File size: 4,729 Bytes
7a43c3b 56830d3 7a43c3b 56830d3 7a43c3b 24e0664 cbf7f0b 7a43c3b cbf7f0b 7a43c3b cbf7f0b a9ac24f cbf7f0b a9ac24f cbf7f0b a9ac24f 7a43c3b 70378de 7a43c3b 56830d3 7a43c3b 56830d3 7a43c3b cbf7f0b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 |
---
language:
- sl
license: apache-2.0
tags:
- automatic-speech-recognition
- generated_from_trainer
- hf-asr-leaderboard
- mozilla-foundation/common_voice_8_0
- robust-speech-event
datasets:
- mozilla-foundation/common_voice_8_0
model-index:
- name: XLS-R-300M - Slovenian
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Common Voice 8
type: mozilla-foundation/common_voice_8_0
args: sl
metrics:
- name: Test WER
type: wer
value: 12.736
- name: Test CER
type: cer
value: 3.605
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Robust Speech Event - Dev Data
type: speech-recognition-community-v2/dev_data
args: sl
metrics:
- name: Test WER
type: wer
value: 45.587
- name: Test CER
type: cer
value: 20.886
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Robust Speech Event - Test Data
type: speech-recognition-community-v2/eval_data
args: sl
metrics:
- name: Test WER
type: wer
value: 45.42
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# XLS-R-300M - Slovenian
This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the MOZILLA-FOUNDATION/COMMON_VOICE_8_0 - SL dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2578
- Wer: 0.2273
## 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: 7.5e-05
- train_batch_size: 32
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 60.0
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|:-------------:|:-----:|:----:|:---------------:|:------:|
| 3.1829 | 4.88 | 400 | 3.1228 | 1.0 |
| 2.8675 | 9.76 | 800 | 2.8616 | 0.9993 |
| 1.583 | 14.63 | 1200 | 0.6392 | 0.6239 |
| 1.1959 | 19.51 | 1600 | 0.3602 | 0.3651 |
| 1.0276 | 24.39 | 2000 | 0.3021 | 0.2981 |
| 0.9671 | 29.27 | 2400 | 0.2872 | 0.2739 |
| 0.873 | 34.15 | 2800 | 0.2593 | 0.2459 |
| 0.8513 | 39.02 | 3200 | 0.2617 | 0.2473 |
| 0.8132 | 43.9 | 3600 | 0.2548 | 0.2426 |
| 0.7935 | 48.78 | 4000 | 0.2637 | 0.2353 |
| 0.7565 | 53.66 | 4400 | 0.2629 | 0.2322 |
| 0.7359 | 58.54 | 4800 | 0.2579 | 0.2253 |
### Framework versions
- Transformers 4.17.0.dev0
- Pytorch 1.10.2+cu102
- Datasets 1.18.2.dev0
- Tokenizers 0.11.0
#### Evaluation Commands
1. To evaluate on `mozilla-foundation/common_voice_8_0` with split `test`
```bash
python eval.py --model_id anuragshas/wav2vec2-xls-r-300m-sl-cv8-with-lm --dataset mozilla-foundation/common_voice_8_0 --config sl --split test
```
2. To evaluate on `speech-recognition-community-v2/dev_data`
```bash
python eval.py --model_id anuragshas/wav2vec2-xls-r-300m-sl-cv8-with-lm --dataset speech-recognition-community-v2/dev_data --config sl --split validation --chunk_length_s 5.0 --stride_length_s 1.0
```
### Inference With LM
```python
import torch
from datasets import load_dataset
from transformers import AutoModelForCTC, AutoProcessor
import torchaudio.functional as F
model_id = "anuragshas/wav2vec2-xls-r-300m-sl-cv8-with-lm"
sample_iter = iter(load_dataset("mozilla-foundation/common_voice_8_0", "sl", split="test", streaming=True, use_auth_token=True))
sample = next(sample_iter)
resampled_audio = F.resample(torch.tensor(sample["audio"]["array"]), 48_000, 16_000).numpy()
model = AutoModelForCTC.from_pretrained(model_id)
processor = AutoProcessor.from_pretrained(model_id)
input_values = processor(resampled_audio, return_tensors="pt").input_values
with torch.no_grad():
logits = model(input_values).logits
transcription = processor.batch_decode(logits.numpy()).text
# => "zmago je divje od letel s helikopterjem visoko vzrak"
```
### Eval results on Common Voice 8 "test" (WER):
| Without LM | With LM (run `./eval.py`) |
|---|---|
| 19.938 | 12.736 | |