qanastek's picture
Upload 43 files
1a0b1a9
raw
history blame contribute delete
No virus
21.4 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 17.0,
"global_step": 86615,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 5e-09,
"loss": 10.525,
"step": 1
},
{
"epoch": 0.1,
"learning_rate": 2.5e-06,
"loss": 9.5219,
"step": 500
},
{
"epoch": 0.2,
"learning_rate": 5e-06,
"loss": 7.6948,
"step": 1000
},
{
"epoch": 0.29,
"learning_rate": 7.5e-06,
"loss": 6.4159,
"step": 1500
},
{
"epoch": 0.39,
"learning_rate": 1e-05,
"loss": 6.1298,
"step": 2000
},
{
"epoch": 0.49,
"learning_rate": 1.25e-05,
"loss": 6.0017,
"step": 2500
},
{
"epoch": 0.59,
"learning_rate": 1.5e-05,
"loss": 5.9201,
"step": 3000
},
{
"epoch": 0.69,
"learning_rate": 1.75e-05,
"loss": 5.8528,
"step": 3500
},
{
"epoch": 0.79,
"learning_rate": 2e-05,
"loss": 5.8017,
"step": 4000
},
{
"epoch": 0.88,
"learning_rate": 2.25e-05,
"loss": 5.7573,
"step": 4500
},
{
"epoch": 0.98,
"learning_rate": 2.5e-05,
"loss": 5.7162,
"step": 5000
},
{
"epoch": 1.08,
"learning_rate": 2.7495000000000004e-05,
"loss": 5.6827,
"step": 5500
},
{
"epoch": 1.18,
"learning_rate": 2.9995e-05,
"loss": 5.6557,
"step": 6000
},
{
"epoch": 1.28,
"learning_rate": 3.2495000000000007e-05,
"loss": 5.6229,
"step": 6500
},
{
"epoch": 1.37,
"learning_rate": 3.4995e-05,
"loss": 5.599,
"step": 7000
},
{
"epoch": 1.47,
"learning_rate": 3.7495e-05,
"loss": 5.5785,
"step": 7500
},
{
"epoch": 1.57,
"learning_rate": 3.9995000000000006e-05,
"loss": 5.5579,
"step": 8000
},
{
"epoch": 1.67,
"learning_rate": 4.2495e-05,
"loss": 5.5374,
"step": 8500
},
{
"epoch": 1.77,
"learning_rate": 4.4995000000000005e-05,
"loss": 5.5223,
"step": 9000
},
{
"epoch": 1.86,
"learning_rate": 4.7495e-05,
"loss": 5.5075,
"step": 9500
},
{
"epoch": 1.96,
"learning_rate": 4.9995000000000005e-05,
"loss": 5.4966,
"step": 10000
},
{
"epoch": 2.06,
"learning_rate": 4.9987721893491125e-05,
"loss": 5.4806,
"step": 10500
},
{
"epoch": 2.16,
"learning_rate": 4.997539447731755e-05,
"loss": 5.4673,
"step": 11000
},
{
"epoch": 2.26,
"learning_rate": 4.996306706114399e-05,
"loss": 5.4577,
"step": 11500
},
{
"epoch": 2.36,
"learning_rate": 4.9950739644970415e-05,
"loss": 5.4477,
"step": 12000
},
{
"epoch": 2.45,
"learning_rate": 4.993841222879685e-05,
"loss": 5.4356,
"step": 12500
},
{
"epoch": 2.55,
"learning_rate": 4.9926084812623276e-05,
"loss": 5.4225,
"step": 13000
},
{
"epoch": 2.65,
"learning_rate": 4.9913757396449704e-05,
"loss": 5.4196,
"step": 13500
},
{
"epoch": 2.75,
"learning_rate": 4.990142998027614e-05,
"loss": 5.4087,
"step": 14000
},
{
"epoch": 2.85,
"learning_rate": 4.9889127218934914e-05,
"loss": 5.4021,
"step": 14500
},
{
"epoch": 2.94,
"learning_rate": 4.987679980276134e-05,
"loss": 5.3944,
"step": 15000
},
{
"epoch": 3.04,
"learning_rate": 4.986449704142012e-05,
"loss": 5.3828,
"step": 15500
},
{
"epoch": 3.14,
"learning_rate": 4.985216962524655e-05,
"loss": 5.3786,
"step": 16000
},
{
"epoch": 3.24,
"learning_rate": 4.983984220907298e-05,
"loss": 5.3733,
"step": 16500
},
{
"epoch": 3.34,
"learning_rate": 4.982751479289941e-05,
"loss": 5.3666,
"step": 17000
},
{
"epoch": 3.43,
"learning_rate": 4.981518737672584e-05,
"loss": 5.3626,
"step": 17500
},
{
"epoch": 3.53,
"learning_rate": 4.9802859960552274e-05,
"loss": 5.353,
"step": 18000
},
{
"epoch": 3.63,
"learning_rate": 4.97905325443787e-05,
"loss": 5.3496,
"step": 18500
},
{
"epoch": 3.73,
"learning_rate": 4.977820512820513e-05,
"loss": 5.3454,
"step": 19000
},
{
"epoch": 3.83,
"learning_rate": 4.976590236686391e-05,
"loss": 5.3396,
"step": 19500
},
{
"epoch": 3.93,
"learning_rate": 4.975359960552268e-05,
"loss": 5.3352,
"step": 20000
},
{
"epoch": 4.02,
"learning_rate": 4.974129684418146e-05,
"loss": 5.3312,
"step": 20500
},
{
"epoch": 4.12,
"learning_rate": 4.9728969428007896e-05,
"loss": 5.3245,
"step": 21000
},
{
"epoch": 4.22,
"learning_rate": 4.9716642011834323e-05,
"loss": 5.3126,
"step": 21500
},
{
"epoch": 4.32,
"learning_rate": 4.970431459566075e-05,
"loss": 5.0663,
"step": 22000
},
{
"epoch": 4.42,
"learning_rate": 4.969198717948718e-05,
"loss": 4.7802,
"step": 22500
},
{
"epoch": 4.51,
"learning_rate": 4.967965976331361e-05,
"loss": 4.5089,
"step": 23000
},
{
"epoch": 4.61,
"learning_rate": 4.966733234714005e-05,
"loss": 4.2703,
"step": 23500
},
{
"epoch": 4.71,
"learning_rate": 4.965500493096647e-05,
"loss": 4.0607,
"step": 24000
},
{
"epoch": 4.81,
"learning_rate": 4.96426775147929e-05,
"loss": 3.8757,
"step": 24500
},
{
"epoch": 4.91,
"learning_rate": 4.963037475345168e-05,
"loss": 3.5749,
"step": 25000
},
{
"epoch": 5.0,
"learning_rate": 4.961807199211045e-05,
"loss": 3.0471,
"step": 25500
},
{
"epoch": 5.1,
"learning_rate": 4.960574457593689e-05,
"loss": 2.5977,
"step": 26000
},
{
"epoch": 5.2,
"learning_rate": 4.9593417159763314e-05,
"loss": 2.1306,
"step": 26500
},
{
"epoch": 5.3,
"learning_rate": 4.958108974358975e-05,
"loss": 1.9047,
"step": 27000
},
{
"epoch": 5.4,
"learning_rate": 4.9568762327416176e-05,
"loss": 1.7619,
"step": 27500
},
{
"epoch": 5.5,
"learning_rate": 4.9556434911242603e-05,
"loss": 1.6574,
"step": 28000
},
{
"epoch": 5.59,
"learning_rate": 4.954410749506904e-05,
"loss": 1.5785,
"step": 28500
},
{
"epoch": 5.69,
"learning_rate": 4.9531780078895465e-05,
"loss": 1.5168,
"step": 29000
},
{
"epoch": 5.79,
"learning_rate": 4.951945266272189e-05,
"loss": 1.467,
"step": 29500
},
{
"epoch": 5.89,
"learning_rate": 4.950712524654833e-05,
"loss": 1.4234,
"step": 30000
},
{
"epoch": 5.99,
"learning_rate": 4.94948224852071e-05,
"loss": 1.3853,
"step": 30500
},
{
"epoch": 6.08,
"learning_rate": 4.948251972386588e-05,
"loss": 1.3438,
"step": 31000
},
{
"epoch": 6.18,
"learning_rate": 4.947019230769231e-05,
"loss": 1.2958,
"step": 31500
},
{
"epoch": 6.28,
"learning_rate": 4.945786489151874e-05,
"loss": 1.241,
"step": 32000
},
{
"epoch": 6.38,
"learning_rate": 4.9445537475345174e-05,
"loss": 1.1933,
"step": 32500
},
{
"epoch": 6.48,
"learning_rate": 4.9433210059171594e-05,
"loss": 1.1561,
"step": 33000
},
{
"epoch": 6.58,
"learning_rate": 4.942088264299803e-05,
"loss": 1.1211,
"step": 33500
},
{
"epoch": 6.67,
"learning_rate": 4.940855522682446e-05,
"loss": 1.097,
"step": 34000
},
{
"epoch": 6.77,
"learning_rate": 4.939622781065089e-05,
"loss": 1.0713,
"step": 34500
},
{
"epoch": 6.87,
"learning_rate": 4.9383925049309666e-05,
"loss": 1.0513,
"step": 35000
},
{
"epoch": 6.97,
"learning_rate": 4.937159763313609e-05,
"loss": 1.0334,
"step": 35500
},
{
"epoch": 7.07,
"learning_rate": 4.9359294871794875e-05,
"loss": 1.012,
"step": 36000
},
{
"epoch": 7.16,
"learning_rate": 4.934696745562131e-05,
"loss": 0.9953,
"step": 36500
},
{
"epoch": 7.26,
"learning_rate": 4.933464003944773e-05,
"loss": 0.9811,
"step": 37000
},
{
"epoch": 7.36,
"learning_rate": 4.9322312623274164e-05,
"loss": 0.9679,
"step": 37500
},
{
"epoch": 7.46,
"learning_rate": 4.930998520710059e-05,
"loss": 0.9546,
"step": 38000
},
{
"epoch": 7.56,
"learning_rate": 4.9297657790927026e-05,
"loss": 0.9435,
"step": 38500
},
{
"epoch": 7.65,
"learning_rate": 4.9285330374753454e-05,
"loss": 0.9284,
"step": 39000
},
{
"epoch": 7.75,
"learning_rate": 4.927300295857988e-05,
"loss": 0.9222,
"step": 39500
},
{
"epoch": 7.85,
"learning_rate": 4.926070019723866e-05,
"loss": 0.913,
"step": 40000
},
{
"epoch": 7.95,
"learning_rate": 4.924837278106509e-05,
"loss": 0.9039,
"step": 40500
},
{
"epoch": 8.05,
"learning_rate": 4.9236070019723866e-05,
"loss": 0.8952,
"step": 41000
},
{
"epoch": 8.15,
"learning_rate": 4.92237426035503e-05,
"loss": 0.8863,
"step": 41500
},
{
"epoch": 8.24,
"learning_rate": 4.921141518737673e-05,
"loss": 0.8774,
"step": 42000
},
{
"epoch": 8.34,
"learning_rate": 4.9199087771203155e-05,
"loss": 0.8715,
"step": 42500
},
{
"epoch": 8.44,
"learning_rate": 4.918676035502959e-05,
"loss": 0.8636,
"step": 43000
},
{
"epoch": 8.54,
"learning_rate": 4.917443293885602e-05,
"loss": 0.8558,
"step": 43500
},
{
"epoch": 8.64,
"learning_rate": 4.916210552268245e-05,
"loss": 0.8472,
"step": 44000
},
{
"epoch": 8.73,
"learning_rate": 4.914977810650888e-05,
"loss": 0.8439,
"step": 44500
},
{
"epoch": 8.83,
"learning_rate": 4.9137475345167654e-05,
"loss": 0.8341,
"step": 45000
},
{
"epoch": 8.93,
"learning_rate": 4.912514792899409e-05,
"loss": 0.8298,
"step": 45500
},
{
"epoch": 9.03,
"learning_rate": 4.9112820512820516e-05,
"loss": 0.8235,
"step": 46000
},
{
"epoch": 9.13,
"learning_rate": 4.910049309664694e-05,
"loss": 0.8192,
"step": 46500
},
{
"epoch": 9.22,
"learning_rate": 4.908816568047338e-05,
"loss": 0.8103,
"step": 47000
},
{
"epoch": 9.32,
"learning_rate": 4.907586291913215e-05,
"loss": 0.8077,
"step": 47500
},
{
"epoch": 9.42,
"learning_rate": 4.906353550295858e-05,
"loss": 0.8009,
"step": 48000
},
{
"epoch": 9.52,
"learning_rate": 4.905120808678501e-05,
"loss": 0.7984,
"step": 48500
},
{
"epoch": 9.62,
"learning_rate": 4.903888067061144e-05,
"loss": 0.7924,
"step": 49000
},
{
"epoch": 9.72,
"learning_rate": 4.9026553254437876e-05,
"loss": 0.7877,
"step": 49500
},
{
"epoch": 9.81,
"learning_rate": 4.9014225838264304e-05,
"loss": 0.7824,
"step": 50000
},
{
"epoch": 9.91,
"learning_rate": 4.900192307692308e-05,
"loss": 0.7794,
"step": 50500
},
{
"epoch": 10.01,
"learning_rate": 4.898962031558186e-05,
"loss": 0.7749,
"step": 51000
},
{
"epoch": 10.11,
"learning_rate": 4.897729289940829e-05,
"loss": 0.7697,
"step": 51500
},
{
"epoch": 10.21,
"learning_rate": 4.8964965483234716e-05,
"loss": 0.7651,
"step": 52000
},
{
"epoch": 10.3,
"learning_rate": 4.8952638067061144e-05,
"loss": 0.7603,
"step": 52500
},
{
"epoch": 10.4,
"learning_rate": 4.894031065088758e-05,
"loss": 0.7583,
"step": 53000
},
{
"epoch": 10.5,
"learning_rate": 4.892798323471401e-05,
"loss": 0.755,
"step": 53500
},
{
"epoch": 10.6,
"learning_rate": 4.891565581854043e-05,
"loss": 0.7497,
"step": 54000
},
{
"epoch": 10.7,
"learning_rate": 4.890332840236687e-05,
"loss": 0.7457,
"step": 54500
},
{
"epoch": 10.79,
"learning_rate": 4.8891000986193295e-05,
"loss": 0.7429,
"step": 55000
},
{
"epoch": 10.89,
"learning_rate": 4.887869822485207e-05,
"loss": 0.7381,
"step": 55500
},
{
"epoch": 10.99,
"learning_rate": 4.8866370808678504e-05,
"loss": 0.7369,
"step": 56000
},
{
"epoch": 11.09,
"learning_rate": 4.885406804733728e-05,
"loss": 0.7321,
"step": 56500
},
{
"epoch": 11.19,
"learning_rate": 4.8841740631163714e-05,
"loss": 0.7293,
"step": 57000
},
{
"epoch": 11.29,
"learning_rate": 4.882941321499014e-05,
"loss": 0.7258,
"step": 57500
},
{
"epoch": 11.38,
"learning_rate": 4.881708579881657e-05,
"loss": 0.7243,
"step": 58000
},
{
"epoch": 11.48,
"learning_rate": 4.8804758382643e-05,
"loss": 0.7207,
"step": 58500
},
{
"epoch": 11.58,
"learning_rate": 4.879243096646943e-05,
"loss": 0.716,
"step": 59000
},
{
"epoch": 11.68,
"learning_rate": 4.878010355029586e-05,
"loss": 0.7123,
"step": 59500
},
{
"epoch": 11.78,
"learning_rate": 4.876777613412229e-05,
"loss": 0.7106,
"step": 60000
},
{
"epoch": 11.87,
"learning_rate": 4.875544871794872e-05,
"loss": 0.7086,
"step": 60500
},
{
"epoch": 11.97,
"learning_rate": 4.8743121301775154e-05,
"loss": 0.7058,
"step": 61000
},
{
"epoch": 12.07,
"learning_rate": 4.873084319526628e-05,
"loss": 0.7038,
"step": 61500
},
{
"epoch": 12.17,
"learning_rate": 4.8718515779092704e-05,
"loss": 0.7001,
"step": 62000
},
{
"epoch": 12.27,
"learning_rate": 4.870618836291914e-05,
"loss": 0.6989,
"step": 62500
},
{
"epoch": 12.37,
"learning_rate": 4.869386094674556e-05,
"loss": 0.6932,
"step": 63000
},
{
"epoch": 12.46,
"learning_rate": 4.8681533530571994e-05,
"loss": 0.6922,
"step": 63500
},
{
"epoch": 12.56,
"learning_rate": 4.866920611439843e-05,
"loss": 0.6894,
"step": 64000
},
{
"epoch": 12.66,
"learning_rate": 4.8656878698224855e-05,
"loss": 0.6862,
"step": 64500
},
{
"epoch": 12.76,
"learning_rate": 4.864455128205128e-05,
"loss": 0.6846,
"step": 65000
},
{
"epoch": 12.86,
"learning_rate": 4.863224852071006e-05,
"loss": 0.6832,
"step": 65500
},
{
"epoch": 12.95,
"learning_rate": 4.861994575936884e-05,
"loss": 0.6807,
"step": 66000
},
{
"epoch": 13.05,
"learning_rate": 4.8607642998027616e-05,
"loss": 0.678,
"step": 66500
},
{
"epoch": 13.15,
"learning_rate": 4.859531558185404e-05,
"loss": 0.6761,
"step": 67000
},
{
"epoch": 13.25,
"learning_rate": 4.858298816568048e-05,
"loss": 0.6738,
"step": 67500
},
{
"epoch": 13.35,
"learning_rate": 4.8570660749506905e-05,
"loss": 0.6713,
"step": 68000
},
{
"epoch": 13.44,
"learning_rate": 4.855833333333333e-05,
"loss": 0.6697,
"step": 68500
},
{
"epoch": 13.54,
"learning_rate": 4.8546005917159767e-05,
"loss": 0.6667,
"step": 69000
},
{
"epoch": 13.64,
"learning_rate": 4.8533678500986194e-05,
"loss": 0.6653,
"step": 69500
},
{
"epoch": 13.74,
"learning_rate": 4.852135108481263e-05,
"loss": 0.6634,
"step": 70000
},
{
"epoch": 13.84,
"learning_rate": 4.8509023668639056e-05,
"loss": 0.6603,
"step": 70500
},
{
"epoch": 13.94,
"learning_rate": 4.849669625246548e-05,
"loss": 0.6591,
"step": 71000
},
{
"epoch": 14.03,
"learning_rate": 4.8484393491124265e-05,
"loss": 0.6573,
"step": 71500
},
{
"epoch": 14.13,
"learning_rate": 4.847206607495069e-05,
"loss": 0.6543,
"step": 72000
},
{
"epoch": 14.23,
"learning_rate": 4.845973865877712e-05,
"loss": 0.6525,
"step": 72500
},
{
"epoch": 14.33,
"learning_rate": 4.8447411242603555e-05,
"loss": 0.6518,
"step": 73000
},
{
"epoch": 14.43,
"learning_rate": 4.843508382642998e-05,
"loss": 0.6494,
"step": 73500
},
{
"epoch": 14.52,
"learning_rate": 4.8422756410256416e-05,
"loss": 0.6478,
"step": 74000
},
{
"epoch": 14.62,
"learning_rate": 4.841042899408284e-05,
"loss": 0.646,
"step": 74500
},
{
"epoch": 14.72,
"learning_rate": 4.839810157790927e-05,
"loss": 0.6438,
"step": 75000
},
{
"epoch": 14.82,
"learning_rate": 4.8385774161735706e-05,
"loss": 0.6421,
"step": 75500
},
{
"epoch": 14.92,
"learning_rate": 4.837349605522682e-05,
"loss": 0.6417,
"step": 76000
},
{
"epoch": 15.01,
"learning_rate": 4.8361168639053256e-05,
"loss": 0.6416,
"step": 76500
},
{
"epoch": 15.11,
"learning_rate": 4.834884122287969e-05,
"loss": 0.637,
"step": 77000
},
{
"epoch": 15.21,
"learning_rate": 4.833651380670612e-05,
"loss": 0.6362,
"step": 77500
},
{
"epoch": 15.31,
"learning_rate": 4.8324186390532545e-05,
"loss": 0.6355,
"step": 78000
},
{
"epoch": 15.41,
"learning_rate": 4.831185897435897e-05,
"loss": 0.6336,
"step": 78500
},
{
"epoch": 15.51,
"learning_rate": 4.829953155818541e-05,
"loss": 0.6304,
"step": 79000
},
{
"epoch": 15.6,
"learning_rate": 4.828720414201184e-05,
"loss": 0.6306,
"step": 79500
},
{
"epoch": 15.7,
"learning_rate": 4.827487672583826e-05,
"loss": 0.6271,
"step": 80000
},
{
"epoch": 15.8,
"learning_rate": 4.8262573964497044e-05,
"loss": 0.6278,
"step": 80500
},
{
"epoch": 15.9,
"learning_rate": 4.825024654832347e-05,
"loss": 0.6245,
"step": 81000
},
{
"epoch": 16.0,
"learning_rate": 4.8237919132149906e-05,
"loss": 0.6236,
"step": 81500
},
{
"epoch": 16.09,
"learning_rate": 4.822561637080868e-05,
"loss": 0.622,
"step": 82000
},
{
"epoch": 16.19,
"learning_rate": 4.821328895463511e-05,
"loss": 0.6209,
"step": 82500
},
{
"epoch": 16.29,
"learning_rate": 4.820096153846154e-05,
"loss": 0.6187,
"step": 83000
},
{
"epoch": 16.39,
"learning_rate": 4.818863412228797e-05,
"loss": 0.6189,
"step": 83500
},
{
"epoch": 16.49,
"learning_rate": 4.8176331360946746e-05,
"loss": 0.6159,
"step": 84000
},
{
"epoch": 16.58,
"learning_rate": 4.816400394477318e-05,
"loss": 0.615,
"step": 84500
},
{
"epoch": 16.68,
"learning_rate": 4.815167652859961e-05,
"loss": 0.6155,
"step": 85000
},
{
"epoch": 16.78,
"learning_rate": 4.8139349112426035e-05,
"loss": 0.6121,
"step": 85500
},
{
"epoch": 16.88,
"learning_rate": 4.812702169625247e-05,
"loss": 0.6115,
"step": 86000
},
{
"epoch": 16.98,
"learning_rate": 4.8114718934911245e-05,
"loss": 0.6113,
"step": 86500
}
],
"max_steps": 2038000,
"num_train_epochs": 400,
"total_flos": 2.3340589932538233e+19,
"trial_name": null,
"trial_params": null
}