, + "learning_rate": 1.5117788944723619e-05, + "loss": 1.4102, + "step": 24800 + }, + { + "epoch": 1.25, + "grad_norm": 4.181049823760986, + "learning_rate": 1.5097688442211057e-05, + "loss": 1.3777, + "step": 24900 + }, + { + "epoch": 1.25, + "grad_norm": 5.829825401306152, + "learning_rate": 1.5077587939698495e-05, + "loss": 1.4278, + "step": 25000 + }, + { + "epoch": 1.25, + "grad_norm": 4.409423351287842, + "learning_rate": 1.505748743718593e-05, + "loss": 1.3751, + "step": 25100 + }, + { + "epoch": 1.26, + "grad_norm": 5.290346145629883, + "learning_rate": 1.5037386934673369e-05, + "loss": 1.3752, + "step": 25200 + }, + { + "epoch": 1.27, + "grad_norm": 6.922583103179932, + "learning_rate": 1.5017286432160805e-05, + "loss": 1.3756, + "step": 25300 + }, + { + "epoch": 1.27, + "grad_norm": 4.969797134399414, + "learning_rate": 1.4997185929648241e-05, + "loss": 1.3808, + "step": 25400 + }, + { + "epoch": 1.27, + "grad_norm": 4.4493184089660645, + "learning_rate": 1.4977085427135681e-05, + "loss": 1.3836, + "step": 25500 + }, + { + "epoch": 1.28, + "grad_norm": 3.4044313430786133, + "learning_rate": 1.4956984924623117e-05, + "loss": 1.3737, + "step": 25600 + }, + { + "epoch": 1.28, + "grad_norm": 3.3968327045440674, + "learning_rate": 1.4936884422110554e-05, + "loss": 1.3776, + "step": 25700 + }, + { + "epoch": 1.29, + "grad_norm": 3.016774892807007, + "learning_rate": 1.491678391959799e-05, + "loss": 1.3642, + "step": 25800 + }, + { + "epoch": 1.29, + "grad_norm": 6.324804306030273, + "learning_rate": 1.489668341708543e-05, + "loss": 1.3838, + "step": 25900 + }, + { + "epoch": 1.3, + "grad_norm": 3.8945064544677734, + "learning_rate": 1.4876582914572866e-05, + "loss": 1.339, + "step": 26000 + }, + { + "epoch": 1.3, + "grad_norm": 6.601470470428467, + "learning_rate": 1.4856482412060302e-05, + "loss": 1.3609, + "step": 26100 + }, + { + "epoch": 1.31, + "grad_norm": 5.291379928588867, + "learning_rate": 1.4836381909547738e-05, + "loss": 1.3829, + "step": 26200 + }, + { + "epoch": 1.31, + "grad_norm": 5.4891462326049805, + "learning_rate": 1.4816281407035178e-05, + "loss": 1.3733, + "step": 26300 + }, + { + "epoch": 1.32, + "grad_norm": 4.400446891784668, + "learning_rate": 1.4796180904522614e-05, + "loss": 1.3601, + "step": 26400 + }, + { + "epoch": 1.32, + "grad_norm": 5.4064860343933105, + "learning_rate": 1.477608040201005e-05, + "loss": 1.3922, + "step": 26500 + }, + { + "epoch": 1.33, + "grad_norm": 6.4848737716674805, + "learning_rate": 1.4755979899497489e-05, + "loss": 1.3839, + "step": 26600 + }, + { + "epoch": 1.33, + "grad_norm": 3.8651046752929688, + "learning_rate": 1.4735879396984927e-05, + "loss": 1.3798, + "step": 26700 + }, + { + "epoch": 1.34, + "grad_norm": 6.21872615814209, + "learning_rate": 1.4715778894472363e-05, + "loss": 1.3771, + "step": 26800 + }, + { + "epoch": 1.34, + "grad_norm": 4.698353290557861, + "learning_rate": 1.46956783919598e-05, + "loss": 1.3794, + "step": 26900 + }, + { + "epoch": 1.35, + "grad_norm": 6.222665309906006, + "learning_rate": 1.4675577889447237e-05, + "loss": 1.3603, + "step": 27000 + }, + { + "epoch": 1.35, + "grad_norm": 5.659895896911621, + "learning_rate": 1.4655477386934675e-05, + "loss": 1.4091, + "step": 27100 + }, + { + "epoch": 1.36, + "grad_norm": 5.340900897979736, + "learning_rate": 1.4635376884422113e-05, + "loss": 1.385, + "step": 27200 + }, + { + "epoch": 1.36, + "grad_norm": 2.827996253967285, + "learning_rate": 1.461527638190955e-05, + "loss": 1.3724, + "step": 27300 + }, + { + "epoch": 1.37, + "grad_norm": 5.637544631958008, + "learning_rate": 1.4595175879396986e-05, + "loss": 1.3911, + "step": 27400 + }, + { + "epoch": 1.38, + "grad_norm": 3.459794282913208, + "learning_rate": 1.4575075376884422e-05, + "loss": 1.3946, + "step": 27500 + }, + { + "epoch": 1.38, + "grad_norm": 6.612933158874512, + "learning_rate": 1.4554974874371862e-05, + "loss": 1.404, + "step": 27600 + }, + { + "epoch": 1.39, + "grad_norm": 4.6960577964782715, + "learning_rate": 1.4534874371859298e-05, + "loss": 1.3571, + "step": 27700 + }, + { + "epoch": 1.39, + "grad_norm": 3.677015781402588, + "learning_rate": 1.4514773869346734e-05, + "loss": 1.4166, + "step": 27800 + }, + { + "epoch": 1.4, + "grad_norm": 4.411760330200195, + "learning_rate": 1.449467336683417e-05, + "loss": 1.3666, + "step": 27900 + }, + { + "epoch": 1.4, + "grad_norm": 4.236432075500488, + "learning_rate": 1.447457286432161e-05, + "loss": 1.3738, + "step": 28000 + }, + { + "epoch": 1.41, + "grad_norm": 7.484130859375, + "learning_rate": 1.4454472361809046e-05, + "loss": 1.3454, + "step": 28100 + }, + { + "epoch": 1.41, + "grad_norm": 4.38557243347168, + "learning_rate": 1.4434371859296483e-05, + "loss": 1.3842, + "step": 28200 + }, + { + "epoch": 1.42, + "grad_norm": 5.947939395904541, + "learning_rate": 1.441427135678392e-05, + "loss": 1.3391, + "step": 28300 + }, + { + "epoch": 1.42, + "grad_norm": 4.879386901855469, + "learning_rate": 1.4394170854271359e-05, + "loss": 1.3337, + "step": 28400 + }, + { + "epoch": 1.43, + "grad_norm": 5.369794845581055, + "learning_rate": 1.4374070351758795e-05, + "loss": 1.371, + "step": 28500 + }, + { + "epoch": 1.43, + "grad_norm": 6.312124252319336, + "learning_rate": 1.4353969849246233e-05, + "loss": 1.3321, + "step": 28600 + }, + { + "epoch": 1.44, + "grad_norm": 5.254230976104736, + "learning_rate": 1.4333869346733669e-05, + "loss": 1.3835, + "step": 28700 + }, + { + "epoch": 1.44, + "grad_norm": 5.279263496398926, + "learning_rate": 1.4313768844221107e-05, + "loss": 1.341, + "step": 28800 + }, + { + "epoch": 1.45, + "grad_norm": 3.9145216941833496, + "learning_rate": 1.4293668341708545e-05, + "loss": 1.3457, + "step": 28900 + }, + { + "epoch": 1.45, + "grad_norm": 18.773277282714844, + "learning_rate": 1.4273768844221107e-05, + "loss": 1.3617, + "step": 29000 + }, + { + "epoch": 1.46, + "grad_norm": 4.489799499511719, + "learning_rate": 1.4253668341708544e-05, + "loss": 1.3604, + "step": 29100 + }, + { + "epoch": 1.46, + "grad_norm": 3.820908784866333, + "learning_rate": 1.423356783919598e-05, + "loss": 1.3688, + "step": 29200 + }, + { + "epoch": 1.47, + "grad_norm": 5.470434188842773, + "learning_rate": 1.4213467336683418e-05, + "loss": 1.4041, + "step": 29300 + }, + { + "epoch": 1.47, + "grad_norm": 2.9653820991516113, + "learning_rate": 1.4193366834170856e-05, + "loss": 1.3593, + "step": 29400 + }, + { + "epoch": 1.48, + "grad_norm": 4.433176517486572, + "learning_rate": 1.4173266331658292e-05, + "loss": 1.3402, + "step": 29500 + }, + { + "epoch": 1.48, + "grad_norm": 3.5363194942474365, + "learning_rate": 1.415316582914573e-05, + "loss": 1.376, + "step": 29600 + }, + { + "epoch": 1.48, + "grad_norm": 6.819579601287842, + "learning_rate": 1.4133065326633166e-05, + "loss": 1.3203, + "step": 29700 + }, + { + "epoch": 1.49, + "grad_norm": 5.506997585296631, + "learning_rate": 1.4112964824120604e-05, + "loss": 1.3234, + "step": 29800 + }, + { + "epoch": 1.5, + "grad_norm": 6.012782573699951, + "learning_rate": 1.4092864321608042e-05, + "loss": 1.3625, + "step": 29900 + }, + { + "epoch": 1.5, + "grad_norm": 2.309823751449585, + "learning_rate": 1.4072763819095479e-05, + "loss": 1.351, + "step": 30000 + }, + { + "epoch": 1.5, + "eval_loss": 1.3539750576019287, + "eval_runtime": 21.8688, + "eval_samples_per_second": 45.727, + "eval_steps_per_second": 5.716, + "step": 30000 + }, + { + "epoch": 1.5, + "grad_norm": 5.257068157196045, + "learning_rate": 1.4052663316582915e-05, + "loss": 1.348, + "step": 30100 + }, + { + "epoch": 1.51, + "grad_norm": 13.082318305969238, + "learning_rate": 1.4032562814070351e-05, + "loss": 1.3791, + "step": 30200 + }, + { + "epoch": 1.52, + "grad_norm": 4.944590091705322, + "learning_rate": 1.4012462311557791e-05, + "loss": 1.3805, + "step": 30300 + }, + { + "epoch": 1.52, + "grad_norm": 4.780072212219238, + "learning_rate": 1.3992361809045227e-05, + "loss": 1.3651, + "step": 30400 + }, + { + "epoch": 1.52, + "grad_norm": 4.359679698944092, + "learning_rate": 1.3972261306532664e-05, + "loss": 1.3118, + "step": 30500 + }, + { + "epoch": 1.53, + "grad_norm": 4.789872646331787, + "learning_rate": 1.3952160804020101e-05, + "loss": 1.3662, + "step": 30600 + }, + { + "epoch": 1.54, + "grad_norm": 4.301767349243164, + "learning_rate": 1.393206030150754e-05, + "loss": 1.3564, + "step": 30700 + }, + { + "epoch": 1.54, + "grad_norm": 4.046327590942383, + "learning_rate": 1.3911959798994976e-05, + "loss": 1.3386, + "step": 30800 + }, + { + "epoch": 1.54, + "grad_norm": 6.321465969085693, + "learning_rate": 1.3891859296482412e-05, + "loss": 1.3544, + "step": 30900 + }, + { + "epoch": 1.55, + "grad_norm": 5.538000106811523, + "learning_rate": 1.387175879396985e-05, + "loss": 1.3466, + "step": 31000 + }, + { + "epoch": 1.56, + "grad_norm": 6.25814151763916, + "learning_rate": 1.3851658291457288e-05, + "loss": 1.3246, + "step": 31100 + }, + { + "epoch": 1.56, + "grad_norm": 5.343544006347656, + "learning_rate": 1.3831758793969849e-05, + "loss": 1.3629, + "step": 31200 + }, + { + "epoch": 1.56, + "grad_norm": 4.390071868896484, + "learning_rate": 1.3811658291457288e-05, + "loss": 1.3211, + "step": 31300 + }, + { + "epoch": 1.57, + "grad_norm": 5.539604187011719, + "learning_rate": 1.3791557788944725e-05, + "loss": 1.3297, + "step": 31400 + }, + { + "epoch": 1.57, + "grad_norm": 4.06265115737915, + "learning_rate": 1.3771457286432161e-05, + "loss": 1.37, + "step": 31500 + }, + { + "epoch": 1.58, + "grad_norm": 4.819797515869141, + "learning_rate": 1.3751356783919599e-05, + "loss": 1.3279, + "step": 31600 + }, + { + "epoch": 1.58, + "grad_norm": 4.675261497497559, + "learning_rate": 1.3731256281407037e-05, + "loss": 1.321, + "step": 31700 + }, + { + "epoch": 1.59, + "grad_norm": 6.112530708312988, + "learning_rate": 1.3711155778894473e-05, + "loss": 1.3321, + "step": 31800 + }, + { + "epoch": 1.59, + "grad_norm": 4.85811185836792, + "learning_rate": 1.3691055276381911e-05, + "loss": 1.36, + "step": 31900 + }, + { + "epoch": 1.6, + "grad_norm": 3.9626624584198, + "learning_rate": 1.3670954773869347e-05, + "loss": 1.3446, + "step": 32000 + }, + { + "epoch": 1.6, + "grad_norm": 4.470461845397949, + "learning_rate": 1.3650854271356785e-05, + "loss": 1.3503, + "step": 32100 + }, + { + "epoch": 1.61, + "grad_norm": 3.1345880031585693, + "learning_rate": 1.3630753768844223e-05, + "loss": 1.322, + "step": 32200 + }, + { + "epoch": 1.61, + "grad_norm": 4.220657825469971, + "learning_rate": 1.361065326633166e-05, + "loss": 1.2922, + "step": 32300 + }, + { + "epoch": 1.62, + "grad_norm": 4.827053546905518, + "learning_rate": 1.3590552763819096e-05, + "loss": 1.3663, + "step": 32400 + }, + { + "epoch": 1.62, + "grad_norm": 3.613919496536255, + "learning_rate": 1.3570452261306536e-05, + "loss": 1.359, + "step": 32500 + }, + { + "epoch": 1.63, + "grad_norm": 6.170840263366699, + "learning_rate": 1.3550351758793972e-05, + "loss": 1.3076, + "step": 32600 + }, + { + "epoch": 1.64, + "grad_norm": 5.604345321655273, + "learning_rate": 1.3530251256281408e-05, + "loss": 1.3335, + "step": 32700 + }, + { + "epoch": 1.64, + "grad_norm": 3.617830514907837, + "learning_rate": 1.3510150753768844e-05, + "loss": 1.3601, + "step": 32800 + }, + { + "epoch": 1.65, + "grad_norm": 5.692416191101074, + "learning_rate": 1.349005025125628e-05, + "loss": 1.3117, + "step": 32900 + }, + { + "epoch": 1.65, + "grad_norm": 6.292971611022949, + "learning_rate": 1.346994974874372e-05, + "loss": 1.3427, + "step": 33000 + }, + { + "epoch": 1.66, + "grad_norm": 4.1335906982421875, + "learning_rate": 1.3449849246231157e-05, + "loss": 1.3593, + "step": 33100 + }, + { + "epoch": 1.66, + "grad_norm": 3.7386412620544434, + "learning_rate": 1.3429748743718593e-05, + "loss": 1.3509, + "step": 33200 + }, + { + "epoch": 1.67, + "grad_norm": 4.687217712402344, + "learning_rate": 1.3409648241206031e-05, + "loss": 1.3331, + "step": 33300 + }, + { + "epoch": 1.67, + "grad_norm": 3.9221270084381104, + "learning_rate": 1.3389547738693469e-05, + "loss": 1.3299, + "step": 33400 + }, + { + "epoch": 1.68, + "grad_norm": 4.383951663970947, + "learning_rate": 1.3369447236180905e-05, + "loss": 1.3455, + "step": 33500 + }, + { + "epoch": 1.68, + "grad_norm": 4.3801140785217285, + "learning_rate": 1.3349346733668343e-05, + "loss": 1.3092, + "step": 33600 + }, + { + "epoch": 1.69, + "grad_norm": 11.051312446594238, + "learning_rate": 1.332924623115578e-05, + "loss": 1.3164, + "step": 33700 + }, + { + "epoch": 1.69, + "grad_norm": 5.16642951965332, + "learning_rate": 1.3309346733668342e-05, + "loss": 1.3463, + "step": 33800 + }, + { + "epoch": 1.69, + "grad_norm": 4.878006935119629, + "learning_rate": 1.328924623115578e-05, + "loss": 1.3452, + "step": 33900 + }, + { + "epoch": 1.7, + "grad_norm": 3.8694944381713867, + "learning_rate": 1.3269145728643218e-05, + "loss": 1.3393, + "step": 34000 + }, + { + "epoch": 1.71, + "grad_norm": 3.1603684425354004, + "learning_rate": 1.3249045226130654e-05, + "loss": 1.3608, + "step": 34100 + }, + { + "epoch": 1.71, + "grad_norm": 7.4358086585998535, + "learning_rate": 1.3228944723618092e-05, + "loss": 1.334, + "step": 34200 + }, + { + "epoch": 1.71, + "grad_norm": 3.47105073928833, + "learning_rate": 1.3208844221105528e-05, + "loss": 1.3394, + "step": 34300 + }, + { + "epoch": 1.72, + "grad_norm": 6.943812847137451, + "learning_rate": 1.3188743718592966e-05, + "loss": 1.3316, + "step": 34400 + }, + { + "epoch": 1.73, + "grad_norm": 3.4260807037353516, + "learning_rate": 1.3168643216080404e-05, + "loss": 1.2544, + "step": 34500 + }, + { + "epoch": 1.73, + "grad_norm": 5.455039978027344, + "learning_rate": 1.314854271356784e-05, + "loss": 1.3122, + "step": 34600 + }, + { + "epoch": 1.73, + "grad_norm": 5.853464603424072, + "learning_rate": 1.3128442211055277e-05, + "loss": 1.3533, + "step": 34700 + }, + { + "epoch": 1.74, + "grad_norm": 6.121273994445801, + "learning_rate": 1.3108341708542715e-05, + "loss": 1.3144, + "step": 34800 + }, + { + "epoch": 1.75, + "grad_norm": 4.645484447479248, + "learning_rate": 1.3088241206030153e-05, + "loss": 1.3154, + "step": 34900 + }, + { + "epoch": 1.75, + "grad_norm": 5.509624004364014, + "learning_rate": 1.3068140703517589e-05, + "loss": 1.3335, + "step": 35000 + }, + { + "epoch": 1.75, + "grad_norm": 4.298194885253906, + "learning_rate": 1.3048040201005025e-05, + "loss": 1.3374, + "step": 35100 + }, + { + "epoch": 1.76, + "grad_norm": 3.8274903297424316, + "learning_rate": 1.3027939698492465e-05, + "loss": 1.3577, + "step": 35200 + }, + { + "epoch": 1.77, + "grad_norm": 3.6994845867156982, + "learning_rate": 1.3007839195979901e-05, + "loss": 1.3439, + "step": 35300 + }, + { + "epoch": 1.77, + "grad_norm": 5.096419334411621, + "learning_rate": 1.2987738693467338e-05, + "loss": 1.3425, + "step": 35400 + }, + { + "epoch": 1.77, + "grad_norm": 5.432124614715576, + "learning_rate": 1.2967638190954774e-05, + "loss": 1.3089, + "step": 35500 + }, + { + "epoch": 1.78, + "grad_norm": 6.754784107208252, + "learning_rate": 1.2947537688442212e-05, + "loss": 1.3577, + "step": 35600 + }, + { + "epoch": 1.79, + "grad_norm": 5.67179012298584, + "learning_rate": 1.292743718592965e-05, + "loss": 1.3158, + "step": 35700 + }, + { + "epoch": 1.79, + "grad_norm": 5.4324517250061035, + "learning_rate": 1.2907336683417086e-05, + "loss": 1.3423, + "step": 35800 + }, + { + "epoch": 1.79, + "grad_norm": 6.366858959197998, + "learning_rate": 1.2887236180904524e-05, + "loss": 1.348, + "step": 35900 + }, + { + "epoch": 1.8, + "grad_norm": 4.625099182128906, + "learning_rate": 1.286713567839196e-05, + "loss": 1.344, + "step": 36000 + }, + { + "epoch": 1.81, + "grad_norm": 4.306463718414307, + "learning_rate": 1.2847035175879398e-05, + "loss": 1.3471, + "step": 36100 + }, + { + "epoch": 1.81, + "grad_norm": 5.759509086608887, + "learning_rate": 1.2826934673366835e-05, + "loss": 1.2994, + "step": 36200 + }, + { + "epoch": 1.81, + "grad_norm": 6.741412162780762, + "learning_rate": 1.2806834170854273e-05, + "loss": 1.3385, + "step": 36300 + }, + { + "epoch": 1.82, + "grad_norm": 8.009258270263672, + "learning_rate": 1.2786733668341709e-05, + "loss": 1.307, + "step": 36400 + }, + { + "epoch": 1.82, + "grad_norm": 4.115825653076172, + "learning_rate": 1.2766633165829147e-05, + "loss": 1.3514, + "step": 36500 + }, + { + "epoch": 1.83, + "grad_norm": 6.015181541442871, + "learning_rate": 1.2746532663316585e-05, + "loss": 1.3126, + "step": 36600 + }, + { + "epoch": 1.83, + "grad_norm": 5.512998104095459, + "learning_rate": 1.2726633165829147e-05, + "loss": 1.3288, + "step": 36700 + }, + { + "epoch": 1.84, + "grad_norm": 4.820090293884277, + "learning_rate": 1.2706532663316583e-05, + "loss": 1.322, + "step": 36800 + }, + { + "epoch": 1.84, + "grad_norm": 5.174429893493652, + "learning_rate": 1.2686432160804021e-05, + "loss": 1.3451, + "step": 36900 + }, + { + "epoch": 1.85, + "grad_norm": 5.226765155792236, + "learning_rate": 1.2666331658291458e-05, + "loss": 1.3362, + "step": 37000 + }, + { + "epoch": 1.85, + "grad_norm": 7.082357406616211, + "learning_rate": 1.2646231155778896e-05, + "loss": 1.3053, + "step": 37100 + }, + { + "epoch": 1.86, + "grad_norm": 5.663987636566162, + "learning_rate": 1.2626130653266334e-05, + "loss": 1.3312, + "step": 37200 + }, + { + "epoch": 1.86, + "grad_norm": 4.99432897567749, + "learning_rate": 1.260603015075377e-05, + "loss": 1.2969, + "step": 37300 + }, + { + "epoch": 1.87, + "grad_norm": 3.6237847805023193, + "learning_rate": 1.2585929648241206e-05, + "loss": 1.3284, + "step": 37400 + }, + { + "epoch": 1.88, + "grad_norm": 4.317774772644043, + "learning_rate": 1.2565829145728646e-05, + "loss": 1.2724, + "step": 37500 + }, + { + "epoch": 1.88, + "grad_norm": 28.135671615600586, + "learning_rate": 1.2545728643216082e-05, + "loss": 1.301, + "step": 37600 + }, + { + "epoch": 1.89, + "grad_norm": 6.461686611175537, + "learning_rate": 1.2525628140703518e-05, + "loss": 1.311, + "step": 37700 + }, + { + "epoch": 1.89, + "grad_norm": 5.122781276702881, + "learning_rate": 1.2505527638190955e-05, + "loss": 1.3124, + "step": 37800 + }, + { + "epoch": 1.9, + "grad_norm": 7.070276737213135, + "learning_rate": 1.2485628140703519e-05, + "loss": 1.3042, + "step": 37900 + }, + { + "epoch": 1.9, + "grad_norm": 5.5672607421875, + "learning_rate": 1.2465527638190955e-05, + "loss": 1.2705, + "step": 38000 + }, + { + "epoch": 1.91, + "grad_norm": 5.857154369354248, + "learning_rate": 1.2445427135678395e-05, + "loss": 1.2842, + "step": 38100 + }, + { + "epoch": 1.91, + "grad_norm": 7.3641133308410645, + "learning_rate": 1.2425326633165831e-05, + "loss": 1.3284, + "step": 38200 + }, + { + "epoch": 1.92, + "grad_norm": 7.004215717315674, + "learning_rate": 1.2405226130653267e-05, + "loss": 1.3106, + "step": 38300 + }, + { + "epoch": 1.92, + "grad_norm": 6.171952247619629, + "learning_rate": 1.2385125628140704e-05, + "loss": 1.2711, + "step": 38400 + }, + { + "epoch": 1.93, + "grad_norm": 4.094524383544922, + "learning_rate": 1.236502512562814e-05, + "loss": 1.2879, + "step": 38500 + }, + { + "epoch": 1.93, + "grad_norm": 6.0571980476379395, + "learning_rate": 1.234492462311558e-05, + "loss": 1.3454, + "step": 38600 + }, + { + "epoch": 1.94, + "grad_norm": 4.057672023773193, + "learning_rate": 1.2324824120603016e-05, + "loss": 1.3275, + "step": 38700 + }, + { + "epoch": 1.94, + "grad_norm": 6.506563663482666, + "learning_rate": 1.2304723618090452e-05, + "loss": 1.2866, + "step": 38800 + }, + { + "epoch": 1.94, + "grad_norm": 5.117976188659668, + "learning_rate": 1.228462311557789e-05, + "loss": 1.2811, + "step": 38900 + }, + { + "epoch": 1.95, + "grad_norm": 4.050692558288574, + "learning_rate": 1.2264522613065328e-05, + "loss": 1.2682, + "step": 39000 + }, + { + "epoch": 1.96, + "grad_norm": 4.216948509216309, + "learning_rate": 1.2244422110552764e-05, + "loss": 1.3079, + "step": 39100 + }, + { + "epoch": 1.96, + "grad_norm": 5.064427375793457, + "learning_rate": 1.2224321608040202e-05, + "loss": 1.3022, + "step": 39200 + }, + { + "epoch": 1.96, + "grad_norm": 6.869637489318848, + "learning_rate": 1.2204221105527639e-05, + "loss": 1.294, + "step": 39300 + }, + { + "epoch": 1.97, + "grad_norm": 4.062154769897461, + "learning_rate": 1.2184120603015077e-05, + "loss": 1.3249, + "step": 39400 + }, + { + "epoch": 1.98, + "grad_norm": 5.1579976081848145, + "learning_rate": 1.2164020100502515e-05, + "loss": 1.3006, + "step": 39500 + }, + { + "epoch": 1.98, + "grad_norm": 6.146691799163818, + "learning_rate": 1.214391959798995e-05, + "loss": 1.317, + "step": 39600 + }, + { + "epoch": 1.98, + "grad_norm": 5.503583908081055, + "learning_rate": 1.2123819095477387e-05, + "loss": 1.2803, + "step": 39700 + }, + { + "epoch": 1.99, + "grad_norm": 5.574082374572754, + "learning_rate": 1.2103718592964827e-05, + "loss": 1.3149, + "step": 39800 + }, + { + "epoch": 2.0, + "grad_norm": 7.74934720993042, + "learning_rate": 1.2083618090452263e-05, + "loss": 1.3085, + "step": 39900 + }, + { + "epoch": 2.0, + "grad_norm": 5.324882984161377, + "learning_rate": 1.20635175879397e-05, + "loss": 1.2662, + "step": 40000 + }, + { + "epoch": 2.0, + "eval_loss": 1.311901330947876, + "eval_runtime": 21.8453, + "eval_samples_per_second": 45.776, + "eval_steps_per_second": 5.722, + "step": 40000 + }, + { + "epoch": 2.0, + "grad_norm": 4.58272647857666, + "learning_rate": 1.2043417085427136e-05, + "loss": 1.251, + "step": 40100 + }, + { + "epoch": 2.01, + "grad_norm": 6.063676357269287, + "learning_rate": 1.2023316582914575e-05, + "loss": 1.3191, + "step": 40200 + }, + { + "epoch": 2.02, + "grad_norm": 6.219407558441162, + "learning_rate": 1.2003216080402012e-05, + "loss": 1.2535, + "step": 40300 + }, + { + "epoch": 2.02, + "grad_norm": 4.881536960601807, + "learning_rate": 1.1983115577889448e-05, + "loss": 1.3005, + "step": 40400 + }, + { + "epoch": 2.02, + "grad_norm": 2.7239251136779785, + "learning_rate": 1.1963015075376884e-05, + "loss": 1.2162, + "step": 40500 + }, + { + "epoch": 2.03, + "grad_norm": 4.018795490264893, + "learning_rate": 1.1942914572864324e-05, + "loss": 1.2869, + "step": 40600 + }, + { + "epoch": 2.04, + "grad_norm": 5.884511947631836, + "learning_rate": 1.192281407035176e-05, + "loss": 1.3242, + "step": 40700 + }, + { + "epoch": 2.04, + "grad_norm": 6.765765190124512, + "learning_rate": 1.1902713567839196e-05, + "loss": 1.3089, + "step": 40800 + }, + { + "epoch": 2.04, + "grad_norm": 7.366244316101074, + "learning_rate": 1.1882613065326634e-05, + "loss": 1.2574, + "step": 40900 + }, + { + "epoch": 2.05, + "grad_norm": 4.718629837036133, + "learning_rate": 1.186251256281407e-05, + "loss": 1.2967, + "step": 41000 + }, + { + "epoch": 2.06, + "grad_norm": 3.110818386077881, + "learning_rate": 1.1842412060301509e-05, + "loss": 1.2619, + "step": 41100 + }, + { + "epoch": 2.06, + "grad_norm": 5.5501298904418945, + "learning_rate": 1.1822311557788947e-05, + "loss": 1.2895, + "step": 41200 + }, + { + "epoch": 2.06, + "grad_norm": 4.115253925323486, + "learning_rate": 1.1802412060301509e-05, + "loss": 1.2978, + "step": 41300 + }, + { + "epoch": 2.07, + "grad_norm": 4.711161136627197, + "learning_rate": 1.1782311557788945e-05, + "loss": 1.2846, + "step": 41400 + }, + { + "epoch": 2.08, + "grad_norm": 4.220121383666992, + "learning_rate": 1.1762211055276383e-05, + "loss": 1.2873, + "step": 41500 + }, + { + "epoch": 2.08, + "grad_norm": 2.8940072059631348, + "learning_rate": 1.174211055276382e-05, + "loss": 1.2794, + "step": 41600 + }, + { + "epoch": 2.08, + "grad_norm": 4.298203468322754, + "learning_rate": 1.1722010050251257e-05, + "loss": 1.2412, + "step": 41700 + }, + { + "epoch": 2.09, + "grad_norm": 2.983751058578491, + "learning_rate": 1.1701909547738694e-05, + "loss": 1.2379, + "step": 41800 + }, + { + "epoch": 2.1, + "grad_norm": 5.402541637420654, + "learning_rate": 1.1681809045226132e-05, + "loss": 1.2579, + "step": 41900 + }, + { + "epoch": 2.1, + "grad_norm": 4.412992000579834, + "learning_rate": 1.1661708542713568e-05, + "loss": 1.2576, + "step": 42000 + }, + { + "epoch": 2.1, + "grad_norm": 3.6417946815490723, + "learning_rate": 1.1641608040201006e-05, + "loss": 1.2711, + "step": 42100 + }, + { + "epoch": 2.11, + "grad_norm": 4.4454216957092285, + "learning_rate": 1.1621507537688444e-05, + "loss": 1.2823, + "step": 42200 + }, + { + "epoch": 2.12, + "grad_norm": 5.299724578857422, + "learning_rate": 1.160140703517588e-05, + "loss": 1.2559, + "step": 42300 + }, + { + "epoch": 2.12, + "grad_norm": 4.227545261383057, + "learning_rate": 1.1581306532663317e-05, + "loss": 1.2321, + "step": 42400 + }, + { + "epoch": 2.12, + "grad_norm": 5.005281925201416, + "learning_rate": 1.1561206030150756e-05, + "loss": 1.2988, + "step": 42500 + }, + { + "epoch": 2.13, + "grad_norm": 5.402487754821777, + "learning_rate": 1.1541105527638192e-05, + "loss": 1.2935, + "step": 42600 + }, + { + "epoch": 2.13, + "grad_norm": 4.937114715576172, + "learning_rate": 1.1521005025125629e-05, + "loss": 1.2786, + "step": 42700 + }, + { + "epoch": 2.14, + "grad_norm": 5.078079700469971, + "learning_rate": 1.1500904522613065e-05, + "loss": 1.2735, + "step": 42800 + }, + { + "epoch": 2.15, + "grad_norm": 3.767037868499756, + "learning_rate": 1.1480804020100505e-05, + "loss": 1.2557, + "step": 42900 + }, + { + "epoch": 2.15, + "grad_norm": 5.843142032623291, + "learning_rate": 1.1460703517587941e-05, + "loss": 1.2338, + "step": 43000 + }, + { + "epoch": 2.15, + "grad_norm": 7.4397053718566895, + "learning_rate": 1.1440804020100505e-05, + "loss": 1.2841, + "step": 43100 + }, + { + "epoch": 2.16, + "grad_norm": 5.660872459411621, + "learning_rate": 1.1420703517587941e-05, + "loss": 1.245, + "step": 43200 + }, + { + "epoch": 2.17, + "grad_norm": 5.597433567047119, + "learning_rate": 1.1400603015075378e-05, + "loss": 1.2267, + "step": 43300 + }, + { + "epoch": 2.17, + "grad_norm": 5.053484916687012, + "learning_rate": 1.1380502512562814e-05, + "loss": 1.2734, + "step": 43400 + }, + { + "epoch": 2.17, + "grad_norm": 4.610946178436279, + "learning_rate": 1.1360402010050254e-05, + "loss": 1.2443, + "step": 43500 + }, + { + "epoch": 2.18, + "grad_norm": 4.457014083862305, + "learning_rate": 1.134030150753769e-05, + "loss": 1.2665, + "step": 43600 + }, + { + "epoch": 2.19, + "grad_norm": 4.798270225524902, + "learning_rate": 1.1320201005025126e-05, + "loss": 1.3062, + "step": 43700 + }, + { + "epoch": 2.19, + "grad_norm": 6.110182762145996, + "learning_rate": 1.1300100502512562e-05, + "loss": 1.2652, + "step": 43800 + }, + { + "epoch": 2.19, + "grad_norm": 5.528191089630127, + "learning_rate": 1.128e-05, + "loss": 1.2886, + "step": 43900 + }, + { + "epoch": 2.2, + "grad_norm": 6.199995994567871, + "learning_rate": 1.1259899497487438e-05, + "loss": 1.2431, + "step": 44000 + }, + { + "epoch": 2.21, + "grad_norm": 4.476943492889404, + "learning_rate": 1.1239798994974875e-05, + "loss": 1.292, + "step": 44100 + }, + { + "epoch": 2.21, + "grad_norm": 4.873668670654297, + "learning_rate": 1.1219698492462313e-05, + "loss": 1.3071, + "step": 44200 + }, + { + "epoch": 2.21, + "grad_norm": 4.769585609436035, + "learning_rate": 1.1199597989949749e-05, + "loss": 1.2568, + "step": 44300 + }, + { + "epoch": 2.22, + "grad_norm": 3.5369479656219482, + "learning_rate": 1.1179497487437187e-05, + "loss": 1.2472, + "step": 44400 + }, + { + "epoch": 2.23, + "grad_norm": 3.9671337604522705, + "learning_rate": 1.1159396984924625e-05, + "loss": 1.2917, + "step": 44500 + }, + { + "epoch": 2.23, + "grad_norm": 5.720705986022949, + "learning_rate": 1.1139296482412061e-05, + "loss": 1.235, + "step": 44600 + }, + { + "epoch": 2.23, + "grad_norm": 5.337419509887695, + "learning_rate": 1.1119195979899497e-05, + "loss": 1.3176, + "step": 44700 + }, + { + "epoch": 2.24, + "grad_norm": 5.101902961730957, + "learning_rate": 1.1099095477386937e-05, + "loss": 1.3028, + "step": 44800 + }, + { + "epoch": 2.25, + "grad_norm": 4.49253511428833, + "learning_rate": 1.1078994974874373e-05, + "loss": 1.2426, + "step": 44900 + }, + { + "epoch": 2.25, + "grad_norm": 6.949131011962891, + "learning_rate": 1.105889447236181e-05, + "loss": 1.2352, + "step": 45000 + }, + { + "epoch": 2.25, + "grad_norm": 3.002480983734131, + "learning_rate": 1.1038793969849246e-05, + "loss": 1.2544, + "step": 45100 + }, + { + "epoch": 2.26, + "grad_norm": 4.93209171295166, + "learning_rate": 1.1018693467336686e-05, + "loss": 1.2528, + "step": 45200 + }, + { + "epoch": 2.27, + "grad_norm": 3.5885465145111084, + "learning_rate": 1.0998592964824122e-05, + "loss": 1.2395, + "step": 45300 + }, + { + "epoch": 2.27, + "grad_norm": 6.264705657958984, + "learning_rate": 1.0978492462311558e-05, + "loss": 1.2468, + "step": 45400 + }, + { + "epoch": 2.27, + "grad_norm": 4.435594081878662, + "learning_rate": 1.0958391959798994e-05, + "loss": 1.2523, + "step": 45500 + }, + { + "epoch": 2.28, + "grad_norm": 2.9626126289367676, + "learning_rate": 1.0938291457286434e-05, + "loss": 1.2618, + "step": 45600 + }, + { + "epoch": 2.29, + "grad_norm": 4.376198768615723, + "learning_rate": 1.091819095477387e-05, + "loss": 1.2738, + "step": 45700 + }, + { + "epoch": 2.29, + "grad_norm": 4.058696269989014, + "learning_rate": 1.0898090452261307e-05, + "loss": 1.2787, + "step": 45800 + }, + { + "epoch": 2.29, + "grad_norm": 5.347177982330322, + "learning_rate": 1.087819095477387e-05, + "loss": 1.2778, + "step": 45900 + }, + { + "epoch": 2.3, + "grad_norm": 3.85967755317688, + "learning_rate": 1.0858090452261307e-05, + "loss": 1.2746, + "step": 46000 + }, + { + "epoch": 2.31, + "grad_norm": 5.233943939208984, + "learning_rate": 1.0837989949748743e-05, + "loss": 1.283, + "step": 46100 + }, + { + "epoch": 2.31, + "grad_norm": 6.364080429077148, + "learning_rate": 1.0817889447236183e-05, + "loss": 1.2685, + "step": 46200 + }, + { + "epoch": 2.31, + "grad_norm": 5.601933479309082, + "learning_rate": 1.079778894472362e-05, + "loss": 1.2214, + "step": 46300 + }, + { + "epoch": 2.32, + "grad_norm": 7.273884296417236, + "learning_rate": 1.0777688442211056e-05, + "loss": 1.2522, + "step": 46400 + }, + { + "epoch": 2.33, + "grad_norm": 4.99397611618042, + "learning_rate": 1.0757587939698494e-05, + "loss": 1.2779, + "step": 46500 + }, + { + "epoch": 2.33, + "grad_norm": 6.805306434631348, + "learning_rate": 1.073748743718593e-05, + "loss": 1.2605, + "step": 46600 + }, + { + "epoch": 2.33, + "grad_norm": 5.773606777191162, + "learning_rate": 1.0717386934673368e-05, + "loss": 1.2236, + "step": 46700 + }, + { + "epoch": 2.34, + "grad_norm": 5.045441150665283, + "learning_rate": 1.0697286432160806e-05, + "loss": 1.2629, + "step": 46800 + }, + { + "epoch": 2.34, + "grad_norm": 6.011552333831787, + "learning_rate": 1.0677185929648242e-05, + "loss": 1.2061, + "step": 46900 + }, + { + "epoch": 2.35, + "grad_norm": 6.2456817626953125, + "learning_rate": 1.0657085427135678e-05, + "loss": 1.2532, + "step": 47000 + }, + { + "epoch": 2.35, + "grad_norm": 4.08701229095459, + "learning_rate": 1.0636984924623116e-05, + "loss": 1.2701, + "step": 47100 + }, + { + "epoch": 2.36, + "grad_norm": 4.472239017486572, + "learning_rate": 1.0616884422110554e-05, + "loss": 1.261, + "step": 47200 + }, + { + "epoch": 2.37, + "grad_norm": 5.726062297821045, + "learning_rate": 1.059678391959799e-05, + "loss": 1.2506, + "step": 47300 + }, + { + "epoch": 2.37, + "grad_norm": 5.6060686111450195, + "learning_rate": 1.0576683417085427e-05, + "loss": 1.2472, + "step": 47400 + }, + { + "epoch": 2.38, + "grad_norm": 5.226354122161865, + "learning_rate": 1.0556582914572867e-05, + "loss": 1.2599, + "step": 47500 + }, + { + "epoch": 2.38, + "grad_norm": 6.913018703460693, + "learning_rate": 1.0536482412060303e-05, + "loss": 1.2555, + "step": 47600 + }, + { + "epoch": 2.38, + "grad_norm": 4.932835578918457, + "learning_rate": 1.0516381909547739e-05, + "loss": 1.2658, + "step": 47700 + }, + { + "epoch": 2.39, + "grad_norm": 5.260751724243164, + "learning_rate": 1.0496281407035175e-05, + "loss": 1.3025, + "step": 47800 + }, + { + "epoch": 2.4, + "grad_norm": 4.677539825439453, + "learning_rate": 1.0476180904522615e-05, + "loss": 1.2367, + "step": 47900 + }, + { + "epoch": 2.4, + "grad_norm": 5.679705619812012, + "learning_rate": 1.0456080402010051e-05, + "loss": 1.2693, + "step": 48000 + }, + { + "epoch": 2.41, + "grad_norm": 5.315084934234619, + "learning_rate": 1.0435979899497488e-05, + "loss": 1.3188, + "step": 48100 + }, + { + "epoch": 2.41, + "grad_norm": 5.218776702880859, + "learning_rate": 1.0415879396984926e-05, + "loss": 1.25, + "step": 48200 + }, + { + "epoch": 2.42, + "grad_norm": 3.9905712604522705, + "learning_rate": 1.0395778894472364e-05, + "loss": 1.2318, + "step": 48300 + }, + { + "epoch": 2.42, + "grad_norm": 5.051150798797607, + "learning_rate": 1.03756783919598e-05, + "loss": 1.2564, + "step": 48400 + }, + { + "epoch": 2.42, + "grad_norm": 4.899648666381836, + "learning_rate": 1.0355577889447238e-05, + "loss": 1.2616, + "step": 48500 + }, + { + "epoch": 2.43, + "grad_norm": 5.728534698486328, + "learning_rate": 1.0335477386934674e-05, + "loss": 1.2387, + "step": 48600 + }, + { + "epoch": 2.44, + "grad_norm": 4.747395038604736, + "learning_rate": 1.0315376884422112e-05, + "loss": 1.2352, + "step": 48700 + }, + { + "epoch": 2.44, + "grad_norm": 3.0430312156677246, + "learning_rate": 1.0295276381909548e-05, + "loss": 1.2302, + "step": 48800 + }, + { + "epoch": 2.44, + "grad_norm": 4.847692012786865, + "learning_rate": 1.0275175879396986e-05, + "loss": 1.23, + "step": 48900 + }, + { + "epoch": 2.45, + "grad_norm": 4.695472240447998, + "learning_rate": 1.0255075376884423e-05, + "loss": 1.2543, + "step": 49000 + }, + { + "epoch": 2.46, + "grad_norm": 4.463906764984131, + "learning_rate": 1.0234974874371859e-05, + "loss": 1.2596, + "step": 49100 + }, + { + "epoch": 2.46, + "grad_norm": 4.5606770515441895, + "learning_rate": 1.0214874371859299e-05, + "loss": 1.235, + "step": 49200 + }, + { + "epoch": 2.46, + "grad_norm": 5.294122219085693, + "learning_rate": 1.0194773869346735e-05, + "loss": 1.2505, + "step": 49300 + }, + { + "epoch": 2.47, + "grad_norm": 3.5599803924560547, + "learning_rate": 1.0174673366834171e-05, + "loss": 1.2817, + "step": 49400 + }, + { + "epoch": 2.48, + "grad_norm": 3.720597982406616, + "learning_rate": 1.0154572864321607e-05, + "loss": 1.2628, + "step": 49500 + }, + { + "epoch": 2.48, + "grad_norm": 4.774421215057373, + "learning_rate": 1.0134472361809047e-05, + "loss": 1.2705, + "step": 49600 + }, + { + "epoch": 2.48, + "grad_norm": 5.8144330978393555, + "learning_rate": 1.0114371859296483e-05, + "loss": 1.2982, + "step": 49700 + }, + { + "epoch": 2.49, + "grad_norm": 6.6790385246276855, + "learning_rate": 1.009427135678392e-05, + "loss": 1.2301, + "step": 49800 + }, + { + "epoch": 2.5, + "grad_norm": 4.106939792633057, + "learning_rate": 1.0074170854271358e-05, + "loss": 1.2355, + "step": 49900 + }, + { + "epoch": 2.5, + "grad_norm": 3.354093551635742, + "learning_rate": 1.0054070351758796e-05, + "loss": 1.2459, + "step": 50000 + }, + { + "epoch": 2.5, + "eval_loss": 1.3116555213928223, + "eval_runtime": 21.8654, + "eval_samples_per_second": 45.734, + "eval_steps_per_second": 5.717, + "step": 50000 + }, + { + "epoch": 2.5, + "grad_norm": 5.311431407928467, + "learning_rate": 1.0033969849246232e-05, + "loss": 1.2402, + "step": 50100 + }, + { + "epoch": 2.51, + "grad_norm": 5.163784503936768, + "learning_rate": 1.0013869346733668e-05, + "loss": 1.2329, + "step": 50200 + }, + { + "epoch": 2.52, + "grad_norm": 5.684957981109619, + "learning_rate": 9.993768844221106e-06, + "loss": 1.2403, + "step": 50300 + }, + { + "epoch": 2.52, + "grad_norm": 6.14009952545166, + "learning_rate": 9.973668341708544e-06, + "loss": 1.2217, + "step": 50400 + }, + { + "epoch": 2.52, + "grad_norm": 6.6271796226501465, + "learning_rate": 9.95356783919598e-06, + "loss": 1.2122, + "step": 50500 + }, + { + "epoch": 2.53, + "grad_norm": 4.011633396148682, + "learning_rate": 9.933467336683418e-06, + "loss": 1.2242, + "step": 50600 + }, + { + "epoch": 2.54, + "grad_norm": 4.849850177764893, + "learning_rate": 9.913366834170856e-06, + "loss": 1.2349, + "step": 50700 + }, + { + "epoch": 2.54, + "grad_norm": 3.841789484024048, + "learning_rate": 9.893266331658293e-06, + "loss": 1.2556, + "step": 50800 + }, + { + "epoch": 2.54, + "grad_norm": 7.193374156951904, + "learning_rate": 9.87316582914573e-06, + "loss": 1.2119, + "step": 50900 + }, + { + "epoch": 2.55, + "grad_norm": 3.556542158126831, + "learning_rate": 9.853065326633167e-06, + "loss": 1.2426, + "step": 51000 + }, + { + "epoch": 2.56, + "grad_norm": 6.54746150970459, + "learning_rate": 9.832964824120603e-06, + "loss": 1.1936, + "step": 51100 + }, + { + "epoch": 2.56, + "grad_norm": 4.4405951499938965, + "learning_rate": 9.812864321608041e-06, + "loss": 1.2591, + "step": 51200 + }, + { + "epoch": 2.56, + "grad_norm": 5.398285865783691, + "learning_rate": 9.792763819095477e-06, + "loss": 1.2375, + "step": 51300 + }, + { + "epoch": 2.57, + "grad_norm": 5.835482120513916, + "learning_rate": 9.772864321608041e-06, + "loss": 1.2626, + "step": 51400 + }, + { + "epoch": 2.58, + "grad_norm": 5.3824543952941895, + "learning_rate": 9.752763819095478e-06, + "loss": 1.2145, + "step": 51500 + }, + { + "epoch": 2.58, + "grad_norm": 4.847600936889648, + "learning_rate": 9.732663316582916e-06, + "loss": 1.2699, + "step": 51600 + }, + { + "epoch": 2.58, + "grad_norm": 4.644218921661377, + "learning_rate": 9.712562814070352e-06, + "loss": 1.2669, + "step": 51700 + }, + { + "epoch": 2.59, + "grad_norm": 5.046612739562988, + "learning_rate": 9.69246231155779e-06, + "loss": 1.2425, + "step": 51800 + }, + { + "epoch": 2.59, + "grad_norm": 3.9644362926483154, + "learning_rate": 9.672361809045226e-06, + "loss": 1.2802, + "step": 51900 + }, + { + "epoch": 2.6, + "grad_norm": 4.848786354064941, + "learning_rate": 9.652261306532664e-06, + "loss": 1.2257, + "step": 52000 + }, + { + "epoch": 2.6, + "grad_norm": 5.159448623657227, + "learning_rate": 9.6321608040201e-06, + "loss": 1.2156, + "step": 52100 + }, + { + "epoch": 2.61, + "grad_norm": 6.021341800689697, + "learning_rate": 9.612060301507538e-06, + "loss": 1.2122, + "step": 52200 + }, + { + "epoch": 2.62, + "grad_norm": 7.450246334075928, + "learning_rate": 9.591959798994975e-06, + "loss": 1.2452, + "step": 52300 + }, + { + "epoch": 2.62, + "grad_norm": 4.6322503089904785, + "learning_rate": 9.571859296482413e-06, + "loss": 1.2615, + "step": 52400 + }, + { + "epoch": 2.62, + "grad_norm": 9.119827270507812, + "learning_rate": 9.551758793969849e-06, + "loss": 1.2484, + "step": 52500 + }, + { + "epoch": 2.63, + "grad_norm": 7.022191047668457, + "learning_rate": 9.531658291457287e-06, + "loss": 1.2413, + "step": 52600 + }, + { + "epoch": 2.63, + "grad_norm": 6.008714199066162, + "learning_rate": 9.511557788944725e-06, + "loss": 1.2545, + "step": 52700 + }, + { + "epoch": 2.64, + "grad_norm": 4.883365631103516, + "learning_rate": 9.491457286432161e-06, + "loss": 1.256, + "step": 52800 + }, + { + "epoch": 2.65, + "grad_norm": 4.666494846343994, + "learning_rate": 9.4713567839196e-06, + "loss": 1.212, + "step": 52900 + }, + { + "epoch": 2.65, + "grad_norm": 6.942872524261475, + "learning_rate": 9.451256281407035e-06, + "loss": 1.2539, + "step": 53000 + }, + { + "epoch": 2.66, + "grad_norm": 8.57226848602295, + "learning_rate": 9.431155778894473e-06, + "loss": 1.238, + "step": 53100 + }, + { + "epoch": 2.66, + "grad_norm": 3.5034737586975098, + "learning_rate": 9.411055276381911e-06, + "loss": 1.1608, + "step": 53200 + }, + { + "epoch": 2.67, + "grad_norm": 4.17569637298584, + "learning_rate": 9.390954773869348e-06, + "loss": 1.2335, + "step": 53300 + }, + { + "epoch": 2.67, + "grad_norm": 5.2211012840271, + "learning_rate": 9.370854271356786e-06, + "loss": 1.2664, + "step": 53400 + }, + { + "epoch": 2.67, + "grad_norm": 3.3811118602752686, + "learning_rate": 9.350753768844222e-06, + "loss": 1.2317, + "step": 53500 + }, + { + "epoch": 2.68, + "grad_norm": 6.415603160858154, + "learning_rate": 9.33065326633166e-06, + "loss": 1.2613, + "step": 53600 + }, + { + "epoch": 2.69, + "grad_norm": 4.720609188079834, + "learning_rate": 9.310552763819096e-06, + "loss": 1.2124, + "step": 53700 + }, + { + "epoch": 2.69, + "grad_norm": 5.3697710037231445, + "learning_rate": 9.290452261306533e-06, + "loss": 1.2388, + "step": 53800 + }, + { + "epoch": 2.69, + "grad_norm": 4.376136302947998, + "learning_rate": 9.27035175879397e-06, + "loss": 1.2495, + "step": 53900 + }, + { + "epoch": 2.7, + "grad_norm": 3.973159074783325, + "learning_rate": 9.250251256281407e-06, + "loss": 1.2389, + "step": 54000 + }, + { + "epoch": 2.71, + "grad_norm": 6.894681930541992, + "learning_rate": 9.230150753768845e-06, + "loss": 1.2056, + "step": 54100 + }, + { + "epoch": 2.71, + "grad_norm": 4.781852722167969, + "learning_rate": 9.210050251256281e-06, + "loss": 1.2316, + "step": 54200 + }, + { + "epoch": 2.71, + "grad_norm": 4.408322334289551, + "learning_rate": 9.189949748743719e-06, + "loss": 1.2945, + "step": 54300 + }, + { + "epoch": 2.72, + "grad_norm": 4.875626564025879, + "learning_rate": 9.169849246231157e-06, + "loss": 1.2458, + "step": 54400 + }, + { + "epoch": 2.73, + "grad_norm": 5.88706111907959, + "learning_rate": 9.149748743718593e-06, + "loss": 1.2257, + "step": 54500 + }, + { + "epoch": 2.73, + "grad_norm": 4.785450458526611, + "learning_rate": 9.129648241206031e-06, + "loss": 1.2372, + "step": 54600 + }, + { + "epoch": 2.73, + "grad_norm": 4.651752948760986, + "learning_rate": 9.109547738693468e-06, + "loss": 1.2146, + "step": 54700 + }, + { + "epoch": 2.74, + "grad_norm": 5.303548336029053, + "learning_rate": 9.089447236180905e-06, + "loss": 1.2307, + "step": 54800 + }, + { + "epoch": 2.75, + "grad_norm": 4.032742977142334, + "learning_rate": 9.069346733668343e-06, + "loss": 1.2467, + "step": 54900 + }, + { + "epoch": 2.75, + "grad_norm": 4.288597583770752, + "learning_rate": 9.04924623115578e-06, + "loss": 1.248, + "step": 55000 + }, + { + "epoch": 2.75, + "grad_norm": 5.981525897979736, + "learning_rate": 9.029145728643218e-06, + "loss": 1.2344, + "step": 55100 + }, + { + "epoch": 2.76, + "grad_norm": 4.837640762329102, + "learning_rate": 9.009045226130654e-06, + "loss": 1.2305, + "step": 55200 + }, + { + "epoch": 2.77, + "grad_norm": 5.082337856292725, + "learning_rate": 8.988944723618092e-06, + "loss": 1.2199, + "step": 55300 + }, + { + "epoch": 2.77, + "grad_norm": 5.879444599151611, + "learning_rate": 8.968844221105528e-06, + "loss": 1.2158, + "step": 55400 + }, + { + "epoch": 2.77, + "grad_norm": 4.926747798919678, + "learning_rate": 8.948944723618092e-06, + "loss": 1.2575, + "step": 55500 + }, + { + "epoch": 2.78, + "grad_norm": 4.744166851043701, + "learning_rate": 8.928844221105529e-06, + "loss": 1.2325, + "step": 55600 + }, + { + "epoch": 2.79, + "grad_norm": 5.994776725769043, + "learning_rate": 8.908743718592967e-06, + "loss": 1.2199, + "step": 55700 + }, + { + "epoch": 2.79, + "grad_norm": 4.0552215576171875, + "learning_rate": 8.888643216080403e-06, + "loss": 1.2214, + "step": 55800 + }, + { + "epoch": 2.79, + "grad_norm": 6.152566432952881, + "learning_rate": 8.868542713567841e-06, + "loss": 1.2565, + "step": 55900 + }, + { + "epoch": 2.8, + "grad_norm": 5.718895435333252, + "learning_rate": 8.848442211055277e-06, + "loss": 1.2452, + "step": 56000 + }, + { + "epoch": 2.81, + "grad_norm": 6.39285135269165, + "learning_rate": 8.828341708542715e-06, + "loss": 1.2, + "step": 56100 + }, + { + "epoch": 2.81, + "grad_norm": 6.312516689300537, + "learning_rate": 8.808241206030151e-06, + "loss": 1.2739, + "step": 56200 + }, + { + "epoch": 2.81, + "grad_norm": 5.090448379516602, + "learning_rate": 8.78814070351759e-06, + "loss": 1.2405, + "step": 56300 + }, + { + "epoch": 2.82, + "grad_norm": 3.6773719787597656, + "learning_rate": 8.768040201005026e-06, + "loss": 1.2231, + "step": 56400 + }, + { + "epoch": 2.83, + "grad_norm": 3.831404209136963, + "learning_rate": 8.747939698492462e-06, + "loss": 1.2068, + "step": 56500 + }, + { + "epoch": 2.83, + "grad_norm": 6.460518836975098, + "learning_rate": 8.7278391959799e-06, + "loss": 1.2585, + "step": 56600 + }, + { + "epoch": 2.83, + "grad_norm": 5.495750427246094, + "learning_rate": 8.707738693467336e-06, + "loss": 1.2489, + "step": 56700 + }, + { + "epoch": 2.84, + "grad_norm": 3.785914421081543, + "learning_rate": 8.687638190954774e-06, + "loss": 1.1986, + "step": 56800 + }, + { + "epoch": 2.84, + "grad_norm": 3.443301200866699, + "learning_rate": 8.667537688442212e-06, + "loss": 1.233, + "step": 56900 + }, + { + "epoch": 2.85, + "grad_norm": 7.463976860046387, + "learning_rate": 8.647437185929648e-06, + "loss": 1.214, + "step": 57000 + }, + { + "epoch": 2.85, + "grad_norm": 5.533621788024902, + "learning_rate": 8.627336683417086e-06, + "loss": 1.2116, + "step": 57100 + }, + { + "epoch": 2.86, + "grad_norm": 5.238498210906982, + "learning_rate": 8.607236180904524e-06, + "loss": 1.1792, + "step": 57200 + }, + { + "epoch": 2.87, + "grad_norm": 5.2816996574401855, + "learning_rate": 8.58713567839196e-06, + "loss": 1.2359, + "step": 57300 + }, + { + "epoch": 2.87, + "grad_norm": 4.712213516235352, + "learning_rate": 8.567035175879399e-06, + "loss": 1.1824, + "step": 57400 + }, + { + "epoch": 2.88, + "grad_norm": 3.9056365489959717, + "learning_rate": 8.546934673366835e-06, + "loss": 1.2206, + "step": 57500 + }, + { + "epoch": 2.88, + "grad_norm": 6.587601184844971, + "learning_rate": 8.527035175879397e-06, + "loss": 1.2542, + "step": 57600 + }, + { + "epoch": 2.88, + "grad_norm": 4.349347114562988, + "learning_rate": 8.506934673366835e-06, + "loss": 1.2516, + "step": 57700 + }, + { + "epoch": 2.89, + "grad_norm": 4.775893211364746, + "learning_rate": 8.486834170854272e-06, + "loss": 1.18, + "step": 57800 + }, + { + "epoch": 2.9, + "grad_norm": 4.952343940734863, + "learning_rate": 8.46673366834171e-06, + "loss": 1.2287, + "step": 57900 + }, + { + "epoch": 2.9, + "grad_norm": 5.0424089431762695, + "learning_rate": 8.446834170854272e-06, + "loss": 1.2044, + "step": 58000 + }, + { + "epoch": 2.91, + "grad_norm": 5.429243564605713, + "learning_rate": 8.426733668341708e-06, + "loss": 1.1936, + "step": 58100 + }, + { + "epoch": 2.91, + "grad_norm": 4.514014720916748, + "learning_rate": 8.406633165829146e-06, + "loss": 1.2417, + "step": 58200 + }, + { + "epoch": 2.92, + "grad_norm": 4.364452362060547, + "learning_rate": 8.386532663316584e-06, + "loss": 1.2004, + "step": 58300 + }, + { + "epoch": 2.92, + "grad_norm": 3.5190353393554688, + "learning_rate": 8.36643216080402e-06, + "loss": 1.2144, + "step": 58400 + }, + { + "epoch": 2.92, + "grad_norm": 5.794633865356445, + "learning_rate": 8.346331658291458e-06, + "loss": 1.2072, + "step": 58500 + }, + { + "epoch": 2.93, + "grad_norm": 4.060710430145264, + "learning_rate": 8.326231155778895e-06, + "loss": 1.225, + "step": 58600 + }, + { + "epoch": 2.94, + "grad_norm": 4.3035664558410645, + "learning_rate": 8.306130653266333e-06, + "loss": 1.1935, + "step": 58700 + }, + { + "epoch": 2.94, + "grad_norm": 4.8658246994018555, + "learning_rate": 8.28603015075377e-06, + "loss": 1.2316, + "step": 58800 + }, + { + "epoch": 2.94, + "grad_norm": 3.5524916648864746, + "learning_rate": 8.265929648241207e-06, + "loss": 1.2119, + "step": 58900 + }, + { + "epoch": 2.95, + "grad_norm": 3.966935157775879, + "learning_rate": 8.245829145728645e-06, + "loss": 1.2427, + "step": 59000 + }, + { + "epoch": 2.96, + "grad_norm": 5.453131675720215, + "learning_rate": 8.225728643216081e-06, + "loss": 1.1948, + "step": 59100 + }, + { + "epoch": 2.96, + "grad_norm": 6.029975414276123, + "learning_rate": 8.20562814070352e-06, + "loss": 1.1688, + "step": 59200 + }, + { + "epoch": 2.96, + "grad_norm": 3.311718225479126, + "learning_rate": 8.185527638190955e-06, + "loss": 1.2195, + "step": 59300 + }, + { + "epoch": 2.97, + "grad_norm": 3.63813853263855, + "learning_rate": 8.165427135678393e-06, + "loss": 1.2288, + "step": 59400 + }, + { + "epoch": 2.98, + "grad_norm": 4.470839500427246, + "learning_rate": 8.14532663316583e-06, + "loss": 1.2273, + "step": 59500 + }, + { + "epoch": 2.98, + "grad_norm": 4.462855815887451, + "learning_rate": 8.125226130653266e-06, + "loss": 1.2066, + "step": 59600 + }, + { + "epoch": 2.98, + "grad_norm": 4.757040023803711, + "learning_rate": 8.105125628140704e-06, + "loss": 1.2159, + "step": 59700 + }, + { + "epoch": 2.99, + "grad_norm": 5.637049674987793, + "learning_rate": 8.08502512562814e-06, + "loss": 1.2449, + "step": 59800 + }, + { + "epoch": 3.0, + "grad_norm": 5.578622341156006, + "learning_rate": 8.064924623115578e-06, + "loss": 1.2401, + "step": 59900 + }, + { + "epoch": 3.0, + "grad_norm": 6.322601318359375, + "learning_rate": 8.044824120603014e-06, + "loss": 1.2546, + "step": 60000 + }, + { + "epoch": 3.0, + "eval_loss": 1.232736587524414, + "eval_runtime": 21.8587, + "eval_samples_per_second": 45.748, + "eval_steps_per_second": 5.719, + "step": 60000 + }, + { + "epoch": 3.0, + "grad_norm": 4.387004375457764, + "learning_rate": 8.024723618090452e-06, + "loss": 1.1591, + "step": 60100 + }, + { + "epoch": 3.01, + "grad_norm": 3.7655532360076904, + "learning_rate": 8.00462311557789e-06, + "loss": 1.2148, + "step": 60200 + }, + { + "epoch": 3.02, + "grad_norm": 4.917843341827393, + "learning_rate": 7.984522613065327e-06, + "loss": 1.1924, + "step": 60300 + }, + { + "epoch": 3.02, + "grad_norm": 4.71078634262085, + "learning_rate": 7.964422110552765e-06, + "loss": 1.1649, + "step": 60400 + }, + { + "epoch": 3.02, + "grad_norm": 6.106967449188232, + "learning_rate": 7.944321608040203e-06, + "loss": 1.2001, + "step": 60500 + }, + { + "epoch": 3.03, + "grad_norm": 5.224365711212158, + "learning_rate": 7.924221105527639e-06, + "loss": 1.1912, + "step": 60600 + }, + { + "epoch": 3.04, + "grad_norm": 6.110058784484863, + "learning_rate": 7.904120603015077e-06, + "loss": 1.2004, + "step": 60700 + }, + { + "epoch": 3.04, + "grad_norm": 4.606750965118408, + "learning_rate": 7.884020100502513e-06, + "loss": 1.1824, + "step": 60800 + }, + { + "epoch": 3.04, + "grad_norm": 4.328644275665283, + "learning_rate": 7.863919597989951e-06, + "loss": 1.1818, + "step": 60900 + }, + { + "epoch": 3.05, + "grad_norm": 5.017879009246826, + "learning_rate": 7.843819095477387e-06, + "loss": 1.1843, + "step": 61000 + }, + { + "epoch": 3.06, + "grad_norm": 6.072721481323242, + "learning_rate": 7.823718592964825e-06, + "loss": 1.1876, + "step": 61100 + }, + { + "epoch": 3.06, + "grad_norm": 5.169823169708252, + "learning_rate": 7.803618090452262e-06, + "loss": 1.1762, + "step": 61200 + }, + { + "epoch": 3.06, + "grad_norm": 5.349250793457031, + "learning_rate": 7.7835175879397e-06, + "loss": 1.1541, + "step": 61300 + }, + { + "epoch": 3.07, + "grad_norm": 5.824612140655518, + "learning_rate": 7.763417085427136e-06, + "loss": 1.1696, + "step": 61400 + }, + { + "epoch": 3.08, + "grad_norm": 6.2018938064575195, + "learning_rate": 7.743316582914574e-06, + "loss": 1.1621, + "step": 61500 + }, + { + "epoch": 3.08, + "grad_norm": 4.709869384765625, + "learning_rate": 7.72321608040201e-06, + "loss": 1.1777, + "step": 61600 + }, + { + "epoch": 3.08, + "grad_norm": 4.259114742279053, + "learning_rate": 7.703115577889448e-06, + "loss": 1.1883, + "step": 61700 + }, + { + "epoch": 3.09, + "grad_norm": 5.505044460296631, + "learning_rate": 7.683015075376884e-06, + "loss": 1.1896, + "step": 61800 + }, + { + "epoch": 3.1, + "grad_norm": 5.121050834655762, + "learning_rate": 7.662914572864322e-06, + "loss": 1.1533, + "step": 61900 + }, + { + "epoch": 3.1, + "grad_norm": 3.265988349914551, + "learning_rate": 7.642814070351759e-06, + "loss": 1.1746, + "step": 62000 + }, + { + "epoch": 3.1, + "grad_norm": 4.327176094055176, + "learning_rate": 7.622713567839196e-06, + "loss": 1.2136, + "step": 62100 + }, + { + "epoch": 3.11, + "grad_norm": 6.783113479614258, + "learning_rate": 7.602613065326634e-06, + "loss": 1.1538, + "step": 62200 + }, + { + "epoch": 3.12, + "grad_norm": 6.27109956741333, + "learning_rate": 7.582512562814071e-06, + "loss": 1.1995, + "step": 62300 + }, + { + "epoch": 3.12, + "grad_norm": 6.903465270996094, + "learning_rate": 7.562412060301508e-06, + "loss": 1.1468, + "step": 62400 + }, + { + "epoch": 3.12, + "grad_norm": 4.696254253387451, + "learning_rate": 7.542311557788945e-06, + "loss": 1.199, + "step": 62500 + }, + { + "epoch": 3.13, + "grad_norm": 4.280835151672363, + "learning_rate": 7.522211055276382e-06, + "loss": 1.1616, + "step": 62600 + }, + { + "epoch": 3.13, + "grad_norm": 5.0677924156188965, + "learning_rate": 7.5021105527638195e-06, + "loss": 1.1575, + "step": 62700 + }, + { + "epoch": 3.14, + "grad_norm": 6.276374816894531, + "learning_rate": 7.4820100502512574e-06, + "loss": 1.137, + "step": 62800 + }, + { + "epoch": 3.15, + "grad_norm": 4.777525424957275, + "learning_rate": 7.461909547738694e-06, + "loss": 1.201, + "step": 62900 + }, + { + "epoch": 3.15, + "grad_norm": 4.285521030426025, + "learning_rate": 7.441809045226132e-06, + "loss": 1.1924, + "step": 63000 + }, + { + "epoch": 3.15, + "grad_norm": 6.18180513381958, + "learning_rate": 7.421708542713568e-06, + "loss": 1.1347, + "step": 63100 + }, + { + "epoch": 3.16, + "grad_norm": 6.091145038604736, + "learning_rate": 7.401608040201006e-06, + "loss": 1.1795, + "step": 63200 + }, + { + "epoch": 3.17, + "grad_norm": 5.018629550933838, + "learning_rate": 7.381507537688442e-06, + "loss": 1.1755, + "step": 63300 + }, + { + "epoch": 3.17, + "grad_norm": 5.406840801239014, + "learning_rate": 7.36140703517588e-06, + "loss": 1.1869, + "step": 63400 + }, + { + "epoch": 3.17, + "grad_norm": 4.727605819702148, + "learning_rate": 7.341306532663317e-06, + "loss": 1.1772, + "step": 63500 + }, + { + "epoch": 3.18, + "grad_norm": 5.590334892272949, + "learning_rate": 7.3212060301507544e-06, + "loss": 1.1686, + "step": 63600 + }, + { + "epoch": 3.19, + "grad_norm": 5.156420707702637, + "learning_rate": 7.3011055276381916e-06, + "loss": 1.1696, + "step": 63700 + }, + { + "epoch": 3.19, + "grad_norm": 6.83641242980957, + "learning_rate": 7.2810050251256296e-06, + "loss": 1.1732, + "step": 63800 + }, + { + "epoch": 3.19, + "grad_norm": 4.086230278015137, + "learning_rate": 7.260904522613066e-06, + "loss": 1.1536, + "step": 63900 + }, + { + "epoch": 3.2, + "grad_norm": 7.394796371459961, + "learning_rate": 7.241005025125629e-06, + "loss": 1.1636, + "step": 64000 + }, + { + "epoch": 3.21, + "grad_norm": 6.290234088897705, + "learning_rate": 7.220904522613066e-06, + "loss": 1.1486, + "step": 64100 + }, + { + "epoch": 3.21, + "grad_norm": 5.5817036628723145, + "learning_rate": 7.200804020100503e-06, + "loss": 1.1424, + "step": 64200 + }, + { + "epoch": 3.21, + "grad_norm": 5.681445598602295, + "learning_rate": 7.1807035175879405e-06, + "loss": 1.1591, + "step": 64300 + }, + { + "epoch": 3.22, + "grad_norm": 6.768691062927246, + "learning_rate": 7.160603015075377e-06, + "loss": 1.1293, + "step": 64400 + }, + { + "epoch": 3.23, + "grad_norm": 5.4178595542907715, + "learning_rate": 7.140502512562815e-06, + "loss": 1.1729, + "step": 64500 + }, + { + "epoch": 3.23, + "grad_norm": 4.215164661407471, + "learning_rate": 7.120402010050251e-06, + "loss": 1.1783, + "step": 64600 + }, + { + "epoch": 3.23, + "grad_norm": 5.66365385055542, + "learning_rate": 7.100301507537689e-06, + "loss": 1.1967, + "step": 64700 + }, + { + "epoch": 3.24, + "grad_norm": 5.554622650146484, + "learning_rate": 7.080201005025126e-06, + "loss": 1.1466, + "step": 64800 + }, + { + "epoch": 3.25, + "grad_norm": 5.001458644866943, + "learning_rate": 7.060100502512563e-06, + "loss": 1.1732, + "step": 64900 + }, + { + "epoch": 3.25, + "grad_norm": 2.8027803897857666, + "learning_rate": 7.04e-06, + "loss": 1.152, + "step": 65000 + }, + { + "epoch": 3.25, + "grad_norm": 4.733661651611328, + "learning_rate": 7.019899497487438e-06, + "loss": 1.1155, + "step": 65100 + }, + { + "epoch": 3.26, + "grad_norm": 7.268320083618164, + "learning_rate": 6.999798994974875e-06, + "loss": 1.1703, + "step": 65200 + }, + { + "epoch": 3.27, + "grad_norm": 6.752691745758057, + "learning_rate": 6.979698492462313e-06, + "loss": 1.1986, + "step": 65300 + }, + { + "epoch": 3.27, + "grad_norm": 5.244182586669922, + "learning_rate": 6.959597989949749e-06, + "loss": 1.148, + "step": 65400 + }, + { + "epoch": 3.27, + "grad_norm": 5.481043815612793, + "learning_rate": 6.939497487437187e-06, + "loss": 1.1811, + "step": 65500 + }, + { + "epoch": 3.28, + "grad_norm": 5.892518997192383, + "learning_rate": 6.919396984924623e-06, + "loss": 1.1574, + "step": 65600 + }, + { + "epoch": 3.29, + "grad_norm": 5.347742557525635, + "learning_rate": 6.899296482412061e-06, + "loss": 1.1353, + "step": 65700 + }, + { + "epoch": 3.29, + "grad_norm": 5.078448295593262, + "learning_rate": 6.879195979899498e-06, + "loss": 1.1398, + "step": 65800 + }, + { + "epoch": 3.29, + "grad_norm": 4.15362548828125, + "learning_rate": 6.859095477386935e-06, + "loss": 1.1556, + "step": 65900 + }, + { + "epoch": 3.3, + "grad_norm": 4.748194694519043, + "learning_rate": 6.8389949748743725e-06, + "loss": 1.1929, + "step": 66000 + }, + { + "epoch": 3.31, + "grad_norm": 5.56561803817749, + "learning_rate": 6.81889447236181e-06, + "loss": 1.1907, + "step": 66100 + }, + { + "epoch": 3.31, + "grad_norm": 4.8242316246032715, + "learning_rate": 6.798793969849247e-06, + "loss": 1.166, + "step": 66200 + }, + { + "epoch": 3.31, + "grad_norm": 5.7045087814331055, + "learning_rate": 6.778693467336685e-06, + "loss": 1.1579, + "step": 66300 + }, + { + "epoch": 3.32, + "grad_norm": 4.583883285522461, + "learning_rate": 6.758592964824121e-06, + "loss": 1.1259, + "step": 66400 + }, + { + "epoch": 3.33, + "grad_norm": 5.085745811462402, + "learning_rate": 6.738492462311559e-06, + "loss": 1.1431, + "step": 66500 + }, + { + "epoch": 3.33, + "grad_norm": 4.655329704284668, + "learning_rate": 6.718391959798995e-06, + "loss": 1.1681, + "step": 66600 + }, + { + "epoch": 3.33, + "grad_norm": 4.375367164611816, + "learning_rate": 6.698291457286433e-06, + "loss": 1.1428, + "step": 66700 + }, + { + "epoch": 3.34, + "grad_norm": 5.143255710601807, + "learning_rate": 6.6781909547738695e-06, + "loss": 1.1676, + "step": 66800 + }, + { + "epoch": 3.34, + "grad_norm": 5.463631629943848, + "learning_rate": 6.658090452261307e-06, + "loss": 1.1742, + "step": 66900 + }, + { + "epoch": 3.35, + "grad_norm": 5.112860679626465, + "learning_rate": 6.637989949748745e-06, + "loss": 1.1655, + "step": 67000 + }, + { + "epoch": 3.35, + "grad_norm": 3.88566517829895, + "learning_rate": 6.617889447236181e-06, + "loss": 1.1428, + "step": 67100 + }, + { + "epoch": 3.36, + "grad_norm": 5.075991153717041, + "learning_rate": 6.597788944723619e-06, + "loss": 1.1853, + "step": 67200 + }, + { + "epoch": 3.37, + "grad_norm": 6.4206061363220215, + "learning_rate": 6.577688442211055e-06, + "loss": 1.1622, + "step": 67300 + }, + { + "epoch": 3.37, + "grad_norm": 4.789801597595215, + "learning_rate": 6.557587939698493e-06, + "loss": 1.2021, + "step": 67400 + }, + { + "epoch": 3.38, + "grad_norm": 4.722198486328125, + "learning_rate": 6.53748743718593e-06, + "loss": 1.1515, + "step": 67500 + }, + { + "epoch": 3.38, + "grad_norm": 4.774144649505615, + "learning_rate": 6.517386934673367e-06, + "loss": 1.1429, + "step": 67600 + }, + { + "epoch": 3.38, + "grad_norm": 3.848876953125, + "learning_rate": 6.49748743718593e-06, + "loss": 1.1719, + "step": 67700 + }, + { + "epoch": 3.39, + "grad_norm": 6.2731804847717285, + "learning_rate": 6.477386934673368e-06, + "loss": 1.1513, + "step": 67800 + }, + { + "epoch": 3.4, + "grad_norm": 6.135923385620117, + "learning_rate": 6.457286432160804e-06, + "loss": 1.1758, + "step": 67900 + }, + { + "epoch": 3.4, + "grad_norm": 5.437047958374023, + "learning_rate": 6.437185929648242e-06, + "loss": 1.174, + "step": 68000 + }, + { + "epoch": 3.41, + "grad_norm": 5.043646335601807, + "learning_rate": 6.417085427135678e-06, + "loss": 1.171, + "step": 68100 + }, + { + "epoch": 3.41, + "grad_norm": 4.104462623596191, + "learning_rate": 6.396984924623116e-06, + "loss": 1.1712, + "step": 68200 + }, + { + "epoch": 3.42, + "grad_norm": 2.740678310394287, + "learning_rate": 6.376884422110553e-06, + "loss": 1.1127, + "step": 68300 + }, + { + "epoch": 3.42, + "grad_norm": 5.4752936363220215, + "learning_rate": 6.3567839195979905e-06, + "loss": 1.1971, + "step": 68400 + }, + { + "epoch": 3.42, + "grad_norm": 7.34414529800415, + "learning_rate": 6.336683417085428e-06, + "loss": 1.1714, + "step": 68500 + }, + { + "epoch": 3.43, + "grad_norm": 3.3866333961486816, + "learning_rate": 6.316582914572866e-06, + "loss": 1.1604, + "step": 68600 + }, + { + "epoch": 3.44, + "grad_norm": 5.284789085388184, + "learning_rate": 6.296482412060302e-06, + "loss": 1.1792, + "step": 68700 + }, + { + "epoch": 3.44, + "grad_norm": 4.47866678237915, + "learning_rate": 6.27638190954774e-06, + "loss": 1.1343, + "step": 68800 + }, + { + "epoch": 3.44, + "grad_norm": 6.508190631866455, + "learning_rate": 6.256281407035176e-06, + "loss": 1.1286, + "step": 68900 + }, + { + "epoch": 3.45, + "grad_norm": 5.973139762878418, + "learning_rate": 6.236180904522614e-06, + "loss": 1.1774, + "step": 69000 + }, + { + "epoch": 3.46, + "grad_norm": 4.717242240905762, + "learning_rate": 6.21608040201005e-06, + "loss": 1.1273, + "step": 69100 + }, + { + "epoch": 3.46, + "grad_norm": 5.430871486663818, + "learning_rate": 6.195979899497488e-06, + "loss": 1.1764, + "step": 69200 + }, + { + "epoch": 3.46, + "grad_norm": 4.484432697296143, + "learning_rate": 6.1758793969849255e-06, + "loss": 1.154, + "step": 69300 + }, + { + "epoch": 3.47, + "grad_norm": 4.041011333465576, + "learning_rate": 6.155778894472362e-06, + "loss": 1.1614, + "step": 69400 + }, + { + "epoch": 3.48, + "grad_norm": 4.026901721954346, + "learning_rate": 6.1356783919598e-06, + "loss": 1.1376, + "step": 69500 + }, + { + "epoch": 3.48, + "grad_norm": 6.2372660636901855, + "learning_rate": 6.115577889447236e-06, + "loss": 1.1556, + "step": 69600 + }, + { + "epoch": 3.48, + "grad_norm": 5.324029445648193, + "learning_rate": 6.095477386934674e-06, + "loss": 1.1533, + "step": 69700 + }, + { + "epoch": 3.49, + "grad_norm": 7.105170726776123, + "learning_rate": 6.07537688442211e-06, + "loss": 1.1895, + "step": 69800 + }, + { + "epoch": 3.5, + "grad_norm": 23.556692123413086, + "learning_rate": 6.055276381909548e-06, + "loss": 1.1569, + "step": 69900 + }, + { + "epoch": 3.5, + "grad_norm": 5.792446613311768, + "learning_rate": 6.035175879396985e-06, + "loss": 1.1488, + "step": 70000 + }, + { + "epoch": 3.5, + "eval_loss": 1.2055375576019287, + "eval_runtime": 21.7807, + "eval_samples_per_second": 45.912, + "eval_steps_per_second": 5.739, + "step": 70000 + }, + { + "epoch": 3.5, + "grad_norm": 5.223493576049805, + "learning_rate": 6.0150753768844225e-06, + "loss": 1.144, + "step": 70100 + }, + { + "epoch": 3.51, + "grad_norm": 4.998097896575928, + "learning_rate": 5.99497487437186e-06, + "loss": 1.154, + "step": 70200 + }, + { + "epoch": 3.52, + "grad_norm": 6.396411895751953, + "learning_rate": 5.975075376884423e-06, + "loss": 1.1398, + "step": 70300 + }, + { + "epoch": 3.52, + "grad_norm": 4.5480427742004395, + "learning_rate": 5.954974874371859e-06, + "loss": 1.1753, + "step": 70400 + }, + { + "epoch": 3.52, + "grad_norm": 5.8266167640686035, + "learning_rate": 5.934874371859297e-06, + "loss": 1.1481, + "step": 70500 + }, + { + "epoch": 3.53, + "grad_norm": 3.3341739177703857, + "learning_rate": 5.9147738693467334e-06, + "loss": 1.1801, + "step": 70600 + }, + { + "epoch": 3.54, + "grad_norm": 5.731110572814941, + "learning_rate": 5.894673366834171e-06, + "loss": 1.1945, + "step": 70700 + }, + { + "epoch": 3.54, + "grad_norm": 5.0252251625061035, + "learning_rate": 5.8745728643216085e-06, + "loss": 1.1171, + "step": 70800 + }, + { + "epoch": 3.54, + "grad_norm": 6.497035503387451, + "learning_rate": 5.854472361809046e-06, + "loss": 1.1299, + "step": 70900 + }, + { + "epoch": 3.55, + "grad_norm": 3.014439582824707, + "learning_rate": 5.834371859296483e-06, + "loss": 1.1135, + "step": 71000 + }, + { + "epoch": 3.56, + "grad_norm": 6.145033359527588, + "learning_rate": 5.814271356783921e-06, + "loss": 1.1544, + "step": 71100 + }, + { + "epoch": 3.56, + "grad_norm": 4.730653285980225, + "learning_rate": 5.794170854271357e-06, + "loss": 1.1501, + "step": 71200 + }, + { + "epoch": 3.56, + "grad_norm": 4.570452690124512, + "learning_rate": 5.774070351758795e-06, + "loss": 1.1638, + "step": 71300 + }, + { + "epoch": 3.57, + "grad_norm": 3.947618007659912, + "learning_rate": 5.753969849246231e-06, + "loss": 1.1436, + "step": 71400 + }, + { + "epoch": 3.58, + "grad_norm": 6.833681106567383, + "learning_rate": 5.733869346733669e-06, + "loss": 1.1271, + "step": 71500 + }, + { + "epoch": 3.58, + "grad_norm": 4.837987422943115, + "learning_rate": 5.7137688442211056e-06, + "loss": 1.1503, + "step": 71600 + }, + { + "epoch": 3.58, + "grad_norm": 4.822892189025879, + "learning_rate": 5.6936683417085435e-06, + "loss": 1.1089, + "step": 71700 + }, + { + "epoch": 3.59, + "grad_norm": 5.022984027862549, + "learning_rate": 5.673567839195981e-06, + "loss": 1.1562, + "step": 71800 + }, + { + "epoch": 3.59, + "grad_norm": 5.105147838592529, + "learning_rate": 5.653467336683418e-06, + "loss": 1.1174, + "step": 71900 + }, + { + "epoch": 3.6, + "grad_norm": 4.37985372543335, + "learning_rate": 5.633366834170855e-06, + "loss": 1.1627, + "step": 72000 + }, + { + "epoch": 3.6, + "grad_norm": 3.854820966720581, + "learning_rate": 5.613266331658291e-06, + "loss": 1.1316, + "step": 72100 + }, + { + "epoch": 3.61, + "grad_norm": 7.305357933044434, + "learning_rate": 5.593165829145729e-06, + "loss": 1.1827, + "step": 72200 + }, + { + "epoch": 3.62, + "grad_norm": 7.693294048309326, + "learning_rate": 5.5730653266331654e-06, + "loss": 1.1649, + "step": 72300 + }, + { + "epoch": 3.62, + "grad_norm": 4.14479398727417, + "learning_rate": 5.552964824120603e-06, + "loss": 1.1844, + "step": 72400 + }, + { + "epoch": 3.62, + "grad_norm": 6.665209770202637, + "learning_rate": 5.5328643216080405e-06, + "loss": 1.1304, + "step": 72500 + }, + { + "epoch": 3.63, + "grad_norm": 4.629899978637695, + "learning_rate": 5.512763819095478e-06, + "loss": 1.1318, + "step": 72600 + }, + { + "epoch": 3.63, + "grad_norm": 3.4445009231567383, + "learning_rate": 5.492663316582915e-06, + "loss": 1.1532, + "step": 72700 + }, + { + "epoch": 3.64, + "grad_norm": 3.96806001663208, + "learning_rate": 5.472562814070353e-06, + "loss": 1.1332, + "step": 72800 + }, + { + "epoch": 3.65, + "grad_norm": 7.256198406219482, + "learning_rate": 5.452462311557789e-06, + "loss": 1.1339, + "step": 72900 + }, + { + "epoch": 3.65, + "grad_norm": 4.4425458908081055, + "learning_rate": 5.432361809045227e-06, + "loss": 1.1615, + "step": 73000 + }, + { + "epoch": 3.66, + "grad_norm": 6.114246368408203, + "learning_rate": 5.412261306532663e-06, + "loss": 1.1359, + "step": 73100 + }, + { + "epoch": 3.66, + "grad_norm": 3.9182209968566895, + "learning_rate": 5.392160804020101e-06, + "loss": 1.1574, + "step": 73200 + }, + { + "epoch": 3.67, + "grad_norm": 6.088989734649658, + "learning_rate": 5.3720603015075376e-06, + "loss": 1.0938, + "step": 73300 + }, + { + "epoch": 3.67, + "grad_norm": 6.2887163162231445, + "learning_rate": 5.3519597989949755e-06, + "loss": 1.1546, + "step": 73400 + }, + { + "epoch": 3.67, + "grad_norm": 5.033719539642334, + "learning_rate": 5.331859296482413e-06, + "loss": 1.1177, + "step": 73500 + }, + { + "epoch": 3.68, + "grad_norm": 6.611480236053467, + "learning_rate": 5.31175879396985e-06, + "loss": 1.1238, + "step": 73600 + }, + { + "epoch": 3.69, + "grad_norm": 7.612136363983154, + "learning_rate": 5.291658291457287e-06, + "loss": 1.1494, + "step": 73700 + }, + { + "epoch": 3.69, + "grad_norm": 3.842085599899292, + "learning_rate": 5.271557788944725e-06, + "loss": 1.1585, + "step": 73800 + }, + { + "epoch": 3.69, + "grad_norm": 4.859694004058838, + "learning_rate": 5.251457286432161e-06, + "loss": 1.1299, + "step": 73900 + }, + { + "epoch": 3.7, + "grad_norm": 6.1673808097839355, + "learning_rate": 5.231356783919599e-06, + "loss": 1.1242, + "step": 74000 + }, + { + "epoch": 3.71, + "grad_norm": 4.361232280731201, + "learning_rate": 5.2112562814070354e-06, + "loss": 1.1545, + "step": 74100 + }, + { + "epoch": 3.71, + "grad_norm": 5.18151330947876, + "learning_rate": 5.191155778894473e-06, + "loss": 1.0794, + "step": 74200 + }, + { + "epoch": 3.71, + "grad_norm": 5.049399375915527, + "learning_rate": 5.17105527638191e-06, + "loss": 1.136, + "step": 74300 + }, + { + "epoch": 3.72, + "grad_norm": 4.28516149520874, + "learning_rate": 5.150954773869347e-06, + "loss": 1.177, + "step": 74400 + }, + { + "epoch": 3.73, + "grad_norm": 3.71186900138855, + "learning_rate": 5.13105527638191e-06, + "loss": 1.1747, + "step": 74500 + }, + { + "epoch": 3.73, + "grad_norm": 8.269830703735352, + "learning_rate": 5.110954773869348e-06, + "loss": 1.1494, + "step": 74600 + }, + { + "epoch": 3.73, + "grad_norm": 4.453472137451172, + "learning_rate": 5.09105527638191e-06, + "loss": 1.1511, + "step": 74700 + }, + { + "epoch": 3.74, + "grad_norm": 6.132487773895264, + "learning_rate": 5.070954773869348e-06, + "loss": 1.1604, + "step": 74800 + }, + { + "epoch": 3.75, + "grad_norm": 4.084425926208496, + "learning_rate": 5.050854271356785e-06, + "loss": 1.1582, + "step": 74900 + }, + { + "epoch": 3.75, + "grad_norm": 4.719122886657715, + "learning_rate": 5.030753768844222e-06, + "loss": 1.0998, + "step": 75000 + }, + { + "epoch": 3.75, + "grad_norm": 6.886116981506348, + "learning_rate": 5.010854271356784e-06, + "loss": 1.1402, + "step": 75100 + }, + { + "epoch": 3.76, + "grad_norm": 3.9882471561431885, + "learning_rate": 4.990753768844221e-06, + "loss": 1.1163, + "step": 75200 + }, + { + "epoch": 3.77, + "grad_norm": 7.415084362030029, + "learning_rate": 4.9706532663316585e-06, + "loss": 1.1145, + "step": 75300 + }, + { + "epoch": 3.77, + "grad_norm": 4.16004753112793, + "learning_rate": 4.950552763819096e-06, + "loss": 1.1071, + "step": 75400 + }, + { + "epoch": 3.77, + "grad_norm": 5.258670806884766, + "learning_rate": 4.930452261306533e-06, + "loss": 1.1242, + "step": 75500 + }, + { + "epoch": 3.78, + "grad_norm": 5.320291519165039, + "learning_rate": 4.910351758793971e-06, + "loss": 1.1388, + "step": 75600 + }, + { + "epoch": 3.79, + "grad_norm": 4.200166702270508, + "learning_rate": 4.890251256281408e-06, + "loss": 1.1395, + "step": 75700 + }, + { + "epoch": 3.79, + "grad_norm": 4.569030284881592, + "learning_rate": 4.870150753768845e-06, + "loss": 1.1397, + "step": 75800 + }, + { + "epoch": 3.79, + "grad_norm": 4.229086875915527, + "learning_rate": 4.850050251256282e-06, + "loss": 1.1705, + "step": 75900 + }, + { + "epoch": 3.8, + "grad_norm": 5.543234825134277, + "learning_rate": 4.829949748743719e-06, + "loss": 1.1403, + "step": 76000 + }, + { + "epoch": 3.81, + "grad_norm": 4.94819974899292, + "learning_rate": 4.809849246231156e-06, + "loss": 1.1395, + "step": 76100 + }, + { + "epoch": 3.81, + "grad_norm": 3.889681577682495, + "learning_rate": 4.7897487437185935e-06, + "loss": 1.137, + "step": 76200 + }, + { + "epoch": 3.81, + "grad_norm": 5.113078594207764, + "learning_rate": 4.769648241206031e-06, + "loss": 1.1171, + "step": 76300 + }, + { + "epoch": 3.82, + "grad_norm": 4.799468994140625, + "learning_rate": 4.749547738693468e-06, + "loss": 1.1235, + "step": 76400 + }, + { + "epoch": 3.83, + "grad_norm": 4.0734453201293945, + "learning_rate": 4.729447236180905e-06, + "loss": 1.1283, + "step": 76500 + }, + { + "epoch": 3.83, + "grad_norm": 4.847392559051514, + "learning_rate": 4.709346733668342e-06, + "loss": 1.1932, + "step": 76600 + }, + { + "epoch": 3.83, + "grad_norm": 6.197125434875488, + "learning_rate": 4.689246231155779e-06, + "loss": 1.144, + "step": 76700 + }, + { + "epoch": 3.84, + "grad_norm": 5.93238639831543, + "learning_rate": 4.669145728643216e-06, + "loss": 1.0989, + "step": 76800 + }, + { + "epoch": 3.84, + "grad_norm": 5.202603340148926, + "learning_rate": 4.649045226130653e-06, + "loss": 1.1312, + "step": 76900 + }, + { + "epoch": 3.85, + "grad_norm": 5.290008544921875, + "learning_rate": 4.6289447236180905e-06, + "loss": 1.096, + "step": 77000 + }, + { + "epoch": 3.85, + "grad_norm": 4.39517879486084, + "learning_rate": 4.608844221105528e-06, + "loss": 1.1948, + "step": 77100 + }, + { + "epoch": 3.86, + "grad_norm": 6.785552024841309, + "learning_rate": 4.588743718592965e-06, + "loss": 1.139, + "step": 77200 + }, + { + "epoch": 3.87, + "grad_norm": 5.2934370040893555, + "learning_rate": 4.568643216080402e-06, + "loss": 1.1152, + "step": 77300 + }, + { + "epoch": 3.87, + "grad_norm": 5.2376604080200195, + "learning_rate": 4.54854271356784e-06, + "loss": 1.1464, + "step": 77400 + }, + { + "epoch": 3.88, + "grad_norm": 6.19007682800293, + "learning_rate": 4.528442211055277e-06, + "loss": 1.1199, + "step": 77500 + }, + { + "epoch": 3.88, + "grad_norm": 5.796671390533447, + "learning_rate": 4.508341708542714e-06, + "loss": 1.1399, + "step": 77600 + }, + { + "epoch": 3.88, + "grad_norm": 5.685388565063477, + "learning_rate": 4.488241206030151e-06, + "loss": 1.1485, + "step": 77700 + }, + { + "epoch": 3.89, + "grad_norm": 6.502816677093506, + "learning_rate": 4.468140703517588e-06, + "loss": 1.15, + "step": 77800 + }, + { + "epoch": 3.9, + "grad_norm": 4.437497138977051, + "learning_rate": 4.4480402010050255e-06, + "loss": 1.1344, + "step": 77900 + }, + { + "epoch": 3.9, + "grad_norm": 6.776554107666016, + "learning_rate": 4.427939698492463e-06, + "loss": 1.1486, + "step": 78000 + }, + { + "epoch": 3.91, + "grad_norm": 3.9491071701049805, + "learning_rate": 4.4078391959799e-06, + "loss": 1.1269, + "step": 78100 + }, + { + "epoch": 3.91, + "grad_norm": 5.485203266143799, + "learning_rate": 4.387738693467337e-06, + "loss": 1.1539, + "step": 78200 + }, + { + "epoch": 3.92, + "grad_norm": 7.40858793258667, + "learning_rate": 4.367638190954774e-06, + "loss": 1.1281, + "step": 78300 + }, + { + "epoch": 3.92, + "grad_norm": 4.498636722564697, + "learning_rate": 4.347537688442212e-06, + "loss": 1.1912, + "step": 78400 + }, + { + "epoch": 3.92, + "grad_norm": 6.4472856521606445, + "learning_rate": 4.327437185929649e-06, + "loss": 1.1276, + "step": 78500 + }, + { + "epoch": 3.93, + "grad_norm": 6.126656532287598, + "learning_rate": 4.307336683417086e-06, + "loss": 1.1329, + "step": 78600 + }, + { + "epoch": 3.94, + "grad_norm": 5.280217170715332, + "learning_rate": 4.287236180904523e-06, + "loss": 1.093, + "step": 78700 + }, + { + "epoch": 3.94, + "grad_norm": 6.618311405181885, + "learning_rate": 4.2671356783919605e-06, + "loss": 1.1215, + "step": 78800 + }, + { + "epoch": 3.94, + "grad_norm": 5.225731372833252, + "learning_rate": 4.247035175879397e-06, + "loss": 1.1252, + "step": 78900 + }, + { + "epoch": 3.95, + "grad_norm": 6.615197658538818, + "learning_rate": 4.226934673366834e-06, + "loss": 1.0734, + "step": 79000 + }, + { + "epoch": 3.96, + "grad_norm": 5.426534175872803, + "learning_rate": 4.206834170854272e-06, + "loss": 1.0851, + "step": 79100 + }, + { + "epoch": 3.96, + "grad_norm": 3.9240429401397705, + "learning_rate": 4.186733668341709e-06, + "loss": 1.1033, + "step": 79200 + }, + { + "epoch": 3.96, + "grad_norm": 4.4714250564575195, + "learning_rate": 4.166633165829146e-06, + "loss": 1.1347, + "step": 79300 + }, + { + "epoch": 3.97, + "grad_norm": 10.831876754760742, + "learning_rate": 4.146532663316583e-06, + "loss": 1.0971, + "step": 79400 + }, + { + "epoch": 3.98, + "grad_norm": 5.725513458251953, + "learning_rate": 4.12643216080402e-06, + "loss": 1.1346, + "step": 79500 + }, + { + "epoch": 3.98, + "grad_norm": 6.024409770965576, + "learning_rate": 4.1063316582914575e-06, + "loss": 1.1551, + "step": 79600 + }, + { + "epoch": 3.98, + "grad_norm": 3.672581195831299, + "learning_rate": 4.086231155778895e-06, + "loss": 1.1365, + "step": 79700 + }, + { + "epoch": 3.99, + "grad_norm": 5.520185470581055, + "learning_rate": 4.066130653266332e-06, + "loss": 1.1321, + "step": 79800 + }, + { + "epoch": 4.0, + "grad_norm": 6.262924671173096, + "learning_rate": 4.046030150753769e-06, + "loss": 1.1062, + "step": 79900 + }, + { + "epoch": 4.0, + "grad_norm": 6.248514652252197, + "learning_rate": 4.025929648241206e-06, + "loss": 1.0963, + "step": 80000 + }, + { + "epoch": 4.0, + "eval_loss": 1.1962475776672363, + "eval_runtime": 21.7746, + "eval_samples_per_second": 45.925, + "eval_steps_per_second": 5.741, + "step": 80000 + }, + { + "epoch": 4.0, + "grad_norm": 4.919951915740967, + "learning_rate": 4.005829145728643e-06, + "loss": 1.0946, + "step": 80100 + }, + { + "epoch": 4.01, + "grad_norm": 4.365044593811035, + "learning_rate": 3.985728643216081e-06, + "loss": 1.0987, + "step": 80200 + }, + { + "epoch": 4.01, + "grad_norm": 9.315962791442871, + "learning_rate": 3.965628140703518e-06, + "loss": 1.0575, + "step": 80300 + }, + { + "epoch": 4.02, + "grad_norm": 10.50012493133545, + "learning_rate": 3.945527638190955e-06, + "loss": 1.069, + "step": 80400 + }, + { + "epoch": 4.03, + "grad_norm": 8.853384017944336, + "learning_rate": 3.9254271356783925e-06, + "loss": 1.0638, + "step": 80500 + }, + { + "epoch": 4.03, + "grad_norm": 5.290276527404785, + "learning_rate": 3.90532663316583e-06, + "loss": 1.0422, + "step": 80600 + }, + { + "epoch": 4.04, + "grad_norm": 5.553296089172363, + "learning_rate": 3.885226130653267e-06, + "loss": 1.0517, + "step": 80700 + }, + { + "epoch": 4.04, + "grad_norm": 5.262092113494873, + "learning_rate": 3.865125628140704e-06, + "loss": 1.1184, + "step": 80800 + }, + { + "epoch": 4.04, + "grad_norm": 8.444523811340332, + "learning_rate": 3.845025125628141e-06, + "loss": 1.0861, + "step": 80900 + }, + { + "epoch": 4.05, + "grad_norm": 5.442756652832031, + "learning_rate": 3.824924623115578e-06, + "loss": 1.0559, + "step": 81000 + }, + { + "epoch": 4.05, + "grad_norm": 5.926506519317627, + "learning_rate": 3.8050251256281414e-06, + "loss": 1.0482, + "step": 81100 + }, + { + "epoch": 4.06, + "grad_norm": 5.596924781799316, + "learning_rate": 3.7849246231155785e-06, + "loss": 1.071, + "step": 81200 + }, + { + "epoch": 4.07, + "grad_norm": 8.789709091186523, + "learning_rate": 3.7648241206030156e-06, + "loss": 1.0796, + "step": 81300 + }, + { + "epoch": 4.07, + "grad_norm": 5.048920631408691, + "learning_rate": 3.7447236180904528e-06, + "loss": 1.0602, + "step": 81400 + }, + { + "epoch": 4.08, + "grad_norm": 6.4080705642700195, + "learning_rate": 3.72462311557789e-06, + "loss": 1.0481, + "step": 81500 + }, + { + "epoch": 4.08, + "grad_norm": 6.361392974853516, + "learning_rate": 3.7045226130653266e-06, + "loss": 1.0393, + "step": 81600 + }, + { + "epoch": 4.08, + "grad_norm": 7.596906661987305, + "learning_rate": 3.6844221105527637e-06, + "loss": 1.0212, + "step": 81700 + }, + { + "epoch": 4.09, + "grad_norm": 5.352015972137451, + "learning_rate": 3.6643216080402013e-06, + "loss": 1.0159, + "step": 81800 + }, + { + "epoch": 4.09, + "grad_norm": 5.569393157958984, + "learning_rate": 3.6442211055276384e-06, + "loss": 1.0516, + "step": 81900 + }, + { + "epoch": 4.1, + "grad_norm": 5.463687419891357, + "learning_rate": 3.6241206030150755e-06, + "loss": 1.0729, + "step": 82000 + }, + { + "epoch": 4.11, + "grad_norm": 11.406126976013184, + "learning_rate": 3.6040201005025127e-06, + "loss": 1.0389, + "step": 82100 + }, + { + "epoch": 4.11, + "grad_norm": 6.264597415924072, + "learning_rate": 3.58391959798995e-06, + "loss": 1.0613, + "step": 82200 + }, + { + "epoch": 4.12, + "grad_norm": 6.288965225219727, + "learning_rate": 3.5638190954773873e-06, + "loss": 1.0486, + "step": 82300 + }, + { + "epoch": 4.12, + "grad_norm": 5.622555732727051, + "learning_rate": 3.5437185929648245e-06, + "loss": 1.0665, + "step": 82400 + }, + { + "epoch": 4.12, + "grad_norm": 4.834249973297119, + "learning_rate": 3.5236180904522616e-06, + "loss": 1.074, + "step": 82500 + }, + { + "epoch": 4.13, + "grad_norm": 8.10172176361084, + "learning_rate": 3.5035175879396987e-06, + "loss": 1.0272, + "step": 82600 + }, + { + "epoch": 4.13, + "grad_norm": 5.477063179016113, + "learning_rate": 3.483417085427136e-06, + "loss": 1.0405, + "step": 82700 + }, + { + "epoch": 4.14, + "grad_norm": 4.51005220413208, + "learning_rate": 3.463316582914573e-06, + "loss": 1.0339, + "step": 82800 + }, + { + "epoch": 4.14, + "grad_norm": 7.996946811676025, + "learning_rate": 3.4432160804020105e-06, + "loss": 1.0159, + "step": 82900 + }, + { + "epoch": 4.15, + "grad_norm": 6.292382717132568, + "learning_rate": 3.4231155778894477e-06, + "loss": 1.085, + "step": 83000 + }, + { + "epoch": 4.16, + "grad_norm": 8.48137092590332, + "learning_rate": 3.4030150753768848e-06, + "loss": 1.047, + "step": 83100 + }, + { + "epoch": 4.16, + "grad_norm": 6.187898635864258, + "learning_rate": 3.382914572864322e-06, + "loss": 1.0577, + "step": 83200 + }, + { + "epoch": 4.17, + "grad_norm": 4.8612775802612305, + "learning_rate": 3.3630150753768847e-06, + "loss": 1.0273, + "step": 83300 + }, + { + "epoch": 4.17, + "grad_norm": 5.060107231140137, + "learning_rate": 3.342914572864322e-06, + "loss": 1.0998, + "step": 83400 + }, + { + "epoch": 4.17, + "grad_norm": 4.8297505378723145, + "learning_rate": 3.322814070351759e-06, + "loss": 1.0821, + "step": 83500 + }, + { + "epoch": 4.18, + "grad_norm": 4.36783504486084, + "learning_rate": 3.3027135678391965e-06, + "loss": 1.0591, + "step": 83600 + }, + { + "epoch": 4.18, + "grad_norm": 7.225709438323975, + "learning_rate": 3.2826130653266337e-06, + "loss": 1.0245, + "step": 83700 + }, + { + "epoch": 4.19, + "grad_norm": 5.521580219268799, + "learning_rate": 3.262512562814071e-06, + "loss": 1.0131, + "step": 83800 + }, + { + "epoch": 4.2, + "grad_norm": 4.925292491912842, + "learning_rate": 3.242412060301508e-06, + "loss": 1.023, + "step": 83900 + }, + { + "epoch": 4.2, + "grad_norm": 5.887296676635742, + "learning_rate": 3.222311557788945e-06, + "loss": 1.0233, + "step": 84000 + }, + { + "epoch": 4.21, + "grad_norm": 8.528314590454102, + "learning_rate": 3.2022110552763826e-06, + "loss": 1.0593, + "step": 84100 + }, + { + "epoch": 4.21, + "grad_norm": 5.414628982543945, + "learning_rate": 3.182110552763819e-06, + "loss": 1.0526, + "step": 84200 + }, + { + "epoch": 4.21, + "grad_norm": 8.320696830749512, + "learning_rate": 3.1620100502512564e-06, + "loss": 1.061, + "step": 84300 + }, + { + "epoch": 4.22, + "grad_norm": 7.6390767097473145, + "learning_rate": 3.1419095477386936e-06, + "loss": 1.0789, + "step": 84400 + }, + { + "epoch": 4.22, + "grad_norm": 6.141530990600586, + "learning_rate": 3.1218090452261307e-06, + "loss": 1.0438, + "step": 84500 + }, + { + "epoch": 4.23, + "grad_norm": 5.907855987548828, + "learning_rate": 3.101708542713568e-06, + "loss": 1.068, + "step": 84600 + }, + { + "epoch": 4.24, + "grad_norm": 8.13024616241455, + "learning_rate": 3.081608040201005e-06, + "loss": 1.0612, + "step": 84700 + }, + { + "epoch": 4.24, + "grad_norm": 4.439888954162598, + "learning_rate": 3.0615075376884425e-06, + "loss": 1.0589, + "step": 84800 + }, + { + "epoch": 4.25, + "grad_norm": 5.636837005615234, + "learning_rate": 3.0414070351758796e-06, + "loss": 1.0306, + "step": 84900 + }, + { + "epoch": 4.25, + "grad_norm": 4.6407389640808105, + "learning_rate": 3.0213065326633168e-06, + "loss": 1.0509, + "step": 85000 + }, + { + "epoch": 4.25, + "grad_norm": 5.515079975128174, + "learning_rate": 3.001206030150754e-06, + "loss": 1.0273, + "step": 85100 + }, + { + "epoch": 4.26, + "grad_norm": 5.22351598739624, + "learning_rate": 2.981105527638191e-06, + "loss": 1.0579, + "step": 85200 + }, + { + "epoch": 4.26, + "grad_norm": 6.4437479972839355, + "learning_rate": 2.961206030150754e-06, + "loss": 1.0435, + "step": 85300 + }, + { + "epoch": 4.27, + "grad_norm": 6.850719928741455, + "learning_rate": 2.941105527638191e-06, + "loss": 1.0487, + "step": 85400 + }, + { + "epoch": 4.28, + "grad_norm": 4.1141276359558105, + "learning_rate": 2.9210050251256285e-06, + "loss": 1.0124, + "step": 85500 + }, + { + "epoch": 4.28, + "grad_norm": 6.114626884460449, + "learning_rate": 2.9009045226130656e-06, + "loss": 1.0146, + "step": 85600 + }, + { + "epoch": 4.29, + "grad_norm": 5.943951606750488, + "learning_rate": 2.8808040201005028e-06, + "loss": 1.0484, + "step": 85700 + }, + { + "epoch": 4.29, + "grad_norm": 6.745482444763184, + "learning_rate": 2.86070351758794e-06, + "loss": 1.0353, + "step": 85800 + }, + { + "epoch": 4.29, + "grad_norm": 6.165544033050537, + "learning_rate": 2.840603015075377e-06, + "loss": 1.0572, + "step": 85900 + }, + { + "epoch": 4.3, + "grad_norm": 5.070137023925781, + "learning_rate": 2.8205025125628146e-06, + "loss": 1.0607, + "step": 86000 + }, + { + "epoch": 4.3, + "grad_norm": 5.177759647369385, + "learning_rate": 2.8004020100502517e-06, + "loss": 1.0589, + "step": 86100 + }, + { + "epoch": 4.31, + "grad_norm": 5.926203727722168, + "learning_rate": 2.780301507537689e-06, + "loss": 1.0728, + "step": 86200 + }, + { + "epoch": 4.32, + "grad_norm": 4.766726970672607, + "learning_rate": 2.760201005025126e-06, + "loss": 1.066, + "step": 86300 + }, + { + "epoch": 4.32, + "grad_norm": 7.0791401863098145, + "learning_rate": 2.740100502512563e-06, + "loss": 1.0381, + "step": 86400 + }, + { + "epoch": 4.33, + "grad_norm": 6.904074668884277, + "learning_rate": 2.7200000000000002e-06, + "loss": 1.0341, + "step": 86500 + }, + { + "epoch": 4.33, + "grad_norm": 7.680102348327637, + "learning_rate": 2.700100502512563e-06, + "loss": 1.0288, + "step": 86600 + }, + { + "epoch": 4.33, + "grad_norm": 7.589695930480957, + "learning_rate": 2.68e-06, + "loss": 1.0677, + "step": 86700 + }, + { + "epoch": 4.34, + "grad_norm": 4.686061382293701, + "learning_rate": 2.6598994974874377e-06, + "loss": 1.0213, + "step": 86800 + }, + { + "epoch": 4.34, + "grad_norm": 5.1108269691467285, + "learning_rate": 2.639798994974875e-06, + "loss": 1.0501, + "step": 86900 + }, + { + "epoch": 4.35, + "grad_norm": 6.551150798797607, + "learning_rate": 2.619698492462312e-06, + "loss": 1.0598, + "step": 87000 + }, + { + "epoch": 4.36, + "grad_norm": 6.834580898284912, + "learning_rate": 2.599597989949749e-06, + "loss": 1.0757, + "step": 87100 + }, + { + "epoch": 4.36, + "grad_norm": 4.995261192321777, + "learning_rate": 2.579497487437186e-06, + "loss": 1.0653, + "step": 87200 + }, + { + "epoch": 4.37, + "grad_norm": 5.614989757537842, + "learning_rate": 2.559396984924623e-06, + "loss": 1.0412, + "step": 87300 + }, + { + "epoch": 4.37, + "grad_norm": 12.592680931091309, + "learning_rate": 2.5392964824120605e-06, + "loss": 1.1065, + "step": 87400 + }, + { + "epoch": 4.38, + "grad_norm": 7.026024341583252, + "learning_rate": 2.5191959798994976e-06, + "loss": 1.064, + "step": 87500 + }, + { + "epoch": 4.38, + "grad_norm": 6.075479507446289, + "learning_rate": 2.499095477386935e-06, + "loss": 1.0559, + "step": 87600 + }, + { + "epoch": 4.38, + "grad_norm": 7.670622825622559, + "learning_rate": 2.4789949748743723e-06, + "loss": 1.0173, + "step": 87700 + }, + { + "epoch": 4.39, + "grad_norm": 6.775985240936279, + "learning_rate": 2.458894472361809e-06, + "loss": 1.0192, + "step": 87800 + }, + { + "epoch": 4.39, + "grad_norm": 6.7632222175598145, + "learning_rate": 2.438793969849246e-06, + "loss": 1.0403, + "step": 87900 + }, + { + "epoch": 4.4, + "grad_norm": 6.712850570678711, + "learning_rate": 2.4186934673366837e-06, + "loss": 1.0408, + "step": 88000 + }, + { + "epoch": 4.41, + "grad_norm": 4.842583656311035, + "learning_rate": 2.398592964824121e-06, + "loss": 1.0711, + "step": 88100 + }, + { + "epoch": 4.41, + "grad_norm": 6.134751796722412, + "learning_rate": 2.378492462311558e-06, + "loss": 1.0085, + "step": 88200 + }, + { + "epoch": 4.42, + "grad_norm": 5.06552791595459, + "learning_rate": 2.358391959798995e-06, + "loss": 1.0088, + "step": 88300 + }, + { + "epoch": 4.42, + "grad_norm": 6.872971057891846, + "learning_rate": 2.338291457286432e-06, + "loss": 1.0492, + "step": 88400 + }, + { + "epoch": 4.42, + "grad_norm": 5.2886528968811035, + "learning_rate": 2.3181909547738697e-06, + "loss": 1.0699, + "step": 88500 + }, + { + "epoch": 4.43, + "grad_norm": 7.221102237701416, + "learning_rate": 2.298090452261307e-06, + "loss": 1.0505, + "step": 88600 + }, + { + "epoch": 4.43, + "grad_norm": 7.388364791870117, + "learning_rate": 2.277989949748744e-06, + "loss": 1.0127, + "step": 88700 + }, + { + "epoch": 4.44, + "grad_norm": 11.533583641052246, + "learning_rate": 2.257889447236181e-06, + "loss": 1.055, + "step": 88800 + }, + { + "epoch": 4.45, + "grad_norm": 5.406151294708252, + "learning_rate": 2.2377889447236182e-06, + "loss": 1.0848, + "step": 88900 + }, + { + "epoch": 4.45, + "grad_norm": 8.093517303466797, + "learning_rate": 2.2176884422110554e-06, + "loss": 1.0058, + "step": 89000 + }, + { + "epoch": 4.46, + "grad_norm": 7.073362827301025, + "learning_rate": 2.1975879396984925e-06, + "loss": 1.0233, + "step": 89100 + }, + { + "epoch": 4.46, + "grad_norm": 6.263842582702637, + "learning_rate": 2.1776884422110558e-06, + "loss": 1.0259, + "step": 89200 + }, + { + "epoch": 4.46, + "grad_norm": 4.668169975280762, + "learning_rate": 2.1575879396984925e-06, + "loss": 1.0641, + "step": 89300 + }, + { + "epoch": 4.47, + "grad_norm": 5.550297737121582, + "learning_rate": 2.1374874371859296e-06, + "loss": 1.0768, + "step": 89400 + }, + { + "epoch": 4.47, + "grad_norm": 6.35700798034668, + "learning_rate": 2.1173869346733667e-06, + "loss": 1.0237, + "step": 89500 + }, + { + "epoch": 4.48, + "grad_norm": 8.346928596496582, + "learning_rate": 2.0972864321608043e-06, + "loss": 1.0265, + "step": 89600 + }, + { + "epoch": 4.49, + "grad_norm": 5.672070026397705, + "learning_rate": 2.0771859296482414e-06, + "loss": 1.0309, + "step": 89700 + }, + { + "epoch": 4.49, + "grad_norm": 7.057281494140625, + "learning_rate": 2.0570854271356785e-06, + "loss": 1.0131, + "step": 89800 + }, + { + "epoch": 4.5, + "grad_norm": 5.305558204650879, + "learning_rate": 2.0369849246231156e-06, + "loss": 1.075, + "step": 89900 + }, + { + "epoch": 4.5, + "grad_norm": 7.687346935272217, + "learning_rate": 2.0168844221105528e-06, + "loss": 1.0525, + "step": 90000 + }, + { + "epoch": 4.5, + "eval_loss": 1.240631341934204, + "eval_runtime": 21.7808, + "eval_samples_per_second": 45.912, + "eval_steps_per_second": 5.739, + "step": 90000 + }, + { + "epoch": 4.5, + "grad_norm": 4.474082946777344, + "learning_rate": 1.9967839195979903e-06, + "loss": 1.0245, + "step": 90100 + }, + { + "epoch": 4.51, + "grad_norm": 7.113109588623047, + "learning_rate": 1.9766834170854275e-06, + "loss": 1.0475, + "step": 90200 + }, + { + "epoch": 4.51, + "grad_norm": 5.945183753967285, + "learning_rate": 1.9565829145728646e-06, + "loss": 1.066, + "step": 90300 + }, + { + "epoch": 4.52, + "grad_norm": 5.280254364013672, + "learning_rate": 1.9364824120603017e-06, + "loss": 1.0428, + "step": 90400 + }, + { + "epoch": 4.53, + "grad_norm": 4.7843098640441895, + "learning_rate": 1.916381909547739e-06, + "loss": 1.0468, + "step": 90500 + }, + { + "epoch": 4.53, + "grad_norm": 6.668610572814941, + "learning_rate": 1.896281407035176e-06, + "loss": 1.0357, + "step": 90600 + }, + { + "epoch": 4.54, + "grad_norm": 6.481570720672607, + "learning_rate": 1.876180904522613e-06, + "loss": 1.0502, + "step": 90700 + }, + { + "epoch": 4.54, + "grad_norm": 6.947873115539551, + "learning_rate": 1.8560804020100504e-06, + "loss": 1.0129, + "step": 90800 + }, + { + "epoch": 4.54, + "grad_norm": 5.446567058563232, + "learning_rate": 1.8359798994974876e-06, + "loss": 1.0222, + "step": 90900 + }, + { + "epoch": 4.55, + "grad_norm": 6.875363349914551, + "learning_rate": 1.8158793969849247e-06, + "loss": 1.0444, + "step": 91000 + }, + { + "epoch": 4.55, + "grad_norm": 5.002197742462158, + "learning_rate": 1.795778894472362e-06, + "loss": 1.0339, + "step": 91100 + }, + { + "epoch": 4.56, + "grad_norm": 6.167435169219971, + "learning_rate": 1.7756783919597991e-06, + "loss": 0.9916, + "step": 91200 + }, + { + "epoch": 4.56, + "grad_norm": 4.765905380249023, + "learning_rate": 1.7555778894472365e-06, + "loss": 1.0356, + "step": 91300 + }, + { + "epoch": 4.57, + "grad_norm": 9.28632926940918, + "learning_rate": 1.7354773869346736e-06, + "loss": 1.052, + "step": 91400 + }, + { + "epoch": 4.58, + "grad_norm": 5.864830017089844, + "learning_rate": 1.7153768844221107e-06, + "loss": 1.014, + "step": 91500 + }, + { + "epoch": 4.58, + "grad_norm": 10.017616271972656, + "learning_rate": 1.6952763819095477e-06, + "loss": 1.0347, + "step": 91600 + }, + { + "epoch": 4.58, + "grad_norm": 6.52495813369751, + "learning_rate": 1.675175879396985e-06, + "loss": 1.0094, + "step": 91700 + }, + { + "epoch": 4.59, + "grad_norm": 5.506436347961426, + "learning_rate": 1.6550753768844221e-06, + "loss": 1.0306, + "step": 91800 + }, + { + "epoch": 4.59, + "grad_norm": 6.378169059753418, + "learning_rate": 1.6349748743718595e-06, + "loss": 1.0481, + "step": 91900 + }, + { + "epoch": 4.6, + "grad_norm": 7.866368293762207, + "learning_rate": 1.6148743718592966e-06, + "loss": 1.0433, + "step": 92000 + }, + { + "epoch": 4.61, + "grad_norm": 3.3074705600738525, + "learning_rate": 1.5947738693467337e-06, + "loss": 1.0255, + "step": 92100 + }, + { + "epoch": 4.61, + "grad_norm": 5.663514137268066, + "learning_rate": 1.574673366834171e-06, + "loss": 1.0066, + "step": 92200 + }, + { + "epoch": 4.62, + "grad_norm": 5.797703266143799, + "learning_rate": 1.5545728643216082e-06, + "loss": 1.0223, + "step": 92300 + }, + { + "epoch": 4.62, + "grad_norm": 5.430294513702393, + "learning_rate": 1.5344723618090453e-06, + "loss": 1.0361, + "step": 92400 + }, + { + "epoch": 4.62, + "grad_norm": 7.778083324432373, + "learning_rate": 1.5143718592964826e-06, + "loss": 1.0128, + "step": 92500 + }, + { + "epoch": 4.63, + "grad_norm": 8.762657165527344, + "learning_rate": 1.4942713567839198e-06, + "loss": 0.9764, + "step": 92600 + }, + { + "epoch": 4.63, + "grad_norm": 6.608733654022217, + "learning_rate": 1.4741708542713571e-06, + "loss": 1.0355, + "step": 92700 + }, + { + "epoch": 4.64, + "grad_norm": 4.251249313354492, + "learning_rate": 1.454070351758794e-06, + "loss": 1.0383, + "step": 92800 + }, + { + "epoch": 4.64, + "grad_norm": 5.131290912628174, + "learning_rate": 1.4339698492462312e-06, + "loss": 1.0118, + "step": 92900 + }, + { + "epoch": 4.65, + "grad_norm": 5.982537746429443, + "learning_rate": 1.4138693467336683e-06, + "loss": 1.0031, + "step": 93000 + }, + { + "epoch": 4.66, + "grad_norm": 5.640321254730225, + "learning_rate": 1.3937688442211056e-06, + "loss": 0.9834, + "step": 93100 + }, + { + "epoch": 4.66, + "grad_norm": 4.641716003417969, + "learning_rate": 1.3738693467336682e-06, + "loss": 1.0313, + "step": 93200 + }, + { + "epoch": 4.67, + "grad_norm": 4.14018440246582, + "learning_rate": 1.3537688442211056e-06, + "loss": 1.0132, + "step": 93300 + }, + { + "epoch": 4.67, + "grad_norm": 5.720813274383545, + "learning_rate": 1.3336683417085427e-06, + "loss": 1.0398, + "step": 93400 + }, + { + "epoch": 4.67, + "grad_norm": 6.412359714508057, + "learning_rate": 1.31356783919598e-06, + "loss": 0.9922, + "step": 93500 + }, + { + "epoch": 4.68, + "grad_norm": 6.955712795257568, + "learning_rate": 1.2934673366834172e-06, + "loss": 1.0396, + "step": 93600 + }, + { + "epoch": 4.69, + "grad_norm": 9.27692699432373, + "learning_rate": 1.2733668341708543e-06, + "loss": 1.0883, + "step": 93700 + }, + { + "epoch": 4.69, + "grad_norm": 5.846386909484863, + "learning_rate": 1.2532663316582916e-06, + "loss": 0.9981, + "step": 93800 + }, + { + "epoch": 4.7, + "grad_norm": 7.700655937194824, + "learning_rate": 1.2331658291457288e-06, + "loss": 1.0766, + "step": 93900 + }, + { + "epoch": 4.7, + "grad_norm": 6.861111640930176, + "learning_rate": 1.213065326633166e-06, + "loss": 1.0677, + "step": 94000 + }, + { + "epoch": 4.71, + "grad_norm": 7.702520370483398, + "learning_rate": 1.192964824120603e-06, + "loss": 1.092, + "step": 94100 + }, + { + "epoch": 4.71, + "grad_norm": 7.238519668579102, + "learning_rate": 1.1728643216080404e-06, + "loss": 1.0195, + "step": 94200 + }, + { + "epoch": 4.71, + "grad_norm": 5.9995646476745605, + "learning_rate": 1.1527638190954775e-06, + "loss": 1.0125, + "step": 94300 + }, + { + "epoch": 4.72, + "grad_norm": 8.999128341674805, + "learning_rate": 1.1326633165829146e-06, + "loss": 1.0342, + "step": 94400 + }, + { + "epoch": 4.72, + "grad_norm": 8.37474536895752, + "learning_rate": 1.112562814070352e-06, + "loss": 1.0151, + "step": 94500 + }, + { + "epoch": 4.73, + "grad_norm": 7.018558979034424, + "learning_rate": 1.092462311557789e-06, + "loss": 0.9854, + "step": 94600 + }, + { + "epoch": 4.74, + "grad_norm": 5.188572883605957, + "learning_rate": 1.0723618090452262e-06, + "loss": 0.9934, + "step": 94700 + }, + { + "epoch": 4.74, + "grad_norm": 5.260889530181885, + "learning_rate": 1.0522613065326633e-06, + "loss": 0.9982, + "step": 94800 + }, + { + "epoch": 4.75, + "grad_norm": 8.045933723449707, + "learning_rate": 1.0321608040201007e-06, + "loss": 1.0473, + "step": 94900 + }, + { + "epoch": 4.75, + "grad_norm": 8.305715560913086, + "learning_rate": 1.0120603015075378e-06, + "loss": 0.9831, + "step": 95000 + }, + { + "epoch": 4.75, + "grad_norm": 7.344651222229004, + "learning_rate": 9.91959798994975e-07, + "loss": 1.0251, + "step": 95100 + }, + { + "epoch": 4.76, + "grad_norm": 8.983135223388672, + "learning_rate": 9.71859296482412e-07, + "loss": 1.0084, + "step": 95200 + }, + { + "epoch": 4.76, + "grad_norm": 8.123686790466309, + "learning_rate": 9.51959798994975e-07, + "loss": 1.0136, + "step": 95300 + }, + { + "epoch": 4.77, + "grad_norm": 6.7493462562561035, + "learning_rate": 9.318592964824122e-07, + "loss": 0.9987, + "step": 95400 + }, + { + "epoch": 4.78, + "grad_norm": 8.338164329528809, + "learning_rate": 9.117587939698493e-07, + "loss": 0.99, + "step": 95500 + }, + { + "epoch": 4.78, + "grad_norm": 4.540625095367432, + "learning_rate": 8.916582914572865e-07, + "loss": 1.0015, + "step": 95600 + }, + { + "epoch": 4.79, + "grad_norm": 4.909175872802734, + "learning_rate": 8.715577889447237e-07, + "loss": 1.0395, + "step": 95700 + }, + { + "epoch": 4.79, + "grad_norm": 7.736073970794678, + "learning_rate": 8.514572864321608e-07, + "loss": 1.0005, + "step": 95800 + }, + { + "epoch": 4.79, + "grad_norm": 5.298911094665527, + "learning_rate": 8.313567839195981e-07, + "loss": 1.0283, + "step": 95900 + }, + { + "epoch": 4.8, + "grad_norm": 8.024383544921875, + "learning_rate": 8.112562814070353e-07, + "loss": 1.0152, + "step": 96000 + }, + { + "epoch": 4.8, + "grad_norm": 6.19573974609375, + "learning_rate": 7.911557788944723e-07, + "loss": 1.0195, + "step": 96100 + }, + { + "epoch": 4.81, + "grad_norm": 7.0770182609558105, + "learning_rate": 7.710552763819096e-07, + "loss": 1.0285, + "step": 96200 + }, + { + "epoch": 4.81, + "grad_norm": 5.578373908996582, + "learning_rate": 7.509547738693468e-07, + "loss": 0.9974, + "step": 96300 + }, + { + "epoch": 4.82, + "grad_norm": 6.602869033813477, + "learning_rate": 7.30854271356784e-07, + "loss": 1.0019, + "step": 96400 + }, + { + "epoch": 4.83, + "grad_norm": 8.442864418029785, + "learning_rate": 7.107537688442212e-07, + "loss": 1.035, + "step": 96500 + }, + { + "epoch": 4.83, + "grad_norm": 7.632187366485596, + "learning_rate": 6.906532663316584e-07, + "loss": 1.0316, + "step": 96600 + }, + { + "epoch": 4.83, + "grad_norm": 9.547595024108887, + "learning_rate": 6.705527638190955e-07, + "loss": 0.9944, + "step": 96700 + }, + { + "epoch": 4.84, + "grad_norm": 7.9466938972473145, + "learning_rate": 6.504522613065326e-07, + "loss": 1.0426, + "step": 96800 + }, + { + "epoch": 4.84, + "grad_norm": 4.955949783325195, + "learning_rate": 6.303517587939699e-07, + "loss": 1.0797, + "step": 96900 + }, + { + "epoch": 4.85, + "grad_norm": 8.598955154418945, + "learning_rate": 6.102512562814071e-07, + "loss": 1.0233, + "step": 97000 + }, + { + "epoch": 4.86, + "grad_norm": 5.320690631866455, + "learning_rate": 5.901507537688442e-07, + "loss": 1.0028, + "step": 97100 + }, + { + "epoch": 4.86, + "grad_norm": 7.243544101715088, + "learning_rate": 5.700502512562815e-07, + "loss": 0.9832, + "step": 97200 + }, + { + "epoch": 4.87, + "grad_norm": 6.318332672119141, + "learning_rate": 5.501507537688443e-07, + "loss": 1.0295, + "step": 97300 + }, + { + "epoch": 4.87, + "grad_norm": 5.297775745391846, + "learning_rate": 5.300502512562814e-07, + "loss": 1.0161, + "step": 97400 + }, + { + "epoch": 4.88, + "grad_norm": 5.717208385467529, + "learning_rate": 5.099497487437187e-07, + "loss": 1.0508, + "step": 97500 + }, + { + "epoch": 4.88, + "grad_norm": 6.545378684997559, + "learning_rate": 4.898492462311558e-07, + "loss": 1.039, + "step": 97600 + }, + { + "epoch": 4.88, + "grad_norm": 7.0295867919921875, + "learning_rate": 4.69748743718593e-07, + "loss": 0.9963, + "step": 97700 + }, + { + "epoch": 4.89, + "grad_norm": 6.685591697692871, + "learning_rate": 4.4964824120603015e-07, + "loss": 0.9845, + "step": 97800 + }, + { + "epoch": 4.89, + "grad_norm": 8.319560050964355, + "learning_rate": 4.295477386934674e-07, + "loss": 0.9951, + "step": 97900 + }, + { + "epoch": 4.9, + "grad_norm": 5.060975551605225, + "learning_rate": 4.094472361809045e-07, + "loss": 0.9893, + "step": 98000 + }, + { + "epoch": 4.91, + "grad_norm": 7.493504524230957, + "learning_rate": 3.8934673366834175e-07, + "loss": 0.9722, + "step": 98100 + }, + { + "epoch": 4.91, + "grad_norm": 6.903368949890137, + "learning_rate": 3.6924623115577893e-07, + "loss": 1.0248, + "step": 98200 + }, + { + "epoch": 4.92, + "grad_norm": 7.836544036865234, + "learning_rate": 3.4914572864321606e-07, + "loss": 1.0075, + "step": 98300 + }, + { + "epoch": 4.92, + "grad_norm": 7.111424446105957, + "learning_rate": 3.290452261306533e-07, + "loss": 0.9776, + "step": 98400 + }, + { + "epoch": 4.92, + "grad_norm": 5.424601078033447, + "learning_rate": 3.0894472361809047e-07, + "loss": 1.0235, + "step": 98500 + }, + { + "epoch": 4.93, + "grad_norm": 5.190994739532471, + "learning_rate": 2.8884422110552765e-07, + "loss": 1.0375, + "step": 98600 + }, + { + "epoch": 4.94, + "grad_norm": 7.971541404724121, + "learning_rate": 2.6874371859296483e-07, + "loss": 1.0049, + "step": 98700 + }, + { + "epoch": 4.94, + "grad_norm": 8.369765281677246, + "learning_rate": 2.48643216080402e-07, + "loss": 1.0028, + "step": 98800 + }, + { + "epoch": 4.95, + "grad_norm": 5.76533842086792, + "learning_rate": 2.2854271356783922e-07, + "loss": 0.9663, + "step": 98900 + }, + { + "epoch": 4.95, + "grad_norm": 8.707147598266602, + "learning_rate": 2.084422110552764e-07, + "loss": 1.0048, + "step": 99000 + }, + { + "epoch": 4.96, + "grad_norm": 7.287750720977783, + "learning_rate": 1.8834170854271358e-07, + "loss": 0.9777, + "step": 99100 + }, + { + "epoch": 4.96, + "grad_norm": 6.303613185882568, + "learning_rate": 1.6824120603015078e-07, + "loss": 1.0266, + "step": 99200 + }, + { + "epoch": 4.96, + "grad_norm": 6.6418776512146, + "learning_rate": 1.4814070351758796e-07, + "loss": 1.0179, + "step": 99300 + }, + { + "epoch": 4.97, + "grad_norm": 7.093784332275391, + "learning_rate": 1.2824120603015077e-07, + "loss": 1.0051, + "step": 99400 + }, + { + "epoch": 4.97, + "grad_norm": 9.0274658203125, + "learning_rate": 1.0814070351758795e-07, + "loss": 1.0022, + "step": 99500 + }, + { + "epoch": 4.98, + "grad_norm": 9.27753734588623, + "learning_rate": 8.804020100502513e-08, + "loss": 1.0176, + "step": 99600 + }, + { + "epoch": 4.99, + "grad_norm": 7.9141364097595215, + "learning_rate": 6.793969849246231e-08, + "loss": 0.9695, + "step": 99700 + }, + { + "epoch": 4.99, + "grad_norm": 6.786277770996094, + "learning_rate": 4.7839195979899497e-08, + "loss": 1.0346, + "step": 99800 + }, + { + "epoch": 5.0, + "grad_norm": 6.178213119506836, + "learning_rate": 2.7738693467336683e-08, + "loss": 0.987, + "step": 99900 + }, + { + "epoch": 5.0, + "grad_norm": 5.494964599609375, + "learning_rate": 7.63819095477387e-09, + "loss": 0.9787, + "step": 100000 + }, + { + "epoch": 5.0, + "eval_loss": 1.162515640258789, + "eval_runtime": 21.775, + "eval_samples_per_second": 45.924, + "eval_steps_per_second": 5.741, + "step": 100000 + } + ], + "logging_steps": 100, + "max_steps": 100000, + "num_input_tokens_seen": 0, + "num_train_epochs": 5, + "save_steps": 10000, + "total_flos": 1.1800273747968e+18, + "train_batch_size": 8, + "trial_name": null, + "trial_params": null +}