wav2vec2-large-xlsr-georgian
This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0480
- Wer: 0.2203
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.002
- train_batch_size: 32
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 200
- num_epochs: 5
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
7.887 | 0.03 | 100 | 3.6607 | 1.0 |
4.0646 | 0.05 | 200 | 3.1735 | 1.0 |
3.7509 | 0.08 | 300 | 3.2336 | 1.0 |
3.8832 | 0.11 | 400 | 3.1474 | 1.0 |
3.9646 | 0.14 | 500 | 3.3345 | 1.0 |
3.1657 | 0.16 | 600 | 3.0552 | 1.0 |
3.0641 | 0.19 | 700 | 3.0919 | 1.0 |
1.9443 | 0.22 | 800 | 1.5124 | 1.0 |
0.6772 | 0.25 | 900 | 0.6392 | 0.8845 |
0.5144 | 0.27 | 1000 | 0.4253 | 0.7256 |
0.4361 | 0.3 | 1100 | 0.3405 | 0.6596 |
0.4239 | 0.33 | 1200 | 0.2857 | 0.5969 |
0.3226 | 0.36 | 1300 | 0.2790 | 0.5880 |
0.3508 | 0.38 | 1400 | 0.2492 | 0.5586 |
0.3091 | 0.41 | 1500 | 0.2572 | 0.5655 |
0.3094 | 0.44 | 1600 | 0.2238 | 0.5122 |
0.305 | 0.47 | 1700 | 0.2339 | 0.5011 |
0.258 | 0.49 | 1800 | 0.2067 | 0.5008 |
0.2673 | 0.52 | 1900 | 0.2201 | 0.5077 |
0.2256 | 0.55 | 2000 | 0.1871 | 0.4633 |
0.2697 | 0.58 | 2100 | 0.1933 | 0.4696 |
0.233 | 0.6 | 2200 | 0.1835 | 0.4520 |
0.2078 | 0.63 | 2300 | 0.1761 | 0.4482 |
0.2721 | 0.66 | 2400 | 0.1742 | 0.4406 |
0.2807 | 0.69 | 2500 | 0.1717 | 0.4376 |
0.2397 | 0.71 | 2600 | 0.1788 | 0.4444 |
0.2672 | 0.74 | 2700 | 0.1534 | 0.4126 |
0.1954 | 0.77 | 2800 | 0.1593 | 0.4085 |
0.2753 | 0.79 | 2900 | 0.1575 | 0.4128 |
0.1975 | 0.82 | 3000 | 0.1435 | 0.3926 |
0.2282 | 0.85 | 3100 | 0.1702 | 0.4271 |
0.2372 | 0.88 | 3200 | 0.1417 | 0.3965 |
0.229 | 0.9 | 3300 | 0.1385 | 0.3860 |
0.2332 | 0.93 | 3400 | 0.1325 | 0.3805 |
0.2074 | 0.96 | 3500 | 0.1309 | 0.375 |
0.1937 | 0.99 | 3600 | 0.1306 | 0.3784 |
0.2211 | 1.01 | 3700 | 0.1402 | 0.3836 |
0.1928 | 1.04 | 3800 | 0.1339 | 0.3714 |
0.176 | 1.07 | 3900 | 0.1319 | 0.3743 |
0.1818 | 1.1 | 4000 | 0.1326 | 0.3760 |
0.2191 | 1.12 | 4100 | 0.1312 | 0.3768 |
0.1848 | 1.15 | 4200 | 0.1228 | 0.3599 |
0.1671 | 1.18 | 4300 | 0.1289 | 0.3731 |
0.1759 | 1.21 | 4400 | 0.1245 | 0.3632 |
0.1783 | 1.23 | 4500 | 0.1267 | 0.3545 |
0.1576 | 1.26 | 4600 | 0.1280 | 0.3581 |
0.1726 | 1.29 | 4700 | 0.1215 | 0.3541 |
0.1702 | 1.32 | 4800 | 0.1342 | 0.3690 |
0.2667 | 1.34 | 4900 | 0.1256 | 0.3635 |
0.232 | 1.37 | 5000 | 0.1228 | 0.3580 |
0.2012 | 1.4 | 5100 | 0.1183 | 0.3476 |
0.1862 | 1.42 | 5200 | 0.1210 | 0.3535 |
0.1947 | 1.45 | 5300 | 0.1241 | 0.3514 |
0.1806 | 1.48 | 5400 | 0.1151 | 0.3399 |
0.1779 | 1.51 | 5500 | 0.1127 | 0.3426 |
0.176 | 1.53 | 5600 | 0.1105 | 0.3446 |
0.1387 | 1.56 | 5700 | 0.1019 | 0.3301 |
0.1621 | 1.59 | 5800 | 0.1038 | 0.3309 |
0.1527 | 1.62 | 5900 | 0.1104 | 0.3398 |
0.1748 | 1.64 | 6000 | 0.1071 | 0.3348 |
0.165 | 1.67 | 6100 | 0.1092 | 0.3380 |
0.1384 | 1.7 | 6200 | 0.1063 | 0.3349 |
0.1859 | 1.73 | 6300 | 0.1058 | 0.3250 |
0.1919 | 1.75 | 6400 | 0.1080 | 0.3413 |
0.1425 | 1.78 | 6500 | 0.1050 | 0.3284 |
0.1571 | 1.81 | 6600 | 0.1002 | 0.3244 |
0.1591 | 1.84 | 6700 | 0.0999 | 0.3238 |
0.1615 | 1.86 | 6800 | 0.0961 | 0.3157 |
0.148 | 1.89 | 6900 | 0.0993 | 0.3197 |
0.1893 | 1.92 | 7000 | 0.0963 | 0.3219 |
0.1586 | 1.95 | 7100 | 0.0970 | 0.3227 |
0.1932 | 1.97 | 7200 | 0.0955 | 0.3200 |
0.172 | 2.0 | 7300 | 0.0930 | 0.3159 |
0.1617 | 2.03 | 7400 | 0.0972 | 0.3170 |
0.1644 | 2.06 | 7500 | 0.0923 | 0.3117 |
0.1098 | 2.08 | 7600 | 0.1001 | 0.3186 |
0.1456 | 2.11 | 7700 | 0.1028 | 0.3262 |
0.1518 | 2.14 | 7800 | 0.1018 | 0.3177 |
0.1485 | 2.16 | 7900 | 0.0986 | 0.3153 |
0.1411 | 2.19 | 8000 | 0.0938 | 0.3117 |
0.1439 | 2.22 | 8100 | 0.0937 | 0.3124 |
0.1469 | 2.25 | 8200 | 0.0919 | 0.3096 |
0.1476 | 2.27 | 8300 | 0.0939 | 0.3099 |
0.1678 | 2.3 | 8400 | 0.0926 | 0.3095 |
0.1705 | 2.33 | 8500 | 0.0973 | 0.3167 |
0.1323 | 2.36 | 8600 | 0.0910 | 0.3055 |
0.1258 | 2.38 | 8700 | 0.0872 | 0.3017 |
0.1536 | 2.41 | 8800 | 0.0963 | 0.3103 |
0.1628 | 2.44 | 8900 | 0.0871 | 0.3045 |
0.1504 | 2.47 | 9000 | 0.0898 | 0.3038 |
0.1301 | 2.49 | 9100 | 0.0856 | 0.2966 |
0.1488 | 2.52 | 9200 | 0.0846 | 0.2978 |
0.1621 | 2.55 | 9300 | 0.0855 | 0.3024 |
0.1453 | 2.58 | 9400 | 0.0830 | 0.2940 |
0.13 | 2.6 | 9500 | 0.0824 | 0.2920 |
0.1225 | 2.63 | 9600 | 0.0830 | 0.2920 |
0.1337 | 2.66 | 9700 | 0.0838 | 0.2914 |
0.1192 | 2.69 | 9800 | 0.0846 | 0.2992 |
0.1478 | 2.71 | 9900 | 0.0794 | 0.2921 |
0.1188 | 2.74 | 10000 | 0.0771 | 0.2875 |
0.2217 | 2.77 | 10100 | 0.0840 | 0.3002 |
0.1169 | 2.79 | 10200 | 0.0769 | 0.2876 |
0.1334 | 2.82 | 10300 | 0.0782 | 0.2875 |
0.1623 | 2.85 | 10400 | 0.0833 | 0.2948 |
0.2 | 2.88 | 10500 | 0.0772 | 0.2856 |
0.1288 | 2.9 | 10600 | 0.0773 | 0.2854 |
0.1201 | 2.93 | 10700 | 0.0782 | 0.2885 |
0.1467 | 2.96 | 10800 | 0.0778 | 0.2864 |
0.1452 | 2.99 | 10900 | 0.0747 | 0.2825 |
0.1182 | 3.01 | 11000 | 0.0759 | 0.2812 |
0.1078 | 3.04 | 11100 | 0.0744 | 0.2771 |
0.1426 | 3.07 | 11200 | 0.0797 | 0.2883 |
0.1322 | 3.1 | 11300 | 0.0765 | 0.2797 |
0.1655 | 3.12 | 11400 | 0.0743 | 0.2885 |
0.1243 | 3.15 | 11500 | 0.0744 | 0.2792 |
0.1724 | 3.18 | 11600 | 0.0749 | 0.2778 |
0.1136 | 3.21 | 11700 | 0.0730 | 0.2764 |
0.1428 | 3.23 | 11800 | 0.0776 | 0.2836 |
0.1189 | 3.26 | 11900 | 0.0788 | 0.2893 |
0.1065 | 3.29 | 12000 | 0.0732 | 0.2746 |
0.14 | 3.32 | 12100 | 0.0720 | 0.2756 |
2.7696 | 3.34 | 12200 | 2.8195 | 1.0 |
0.7388 | 3.37 | 12300 | 0.4293 | 0.7221 |
0.1829 | 3.4 | 12400 | 0.0925 | 0.2933 |
0.1534 | 3.42 | 12500 | 0.0794 | 0.2763 |
0.134 | 3.45 | 12600 | 0.0768 | 0.2719 |
0.1095 | 3.48 | 12700 | 0.0716 | 0.2641 |
0.1185 | 3.51 | 12800 | 0.0702 | 0.2603 |
0.2156 | 3.53 | 12900 | 0.1122 | 0.3044 |
0.1246 | 3.56 | 13000 | 0.0792 | 0.2679 |
0.1023 | 3.59 | 13100 | 0.0749 | 0.2646 |
0.1302 | 3.62 | 13200 | 0.0696 | 0.2569 |
0.1168 | 3.64 | 13300 | 0.0675 | 0.2539 |
0.0907 | 3.67 | 13400 | 0.0661 | 0.2495 |
0.1111 | 3.7 | 13500 | 0.0642 | 0.2493 |
0.1003 | 3.73 | 13600 | 0.0626 | 0.2463 |
0.12 | 3.75 | 13700 | 0.0617 | 0.2453 |
0.1268 | 3.78 | 13800 | 0.0607 | 0.2447 |
0.0903 | 3.81 | 13900 | 0.0599 | 0.2430 |
0.0809 | 3.84 | 14000 | 0.0589 | 0.2426 |
0.1066 | 3.86 | 14100 | 0.0576 | 0.2392 |
0.1015 | 3.89 | 14200 | 0.0569 | 0.2374 |
0.1035 | 3.92 | 14300 | 0.0554 | 0.2365 |
0.084 | 3.95 | 14400 | 0.0549 | 0.2338 |
0.0774 | 3.97 | 14500 | 0.0541 | 0.2333 |
0.0928 | 4.0 | 14600 | 0.0542 | 0.2342 |
0.095 | 4.03 | 14700 | 0.0536 | 0.2352 |
0.0941 | 4.05 | 14800 | 0.0533 | 0.2321 |
0.0681 | 4.08 | 14900 | 0.0521 | 0.2294 |
0.0681 | 4.11 | 15000 | 0.0512 | 0.2291 |
0.073 | 4.14 | 15100 | 0.0506 | 0.2275 |
0.0918 | 4.16 | 15200 | 0.0506 | 0.2279 |
0.0751 | 4.19 | 15300 | 0.0507 | 0.2283 |
0.0952 | 4.22 | 15400 | 0.0496 | 0.2295 |
0.0839 | 4.25 | 15500 | 0.0487 | 0.2257 |
0.1116 | 4.27 | 15600 | 0.0498 | 0.2258 |
0.0775 | 4.3 | 15700 | 0.0487 | 0.2250 |
0.0801 | 4.33 | 15800 | 0.0483 | 0.2251 |
0.111 | 4.36 | 15900 | 0.0483 | 0.2240 |
0.0989 | 4.38 | 16000 | 0.0479 | 0.2220 |
0.0885 | 4.41 | 16100 | 0.0472 | 0.2224 |
0.0755 | 4.44 | 16200 | 0.0472 | 0.2211 |
0.0769 | 4.47 | 16300 | 0.0467 | 0.2206 |
0.1033 | 4.49 | 16400 | 0.0464 | 0.2215 |
0.082 | 4.52 | 16500 | 0.0464 | 0.2208 |
0.0847 | 4.55 | 16600 | 0.0459 | 0.2201 |
0.1126 | 4.58 | 16700 | 0.0452 | 0.2204 |
1.7417 | 4.6 | 16800 | 1.3754 | 1.0985 |
0.1646 | 4.63 | 16900 | 0.0896 | 0.2682 |
0.1214 | 4.66 | 17000 | 0.0622 | 0.2345 |
0.1068 | 4.68 | 17100 | 0.0575 | 0.2300 |
0.1223 | 4.71 | 17200 | 0.0548 | 0.2269 |
0.1038 | 4.74 | 17300 | 0.0527 | 0.2261 |
0.1092 | 4.77 | 17400 | 0.0520 | 0.2252 |
0.0931 | 4.79 | 17500 | 0.0516 | 0.2253 |
0.132 | 4.82 | 17600 | 0.0509 | 0.2241 |
0.09 | 4.85 | 17700 | 0.0504 | 0.2231 |
0.0921 | 4.88 | 17800 | 0.0497 | 0.2225 |
0.0765 | 4.9 | 17900 | 0.0492 | 0.2214 |
0.0999 | 4.93 | 18000 | 0.0487 | 0.2209 |
0.0831 | 4.96 | 18100 | 0.0486 | 0.2207 |
0.1 | 4.99 | 18200 | 0.0480 | 0.2203 |
Framework versions
- Transformers 4.34.1
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
- Downloads last month
- 0
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.
Model tree for Amiran13/wav2vec2-large-xlsr-georgian
Base model
facebook/wav2vec2-large-xlsr-53