Commit
•
2f89806
1
Parent(s):
1b1c786
update model card README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,82 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
tags:
|
| 4 |
+
- automatic-speech-recognition
|
| 5 |
+
- Harveenchadha/indic-voice
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
model-index:
|
| 8 |
+
- name: Wav2Vec2_xls_r_openslr_Hi_V2
|
| 9 |
+
results: []
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 13 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 14 |
+
|
| 15 |
+
# Wav2Vec2_xls_r_openslr_Hi_V2
|
| 16 |
+
|
| 17 |
+
This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the Harveenchadha/indic-voice dataset.
|
| 18 |
+
It achieves the following results on the evaluation set:
|
| 19 |
+
- Loss: 0.3184
|
| 20 |
+
- Wer: 0.3104
|
| 21 |
+
- Cer: 0.0958
|
| 22 |
+
|
| 23 |
+
## Model description
|
| 24 |
+
|
| 25 |
+
More information needed
|
| 26 |
+
|
| 27 |
+
## Intended uses & limitations
|
| 28 |
+
|
| 29 |
+
More information needed
|
| 30 |
+
|
| 31 |
+
## Training and evaluation data
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Training procedure
|
| 36 |
+
|
| 37 |
+
### Training hyperparameters
|
| 38 |
+
|
| 39 |
+
The following hyperparameters were used during training:
|
| 40 |
+
- learning_rate: 0.0001
|
| 41 |
+
- train_batch_size: 16
|
| 42 |
+
- eval_batch_size: 8
|
| 43 |
+
- seed: 42
|
| 44 |
+
- gradient_accumulation_steps: 4
|
| 45 |
+
- total_train_batch_size: 64
|
| 46 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 47 |
+
- lr_scheduler_type: linear
|
| 48 |
+
- lr_scheduler_warmup_steps: 200
|
| 49 |
+
- num_epochs: 12
|
| 50 |
+
- mixed_precision_training: Native AMP
|
| 51 |
+
|
| 52 |
+
### Training results
|
| 53 |
+
|
| 54 |
+
| Training Loss | Epoch | Step | Cer | Validation Loss | Wer |
|
| 55 |
+
|:-------------:|:-----:|:----:|:------:|:---------------:|:------:|
|
| 56 |
+
| 7.1097 | 0.48 | 300 | 0.9965 | 3.3989 | 1.0 |
|
| 57 |
+
| 3.0235 | 0.96 | 600 | 0.3163 | 1.3183 | 0.7977 |
|
| 58 |
+
| 1.1419 | 1.44 | 900 | 0.1913 | 0.6416 | 0.5543 |
|
| 59 |
+
| 0.8242 | 1.92 | 1200 | 0.1608 | 0.5063 | 0.4804 |
|
| 60 |
+
| 0.6876 | 2.56 | 1600 | 0.1387 | 0.4401 | 0.4280 |
|
| 61 |
+
| 0.5868 | 3.21 | 2000 | 0.1249 | 0.3940 | 0.3907 |
|
| 62 |
+
| 0.5285 | 3.85 | 2400 | 0.1200 | 0.3661 | 0.3763 |
|
| 63 |
+
| 0.5 | 4.49 | 2800 | 0.3528 | 0.3610 | 0.1136 |
|
| 64 |
+
| 0.4538 | 5.13 | 3200 | 0.3403 | 0.3485 | 0.1086 |
|
| 65 |
+
| 0.4165 | 5.77 | 3600 | 0.3335 | 0.3439 | 0.1062 |
|
| 66 |
+
| 0.3989 | 6.41 | 4000 | 0.3264 | 0.3340 | 0.1036 |
|
| 67 |
+
| 0.3679 | 7.05 | 4400 | 0.3256 | 0.3287 | 0.1013 |
|
| 68 |
+
| 0.3517 | 7.69 | 4800 | 0.3212 | 0.3223 | 0.1002 |
|
| 69 |
+
| 0.3357 | 8.33 | 5200 | 0.3173 | 0.3196 | 0.0986 |
|
| 70 |
+
| 0.3225 | 8.97 | 5600 | 0.3142 | 0.3177 | 0.0985 |
|
| 71 |
+
| 0.3057 | 9.62 | 6000 | 0.3199 | 0.3156 | 0.0975 |
|
| 72 |
+
| 0.2972 | 10.26 | 6400 | 0.3139 | 0.3128 | 0.0967 |
|
| 73 |
+
| 0.2881 | 10.9 | 6800 | 0.3184 | 0.3107 | 0.0957 |
|
| 74 |
+
| 0.2791 | 11.54 | 7200 | 0.3184 | 0.3104 | 0.0958 |
|
| 75 |
+
|
| 76 |
+
|
| 77 |
+
### Framework versions
|
| 78 |
+
|
| 79 |
+
- Transformers 4.17.0.dev0
|
| 80 |
+
- Pytorch 1.10.2+cu102
|
| 81 |
+
- Datasets 1.18.2.dev0
|
| 82 |
+
- Tokenizers 0.11.0
|