wav2vec2-xls-r-300m-mwa-maaloula / trainer_state.json
pburub's picture
Upload 12 files
e579226 verified
raw
history blame
17.7 kB
{
"best_metric": 0.24188034188034188,
"best_model_checkpoint": "/content/drive/MyDrive/model_ft_2/checkpoint-15000",
"epoch": 17.647058823529413,
"global_step": 15000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.12,
"learning_rate": 5.000000000000001e-07,
"loss": 0.3163,
"step": 100
},
{
"epoch": 0.24,
"learning_rate": 1.0000000000000002e-06,
"loss": 0.2851,
"step": 200
},
{
"epoch": 0.35,
"learning_rate": 1.5e-06,
"loss": 0.2783,
"step": 300
},
{
"epoch": 0.47,
"learning_rate": 2.0000000000000003e-06,
"loss": 0.3097,
"step": 400
},
{
"epoch": 0.59,
"learning_rate": 2.5e-06,
"loss": 0.286,
"step": 500
},
{
"epoch": 0.71,
"learning_rate": 3e-06,
"loss": 0.3008,
"step": 600
},
{
"epoch": 0.82,
"learning_rate": 3.5e-06,
"loss": 0.2617,
"step": 700
},
{
"epoch": 0.94,
"learning_rate": 4.000000000000001e-06,
"loss": 0.2949,
"step": 800
},
{
"epoch": 1.06,
"learning_rate": 4.5e-06,
"loss": 0.2744,
"step": 900
},
{
"epoch": 1.18,
"learning_rate": 5e-06,
"loss": 0.2932,
"step": 1000
},
{
"epoch": 1.29,
"learning_rate": 5.500000000000001e-06,
"loss": 0.289,
"step": 1100
},
{
"epoch": 1.41,
"learning_rate": 6e-06,
"loss": 0.2984,
"step": 1200
},
{
"epoch": 1.53,
"learning_rate": 6.5000000000000004e-06,
"loss": 0.3028,
"step": 1300
},
{
"epoch": 1.65,
"learning_rate": 7e-06,
"loss": 0.2607,
"step": 1400
},
{
"epoch": 1.76,
"learning_rate": 7.500000000000001e-06,
"loss": 0.2906,
"step": 1500
},
{
"epoch": 1.88,
"learning_rate": 8.000000000000001e-06,
"loss": 0.2755,
"step": 1600
},
{
"epoch": 2.0,
"learning_rate": 8.5e-06,
"loss": 0.2894,
"step": 1700
},
{
"epoch": 2.12,
"learning_rate": 9e-06,
"loss": 0.282,
"step": 1800
},
{
"epoch": 2.24,
"learning_rate": 9.5e-06,
"loss": 0.2715,
"step": 1900
},
{
"epoch": 2.35,
"learning_rate": 1e-05,
"loss": 0.2845,
"step": 2000
},
{
"epoch": 2.47,
"learning_rate": 1e-05,
"loss": 0.2698,
"step": 2100
},
{
"epoch": 2.59,
"learning_rate": 1e-05,
"loss": 0.2803,
"step": 2200
},
{
"epoch": 2.71,
"learning_rate": 1e-05,
"loss": 0.2812,
"step": 2300
},
{
"epoch": 2.82,
"learning_rate": 1e-05,
"loss": 0.2947,
"step": 2400
},
{
"epoch": 2.94,
"learning_rate": 1e-05,
"loss": 0.2529,
"step": 2500
},
{
"epoch": 3.06,
"learning_rate": 1e-05,
"loss": 0.298,
"step": 2600
},
{
"epoch": 3.18,
"learning_rate": 1e-05,
"loss": 0.2671,
"step": 2700
},
{
"epoch": 3.29,
"learning_rate": 1e-05,
"loss": 0.2619,
"step": 2800
},
{
"epoch": 3.41,
"learning_rate": 1e-05,
"loss": 0.2546,
"step": 2900
},
{
"epoch": 3.53,
"learning_rate": 1e-05,
"loss": 0.2776,
"step": 3000
},
{
"epoch": 3.65,
"learning_rate": 1e-05,
"loss": 0.2869,
"step": 3100
},
{
"epoch": 3.76,
"learning_rate": 1e-05,
"loss": 0.2664,
"step": 3200
},
{
"epoch": 3.88,
"learning_rate": 1e-05,
"loss": 0.2893,
"step": 3300
},
{
"epoch": 4.0,
"learning_rate": 1e-05,
"loss": 0.2691,
"step": 3400
},
{
"epoch": 4.12,
"learning_rate": 1e-05,
"loss": 0.2729,
"step": 3500
},
{
"epoch": 4.24,
"learning_rate": 1e-05,
"loss": 0.2606,
"step": 3600
},
{
"epoch": 4.35,
"learning_rate": 1e-05,
"loss": 0.2731,
"step": 3700
},
{
"epoch": 4.47,
"learning_rate": 1e-05,
"loss": 0.245,
"step": 3800
},
{
"epoch": 4.59,
"learning_rate": 1e-05,
"loss": 0.2644,
"step": 3900
},
{
"epoch": 4.71,
"learning_rate": 1e-05,
"loss": 0.2651,
"step": 4000
},
{
"epoch": 4.82,
"learning_rate": 1e-05,
"loss": 0.2437,
"step": 4100
},
{
"epoch": 4.94,
"learning_rate": 1e-05,
"loss": 0.2566,
"step": 4200
},
{
"epoch": 5.06,
"learning_rate": 1e-05,
"loss": 0.2582,
"step": 4300
},
{
"epoch": 5.18,
"learning_rate": 1e-05,
"loss": 0.2261,
"step": 4400
},
{
"epoch": 5.29,
"learning_rate": 1e-05,
"loss": 0.2685,
"step": 4500
},
{
"epoch": 5.41,
"learning_rate": 1e-05,
"loss": 0.2459,
"step": 4600
},
{
"epoch": 5.53,
"learning_rate": 1e-05,
"loss": 0.2536,
"step": 4700
},
{
"epoch": 5.65,
"learning_rate": 1e-05,
"loss": 0.2592,
"step": 4800
},
{
"epoch": 5.76,
"learning_rate": 1e-05,
"loss": 0.2491,
"step": 4900
},
{
"epoch": 5.88,
"learning_rate": 1e-05,
"loss": 0.2434,
"step": 5000
},
{
"epoch": 5.88,
"eval_cer": 0.0821832842301132,
"eval_loss": 0.575729250907898,
"eval_runtime": 8.2591,
"eval_samples_per_second": 36.324,
"eval_steps_per_second": 18.162,
"eval_wer": 0.25256410256410255,
"step": 5000
},
{
"epoch": 6.0,
"learning_rate": 1e-05,
"loss": 0.2547,
"step": 5100
},
{
"epoch": 6.12,
"learning_rate": 1e-05,
"loss": 0.2459,
"step": 5200
},
{
"epoch": 6.24,
"learning_rate": 1e-05,
"loss": 0.2566,
"step": 5300
},
{
"epoch": 6.35,
"learning_rate": 1e-05,
"loss": 0.2498,
"step": 5400
},
{
"epoch": 6.47,
"learning_rate": 1e-05,
"loss": 0.2492,
"step": 5500
},
{
"epoch": 6.59,
"learning_rate": 1e-05,
"loss": 0.2491,
"step": 5600
},
{
"epoch": 6.71,
"learning_rate": 1e-05,
"loss": 0.2519,
"step": 5700
},
{
"epoch": 6.82,
"learning_rate": 1e-05,
"loss": 0.2426,
"step": 5800
},
{
"epoch": 6.94,
"learning_rate": 1e-05,
"loss": 0.2175,
"step": 5900
},
{
"epoch": 7.06,
"learning_rate": 1e-05,
"loss": 0.2506,
"step": 6000
},
{
"epoch": 7.18,
"learning_rate": 1e-05,
"loss": 0.2342,
"step": 6100
},
{
"epoch": 7.29,
"learning_rate": 1e-05,
"loss": 0.2418,
"step": 6200
},
{
"epoch": 7.41,
"learning_rate": 1e-05,
"loss": 0.2345,
"step": 6300
},
{
"epoch": 7.53,
"learning_rate": 1e-05,
"loss": 0.2358,
"step": 6400
},
{
"epoch": 7.65,
"learning_rate": 1e-05,
"loss": 0.2367,
"step": 6500
},
{
"epoch": 7.76,
"learning_rate": 1e-05,
"loss": 0.2402,
"step": 6600
},
{
"epoch": 7.88,
"learning_rate": 1e-05,
"loss": 0.2448,
"step": 6700
},
{
"epoch": 8.0,
"learning_rate": 1e-05,
"loss": 0.2458,
"step": 6800
},
{
"epoch": 8.12,
"learning_rate": 1e-05,
"loss": 0.2408,
"step": 6900
},
{
"epoch": 8.24,
"learning_rate": 1e-05,
"loss": 0.2425,
"step": 7000
},
{
"epoch": 8.35,
"learning_rate": 1e-05,
"loss": 0.2302,
"step": 7100
},
{
"epoch": 8.47,
"learning_rate": 1e-05,
"loss": 0.2426,
"step": 7200
},
{
"epoch": 8.59,
"learning_rate": 1e-05,
"loss": 0.2101,
"step": 7300
},
{
"epoch": 8.71,
"learning_rate": 1e-05,
"loss": 0.2328,
"step": 7400
},
{
"epoch": 8.82,
"learning_rate": 1e-05,
"loss": 0.2457,
"step": 7500
},
{
"epoch": 8.94,
"learning_rate": 1e-05,
"loss": 0.2351,
"step": 7600
},
{
"epoch": 9.06,
"learning_rate": 1e-05,
"loss": 0.2483,
"step": 7700
},
{
"epoch": 9.18,
"learning_rate": 1e-05,
"loss": 0.2135,
"step": 7800
},
{
"epoch": 9.29,
"learning_rate": 1e-05,
"loss": 0.2227,
"step": 7900
},
{
"epoch": 9.41,
"learning_rate": 1e-05,
"loss": 0.2342,
"step": 8000
},
{
"epoch": 9.53,
"learning_rate": 1e-05,
"loss": 0.2534,
"step": 8100
},
{
"epoch": 9.65,
"learning_rate": 1e-05,
"loss": 0.2368,
"step": 8200
},
{
"epoch": 9.76,
"learning_rate": 1e-05,
"loss": 0.2531,
"step": 8300
},
{
"epoch": 9.88,
"learning_rate": 1e-05,
"loss": 0.2474,
"step": 8400
},
{
"epoch": 10.0,
"learning_rate": 1e-05,
"loss": 0.2049,
"step": 8500
},
{
"epoch": 10.12,
"learning_rate": 1e-05,
"loss": 0.231,
"step": 8600
},
{
"epoch": 10.24,
"learning_rate": 1e-05,
"loss": 0.2384,
"step": 8700
},
{
"epoch": 10.35,
"learning_rate": 1e-05,
"loss": 0.2363,
"step": 8800
},
{
"epoch": 10.47,
"learning_rate": 1e-05,
"loss": 0.2125,
"step": 8900
},
{
"epoch": 10.59,
"learning_rate": 1e-05,
"loss": 0.224,
"step": 9000
},
{
"epoch": 10.71,
"learning_rate": 1e-05,
"loss": 0.2301,
"step": 9100
},
{
"epoch": 10.82,
"learning_rate": 1e-05,
"loss": 0.2598,
"step": 9200
},
{
"epoch": 10.94,
"learning_rate": 1e-05,
"loss": 0.2121,
"step": 9300
},
{
"epoch": 11.06,
"learning_rate": 1e-05,
"loss": 0.2444,
"step": 9400
},
{
"epoch": 11.18,
"learning_rate": 1e-05,
"loss": 0.2409,
"step": 9500
},
{
"epoch": 11.29,
"learning_rate": 1e-05,
"loss": 0.2274,
"step": 9600
},
{
"epoch": 11.41,
"learning_rate": 1e-05,
"loss": 0.2246,
"step": 9700
},
{
"epoch": 11.53,
"learning_rate": 1e-05,
"loss": 0.2256,
"step": 9800
},
{
"epoch": 11.65,
"learning_rate": 1e-05,
"loss": 0.2244,
"step": 9900
},
{
"epoch": 11.76,
"learning_rate": 1e-05,
"loss": 0.2246,
"step": 10000
},
{
"epoch": 11.76,
"eval_cer": 0.08288106683206699,
"eval_loss": 0.6029604077339172,
"eval_runtime": 8.1986,
"eval_samples_per_second": 36.592,
"eval_steps_per_second": 18.296,
"eval_wer": 0.2551282051282051,
"step": 10000
},
{
"epoch": 11.88,
"learning_rate": 9.903e-06,
"loss": 0.2299,
"step": 10100
},
{
"epoch": 12.0,
"learning_rate": 9.803e-06,
"loss": 0.2131,
"step": 10200
},
{
"epoch": 12.12,
"learning_rate": 9.703000000000002e-06,
"loss": 0.2134,
"step": 10300
},
{
"epoch": 12.24,
"learning_rate": 9.603000000000001e-06,
"loss": 0.2388,
"step": 10400
},
{
"epoch": 12.35,
"learning_rate": 9.503e-06,
"loss": 0.2209,
"step": 10500
},
{
"epoch": 12.47,
"learning_rate": 9.403000000000001e-06,
"loss": 0.2347,
"step": 10600
},
{
"epoch": 12.59,
"learning_rate": 9.303e-06,
"loss": 0.2302,
"step": 10700
},
{
"epoch": 12.71,
"learning_rate": 9.203000000000002e-06,
"loss": 0.2174,
"step": 10800
},
{
"epoch": 12.82,
"learning_rate": 9.103e-06,
"loss": 0.2224,
"step": 10900
},
{
"epoch": 12.94,
"learning_rate": 9.003e-06,
"loss": 0.2269,
"step": 11000
},
{
"epoch": 13.06,
"learning_rate": 8.903000000000001e-06,
"loss": 0.2426,
"step": 11100
},
{
"epoch": 13.18,
"learning_rate": 8.803e-06,
"loss": 0.209,
"step": 11200
},
{
"epoch": 13.29,
"learning_rate": 8.703e-06,
"loss": 0.2103,
"step": 11300
},
{
"epoch": 13.41,
"learning_rate": 8.603e-06,
"loss": 0.228,
"step": 11400
},
{
"epoch": 13.53,
"learning_rate": 8.503e-06,
"loss": 0.2256,
"step": 11500
},
{
"epoch": 13.65,
"learning_rate": 8.404000000000001e-06,
"loss": 0.211,
"step": 11600
},
{
"epoch": 13.76,
"learning_rate": 8.304e-06,
"loss": 0.1967,
"step": 11700
},
{
"epoch": 13.88,
"learning_rate": 8.204000000000001e-06,
"loss": 0.2417,
"step": 11800
},
{
"epoch": 14.0,
"learning_rate": 8.104e-06,
"loss": 0.2268,
"step": 11900
},
{
"epoch": 14.12,
"learning_rate": 8.004e-06,
"loss": 0.227,
"step": 12000
},
{
"epoch": 14.24,
"learning_rate": 7.904000000000001e-06,
"loss": 0.2186,
"step": 12100
},
{
"epoch": 14.35,
"learning_rate": 7.804e-06,
"loss": 0.2125,
"step": 12200
},
{
"epoch": 14.47,
"learning_rate": 7.704000000000001e-06,
"loss": 0.2235,
"step": 12300
},
{
"epoch": 14.59,
"learning_rate": 7.604e-06,
"loss": 0.2089,
"step": 12400
},
{
"epoch": 14.71,
"learning_rate": 7.5040000000000005e-06,
"loss": 0.1967,
"step": 12500
},
{
"epoch": 14.82,
"learning_rate": 7.404e-06,
"loss": 0.2175,
"step": 12600
},
{
"epoch": 14.94,
"learning_rate": 7.304000000000001e-06,
"loss": 0.2157,
"step": 12700
},
{
"epoch": 15.06,
"learning_rate": 7.204000000000001e-06,
"loss": 0.1907,
"step": 12800
},
{
"epoch": 15.18,
"learning_rate": 7.104000000000001e-06,
"loss": 0.2072,
"step": 12900
},
{
"epoch": 15.29,
"learning_rate": 7.004000000000001e-06,
"loss": 0.2137,
"step": 13000
},
{
"epoch": 15.41,
"learning_rate": 6.904e-06,
"loss": 0.2137,
"step": 13100
},
{
"epoch": 15.53,
"learning_rate": 6.804e-06,
"loss": 0.2097,
"step": 13200
},
{
"epoch": 15.65,
"learning_rate": 6.7040000000000005e-06,
"loss": 0.2007,
"step": 13300
},
{
"epoch": 15.76,
"learning_rate": 6.604000000000001e-06,
"loss": 0.2183,
"step": 13400
},
{
"epoch": 15.88,
"learning_rate": 6.504e-06,
"loss": 0.1844,
"step": 13500
},
{
"epoch": 16.0,
"learning_rate": 6.404e-06,
"loss": 0.225,
"step": 13600
},
{
"epoch": 16.12,
"learning_rate": 6.304e-06,
"loss": 0.2141,
"step": 13700
},
{
"epoch": 16.24,
"learning_rate": 6.204e-06,
"loss": 0.2047,
"step": 13800
},
{
"epoch": 16.35,
"learning_rate": 6.104000000000001e-06,
"loss": 0.2019,
"step": 13900
},
{
"epoch": 16.47,
"learning_rate": 6.004000000000001e-06,
"loss": 0.1923,
"step": 14000
},
{
"epoch": 16.59,
"learning_rate": 5.9040000000000006e-06,
"loss": 0.2213,
"step": 14100
},
{
"epoch": 16.71,
"learning_rate": 5.804000000000001e-06,
"loss": 0.2035,
"step": 14200
},
{
"epoch": 16.82,
"learning_rate": 5.704000000000001e-06,
"loss": 0.2044,
"step": 14300
},
{
"epoch": 16.94,
"learning_rate": 5.604000000000001e-06,
"loss": 0.2055,
"step": 14400
},
{
"epoch": 17.06,
"learning_rate": 5.505000000000001e-06,
"loss": 0.2051,
"step": 14500
},
{
"epoch": 17.18,
"learning_rate": 5.405e-06,
"loss": 0.2114,
"step": 14600
},
{
"epoch": 17.29,
"learning_rate": 5.305e-06,
"loss": 0.2135,
"step": 14700
},
{
"epoch": 17.41,
"learning_rate": 5.205e-06,
"loss": 0.2079,
"step": 14800
},
{
"epoch": 17.53,
"learning_rate": 5.105e-06,
"loss": 0.183,
"step": 14900
},
{
"epoch": 17.65,
"learning_rate": 5.0049999999999995e-06,
"loss": 0.2098,
"step": 15000
},
{
"epoch": 17.65,
"eval_cer": 0.07985734222360055,
"eval_loss": 0.6368189454078674,
"eval_runtime": 7.9866,
"eval_samples_per_second": 37.563,
"eval_steps_per_second": 18.781,
"eval_wer": 0.24188034188034188,
"step": 15000
}
],
"max_steps": 20000,
"num_train_epochs": 24,
"total_flos": 1.2884780636980976e+19,
"trial_name": null,
"trial_params": null
}