Hugging Face's logo

Vkt
/
model-960hfacebook-2022.06.08

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
wav2vec2
Generated from Trainer
Model card Files Metrics Community
Vkt commited on
Commit
a2733fe
·
1 Parent(s): 53e48f9

update model card README.md

Browse files
Files changed (1) hide show
  1. README.md +119 -107
README.md CHANGED
@@ -16,8 +16,8 @@ should probably proofread and complete it, then remove this comment. -->
16
 
17
  This model is a fine-tuned version of [facebook/wav2vec2-large-960h-lv60-self](https://huggingface.co/facebook/wav2vec2-large-960h-lv60-self) on the common_voice dataset.
18
  It achieves the following results on the evaluation set:
19
- - Loss: 0.0245
20
- - Wer: 0.0219
21
 
22
  ## Model description
23
 
@@ -45,117 +45,129 @@ The following hyperparameters were used during training:
45
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
46
  - lr_scheduler_type: linear
47
  - lr_scheduler_warmup_steps: 500
48
- - num_epochs: 30
49
  - mixed_precision_training: Native AMP
50
 
51
  ### Training results
52
 
53
  | Training Loss | Epoch | Step | Validation Loss | Wer |
54
  |:-------------:|:-----:|:-----:|:---------------:|:------:|
55
- | 5.4352 | 0.29 | 400 | 1.0396 | 0.8310 |
56
- | 1.0181 | 0.57 | 800 | 0.6753 | 0.6456 |
57
- | 0.819 | 0.86 | 1200 | 0.5695 | 0.5623 |
58
- | 0.7059 | 1.14 | 1600 | 0.4785 | 0.5009 |
59
- | 0.6281 | 1.43 | 2000 | 0.4822 | 0.4984 |
60
- | 0.6174 | 1.72 | 2400 | 0.4214 | 0.4498 |
61
- | 0.5798 | 2.0 | 2800 | 0.4169 | 0.4263 |
62
- | 0.515 | 2.29 | 3200 | 0.3947 | 0.3937 |
63
- | 0.4942 | 2.57 | 3600 | 0.3713 | 0.3933 |
64
- | 0.4905 | 2.86 | 4000 | 0.3567 | 0.3597 |
65
- | 0.4474 | 3.15 | 4400 | 0.3537 | 0.3566 |
66
- | 0.4258 | 3.43 | 4800 | 0.3351 | 0.3432 |
67
- | 0.4282 | 3.72 | 5200 | 0.3377 | 0.3386 |
68
- | 0.4221 | 4.0 | 5600 | 0.3378 | 0.3362 |
69
- | 0.3752 | 4.29 | 6000 | 0.3252 | 0.3244 |
70
- | 0.3756 | 4.57 | 6400 | 0.3195 | 0.3005 |
71
- | 0.3689 | 4.86 | 6800 | 0.3060 | 0.3086 |
72
- | 0.3474 | 5.15 | 7200 | 0.3236 | 0.2972 |
73
- | 0.3361 | 5.43 | 7600 | 0.2967 | 0.2915 |
74
- | 0.322 | 5.72 | 8000 | 0.2963 | 0.2877 |
75
- | 0.3292 | 6.0 | 8400 | 0.2880 | 0.2755 |
76
- | 0.2912 | 6.29 | 8800 | 0.2871 | 0.2730 |
77
- | 0.2956 | 6.58 | 9200 | 0.2936 | 0.2775 |
78
- | 0.2907 | 6.86 | 9600 | 0.2924 | 0.2687 |
79
- | 0.2748 | 7.15 | 10000 | 0.3002 | 0.2603 |
80
- | 0.2693 | 7.43 | 10400 | 0.2898 | 0.2550 |
81
- | 0.2708 | 7.72 | 10800 | 0.2810 | 0.2550 |
82
- | 0.2656 | 8.01 | 11200 | 0.2767 | 0.2479 |
83
- | 0.2319 | 8.29 | 11600 | 0.2898 | 0.2494 |
84
- | 0.241 | 8.58 | 12000 | 0.2819 | 0.2474 |
85
- | 0.2485 | 8.86 | 12400 | 0.2737 | 0.2503 |
86
- | 0.2299 | 9.15 | 12800 | 0.2813 | 0.2400 |
87
- | 0.2216 | 9.44 | 13200 | 0.2932 | 0.2397 |
88
- | 0.2204 | 9.72 | 13600 | 0.2707 | 0.2369 |
89
- | 0.2225 | 10.01 | 14000 | 0.2791 | 0.2342 |
90
- | 0.1956 | 10.29 | 14400 | 0.2846 | 0.2350 |
91
- | 0.2031 | 10.58 | 14800 | 0.2758 | 0.2423 |
92
- | 0.2111 | 10.86 | 15200 | 0.2729 | 0.2272 |
93
- | 0.193 | 11.15 | 15600 | 0.2706 | 0.2281 |
94
- | 0.1858 | 11.44 | 16000 | 0.2680 | 0.2206 |
95
- | 0.1876 | 11.72 | 16400 | 0.2812 | 0.2235 |
96
- | 0.1873 | 12.01 | 16800 | 0.2663 | 0.2217 |
97
- | 0.1728 | 12.29 | 17200 | 0.2747 | 0.2200 |
98
- | 0.1747 | 12.58 | 17600 | 0.2748 | 0.2194 |
99
- | 0.1753 | 12.87 | 18000 | 0.2753 | 0.2184 |
100
- | 0.1667 | 13.15 | 18400 | 0.2815 | 0.2175 |
101
- | 0.1601 | 13.44 | 18800 | 0.2765 | 0.2159 |
102
- | 0.1569 | 13.72 | 19200 | 0.2659 | 0.2108 |
103
- | 0.164 | 14.01 | 19600 | 0.2716 | 0.2101 |
104
- | 0.1474 | 14.3 | 20000 | 0.2808 | 0.2088 |
105
- | 0.1549 | 14.58 | 20400 | 0.2761 | 0.2111 |
106
- | 0.1551 | 14.87 | 20800 | 0.2631 | 0.2058 |
107
- | 0.1417 | 15.15 | 21200 | 0.2792 | 0.2032 |
108
- | 0.1444 | 15.44 | 21600 | 0.2606 | 0.2050 |
109
- | 0.1422 | 15.73 | 22000 | 0.2589 | 0.2042 |
110
- | 0.139 | 16.01 | 22400 | 0.2803 | 0.2019 |
111
- | 0.1314 | 16.3 | 22800 | 0.2786 | 0.2042 |
112
- | 0.1352 | 16.58 | 23200 | 0.2717 | 0.2014 |
113
- | 0.134 | 16.87 | 23600 | 0.2681 | 0.2027 |
114
- | 0.129 | 17.16 | 24000 | 0.2782 | 0.1964 |
115
- | 0.1209 | 17.44 | 24400 | 0.2769 | 0.1984 |
116
- | 0.1237 | 17.73 | 24800 | 0.2758 | 0.1958 |
117
- | 0.1235 | 18.01 | 25200 | 0.2811 | 0.1922 |
118
- | 0.1135 | 18.3 | 25600 | 0.2781 | 0.1941 |
119
- | 0.1146 | 18.58 | 26000 | 0.2904 | 0.1940 |
120
- | 0.1162 | 18.87 | 26400 | 0.2760 | 0.1908 |
121
- | 0.1177 | 19.16 | 26800 | 0.2781 | 0.1853 |
122
- | 0.1094 | 19.44 | 27200 | 0.2928 | 0.1915 |
123
- | 0.1104 | 19.73 | 27600 | 0.2937 | 0.1872 |
124
- | 0.1082 | 20.01 | 28000 | 0.2818 | 0.1848 |
125
- | 0.0994 | 20.3 | 28400 | 0.2932 | 0.1869 |
126
- | 0.1015 | 20.59 | 28800 | 0.2845 | 0.1877 |
127
- | 0.101 | 20.87 | 29200 | 0.2926 | 0.1874 |
128
- | 0.0946 | 21.16 | 29600 | 0.2994 | 0.1865 |
129
- | 0.0974 | 21.44 | 30000 | 0.2876 | 0.1829 |
130
- | 0.0964 | 21.73 | 30400 | 0.2941 | 0.1793 |
131
- | 0.0919 | 22.02 | 30800 | 0.2814 | 0.1812 |
132
- | 0.0899 | 22.3 | 31200 | 0.2723 | 0.1807 |
133
- | 0.087 | 22.59 | 31600 | 0.2917 | 0.1809 |
134
- | 0.0877 | 22.87 | 32000 | 0.2858 | 0.1787 |
135
- | 0.0901 | 23.16 | 32400 | 0.2820 | 0.1801 |
136
- | 0.0855 | 23.45 | 32800 | 0.2915 | 0.1796 |
137
- | 0.0852 | 23.73 | 33200 | 0.2827 | 0.1794 |
138
- | 0.0881 | 24.02 | 33600 | 0.2688 | 0.1772 |
139
- | 0.0774 | 24.3 | 34000 | 0.2938 | 0.1769 |
140
- | 0.0812 | 24.59 | 34400 | 0.2919 | 0.1793 |
141
- | 0.087 | 24.87 | 34800 | 0.2915 | 0.1756 |
142
- | 0.08 | 25.16 | 35200 | 0.2861 | 0.1769 |
143
- | 0.0781 | 25.45 | 35600 | 0.2931 | 0.1779 |
144
- | 0.0745 | 25.73 | 36000 | 0.2822 | 0.1780 |
145
- | 0.0779 | 26.02 | 36400 | 0.2849 | 0.1741 |
146
- | 0.0741 | 26.3 | 36800 | 0.3052 | 0.1751 |
147
- | 0.0709 | 26.59 | 37200 | 0.3038 | 0.1757 |
148
- | 0.072 | 26.88 | 37600 | 0.3081 | 0.1749 |
149
- | 0.0751 | 27.16 | 38000 | 0.2988 | 0.1743 |
150
- | 0.0671 | 27.45 | 38400 | 0.3021 | 0.1736 |
151
- | 0.0713 | 27.73 | 38800 | 0.3075 | 0.1723 |
152
- | 0.0681 | 28.02 | 39200 | 0.2972 | 0.1698 |
153
- | 0.0689 | 28.31 | 39600 | 0.2920 | 0.1707 |
154
- | 0.095 | 28.59 | 40000 | 0.0253 | 0.0223 |
155
- | 0.0948 | 28.88 | 40400 | 0.0246 | 0.0223 |
156
- | 0.0903 | 29.16 | 40800 | 0.0246 | 0.0221 |
157
- | 0.0892 | 29.45 | 41200 | 0.0246 | 0.0216 |
158
- | 0.0908 | 29.74 | 41600 | 0.0245 | 0.0219 |
 
 
 
 
 
 
 
 
 
 
 
 
159
 
160
 
161
  ### Framework versions
 
16
 
17
  This model is a fine-tuned version of [facebook/wav2vec2-large-960h-lv60-self](https://huggingface.co/facebook/wav2vec2-large-960h-lv60-self) on the common_voice dataset.
18
  It achieves the following results on the evaluation set:
19
+ - Loss: 0.2907
20
+ - Wer: 0.1804
21
 
22
  ## Model description
23
 
 
45
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
46
  - lr_scheduler_type: linear
47
  - lr_scheduler_warmup_steps: 500
48
+ - num_epochs: 25
49
  - mixed_precision_training: Native AMP
50
 
51
  ### Training results
52
 
53
  | Training Loss | Epoch | Step | Validation Loss | Wer |
54
  |:-------------:|:-----:|:-----:|:---------------:|:------:|
55
+ | 6.7634 | 0.21 | 300 | 2.9743 | 0.9998 |
56
+ | 1.6536 | 0.43 | 600 | 0.8605 | 0.7529 |
57
+ | 0.9823 | 0.64 | 900 | 0.6600 | 0.6286 |
58
+ | 0.8708 | 0.86 | 1200 | 0.5780 | 0.5736 |
59
+ | 0.7878 | 1.07 | 1500 | 0.5386 | 0.5326 |
60
+ | 0.7033 | 1.29 | 1800 | 0.4986 | 0.4992 |
61
+ | 0.681 | 1.5 | 2100 | 0.4575 | 0.4778 |
62
+ | 0.6537 | 1.72 | 2400 | 0.4591 | 0.4482 |
63
+ | 0.6263 | 1.93 | 2700 | 0.4317 | 0.4353 |
64
+ | 0.5811 | 2.14 | 3000 | 0.4149 | 0.4159 |
65
+ | 0.5565 | 2.36 | 3300 | 0.4170 | 0.3956 |
66
+ | 0.5501 | 2.57 | 3600 | 0.4007 | 0.3929 |
67
+ | 0.5444 | 2.79 | 3900 | 0.3930 | 0.3851 |
68
+ | 0.5177 | 3.0 | 4200 | 0.4006 | 0.3630 |
69
+ | 0.4682 | 3.22 | 4500 | 0.3707 | 0.3713 |
70
+ | 0.4805 | 3.43 | 4800 | 0.3564 | 0.3583 |
71
+ | 0.4715 | 3.65 | 5100 | 0.3596 | 0.3434 |
72
+ | 0.4482 | 3.86 | 5400 | 0.3555 | 0.3394 |
73
+ | 0.4407 | 4.07 | 5700 | 0.3680 | 0.3312 |
74
+ | 0.4134 | 4.29 | 6000 | 0.3534 | 0.3328 |
75
+ | 0.4165 | 4.5 | 6300 | 0.3294 | 0.3259 |
76
+ | 0.4196 | 4.72 | 6600 | 0.3353 | 0.3214 |
77
+ | 0.4117 | 4.93 | 6900 | 0.3266 | 0.3211 |
78
+ | 0.3847 | 5.15 | 7200 | 0.3365 | 0.3156 |
79
+ | 0.3687 | 5.36 | 7500 | 0.3233 | 0.3014 |
80
+ | 0.376 | 5.58 | 7800 | 0.3345 | 0.2979 |
81
+ | 0.3732 | 5.79 | 8100 | 0.3105 | 0.2882 |
82
+ | 0.3705 | 6.0 | 8400 | 0.3252 | 0.2935 |
83
+ | 0.3311 | 6.22 | 8700 | 0.3266 | 0.2911 |
84
+ | 0.3386 | 6.43 | 9000 | 0.2975 | 0.2765 |
85
+ | 0.337 | 6.65 | 9300 | 0.3070 | 0.2826 |
86
+ | 0.3458 | 6.86 | 9600 | 0.3090 | 0.2766 |
87
+ | 0.3218 | 7.08 | 9900 | 0.3117 | 0.2748 |
88
+ | 0.3041 | 7.29 | 10200 | 0.2989 | 0.2651 |
89
+ | 0.3031 | 7.51 | 10500 | 0.3210 | 0.2672 |
90
+ | 0.3037 | 7.72 | 10800 | 0.3040 | 0.2667 |
91
+ | 0.3126 | 7.93 | 11100 | 0.2867 | 0.2613 |
92
+ | 0.3005 | 8.15 | 11400 | 0.3075 | 0.2610 |
93
+ | 0.2802 | 8.36 | 11700 | 0.3129 | 0.2608 |
94
+ | 0.2785 | 8.58 | 12000 | 0.3002 | 0.2579 |
95
+ | 0.2788 | 8.79 | 12300 | 0.3063 | 0.2476 |
96
+ | 0.286 | 9.01 | 12600 | 0.2971 | 0.2495 |
97
+ | 0.2534 | 9.22 | 12900 | 0.2766 | 0.2452 |
98
+ | 0.2542 | 9.44 | 13200 | 0.2893 | 0.2405 |
99
+ | 0.2576 | 9.65 | 13500 | 0.3038 | 0.2518 |
100
+ | 0.2552 | 9.86 | 13800 | 0.2851 | 0.2429 |
101
+ | 0.2487 | 10.08 | 14100 | 0.2858 | 0.2356 |
102
+ | 0.2441 | 10.29 | 14400 | 0.2999 | 0.2364 |
103
+ | 0.2345 | 10.51 | 14700 | 0.2907 | 0.2373 |
104
+ | 0.2352 | 10.72 | 15000 | 0.2885 | 0.2402 |
105
+ | 0.2464 | 10.94 | 15300 | 0.2896 | 0.2339 |
106
+ | 0.2219 | 11.15 | 15600 | 0.2999 | 0.2351 |
107
+ | 0.2257 | 11.37 | 15900 | 0.2930 | 0.2326 |
108
+ | 0.2184 | 11.58 | 16200 | 0.2980 | 0.2353 |
109
+ | 0.2182 | 11.79 | 16500 | 0.2832 | 0.2296 |
110
+ | 0.2224 | 12.01 | 16800 | 0.2797 | 0.2285 |
111
+ | 0.1991 | 12.22 | 17100 | 0.2810 | 0.2296 |
112
+ | 0.1993 | 12.44 | 17400 | 0.2949 | 0.2253 |
113
+ | 0.2042 | 12.65 | 17700 | 0.2864 | 0.2207 |
114
+ | 0.2083 | 12.87 | 18000 | 0.2860 | 0.2278 |
115
+ | 0.1998 | 13.08 | 18300 | 0.2872 | 0.2232 |
116
+ | 0.1919 | 13.3 | 18600 | 0.2894 | 0.2247 |
117
+ | 0.1925 | 13.51 | 18900 | 0.3007 | 0.2234 |
118
+ | 0.1966 | 13.72 | 19200 | 0.2831 | 0.2176 |
119
+ | 0.1942 | 13.94 | 19500 | 0.2811 | 0.2161 |
120
+ | 0.1778 | 14.15 | 19800 | 0.2901 | 0.2196 |
121
+ | 0.1755 | 14.37 | 20100 | 0.2864 | 0.2188 |
122
+ | 0.1795 | 14.58 | 20400 | 0.2927 | 0.2170 |
123
+ | 0.1817 | 14.8 | 20700 | 0.2846 | 0.2156 |
124
+ | 0.1754 | 15.01 | 21000 | 0.3036 | 0.2137 |
125
+ | 0.1674 | 15.23 | 21300 | 0.2876 | 0.2156 |
126
+ | 0.171 | 15.44 | 21600 | 0.2812 | 0.2106 |
127
+ | 0.1603 | 15.65 | 21900 | 0.2692 | 0.2093 |
128
+ | 0.1663 | 15.87 | 22200 | 0.2745 | 0.2094 |
129
+ | 0.1608 | 16.08 | 22500 | 0.2807 | 0.2043 |
130
+ | 0.1555 | 16.3 | 22800 | 0.2872 | 0.2036 |
131
+ | 0.1546 | 16.51 | 23100 | 0.2837 | 0.2049 |
132
+ | 0.1515 | 16.73 | 23400 | 0.2746 | 0.2031 |
133
+ | 0.1571 | 16.94 | 23700 | 0.2767 | 0.2047 |
134
+ | 0.1498 | 17.16 | 24000 | 0.2837 | 0.2050 |
135
+ | 0.143 | 17.37 | 24300 | 0.2745 | 0.2038 |
136
+ | 0.1471 | 17.58 | 24600 | 0.2787 | 0.2004 |
137
+ | 0.1442 | 17.8 | 24900 | 0.2779 | 0.2005 |
138
+ | 0.1481 | 18.01 | 25200 | 0.2906 | 0.2021 |
139
+ | 0.1318 | 18.23 | 25500 | 0.2936 | 0.1991 |
140
+ | 0.1396 | 18.44 | 25800 | 0.2913 | 0.1984 |
141
+ | 0.144 | 18.66 | 26100 | 0.2806 | 0.1953 |
142
+ | 0.1341 | 18.87 | 26400 | 0.2896 | 0.1972 |
143
+ | 0.1375 | 19.09 | 26700 | 0.2937 | 0.2002 |
144
+ | 0.1286 | 19.3 | 27000 | 0.2929 | 0.1954 |
145
+ | 0.1242 | 19.51 | 27300 | 0.2968 | 0.1962 |
146
+ | 0.1305 | 19.73 | 27600 | 0.2879 | 0.1944 |
147
+ | 0.1287 | 19.94 | 27900 | 0.2850 | 0.1937 |
148
+ | 0.1286 | 20.16 | 28200 | 0.2910 | 0.1961 |
149
+ | 0.121 | 20.37 | 28500 | 0.2908 | 0.1912 |
150
+ | 0.1264 | 20.59 | 28800 | 0.2853 | 0.1904 |
151
+ | 0.1238 | 20.8 | 29100 | 0.2913 | 0.1926 |
152
+ | 0.117 | 21.02 | 29400 | 0.2907 | 0.1922 |
153
+ | 0.1154 | 21.23 | 29700 | 0.2902 | 0.1888 |
154
+ | 0.1142 | 21.44 | 30000 | 0.2854 | 0.1907 |
155
+ | 0.1168 | 21.66 | 30300 | 0.2918 | 0.1873 |
156
+ | 0.1168 | 21.87 | 30600 | 0.2897 | 0.1873 |
157
+ | 0.1105 | 22.09 | 30900 | 0.2951 | 0.1856 |
158
+ | 0.1134 | 22.3 | 31200 | 0.2842 | 0.1847 |
159
+ | 0.1111 | 22.52 | 31500 | 0.2884 | 0.1829 |
160
+ | 0.1088 | 22.73 | 31800 | 0.2991 | 0.1840 |
161
+ | 0.1139 | 22.94 | 32100 | 0.2876 | 0.1839 |
162
+ | 0.1078 | 23.16 | 32400 | 0.2899 | 0.1830 |
163
+ | 0.1087 | 23.37 | 32700 | 0.2927 | 0.1803 |
164
+ | 0.1076 | 23.59 | 33000 | 0.2924 | 0.1801 |
165
+ | 0.11 | 23.8 | 33300 | 0.2877 | 0.1804 |
166
+ | 0.1067 | 24.02 | 33600 | 0.2918 | 0.1799 |
167
+ | 0.1104 | 24.23 | 33900 | 0.2908 | 0.1809 |
168
+ | 0.1023 | 24.45 | 34200 | 0.2939 | 0.1807 |
169
+ | 0.0993 | 24.66 | 34500 | 0.2925 | 0.1802 |
170
+ | 0.1053 | 24.87 | 34800 | 0.2907 | 0.1804 |
171
 
172
 
173
  ### Framework versions