{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 6.0,
"eval_steps": 500,
"global_step": 94368,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 0.0,
"loss": 8.4152,
"step": 1
},
{
"epoch": 0.03,
"learning_rate": 4.950000000000001e-06,
"loss": 3.4041,
"step": 500
},
{
"epoch": 0.06,
"learning_rate": 9.950000000000001e-06,
"loss": 1.6651,
"step": 1000
},
{
"epoch": 0.1,
"learning_rate": 1.4950000000000001e-05,
"loss": 1.5062,
"step": 1500
},
{
"epoch": 0.13,
"learning_rate": 1.995e-05,
"loss": 1.4128,
"step": 2000
},
{
"epoch": 0.16,
"learning_rate": 2.495e-05,
"loss": 1.347,
"step": 2500
},
{
"epoch": 0.19,
"learning_rate": 2.995e-05,
"loss": 1.2902,
"step": 3000
},
{
"epoch": 0.22,
"learning_rate": 3.495e-05,
"loss": 1.2457,
"step": 3500
},
{
"epoch": 0.25,
"learning_rate": 3.995e-05,
"loss": 1.2037,
"step": 4000
},
{
"epoch": 0.29,
"learning_rate": 4.495e-05,
"loss": 1.1728,
"step": 4500
},
{
"epoch": 0.32,
"learning_rate": 4.995e-05,
"loss": 1.1493,
"step": 5000
},
{
"epoch": 0.35,
"learning_rate": 5.495e-05,
"loss": 1.1211,
"step": 5500
},
{
"epoch": 0.38,
"learning_rate": 5.995000000000001e-05,
"loss": 1.0958,
"step": 6000
},
{
"epoch": 0.41,
"learning_rate": 6.494999999999999e-05,
"loss": 1.0736,
"step": 6500
},
{
"epoch": 0.45,
"learning_rate": 6.995e-05,
"loss": 1.0591,
"step": 7000
},
{
"epoch": 0.48,
"learning_rate": 7.494e-05,
"loss": 1.0399,
"step": 7500
},
{
"epoch": 0.51,
"learning_rate": 7.994000000000001e-05,
"loss": 1.0244,
"step": 8000
},
{
"epoch": 0.54,
"learning_rate": 8.494000000000001e-05,
"loss": 1.0128,
"step": 8500
},
{
"epoch": 0.57,
"learning_rate": 8.994e-05,
"loss": 1.0026,
"step": 9000
},
{
"epoch": 0.6,
"learning_rate": 9.494000000000001e-05,
"loss": 0.9945,
"step": 9500
},
{
"epoch": 0.64,
"learning_rate": 9.994e-05,
"loss": 0.9798,
"step": 10000
},
{
"epoch": 0.67,
"learning_rate": 9.941446994120994e-05,
"loss": 0.9709,
"step": 10500
},
{
"epoch": 0.7,
"learning_rate": 9.882182818130097e-05,
"loss": 0.9623,
"step": 11000
},
{
"epoch": 0.73,
"learning_rate": 9.8229186421392e-05,
"loss": 0.9505,
"step": 11500
},
{
"epoch": 0.76,
"learning_rate": 9.763654466148303e-05,
"loss": 0.9428,
"step": 12000
},
{
"epoch": 0.79,
"learning_rate": 9.704390290157406e-05,
"loss": 0.9323,
"step": 12500
},
{
"epoch": 0.83,
"learning_rate": 9.645244642518491e-05,
"loss": 0.9265,
"step": 13000
},
{
"epoch": 0.86,
"learning_rate": 9.585980466527594e-05,
"loss": 0.9197,
"step": 13500
},
{
"epoch": 0.89,
"learning_rate": 9.526716290536697e-05,
"loss": 0.9122,
"step": 14000
},
{
"epoch": 0.92,
"learning_rate": 9.4674521145458e-05,
"loss": 0.9065,
"step": 14500
},
{
"epoch": 0.95,
"learning_rate": 9.408187938554904e-05,
"loss": 0.9023,
"step": 15000
},
{
"epoch": 0.99,
"learning_rate": 9.348923762564006e-05,
"loss": 0.8965,
"step": 15500
},
{
"epoch": 1.02,
"learning_rate": 9.289659586573108e-05,
"loss": 0.8901,
"step": 16000
},
{
"epoch": 1.05,
"learning_rate": 9.230395410582213e-05,
"loss": 0.8822,
"step": 16500
},
{
"epoch": 1.08,
"learning_rate": 9.171131234591315e-05,
"loss": 0.8767,
"step": 17000
},
{
"epoch": 1.11,
"learning_rate": 9.111867058600417e-05,
"loss": 0.8711,
"step": 17500
},
{
"epoch": 1.14,
"learning_rate": 9.052721410961502e-05,
"loss": 0.8692,
"step": 18000
},
{
"epoch": 1.18,
"learning_rate": 8.993457234970606e-05,
"loss": 0.8655,
"step": 18500
},
{
"epoch": 1.21,
"learning_rate": 8.934193058979709e-05,
"loss": 0.861,
"step": 19000
},
{
"epoch": 1.24,
"learning_rate": 8.874928882988811e-05,
"loss": 0.8566,
"step": 19500
},
{
"epoch": 1.27,
"learning_rate": 8.815664706997914e-05,
"loss": 0.8509,
"step": 20000
},
{
"epoch": 1.3,
"learning_rate": 8.756519059359e-05,
"loss": 0.8492,
"step": 20500
},
{
"epoch": 1.34,
"learning_rate": 8.697373411720084e-05,
"loss": 0.8475,
"step": 21000
},
{
"epoch": 1.37,
"learning_rate": 8.638109235729187e-05,
"loss": 0.8471,
"step": 21500
},
{
"epoch": 1.4,
"learning_rate": 8.578845059738289e-05,
"loss": 0.8393,
"step": 22000
},
{
"epoch": 1.43,
"learning_rate": 8.519580883747393e-05,
"loss": 0.8341,
"step": 22500
},
{
"epoch": 1.46,
"learning_rate": 8.460316707756496e-05,
"loss": 0.8332,
"step": 23000
},
{
"epoch": 1.49,
"learning_rate": 8.401052531765598e-05,
"loss": 0.8295,
"step": 23500
},
{
"epoch": 1.53,
"learning_rate": 8.341788355774702e-05,
"loss": 0.8295,
"step": 24000
},
{
"epoch": 1.56,
"learning_rate": 8.282524179783805e-05,
"loss": 0.8237,
"step": 24500
},
{
"epoch": 1.59,
"learning_rate": 8.22337853214489e-05,
"loss": 0.8213,
"step": 25000
},
{
"epoch": 1.62,
"learning_rate": 8.164232884505975e-05,
"loss": 0.8187,
"step": 25500
},
{
"epoch": 1.65,
"learning_rate": 8.104968708515077e-05,
"loss": 0.8154,
"step": 26000
},
{
"epoch": 1.68,
"learning_rate": 8.04570453252418e-05,
"loss": 0.8133,
"step": 26500
},
{
"epoch": 1.72,
"learning_rate": 7.986440356533284e-05,
"loss": 0.8064,
"step": 27000
},
{
"epoch": 1.75,
"learning_rate": 7.927176180542386e-05,
"loss": 0.8087,
"step": 27500
},
{
"epoch": 1.78,
"learning_rate": 7.86803053290347e-05,
"loss": 0.809,
"step": 28000
},
{
"epoch": 1.81,
"learning_rate": 7.808766356912574e-05,
"loss": 0.8087,
"step": 28500
},
{
"epoch": 1.84,
"learning_rate": 7.749502180921678e-05,
"loss": 0.8018,
"step": 29000
},
{
"epoch": 1.88,
"learning_rate": 7.690356533282762e-05,
"loss": 0.7998,
"step": 29500
},
{
"epoch": 1.91,
"learning_rate": 7.631092357291864e-05,
"loss": 0.7984,
"step": 30000
},
{
"epoch": 1.94,
"learning_rate": 7.571828181300968e-05,
"loss": 0.7941,
"step": 30500
},
{
"epoch": 1.97,
"learning_rate": 7.512564005310071e-05,
"loss": 0.7965,
"step": 31000
},
{
"epoch": 2.0,
"learning_rate": 7.453299829319173e-05,
"loss": 0.7941,
"step": 31500
},
{
"epoch": 2.03,
"learning_rate": 7.394035653328276e-05,
"loss": 0.7901,
"step": 32000
},
{
"epoch": 2.07,
"learning_rate": 7.33477147733738e-05,
"loss": 0.7855,
"step": 32500
},
{
"epoch": 2.1,
"learning_rate": 7.275507301346483e-05,
"loss": 0.7825,
"step": 33000
},
{
"epoch": 2.13,
"learning_rate": 7.216243125355585e-05,
"loss": 0.7848,
"step": 33500
},
{
"epoch": 2.16,
"learning_rate": 7.156978949364688e-05,
"loss": 0.779,
"step": 34000
},
{
"epoch": 2.19,
"learning_rate": 7.097714773373792e-05,
"loss": 0.7778,
"step": 34500
},
{
"epoch": 2.23,
"learning_rate": 7.038450597382894e-05,
"loss": 0.7789,
"step": 35000
},
{
"epoch": 2.26,
"learning_rate": 6.979186421391997e-05,
"loss": 0.7749,
"step": 35500
},
{
"epoch": 2.29,
"learning_rate": 6.920040773753082e-05,
"loss": 0.7719,
"step": 36000
},
{
"epoch": 2.32,
"learning_rate": 6.860776597762185e-05,
"loss": 0.7755,
"step": 36500
},
{
"epoch": 2.35,
"learning_rate": 6.801512421771288e-05,
"loss": 0.7708,
"step": 37000
},
{
"epoch": 2.38,
"learning_rate": 6.742248245780391e-05,
"loss": 0.7677,
"step": 37500
},
{
"epoch": 2.42,
"learning_rate": 6.682984069789495e-05,
"loss": 0.766,
"step": 38000
},
{
"epoch": 2.45,
"learning_rate": 6.623719893798597e-05,
"loss": 0.767,
"step": 38500
},
{
"epoch": 2.48,
"learning_rate": 6.5644557178077e-05,
"loss": 0.7628,
"step": 39000
},
{
"epoch": 2.51,
"learning_rate": 6.505191541816802e-05,
"loss": 0.7638,
"step": 39500
},
{
"epoch": 2.54,
"learning_rate": 6.445927365825906e-05,
"loss": 0.7631,
"step": 40000
},
{
"epoch": 2.58,
"learning_rate": 6.38666318983501e-05,
"loss": 0.7587,
"step": 40500
},
{
"epoch": 2.61,
"learning_rate": 6.327636070548075e-05,
"loss": 0.7595,
"step": 41000
},
{
"epoch": 2.64,
"learning_rate": 6.268371894557177e-05,
"loss": 0.7599,
"step": 41500
},
{
"epoch": 2.67,
"learning_rate": 6.209107718566282e-05,
"loss": 0.7583,
"step": 42000
},
{
"epoch": 2.7,
"learning_rate": 6.149843542575384e-05,
"loss": 0.7534,
"step": 42500
},
{
"epoch": 2.73,
"learning_rate": 6.090579366584487e-05,
"loss": 0.7564,
"step": 43000
},
{
"epoch": 2.77,
"learning_rate": 6.0313151905935906e-05,
"loss": 0.7527,
"step": 43500
},
{
"epoch": 2.8,
"learning_rate": 5.9720510146026934e-05,
"loss": 0.7526,
"step": 44000
},
{
"epoch": 2.83,
"learning_rate": 5.9127868386117956e-05,
"loss": 0.7519,
"step": 44500
},
{
"epoch": 2.86,
"learning_rate": 5.85364119097288e-05,
"loss": 0.749,
"step": 45000
},
{
"epoch": 2.89,
"learning_rate": 5.7943770149819844e-05,
"loss": 0.7492,
"step": 45500
},
{
"epoch": 2.92,
"learning_rate": 5.7351128389910865e-05,
"loss": 0.7468,
"step": 46000
},
{
"epoch": 2.96,
"learning_rate": 5.6758486630001894e-05,
"loss": 0.745,
"step": 46500
},
{
"epoch": 2.99,
"learning_rate": 5.616584487009293e-05,
"loss": 0.7479,
"step": 47000
},
{
"epoch": 3.02,
"learning_rate": 5.557320311018396e-05,
"loss": 0.7431,
"step": 47500
},
{
"epoch": 3.05,
"learning_rate": 5.4980561350274985e-05,
"loss": 0.7406,
"step": 48000
},
{
"epoch": 3.08,
"learning_rate": 5.438791959036602e-05,
"loss": 0.7401,
"step": 48500
},
{
"epoch": 3.12,
"learning_rate": 5.379527783045705e-05,
"loss": 0.7379,
"step": 49000
},
{
"epoch": 3.15,
"learning_rate": 5.3203821354067895e-05,
"loss": 0.7375,
"step": 49500
},
{
"epoch": 3.18,
"learning_rate": 5.261117959415892e-05,
"loss": 0.7339,
"step": 50000
},
{
"epoch": 3.21,
"learning_rate": 5.201853783424996e-05,
"loss": 0.7349,
"step": 50500
},
{
"epoch": 3.24,
"learning_rate": 5.1425896074340986e-05,
"loss": 0.7348,
"step": 51000
},
{
"epoch": 3.27,
"learning_rate": 5.083443959795183e-05,
"loss": 0.7319,
"step": 51500
},
{
"epoch": 3.31,
"learning_rate": 5.024179783804286e-05,
"loss": 0.7271,
"step": 52000
},
{
"epoch": 3.34,
"learning_rate": 4.964915607813389e-05,
"loss": 0.7309,
"step": 52500
},
{
"epoch": 3.37,
"learning_rate": 4.905769960174474e-05,
"loss": 0.733,
"step": 53000
},
{
"epoch": 3.4,
"learning_rate": 4.846505784183577e-05,
"loss": 0.7277,
"step": 53500
},
{
"epoch": 3.43,
"learning_rate": 4.7872416081926805e-05,
"loss": 0.728,
"step": 54000
},
{
"epoch": 3.47,
"learning_rate": 4.7279774322017827e-05,
"loss": 0.7268,
"step": 54500
},
{
"epoch": 3.5,
"learning_rate": 4.668713256210886e-05,
"loss": 0.7261,
"step": 55000
},
{
"epoch": 3.53,
"learning_rate": 4.609567608571971e-05,
"loss": 0.7251,
"step": 55500
},
{
"epoch": 3.56,
"learning_rate": 4.5503034325810736e-05,
"loss": 0.7261,
"step": 56000
},
{
"epoch": 3.59,
"learning_rate": 4.4910392565901764e-05,
"loss": 0.7231,
"step": 56500
},
{
"epoch": 3.62,
"learning_rate": 4.43177508059928e-05,
"loss": 0.7197,
"step": 57000
},
{
"epoch": 3.66,
"learning_rate": 4.3726294329603646e-05,
"loss": 0.7249,
"step": 57500
},
{
"epoch": 3.69,
"learning_rate": 4.313483785321449e-05,
"loss": 0.7204,
"step": 58000
},
{
"epoch": 3.72,
"learning_rate": 4.254219609330552e-05,
"loss": 0.718,
"step": 58500
},
{
"epoch": 3.75,
"learning_rate": 4.194955433339655e-05,
"loss": 0.7205,
"step": 59000
},
{
"epoch": 3.78,
"learning_rate": 4.1356912573487584e-05,
"loss": 0.7164,
"step": 59500
},
{
"epoch": 3.81,
"learning_rate": 4.076427081357861e-05,
"loss": 0.7194,
"step": 60000
},
{
"epoch": 3.85,
"learning_rate": 4.017162905366964e-05,
"loss": 0.7166,
"step": 60500
},
{
"epoch": 3.88,
"learning_rate": 3.9578987293760675e-05,
"loss": 0.7167,
"step": 61000
},
{
"epoch": 3.91,
"learning_rate": 3.8986345533851696e-05,
"loss": 0.7175,
"step": 61500
},
{
"epoch": 3.94,
"learning_rate": 3.839370377394273e-05,
"loss": 0.7104,
"step": 62000
},
{
"epoch": 3.97,
"learning_rate": 3.780106201403376e-05,
"loss": 0.7116,
"step": 62500
},
{
"epoch": 4.01,
"learning_rate": 3.720842025412479e-05,
"loss": 0.7125,
"step": 63000
},
{
"epoch": 4.04,
"learning_rate": 3.6615778494215816e-05,
"loss": 0.7098,
"step": 63500
},
{
"epoch": 4.07,
"learning_rate": 3.602432201782667e-05,
"loss": 0.708,
"step": 64000
},
{
"epoch": 4.1,
"learning_rate": 3.543168025791769e-05,
"loss": 0.7081,
"step": 64500
},
{
"epoch": 4.13,
"learning_rate": 3.4839038498008725e-05,
"loss": 0.7058,
"step": 65000
},
{
"epoch": 4.16,
"learning_rate": 3.4246396738099754e-05,
"loss": 0.7072,
"step": 65500
},
{
"epoch": 4.2,
"learning_rate": 3.365375497819078e-05,
"loss": 0.7068,
"step": 66000
},
{
"epoch": 4.23,
"learning_rate": 3.306111321828182e-05,
"loss": 0.706,
"step": 66500
},
{
"epoch": 4.26,
"learning_rate": 3.246965674189266e-05,
"loss": 0.705,
"step": 67000
},
{
"epoch": 4.29,
"learning_rate": 3.187701498198369e-05,
"loss": 0.7043,
"step": 67500
},
{
"epoch": 4.32,
"learning_rate": 3.128437322207472e-05,
"loss": 0.7025,
"step": 68000
},
{
"epoch": 4.36,
"learning_rate": 3.0691731462165755e-05,
"loss": 0.7019,
"step": 68500
},
{
"epoch": 4.39,
"learning_rate": 3.0099089702256783e-05,
"loss": 0.703,
"step": 69000
},
{
"epoch": 4.42,
"learning_rate": 2.9506447942347808e-05,
"loss": 0.7023,
"step": 69500
},
{
"epoch": 4.45,
"learning_rate": 2.891380618243884e-05,
"loss": 0.7023,
"step": 70000
},
{
"epoch": 4.48,
"learning_rate": 2.832234970604969e-05,
"loss": 0.7,
"step": 70500
},
{
"epoch": 4.51,
"learning_rate": 2.772970794614072e-05,
"loss": 0.6976,
"step": 71000
},
{
"epoch": 4.55,
"learning_rate": 2.7137066186231745e-05,
"loss": 0.7009,
"step": 71500
},
{
"epoch": 4.58,
"learning_rate": 2.6544424426322777e-05,
"loss": 0.6985,
"step": 72000
},
{
"epoch": 4.61,
"learning_rate": 2.595178266641381e-05,
"loss": 0.6976,
"step": 72500
},
{
"epoch": 4.64,
"learning_rate": 2.5359140906504837e-05,
"loss": 0.6988,
"step": 73000
},
{
"epoch": 4.67,
"learning_rate": 2.4766499146595868e-05,
"loss": 0.6988,
"step": 73500
},
{
"epoch": 4.7,
"learning_rate": 2.4173857386686896e-05,
"loss": 0.6964,
"step": 74000
},
{
"epoch": 4.74,
"learning_rate": 2.3582400910297743e-05,
"loss": 0.6972,
"step": 74500
},
{
"epoch": 4.77,
"learning_rate": 2.2989759150388775e-05,
"loss": 0.6948,
"step": 75000
},
{
"epoch": 4.8,
"learning_rate": 2.2397117390479806e-05,
"loss": 0.6946,
"step": 75500
},
{
"epoch": 4.83,
"learning_rate": 2.1804475630570834e-05,
"loss": 0.6927,
"step": 76000
},
{
"epoch": 4.86,
"learning_rate": 2.1211833870661862e-05,
"loss": 0.6909,
"step": 76500
},
{
"epoch": 4.9,
"learning_rate": 2.0620377394272712e-05,
"loss": 0.6906,
"step": 77000
},
{
"epoch": 4.93,
"learning_rate": 2.002892091788356e-05,
"loss": 0.6929,
"step": 77500
},
{
"epoch": 4.96,
"learning_rate": 1.9436279157974587e-05,
"loss": 0.6917,
"step": 78000
},
{
"epoch": 4.99,
"learning_rate": 1.884363739806562e-05,
"loss": 0.6925,
"step": 78500
},
{
"epoch": 5.02,
"learning_rate": 1.8250995638156647e-05,
"loss": 0.6895,
"step": 79000
},
{
"epoch": 5.05,
"learning_rate": 1.765835387824768e-05,
"loss": 0.686,
"step": 79500
},
{
"epoch": 5.09,
"learning_rate": 1.706571211833871e-05,
"loss": 0.6882,
"step": 80000
},
{
"epoch": 5.12,
"learning_rate": 1.6473070358429738e-05,
"loss": 0.6877,
"step": 80500
},
{
"epoch": 5.15,
"learning_rate": 1.5880428598520766e-05,
"loss": 0.688,
"step": 81000
},
{
"epoch": 5.18,
"learning_rate": 1.5287786838611798e-05,
"loss": 0.6863,
"step": 81500
},
{
"epoch": 5.21,
"learning_rate": 1.4696330362222643e-05,
"loss": 0.6882,
"step": 82000
},
{
"epoch": 5.25,
"learning_rate": 1.4104873885833491e-05,
"loss": 0.6892,
"step": 82500
},
{
"epoch": 5.28,
"learning_rate": 1.351223212592452e-05,
"loss": 0.6878,
"step": 83000
},
{
"epoch": 5.31,
"learning_rate": 1.2919590366015552e-05,
"loss": 0.6841,
"step": 83500
},
{
"epoch": 5.34,
"learning_rate": 1.232694860610658e-05,
"loss": 0.6841,
"step": 84000
},
{
"epoch": 5.37,
"learning_rate": 1.173430684619761e-05,
"loss": 0.6855,
"step": 84500
},
{
"epoch": 5.4,
"learning_rate": 1.1141665086288642e-05,
"loss": 0.6833,
"step": 85000
},
{
"epoch": 5.44,
"learning_rate": 1.054902332637967e-05,
"loss": 0.683,
"step": 85500
},
{
"epoch": 5.47,
"learning_rate": 9.9563815664707e-06,
"loss": 0.6847,
"step": 86000
},
{
"epoch": 5.5,
"learning_rate": 9.36373980656173e-06,
"loss": 0.6845,
"step": 86500
},
{
"epoch": 5.53,
"learning_rate": 8.77109804665276e-06,
"loss": 0.6795,
"step": 87000
},
{
"epoch": 5.56,
"learning_rate": 8.17845628674379e-06,
"loss": 0.6817,
"step": 87500
},
{
"epoch": 5.6,
"learning_rate": 7.585814526834819e-06,
"loss": 0.6796,
"step": 88000
},
{
"epoch": 5.63,
"learning_rate": 6.994358050445667e-06,
"loss": 0.684,
"step": 88500
},
{
"epoch": 5.66,
"learning_rate": 6.401716290536697e-06,
"loss": 0.6833,
"step": 89000
},
{
"epoch": 5.69,
"learning_rate": 5.810259814147544e-06,
"loss": 0.6805,
"step": 89500
},
{
"epoch": 5.72,
"learning_rate": 5.217618054238574e-06,
"loss": 0.6823,
"step": 90000
},
{
"epoch": 5.75,
"learning_rate": 4.626161577849421e-06,
"loss": 0.6786,
"step": 90500
},
{
"epoch": 5.79,
"learning_rate": 4.033519817940452e-06,
"loss": 0.676,
"step": 91000
},
{
"epoch": 5.82,
"learning_rate": 3.440878058031481e-06,
"loss": 0.6811,
"step": 91500
},
{
"epoch": 5.85,
"learning_rate": 2.848236298122511e-06,
"loss": 0.6805,
"step": 92000
},
{
"epoch": 5.88,
"learning_rate": 2.2567798217333587e-06,
"loss": 0.6805,
"step": 92500
},
{
"epoch": 5.91,
"learning_rate": 1.6641380618243886e-06,
"loss": 0.6783,
"step": 93000
},
{
"epoch": 5.94,
"learning_rate": 1.0714963019154182e-06,
"loss": 0.6806,
"step": 93500
},
{
"epoch": 5.98,
"learning_rate": 4.78854542006448e-07,
"loss": 0.6792,
"step": 94000
},
{
"before_init_mem_cpu": 1314717696,
"before_init_mem_gpu": 0,
"epoch": 6.0,
"init_mem_cpu_alloc_delta": 880553984,
"init_mem_cpu_peaked_delta": 93917184,
"init_mem_gpu_alloc_delta": 443396608,
"init_mem_gpu_peaked_delta": 0,
"step": 94368,
"total_flos": 1.9076000524147884e+19,
"train_loss": 0.8106447110522921,
"train_mem_cpu_alloc_delta": 31978672128,
"train_mem_cpu_peaked_delta": 500772864,
"train_mem_gpu_alloc_delta": 1348273664,
"train_mem_gpu_peaked_delta": 64082841600,
"train_runtime": 71211.051,
"train_samples_per_second": 1017.691,
"train_steps_per_second": 1.325
}
],
"logging_steps": 500,
"max_steps": 94368,
"num_train_epochs": 6,
"save_steps": 500,
"total_flos": 1.9076000524147884e+19,
"trial_name": null,
"trial_params": null
}