Model reference · open weights
wav2vec2-large-xlsr-nepali is an open-weight audio or speech model from prajin. AxForge deploys and operates it for you on dedicated EU-owned hardware — with the licence handled where one is required.
Available as managed deployment — configured and operated for you on dedicated EU hardware, quoted per deployment.
What it is
| Released by | prajin |
|---|---|
| Type | Audio & music |
| Task | Speech→text |
| Runs with | transformers |
| Released | 2022-03-02 |
| Popularity | 720 downloads / month |
| Licence | Open weights |
About
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m
The dataset used to train are :
For evaluation on publicly available datasets, can use OpenSLR-43 corpus - Model is not trained on this data which is also available on HuggingFace as train-set where it achieves 27% WER and 8.3% CER with 5 gram language model.
Script to Evaluate the Model on OpenSLR-43 train set :
python3 eval.py --model_id prajin/wav2vec2-large-xlsr-300m-nepali --dataset openslr --config SLR43 --split train --log_outputs
Below evaluation result is the evaluation on separated 10000 samples from total training dataset.
It achieves the following results on the evaluation set Without using Language Model :
With Language model ( 5 gram )
More information needed
More information needed
More information needed
The following hyperparameters were used during training:
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.3468 | 0.04 | 400 | 0.2624 | 0.3479 |
| 0.2792 | 0.08 | 800 | 0.2696 | 0.3490 |
| 0.245 | 0.12 | 1200 | 0.2751 | 0.3502 |
| 0.2453 | 0.16 | 1600 | 0.2754 | 0.3523 |
| 0.2332 | 0.2 | 2000 | 0.2779 | 0.3517 |
| 0.2321 | 0.24 | 2400 | 0.2775 | 0.3528 |
| 0.2708 | 0.28 | 2800 | 0.2764 | 0.3533 |
| 0.2709 | 0.32 | 3200 | 0.2723 | 0.3544 |
| 0.2715 | 0.36 | 3600 | 0.2739 | 0.3545 |
| 0.2732 | 0.4 | 4000 | 0.2707 | 0.3498 |
| 0.2643 | 0.44 | 4400 | 0.2696 | 0.3499 |
| 0.2682 | 0.47 | 4800 | 0.2672 | 0.3492 |
| 0.2687 | 0.51 | 5200 | 0.2644 | 0.3474 |
| 0.269 | 0.55 | 5600 | 0.2619 | 0.3502 |
| 0.2675 | 0.59 | 6000 | 0.2606 | 0.3477 |
| 0.2656 | 0.63 | 6400 | 0.2597 | 0.3463 |
| 0.2667 | 0.67 | 6800 | 0.2607 | 0.3458 |
| 0.2639 | 0.71 | 7200 | 0.2601 | 0.3480 |
| 0.2631 | 0.75 | 7600 | 0.2582 | 0.3447 |
| 0.2589 | 0.79 | 8000 | 0.2577 | 0.3438 |
| 0.2554 | 0.83 | 8400 | 0.2557 | 0.3439 |
| 0.2687 | 0.87 | 8800 | 0.2546 | 0.3438 |
| 0.2574 | 0.91 | 9200 | 0.2537 | 0.3434 |
| 0.2623 | 0.95 | 9600 | 0.2530 | 0.3433 |
| 0.2675 | 0.99 | 10000 | 0.2530 | 0.3426 |
From the published model card. Full card on the HuggingFace links in the sidebar.
Benchmarks
As published on the model card — the maker's own numbers, not measured by AxForge.
| Task | Dataset | Metric | Score |
|---|---|---|---|
| Automatic Speech Recognition | openslr | Test WER | 25.020 |
Using it via the API
Once AxForge deploys wav2vec2-large-xlsr-nepali for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (wav2vec2-large-xlsr-nepali below is illustrative; you get the exact model name on deployment.)
$ curl -sS https://api.axforge.ai/v1/audio/transcriptions \ -H "Authorization: Bearer $AXFORGE_API_KEY" \ -F model="wav2vec2-large-xlsr-nepali" -F file=@audio.mp3
Create an account — your API key is available in the console. 3M free tokens every 30 days with every new account.