mt5-small-mlsum / trainer_state.json
LeoCordoba's picture
commit files to HF hub
d628a5c
raw history blame
No virus
14.7 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 7.0,
"global_step": 58268,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.06,
"learning_rate": 4.957094803322579e-05,
"loss": 2.4741,
"step": 500
},
{
"epoch": 0.12,
"learning_rate": 4.914189606645157e-05,
"loss": 2.4661,
"step": 1000
},
{
"epoch": 0.18,
"learning_rate": 4.871284409967736e-05,
"loss": 2.4505,
"step": 1500
},
{
"epoch": 0.24,
"learning_rate": 4.828379213290314e-05,
"loss": 2.4463,
"step": 2000
},
{
"epoch": 0.3,
"learning_rate": 4.785474016612893e-05,
"loss": 2.4309,
"step": 2500
},
{
"epoch": 0.36,
"learning_rate": 4.742568819935471e-05,
"loss": 2.4286,
"step": 3000
},
{
"epoch": 0.42,
"learning_rate": 4.699663623258049e-05,
"loss": 2.4326,
"step": 3500
},
{
"epoch": 0.48,
"learning_rate": 4.656758426580627e-05,
"loss": 2.424,
"step": 4000
},
{
"epoch": 0.54,
"learning_rate": 4.613853229903206e-05,
"loss": 2.4179,
"step": 4500
},
{
"epoch": 0.6,
"learning_rate": 4.570948033225785e-05,
"loss": 2.4141,
"step": 5000
},
{
"epoch": 0.66,
"learning_rate": 4.528042836548363e-05,
"loss": 2.4121,
"step": 5500
},
{
"epoch": 0.72,
"learning_rate": 4.4851376398709416e-05,
"loss": 2.407,
"step": 6000
},
{
"epoch": 0.78,
"learning_rate": 4.44223244319352e-05,
"loss": 2.399,
"step": 6500
},
{
"epoch": 0.84,
"learning_rate": 4.3993272465160985e-05,
"loss": 2.3935,
"step": 7000
},
{
"epoch": 0.9,
"learning_rate": 4.3564220498386766e-05,
"loss": 2.3822,
"step": 7500
},
{
"epoch": 0.96,
"learning_rate": 4.3135168531612554e-05,
"loss": 2.3822,
"step": 8000
},
{
"epoch": 1.02,
"learning_rate": 4.2706116564838335e-05,
"loss": 2.387,
"step": 8500
},
{
"epoch": 1.08,
"learning_rate": 4.227706459806412e-05,
"loss": 2.3601,
"step": 9000
},
{
"epoch": 1.14,
"learning_rate": 4.1848012631289904e-05,
"loss": 2.3582,
"step": 9500
},
{
"epoch": 1.2,
"learning_rate": 4.1418960664515685e-05,
"loss": 2.3706,
"step": 10000
},
{
"epoch": 1.26,
"learning_rate": 4.098990869774147e-05,
"loss": 2.3669,
"step": 10500
},
{
"epoch": 1.32,
"learning_rate": 4.0560856730967254e-05,
"loss": 2.3518,
"step": 11000
},
{
"epoch": 1.38,
"learning_rate": 4.013180476419304e-05,
"loss": 2.3392,
"step": 11500
},
{
"epoch": 1.44,
"learning_rate": 3.970275279741882e-05,
"loss": 2.3502,
"step": 12000
},
{
"epoch": 1.5,
"learning_rate": 3.927370083064461e-05,
"loss": 2.3437,
"step": 12500
},
{
"epoch": 1.56,
"learning_rate": 3.884464886387039e-05,
"loss": 2.3577,
"step": 13000
},
{
"epoch": 1.62,
"learning_rate": 3.841559689709618e-05,
"loss": 2.3435,
"step": 13500
},
{
"epoch": 1.68,
"learning_rate": 3.798654493032196e-05,
"loss": 2.3456,
"step": 14000
},
{
"epoch": 1.74,
"learning_rate": 3.755749296354775e-05,
"loss": 2.3461,
"step": 14500
},
{
"epoch": 1.8,
"learning_rate": 3.712844099677354e-05,
"loss": 2.3472,
"step": 15000
},
{
"epoch": 1.86,
"learning_rate": 3.669938902999932e-05,
"loss": 2.3362,
"step": 15500
},
{
"epoch": 1.92,
"learning_rate": 3.62703370632251e-05,
"loss": 2.345,
"step": 16000
},
{
"epoch": 1.98,
"learning_rate": 3.584128509645088e-05,
"loss": 2.336,
"step": 16500
},
{
"epoch": 2.04,
"learning_rate": 3.541223312967667e-05,
"loss": 2.3365,
"step": 17000
},
{
"epoch": 2.1,
"learning_rate": 3.498318116290245e-05,
"loss": 2.3245,
"step": 17500
},
{
"epoch": 2.16,
"learning_rate": 3.455412919612824e-05,
"loss": 2.3086,
"step": 18000
},
{
"epoch": 2.22,
"learning_rate": 3.412507722935402e-05,
"loss": 2.3251,
"step": 18500
},
{
"epoch": 2.28,
"learning_rate": 3.3696025262579806e-05,
"loss": 2.32,
"step": 19000
},
{
"epoch": 2.34,
"learning_rate": 3.326697329580559e-05,
"loss": 2.3084,
"step": 19500
},
{
"epoch": 2.4,
"learning_rate": 3.2837921329031375e-05,
"loss": 2.3164,
"step": 20000
},
{
"epoch": 2.46,
"learning_rate": 3.240886936225716e-05,
"loss": 2.3054,
"step": 20500
},
{
"epoch": 2.52,
"learning_rate": 3.1979817395482944e-05,
"loss": 2.3172,
"step": 21000
},
{
"epoch": 2.58,
"learning_rate": 3.1550765428708725e-05,
"loss": 2.31,
"step": 21500
},
{
"epoch": 2.64,
"learning_rate": 3.112171346193451e-05,
"loss": 2.3141,
"step": 22000
},
{
"epoch": 2.7,
"learning_rate": 3.0692661495160294e-05,
"loss": 2.3226,
"step": 22500
},
{
"epoch": 2.76,
"learning_rate": 3.026360952838608e-05,
"loss": 2.3194,
"step": 23000
},
{
"epoch": 2.82,
"learning_rate": 2.9834557561611863e-05,
"loss": 2.3203,
"step": 23500
},
{
"epoch": 2.88,
"learning_rate": 2.9405505594837644e-05,
"loss": 2.3065,
"step": 24000
},
{
"epoch": 2.94,
"learning_rate": 2.8976453628063432e-05,
"loss": 2.3112,
"step": 24500
},
{
"epoch": 3.0,
"learning_rate": 2.854740166128922e-05,
"loss": 2.2985,
"step": 25000
},
{
"epoch": 3.06,
"learning_rate": 2.8118349694515e-05,
"loss": 2.3152,
"step": 25500
},
{
"epoch": 3.12,
"learning_rate": 2.768929772774079e-05,
"loss": 2.2912,
"step": 26000
},
{
"epoch": 3.18,
"learning_rate": 2.726024576096657e-05,
"loss": 2.2759,
"step": 26500
},
{
"epoch": 3.24,
"learning_rate": 2.6831193794192354e-05,
"loss": 2.2952,
"step": 27000
},
{
"epoch": 3.3,
"learning_rate": 2.6402141827418136e-05,
"loss": 2.2971,
"step": 27500
},
{
"epoch": 3.36,
"learning_rate": 2.5973089860643923e-05,
"loss": 2.2987,
"step": 28000
},
{
"epoch": 3.42,
"learning_rate": 2.5544037893869704e-05,
"loss": 2.2859,
"step": 28500
},
{
"epoch": 3.48,
"learning_rate": 2.5114985927095492e-05,
"loss": 2.2779,
"step": 29000
},
{
"epoch": 3.54,
"learning_rate": 2.4685933960321277e-05,
"loss": 2.2878,
"step": 29500
},
{
"epoch": 3.6,
"learning_rate": 2.4256881993547058e-05,
"loss": 2.2762,
"step": 30000
},
{
"epoch": 3.66,
"learning_rate": 2.3827830026772842e-05,
"loss": 2.2728,
"step": 30500
},
{
"epoch": 3.72,
"learning_rate": 2.3398778059998627e-05,
"loss": 2.2898,
"step": 31000
},
{
"epoch": 3.78,
"learning_rate": 2.296972609322441e-05,
"loss": 2.2836,
"step": 31500
},
{
"epoch": 3.84,
"learning_rate": 2.25406741264502e-05,
"loss": 2.2882,
"step": 32000
},
{
"epoch": 3.9,
"learning_rate": 2.211162215967598e-05,
"loss": 2.2778,
"step": 32500
},
{
"epoch": 3.96,
"learning_rate": 2.1682570192901765e-05,
"loss": 2.2793,
"step": 33000
},
{
"epoch": 4.02,
"learning_rate": 2.125351822612755e-05,
"loss": 2.2699,
"step": 33500
},
{
"epoch": 4.08,
"learning_rate": 2.0824466259353334e-05,
"loss": 2.2778,
"step": 34000
},
{
"epoch": 4.14,
"learning_rate": 2.0395414292579118e-05,
"loss": 2.2668,
"step": 34500
},
{
"epoch": 4.2,
"learning_rate": 1.9966362325804903e-05,
"loss": 2.257,
"step": 35000
},
{
"epoch": 4.26,
"learning_rate": 1.9537310359030687e-05,
"loss": 2.2496,
"step": 35500
},
{
"epoch": 4.32,
"learning_rate": 1.9108258392256472e-05,
"loss": 2.2742,
"step": 36000
},
{
"epoch": 4.38,
"learning_rate": 1.8679206425482253e-05,
"loss": 2.2643,
"step": 36500
},
{
"epoch": 4.44,
"learning_rate": 1.825015445870804e-05,
"loss": 2.2517,
"step": 37000
},
{
"epoch": 4.51,
"learning_rate": 1.7821102491933825e-05,
"loss": 2.2677,
"step": 37500
},
{
"epoch": 4.57,
"learning_rate": 1.739205052515961e-05,
"loss": 2.2654,
"step": 38000
},
{
"epoch": 4.63,
"learning_rate": 1.6962998558385394e-05,
"loss": 2.2645,
"step": 38500
},
{
"epoch": 4.69,
"learning_rate": 1.6533946591611175e-05,
"loss": 2.2456,
"step": 39000
},
{
"epoch": 4.75,
"learning_rate": 1.610489462483696e-05,
"loss": 2.2686,
"step": 39500
},
{
"epoch": 4.81,
"learning_rate": 1.5675842658062744e-05,
"loss": 2.2478,
"step": 40000
},
{
"epoch": 4.87,
"learning_rate": 1.5246790691288529e-05,
"loss": 2.2558,
"step": 40500
},
{
"epoch": 4.93,
"learning_rate": 1.4817738724514313e-05,
"loss": 2.2541,
"step": 41000
},
{
"epoch": 4.99,
"learning_rate": 1.4388686757740098e-05,
"loss": 2.236,
"step": 41500
},
{
"epoch": 5.05,
"learning_rate": 1.3959634790965884e-05,
"loss": 2.2543,
"step": 42000
},
{
"epoch": 5.11,
"learning_rate": 1.3530582824191668e-05,
"loss": 2.2479,
"step": 42500
},
{
"epoch": 5.17,
"learning_rate": 1.3101530857417451e-05,
"loss": 2.2525,
"step": 43000
},
{
"epoch": 5.23,
"learning_rate": 1.2672478890643236e-05,
"loss": 2.2506,
"step": 43500
},
{
"epoch": 5.29,
"learning_rate": 1.224342692386902e-05,
"loss": 2.2282,
"step": 44000
},
{
"epoch": 5.35,
"learning_rate": 1.1814374957094803e-05,
"loss": 2.2406,
"step": 44500
},
{
"epoch": 5.41,
"learning_rate": 1.1385322990320587e-05,
"loss": 2.2357,
"step": 45000
},
{
"epoch": 5.47,
"learning_rate": 1.0956271023546374e-05,
"loss": 2.2457,
"step": 45500
},
{
"epoch": 5.53,
"learning_rate": 1.0527219056772156e-05,
"loss": 2.2262,
"step": 46000
},
{
"epoch": 5.59,
"learning_rate": 1.0098167089997941e-05,
"loss": 2.2423,
"step": 46500
},
{
"epoch": 5.65,
"learning_rate": 9.669115123223725e-06,
"loss": 2.2435,
"step": 47000
},
{
"epoch": 5.71,
"learning_rate": 9.24006315644951e-06,
"loss": 2.2277,
"step": 47500
},
{
"epoch": 5.77,
"learning_rate": 8.811011189675294e-06,
"loss": 2.2445,
"step": 48000
},
{
"epoch": 5.83,
"learning_rate": 8.381959222901079e-06,
"loss": 2.2395,
"step": 48500
},
{
"epoch": 5.89,
"learning_rate": 7.952907256126863e-06,
"loss": 2.2506,
"step": 49000
},
{
"epoch": 5.95,
"learning_rate": 7.523855289352647e-06,
"loss": 2.234,
"step": 49500
},
{
"epoch": 6.01,
"learning_rate": 7.0948033225784306e-06,
"loss": 2.2488,
"step": 50000
},
{
"epoch": 6.07,
"learning_rate": 6.665751355804216e-06,
"loss": 2.2242,
"step": 50500
},
{
"epoch": 6.13,
"learning_rate": 6.2366993890299995e-06,
"loss": 2.2297,
"step": 51000
},
{
"epoch": 6.19,
"learning_rate": 5.807647422255784e-06,
"loss": 2.2259,
"step": 51500
},
{
"epoch": 6.25,
"learning_rate": 5.378595455481568e-06,
"loss": 2.2263,
"step": 52000
},
{
"epoch": 6.31,
"learning_rate": 4.949543488707353e-06,
"loss": 2.2214,
"step": 52500
},
{
"epoch": 6.37,
"learning_rate": 4.520491521933137e-06,
"loss": 2.2287,
"step": 53000
},
{
"epoch": 6.43,
"learning_rate": 4.091439555158921e-06,
"loss": 2.2384,
"step": 53500
},
{
"epoch": 6.49,
"learning_rate": 3.662387588384705e-06,
"loss": 2.2319,
"step": 54000
},
{
"epoch": 6.55,
"learning_rate": 3.2333356216104892e-06,
"loss": 2.2328,
"step": 54500
},
{
"epoch": 6.61,
"learning_rate": 2.804283654836274e-06,
"loss": 2.2359,
"step": 55000
},
{
"epoch": 6.67,
"learning_rate": 2.375231688062058e-06,
"loss": 2.2372,
"step": 55500
},
{
"epoch": 6.73,
"learning_rate": 1.9461797212878423e-06,
"loss": 2.2402,
"step": 56000
},
{
"epoch": 6.79,
"learning_rate": 1.5171277545136267e-06,
"loss": 2.2243,
"step": 56500
},
{
"epoch": 6.85,
"learning_rate": 1.088075787739411e-06,
"loss": 2.2267,
"step": 57000
},
{
"epoch": 6.91,
"learning_rate": 6.590238209651953e-07,
"loss": 2.2254,
"step": 57500
},
{
"epoch": 6.97,
"learning_rate": 2.2997185419097963e-07,
"loss": 2.2278,
"step": 58000
},
{
"epoch": 7.0,
"step": 58268,
"total_flos": 3.6314933518540524e+18,
"train_runtime": 64951.4124,
"train_samples_per_second": 0.897
}
],
"max_steps": 58268,
"num_train_epochs": 7,
"total_flos": 3.6314933518540524e+18,
"trial_name": null,
"trial_params": null
}