wav2vec2-asr-africa-base-finetuned-naijavoices-hausa-v0.1
This model is a fine-tuned version of asr-africa/wav2vec2-asr-africa-base on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.2111
- Wer: 0.2137
- Cer: 0.0544
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: 0.0003
- train_batch_size: 128
- eval_batch_size: 64
- seed: 42
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 400
- num_epochs: 100.0
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
---|---|---|---|---|---|
1.2573 | 1.0 | 4339 | 0.3857 | 0.4168 | 0.1103 |
0.8756 | 2.0 | 8678 | 0.3528 | 0.3770 | 0.1002 |
0.8298 | 3.0 | 13017 | 0.3253 | 0.3464 | 0.0914 |
0.7952 | 4.0 | 17356 | 0.3182 | 0.3448 | 0.0922 |
0.7721 | 5.0 | 21695 | 0.3102 | 0.3356 | 0.0894 |
0.7546 | 6.0 | 26034 | 0.3009 | 0.3211 | 0.0852 |
0.7368 | 7.0 | 30373 | 0.2978 | 0.3209 | 0.0853 |
0.725 | 8.0 | 34712 | 0.2880 | 0.3160 | 0.0831 |
0.7114 | 9.0 | 39051 | 0.2830 | 0.3031 | 0.0801 |
0.7004 | 10.0 | 43390 | 0.2813 | 0.2986 | 0.0791 |
0.6864 | 11.0 | 47729 | 0.2744 | 0.2939 | 0.0778 |
0.6771 | 12.0 | 52068 | 0.2697 | 0.2888 | 0.0771 |
0.6713 | 13.0 | 56407 | 0.2715 | 0.2939 | 0.0773 |
0.6662 | 14.0 | 60746 | 0.2672 | 0.2874 | 0.0754 |
0.6557 | 15.0 | 65085 | 0.2664 | 0.2818 | 0.0745 |
0.6471 | 16.0 | 69424 | 0.2624 | 0.2847 | 0.0756 |
0.6368 | 17.0 | 73763 | 0.2580 | 0.2792 | 0.0729 |
0.6284 | 18.0 | 78102 | 0.2537 | 0.2722 | 0.0712 |
0.6246 | 19.0 | 82441 | 0.2531 | 0.2729 | 0.0713 |
0.6192 | 20.0 | 86780 | 0.2509 | 0.2761 | 0.0714 |
0.6121 | 21.0 | 91119 | 0.2495 | 0.2649 | 0.0694 |
0.605 | 22.0 | 95458 | 0.2448 | 0.2652 | 0.0693 |
0.6003 | 23.0 | 99797 | 0.2482 | 0.2605 | 0.0687 |
0.5957 | 24.0 | 104136 | 0.2435 | 0.2610 | 0.0680 |
0.5909 | 25.0 | 108475 | 0.2443 | 0.2637 | 0.0686 |
0.587 | 26.0 | 112814 | 0.2433 | 0.2645 | 0.0692 |
0.5815 | 27.0 | 117153 | 0.2432 | 0.2606 | 0.0680 |
0.5773 | 28.0 | 121492 | 0.2390 | 0.2534 | 0.0664 |
0.5706 | 29.0 | 125831 | 0.2373 | 0.2523 | 0.0662 |
0.5647 | 30.0 | 130170 | 0.2342 | 0.2566 | 0.0661 |
0.5601 | 31.0 | 134509 | 0.2348 | 0.2527 | 0.0655 |
0.5563 | 32.0 | 138848 | 0.2343 | 0.2506 | 0.0653 |
0.5525 | 33.0 | 143187 | 0.2332 | 0.2493 | 0.0645 |
0.549 | 34.0 | 147526 | 0.2345 | 0.2558 | 0.0661 |
0.5443 | 35.0 | 151865 | 0.2307 | 0.2439 | 0.0637 |
0.5423 | 36.0 | 156204 | 0.2306 | 0.2494 | 0.0641 |
0.5365 | 37.0 | 160543 | 0.2288 | 0.2423 | 0.0632 |
0.534 | 38.0 | 164882 | 0.2317 | 0.2492 | 0.0648 |
0.529 | 39.0 | 169221 | 0.2290 | 0.2440 | 0.0639 |
0.5233 | 40.0 | 173560 | 0.2255 | 0.2424 | 0.0626 |
0.5213 | 41.0 | 177899 | 0.2268 | 0.2477 | 0.0636 |
0.5157 | 42.0 | 182238 | 0.2255 | 0.2410 | 0.0622 |
0.512 | 43.0 | 186577 | 0.2248 | 0.2400 | 0.0622 |
0.508 | 44.0 | 190916 | 0.2259 | 0.2411 | 0.0623 |
0.5065 | 45.0 | 195255 | 0.2257 | 0.2380 | 0.0616 |
0.5034 | 46.0 | 199594 | 0.2224 | 0.2382 | 0.0616 |
0.4981 | 47.0 | 203933 | 0.2211 | 0.2353 | 0.0608 |
0.4954 | 48.0 | 208272 | 0.2233 | 0.2424 | 0.0627 |
0.4922 | 49.0 | 212611 | 0.2223 | 0.2361 | 0.0611 |
0.4883 | 50.0 | 216950 | 0.2188 | 0.2369 | 0.0607 |
0.4868 | 51.0 | 221289 | 0.2191 | 0.2309 | 0.0595 |
0.4811 | 52.0 | 225628 | 0.2165 | 0.2316 | 0.0593 |
0.4782 | 53.0 | 229967 | 0.2170 | 0.2283 | 0.0590 |
0.4755 | 54.0 | 234306 | 0.2204 | 0.2293 | 0.0593 |
0.4739 | 55.0 | 238645 | 0.2178 | 0.2333 | 0.0598 |
0.4707 | 56.0 | 242984 | 0.2189 | 0.2334 | 0.0598 |
0.4664 | 57.0 | 247323 | 0.2155 | 0.2309 | 0.0591 |
0.4646 | 58.0 | 251662 | 0.2166 | 0.2305 | 0.0588 |
0.4586 | 59.0 | 256001 | 0.2162 | 0.2257 | 0.0579 |
0.4575 | 60.0 | 260340 | 0.2142 | 0.2263 | 0.0580 |
0.4536 | 61.0 | 264679 | 0.2159 | 0.2256 | 0.0575 |
0.4503 | 62.0 | 269018 | 0.2146 | 0.2263 | 0.0579 |
0.4476 | 63.0 | 273357 | 0.2148 | 0.2259 | 0.0578 |
0.4456 | 64.0 | 277696 | 0.2148 | 0.2212 | 0.0567 |
0.4437 | 65.0 | 282035 | 0.2137 | 0.2227 | 0.0566 |
0.4402 | 66.0 | 286374 | 0.2122 | 0.2235 | 0.0571 |
0.4374 | 67.0 | 290713 | 0.2136 | 0.2199 | 0.0562 |
0.4343 | 68.0 | 295052 | 0.2136 | 0.2197 | 0.0565 |
0.431 | 69.0 | 299391 | 0.2117 | 0.2191 | 0.0561 |
0.4288 | 70.0 | 303730 | 0.2116 | 0.2200 | 0.0560 |
0.4267 | 71.0 | 308069 | 0.2153 | 0.2204 | 0.0564 |
0.4237 | 72.0 | 312408 | 0.2114 | 0.2200 | 0.0562 |
0.4208 | 73.0 | 316747 | 0.2104 | 0.2193 | 0.0558 |
0.419 | 74.0 | 321086 | 0.2121 | 0.2200 | 0.0561 |
0.4154 | 75.0 | 325425 | 0.2106 | 0.2191 | 0.0559 |
0.4126 | 76.0 | 329764 | 0.2121 | 0.2141 | 0.0547 |
0.4107 | 77.0 | 334103 | 0.2097 | 0.2190 | 0.0557 |
0.4078 | 78.0 | 338442 | 0.2107 | 0.2162 | 0.0551 |
0.406 | 79.0 | 342781 | 0.2103 | 0.2134 | 0.0544 |
0.4034 | 80.0 | 347120 | 0.2124 | 0.2135 | 0.0543 |
0.4021 | 81.0 | 351459 | 0.2095 | 0.2133 | 0.0544 |
0.3979 | 82.0 | 355798 | 0.2104 | 0.2146 | 0.0544 |
0.3977 | 83.0 | 360137 | 0.2102 | 0.2131 | 0.0544 |
0.3931 | 84.0 | 364476 | 0.2097 | 0.2157 | 0.0547 |
0.3915 | 85.0 | 368815 | 0.2116 | 0.2132 | 0.0541 |
0.3907 | 86.0 | 373154 | 0.2111 | 0.2137 | 0.0544 |
Framework versions
- Transformers 4.51.3
- Pytorch 2.6.0+cu124
- Datasets 3.5.0
- Tokenizers 0.21.1
- Downloads last month
- 5
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
๐
Ask for provider support
Model tree for asr-africa/wav2vec2-asr-africa-base-finetuned-naijavoices-hausa-v0.1
Base model
asr-africa/wav2vec2-asr-africa-base