coqui / last-checkpoint /trainer_state.json
dq158's picture
Training in progress, epoch 2, checkpoint
1ac2948
raw
history blame
No virus
20.9 kB
{
"best_metric": 3.0691702365875244,
"best_model_checkpoint": "dq158/coqui/checkpoint-80696",
"epoch": 2.0,
"eval_steps": 500,
"global_step": 80696,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 8e-05,
"loss": 4.5486,
"step": 500
},
{
"epoch": 0.02,
"learning_rate": 7.999996629129878e-05,
"loss": 3.9512,
"step": 1000
},
{
"epoch": 0.04,
"learning_rate": 7.999986516525194e-05,
"loss": 3.7313,
"step": 1500
},
{
"epoch": 0.05,
"learning_rate": 7.999969662202991e-05,
"loss": 3.6577,
"step": 2000
},
{
"epoch": 0.06,
"learning_rate": 7.999946066191677e-05,
"loss": 3.5469,
"step": 2500
},
{
"epoch": 0.07,
"learning_rate": 7.999915728531019e-05,
"loss": 3.7429,
"step": 3000
},
{
"epoch": 0.09,
"learning_rate": 7.999878649272152e-05,
"loss": 3.5995,
"step": 3500
},
{
"epoch": 0.1,
"learning_rate": 7.99983482847757e-05,
"loss": 3.4906,
"step": 4000
},
{
"epoch": 0.11,
"learning_rate": 7.999784266221129e-05,
"loss": 3.6012,
"step": 4500
},
{
"epoch": 0.12,
"learning_rate": 7.99972696258805e-05,
"loss": 3.4275,
"step": 5000
},
{
"epoch": 0.14,
"learning_rate": 7.999662917674914e-05,
"loss": 3.5431,
"step": 5500
},
{
"epoch": 0.15,
"learning_rate": 7.999592131589664e-05,
"loss": 3.4628,
"step": 6000
},
{
"epoch": 0.16,
"learning_rate": 7.999514604451606e-05,
"loss": 3.5663,
"step": 6500
},
{
"epoch": 0.17,
"learning_rate": 7.999430336391406e-05,
"loss": 3.4909,
"step": 7000
},
{
"epoch": 0.19,
"learning_rate": 7.999339327551093e-05,
"loss": 3.4715,
"step": 7500
},
{
"epoch": 0.2,
"learning_rate": 7.999241578084059e-05,
"loss": 3.4117,
"step": 8000
},
{
"epoch": 0.21,
"learning_rate": 7.999137088155049e-05,
"loss": 3.5166,
"step": 8500
},
{
"epoch": 0.22,
"learning_rate": 7.999025857940177e-05,
"loss": 3.4116,
"step": 9000
},
{
"epoch": 0.24,
"learning_rate": 7.998907887626916e-05,
"loss": 3.4845,
"step": 9500
},
{
"epoch": 0.25,
"learning_rate": 7.998783177414093e-05,
"loss": 3.5398,
"step": 10000
},
{
"epoch": 0.26,
"learning_rate": 7.998651727511902e-05,
"loss": 3.4251,
"step": 10500
},
{
"epoch": 0.27,
"learning_rate": 7.998513538141894e-05,
"loss": 3.4268,
"step": 11000
},
{
"epoch": 0.29,
"learning_rate": 7.998368609536976e-05,
"loss": 3.4133,
"step": 11500
},
{
"epoch": 0.3,
"learning_rate": 7.998216941941417e-05,
"loss": 3.4393,
"step": 12000
},
{
"epoch": 0.31,
"learning_rate": 7.998058535610842e-05,
"loss": 3.3059,
"step": 12500
},
{
"epoch": 0.32,
"learning_rate": 7.997893390812236e-05,
"loss": 3.3631,
"step": 13000
},
{
"epoch": 0.33,
"learning_rate": 7.997721507823938e-05,
"loss": 3.391,
"step": 13500
},
{
"epoch": 0.35,
"learning_rate": 7.997542886935647e-05,
"loss": 3.3573,
"step": 14000
},
{
"epoch": 0.36,
"learning_rate": 7.997357528448417e-05,
"loss": 3.3698,
"step": 14500
},
{
"epoch": 0.37,
"learning_rate": 7.997165432674656e-05,
"loss": 3.2463,
"step": 15000
},
{
"epoch": 0.38,
"learning_rate": 7.996966599938132e-05,
"loss": 3.3657,
"step": 15500
},
{
"epoch": 0.4,
"learning_rate": 7.996761030573961e-05,
"loss": 3.404,
"step": 16000
},
{
"epoch": 0.41,
"learning_rate": 7.99654872492862e-05,
"loss": 3.3893,
"step": 16500
},
{
"epoch": 0.42,
"learning_rate": 7.996329683359935e-05,
"loss": 3.3423,
"step": 17000
},
{
"epoch": 0.43,
"learning_rate": 7.996103906237084e-05,
"loss": 3.238,
"step": 17500
},
{
"epoch": 0.45,
"learning_rate": 7.995871393940603e-05,
"loss": 3.2994,
"step": 18000
},
{
"epoch": 0.46,
"learning_rate": 7.995632146862375e-05,
"loss": 3.2948,
"step": 18500
},
{
"epoch": 0.47,
"learning_rate": 7.995386165405639e-05,
"loss": 3.2603,
"step": 19000
},
{
"epoch": 0.48,
"learning_rate": 7.995133449984973e-05,
"loss": 3.3748,
"step": 19500
},
{
"epoch": 0.5,
"learning_rate": 7.99487400102632e-05,
"loss": 3.3827,
"step": 20000
},
{
"epoch": 0.51,
"learning_rate": 7.99460781896696e-05,
"loss": 3.2538,
"step": 20500
},
{
"epoch": 0.52,
"learning_rate": 7.994334904255528e-05,
"loss": 3.2942,
"step": 21000
},
{
"epoch": 0.53,
"learning_rate": 7.994055257352003e-05,
"loss": 3.439,
"step": 21500
},
{
"epoch": 0.55,
"learning_rate": 7.993768878727711e-05,
"loss": 3.2844,
"step": 22000
},
{
"epoch": 0.56,
"learning_rate": 7.993475768865324e-05,
"loss": 3.4043,
"step": 22500
},
{
"epoch": 0.57,
"learning_rate": 7.993175928258863e-05,
"loss": 3.2896,
"step": 23000
},
{
"epoch": 0.58,
"learning_rate": 7.992869357413687e-05,
"loss": 3.3379,
"step": 23500
},
{
"epoch": 0.59,
"learning_rate": 7.992556056846502e-05,
"loss": 3.3279,
"step": 24000
},
{
"epoch": 0.61,
"learning_rate": 7.992236027085355e-05,
"loss": 3.3186,
"step": 24500
},
{
"epoch": 0.62,
"learning_rate": 7.991909268669638e-05,
"loss": 3.2858,
"step": 25000
},
{
"epoch": 0.63,
"learning_rate": 7.991575782150078e-05,
"loss": 3.366,
"step": 25500
},
{
"epoch": 0.64,
"learning_rate": 7.991235568088746e-05,
"loss": 3.2549,
"step": 26000
},
{
"epoch": 0.66,
"learning_rate": 7.990888627059051e-05,
"loss": 3.3427,
"step": 26500
},
{
"epoch": 0.67,
"learning_rate": 7.990534959645739e-05,
"loss": 3.2713,
"step": 27000
},
{
"epoch": 0.68,
"learning_rate": 7.990174566444893e-05,
"loss": 3.2461,
"step": 27500
},
{
"epoch": 0.69,
"learning_rate": 7.989807448063936e-05,
"loss": 3.2944,
"step": 28000
},
{
"epoch": 0.71,
"learning_rate": 7.989433605121617e-05,
"loss": 3.3129,
"step": 28500
},
{
"epoch": 0.72,
"learning_rate": 7.989053038248027e-05,
"loss": 3.268,
"step": 29000
},
{
"epoch": 0.73,
"learning_rate": 7.988665748084586e-05,
"loss": 3.3362,
"step": 29500
},
{
"epoch": 0.74,
"learning_rate": 7.988271735284046e-05,
"loss": 3.2817,
"step": 30000
},
{
"epoch": 0.76,
"learning_rate": 7.987871000510492e-05,
"loss": 3.359,
"step": 30500
},
{
"epoch": 0.77,
"learning_rate": 7.987463544439334e-05,
"loss": 3.2771,
"step": 31000
},
{
"epoch": 0.78,
"learning_rate": 7.987049367757314e-05,
"loss": 3.3784,
"step": 31500
},
{
"epoch": 0.79,
"learning_rate": 7.9866284711625e-05,
"loss": 3.3053,
"step": 32000
},
{
"epoch": 0.81,
"learning_rate": 7.986200855364284e-05,
"loss": 3.3062,
"step": 32500
},
{
"epoch": 0.82,
"learning_rate": 7.985766521083388e-05,
"loss": 3.2485,
"step": 33000
},
{
"epoch": 0.83,
"learning_rate": 7.985325469051851e-05,
"loss": 3.2577,
"step": 33500
},
{
"epoch": 0.84,
"learning_rate": 7.98487770001304e-05,
"loss": 3.3197,
"step": 34000
},
{
"epoch": 0.86,
"learning_rate": 7.984423214721638e-05,
"loss": 3.206,
"step": 34500
},
{
"epoch": 0.87,
"learning_rate": 7.983962013943654e-05,
"loss": 3.3556,
"step": 35000
},
{
"epoch": 0.88,
"learning_rate": 7.983494098456408e-05,
"loss": 3.3232,
"step": 35500
},
{
"epoch": 0.89,
"learning_rate": 7.983019469048544e-05,
"loss": 3.2438,
"step": 36000
},
{
"epoch": 0.9,
"learning_rate": 7.982538126520019e-05,
"loss": 3.3,
"step": 36500
},
{
"epoch": 0.92,
"learning_rate": 7.982050071682102e-05,
"loss": 3.3622,
"step": 37000
},
{
"epoch": 0.93,
"learning_rate": 7.981555305357381e-05,
"loss": 3.2414,
"step": 37500
},
{
"epoch": 0.94,
"learning_rate": 7.981053828379751e-05,
"loss": 3.3854,
"step": 38000
},
{
"epoch": 0.95,
"learning_rate": 7.980545641594418e-05,
"loss": 3.2529,
"step": 38500
},
{
"epoch": 0.97,
"learning_rate": 7.9800307458579e-05,
"loss": 3.3274,
"step": 39000
},
{
"epoch": 0.98,
"learning_rate": 7.979509142038018e-05,
"loss": 3.2614,
"step": 39500
},
{
"epoch": 0.99,
"learning_rate": 7.978980831013903e-05,
"loss": 3.3069,
"step": 40000
},
{
"epoch": 1.0,
"eval_bleu": 1.0,
"eval_brevity_penalty": 1.0,
"eval_length_ratio": 1.0,
"eval_loss": 3.12583589553833,
"eval_precisions": [
1.0,
1.0,
1.0,
1.0
],
"eval_reference_length": 4591104,
"eval_runtime": 8030.2897,
"eval_samples_per_second": 1.117,
"eval_steps_per_second": 0.558,
"eval_translation_length": 4591104,
"step": 40348
},
{
"epoch": 1.0,
"learning_rate": 7.978445813675987e-05,
"loss": 3.2673,
"step": 40500
},
{
"epoch": 1.02,
"learning_rate": 7.977904090926009e-05,
"loss": 3.3218,
"step": 41000
},
{
"epoch": 1.03,
"learning_rate": 7.977355663677008e-05,
"loss": 3.1583,
"step": 41500
},
{
"epoch": 1.04,
"learning_rate": 7.97680053285332e-05,
"loss": 3.2518,
"step": 42000
},
{
"epoch": 1.05,
"learning_rate": 7.976238699390583e-05,
"loss": 3.2414,
"step": 42500
},
{
"epoch": 1.07,
"learning_rate": 7.975670164235732e-05,
"loss": 3.2775,
"step": 43000
},
{
"epoch": 1.08,
"learning_rate": 7.975094928346996e-05,
"loss": 3.1861,
"step": 43500
},
{
"epoch": 1.09,
"learning_rate": 7.974512992693896e-05,
"loss": 3.1894,
"step": 44000
},
{
"epoch": 1.1,
"learning_rate": 7.973924358257246e-05,
"loss": 3.2074,
"step": 44500
},
{
"epoch": 1.12,
"learning_rate": 7.973329026029155e-05,
"loss": 3.2695,
"step": 45000
},
{
"epoch": 1.13,
"learning_rate": 7.972726997013014e-05,
"loss": 3.3168,
"step": 45500
},
{
"epoch": 1.14,
"learning_rate": 7.972118272223505e-05,
"loss": 3.2341,
"step": 46000
},
{
"epoch": 1.15,
"learning_rate": 7.971502852686593e-05,
"loss": 3.2581,
"step": 46500
},
{
"epoch": 1.16,
"learning_rate": 7.970880739439528e-05,
"loss": 3.2041,
"step": 47000
},
{
"epoch": 1.18,
"learning_rate": 7.970251933530842e-05,
"loss": 3.2144,
"step": 47500
},
{
"epoch": 1.19,
"learning_rate": 7.969616436020346e-05,
"loss": 3.2718,
"step": 48000
},
{
"epoch": 1.2,
"learning_rate": 7.96897424797913e-05,
"loss": 3.186,
"step": 48500
},
{
"epoch": 1.21,
"learning_rate": 7.968325370489561e-05,
"loss": 3.1968,
"step": 49000
},
{
"epoch": 1.23,
"learning_rate": 7.967669804645278e-05,
"loss": 3.2777,
"step": 49500
},
{
"epoch": 1.24,
"learning_rate": 7.967007551551196e-05,
"loss": 3.2775,
"step": 50000
},
{
"epoch": 1.25,
"learning_rate": 7.9663386123235e-05,
"loss": 3.2174,
"step": 50500
},
{
"epoch": 1.26,
"learning_rate": 7.965662988089642e-05,
"loss": 3.2144,
"step": 51000
},
{
"epoch": 1.28,
"learning_rate": 7.964980679988345e-05,
"loss": 3.3125,
"step": 51500
},
{
"epoch": 1.29,
"learning_rate": 7.964291689169593e-05,
"loss": 3.1662,
"step": 52000
},
{
"epoch": 1.3,
"learning_rate": 7.963596016794635e-05,
"loss": 3.223,
"step": 52500
},
{
"epoch": 1.31,
"learning_rate": 7.962893664035984e-05,
"loss": 3.1838,
"step": 53000
},
{
"epoch": 1.33,
"learning_rate": 7.962184632077407e-05,
"loss": 3.3471,
"step": 53500
},
{
"epoch": 1.34,
"learning_rate": 7.961468922113935e-05,
"loss": 3.1718,
"step": 54000
},
{
"epoch": 1.35,
"learning_rate": 7.960746535351847e-05,
"loss": 3.2135,
"step": 54500
},
{
"epoch": 1.36,
"learning_rate": 7.96001747300868e-05,
"loss": 3.1969,
"step": 55000
},
{
"epoch": 1.38,
"learning_rate": 7.959281736313223e-05,
"loss": 3.2402,
"step": 55500
},
{
"epoch": 1.39,
"learning_rate": 7.958539326505511e-05,
"loss": 3.2797,
"step": 56000
},
{
"epoch": 1.4,
"learning_rate": 7.957790244836829e-05,
"loss": 3.2674,
"step": 56500
},
{
"epoch": 1.41,
"learning_rate": 7.957034492569703e-05,
"loss": 3.1757,
"step": 57000
},
{
"epoch": 1.43,
"learning_rate": 7.956272070977906e-05,
"loss": 3.2459,
"step": 57500
},
{
"epoch": 1.44,
"learning_rate": 7.955502981346449e-05,
"loss": 3.2732,
"step": 58000
},
{
"epoch": 1.45,
"learning_rate": 7.954727224971583e-05,
"loss": 3.1479,
"step": 58500
},
{
"epoch": 1.46,
"learning_rate": 7.953944803160796e-05,
"loss": 3.2534,
"step": 59000
},
{
"epoch": 1.47,
"learning_rate": 7.953155717232809e-05,
"loss": 3.26,
"step": 59500
},
{
"epoch": 1.49,
"learning_rate": 7.952359968517573e-05,
"loss": 3.1955,
"step": 60000
},
{
"epoch": 1.5,
"learning_rate": 7.951557558356275e-05,
"loss": 3.1423,
"step": 60500
},
{
"epoch": 1.51,
"learning_rate": 7.95074848810132e-05,
"loss": 3.2203,
"step": 61000
},
{
"epoch": 1.52,
"learning_rate": 7.949932759116346e-05,
"loss": 3.2274,
"step": 61500
},
{
"epoch": 1.54,
"learning_rate": 7.949110372776213e-05,
"loss": 3.2456,
"step": 62000
},
{
"epoch": 1.55,
"learning_rate": 7.948281330466996e-05,
"loss": 3.2285,
"step": 62500
},
{
"epoch": 1.56,
"learning_rate": 7.947445633585995e-05,
"loss": 3.1965,
"step": 63000
},
{
"epoch": 1.57,
"learning_rate": 7.946603283541722e-05,
"loss": 3.2021,
"step": 63500
},
{
"epoch": 1.59,
"learning_rate": 7.945754281753901e-05,
"loss": 3.1108,
"step": 64000
},
{
"epoch": 1.6,
"learning_rate": 7.944898629653475e-05,
"loss": 3.1778,
"step": 64500
},
{
"epoch": 1.61,
"learning_rate": 7.944036328682583e-05,
"loss": 3.2267,
"step": 65000
},
{
"epoch": 1.62,
"learning_rate": 7.943167380294583e-05,
"loss": 3.1233,
"step": 65500
},
{
"epoch": 1.64,
"learning_rate": 7.942291785954028e-05,
"loss": 3.3035,
"step": 66000
},
{
"epoch": 1.65,
"learning_rate": 7.941409547136677e-05,
"loss": 3.1858,
"step": 66500
},
{
"epoch": 1.66,
"learning_rate": 7.940520665329483e-05,
"loss": 3.122,
"step": 67000
},
{
"epoch": 1.67,
"learning_rate": 7.939625142030604e-05,
"loss": 3.2538,
"step": 67500
},
{
"epoch": 1.69,
"learning_rate": 7.93872297874938e-05,
"loss": 3.1382,
"step": 68000
},
{
"epoch": 1.7,
"learning_rate": 7.937814177006355e-05,
"loss": 3.169,
"step": 68500
},
{
"epoch": 1.71,
"learning_rate": 7.936898738333253e-05,
"loss": 3.0855,
"step": 69000
},
{
"epoch": 1.72,
"learning_rate": 7.935976664272985e-05,
"loss": 3.2122,
"step": 69500
},
{
"epoch": 1.73,
"learning_rate": 7.935047956379646e-05,
"loss": 3.1769,
"step": 70000
},
{
"epoch": 1.75,
"learning_rate": 7.934112616218515e-05,
"loss": 3.2102,
"step": 70500
},
{
"epoch": 1.76,
"learning_rate": 7.933170645366047e-05,
"loss": 3.2092,
"step": 71000
},
{
"epoch": 1.77,
"learning_rate": 7.932222045409874e-05,
"loss": 3.195,
"step": 71500
},
{
"epoch": 1.78,
"learning_rate": 7.931266817948795e-05,
"loss": 3.1975,
"step": 72000
},
{
"epoch": 1.8,
"learning_rate": 7.930304964592788e-05,
"loss": 3.3457,
"step": 72500
},
{
"epoch": 1.81,
"learning_rate": 7.929336486962994e-05,
"loss": 3.0512,
"step": 73000
},
{
"epoch": 1.82,
"learning_rate": 7.928361386691718e-05,
"loss": 3.2071,
"step": 73500
},
{
"epoch": 1.83,
"learning_rate": 7.927379665422429e-05,
"loss": 3.2991,
"step": 74000
},
{
"epoch": 1.85,
"learning_rate": 7.926391324809754e-05,
"loss": 3.1947,
"step": 74500
},
{
"epoch": 1.86,
"learning_rate": 7.925396366519476e-05,
"loss": 3.2052,
"step": 75000
},
{
"epoch": 1.87,
"learning_rate": 7.924394792228533e-05,
"loss": 3.1196,
"step": 75500
},
{
"epoch": 1.88,
"learning_rate": 7.923386603625016e-05,
"loss": 3.1718,
"step": 76000
},
{
"epoch": 1.9,
"learning_rate": 7.922371802408158e-05,
"loss": 3.1563,
"step": 76500
},
{
"epoch": 1.91,
"learning_rate": 7.921350390288342e-05,
"loss": 3.2303,
"step": 77000
},
{
"epoch": 1.92,
"learning_rate": 7.920322368987093e-05,
"loss": 3.1139,
"step": 77500
},
{
"epoch": 1.93,
"learning_rate": 7.919287740237071e-05,
"loss": 3.207,
"step": 78000
},
{
"epoch": 1.95,
"learning_rate": 7.918246505782079e-05,
"loss": 3.1756,
"step": 78500
},
{
"epoch": 1.96,
"learning_rate": 7.917198667377047e-05,
"loss": 3.0951,
"step": 79000
},
{
"epoch": 1.97,
"learning_rate": 7.916144226788041e-05,
"loss": 3.2165,
"step": 79500
},
{
"epoch": 1.98,
"learning_rate": 7.915083185792251e-05,
"loss": 3.1236,
"step": 80000
},
{
"epoch": 2.0,
"learning_rate": 7.914015546177993e-05,
"loss": 3.1342,
"step": 80500
},
{
"epoch": 2.0,
"eval_bleu": 1.0,
"eval_brevity_penalty": 1.0,
"eval_length_ratio": 1.0,
"eval_loss": 3.0691702365875244,
"eval_precisions": [
1.0,
1.0,
1.0,
1.0
],
"eval_reference_length": 4591104,
"eval_runtime": 8264.3842,
"eval_samples_per_second": 1.085,
"eval_steps_per_second": 0.543,
"eval_translation_length": 4591104,
"step": 80696
}
],
"logging_steps": 500,
"max_steps": 1210440,
"num_train_epochs": 30,
"save_steps": 1000,
"total_flos": 1.384937121990574e+18,
"trial_name": null,
"trial_params": null
}