File size: 4,279 Bytes
392116b f9d9877 392116b f9d9877 |
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 |
---
license: apache-2.0
base_model: facebook/wav2vec2-xls-r-300m
tags:
- generated_from_trainer
datasets:
- common_voice_17_0
metrics:
- wer
model-index:
- name: xls-r-300m-hbs-fr-unfrozen-batch16
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: common_voice_17_0
type: common_voice_17_0
config: hsb
split: test
args: hsb
metrics:
- name: Wer
type: wer
value: 0.3959700093720712
---
<!-- 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. -->
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/badr-nlp/xlsr-continual-finetuning/runs/pnc4tk8k)
# xls-r-300m-hbs-fr-unfrozen-batch16
This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the common_voice_17_0 dataset.
It achieves the following results on the evaluation set:
- Loss: 0.7093
- Wer: 0.3960
- Cer: 0.0915
## 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: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 100
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|:-------------:|:-------:|:----:|:---------------:|:------:|:------:|
| 3.5652 | 3.2258 | 100 | 3.3748 | 1.0 | 1.0 |
| 3.2583 | 6.4516 | 200 | 3.2149 | 1.0 | 1.0 |
| 3.1829 | 9.6774 | 300 | 3.1452 | 1.0 | 1.0 |
| 0.7256 | 12.9032 | 400 | 0.7889 | 0.7134 | 0.1766 |
| 0.3062 | 16.1290 | 500 | 0.6745 | 0.6146 | 0.1423 |
| 0.1843 | 19.3548 | 600 | 0.6301 | 0.5265 | 0.1242 |
| 0.1259 | 22.5806 | 700 | 0.6102 | 0.4820 | 0.1121 |
| 0.1386 | 25.8065 | 800 | 0.6702 | 0.4939 | 0.1176 |
| 0.0962 | 29.0323 | 900 | 0.6297 | 0.4806 | 0.1147 |
| 0.069 | 32.2581 | 1000 | 0.6766 | 0.4740 | 0.1113 |
| 0.0779 | 35.4839 | 1100 | 0.6565 | 0.4609 | 0.1075 |
| 0.0715 | 38.7097 | 1200 | 0.6649 | 0.4649 | 0.1103 |
| 0.0448 | 41.9355 | 1300 | 0.6558 | 0.4642 | 0.1094 |
| 0.0552 | 45.1613 | 1400 | 0.6893 | 0.4412 | 0.1035 |
| 0.0396 | 48.3871 | 1500 | 0.7179 | 0.4527 | 0.1041 |
| 0.0592 | 51.6129 | 1600 | 0.6455 | 0.4285 | 0.0976 |
| 0.0509 | 54.8387 | 1700 | 0.6605 | 0.4349 | 0.1005 |
| 0.0665 | 58.0645 | 1800 | 0.7340 | 0.4243 | 0.0991 |
| 0.0391 | 61.2903 | 1900 | 0.7378 | 0.4330 | 0.1018 |
| 0.0974 | 64.5161 | 2000 | 0.6984 | 0.4306 | 0.1003 |
| 0.0344 | 67.7419 | 2100 | 0.6895 | 0.4208 | 0.0974 |
| 0.043 | 70.9677 | 2200 | 0.7214 | 0.4140 | 0.0965 |
| 0.0248 | 74.1935 | 2300 | 0.7242 | 0.4149 | 0.0990 |
| 0.0194 | 77.4194 | 2400 | 0.7233 | 0.4107 | 0.0962 |
| 0.0277 | 80.6452 | 2500 | 0.7247 | 0.4100 | 0.0946 |
| 0.0447 | 83.8710 | 2600 | 0.7078 | 0.4004 | 0.0941 |
| 0.0291 | 87.0968 | 2700 | 0.7073 | 0.4002 | 0.0915 |
| 0.0208 | 90.3226 | 2800 | 0.7121 | 0.4025 | 0.0921 |
| 0.0278 | 93.5484 | 2900 | 0.6998 | 0.3932 | 0.0914 |
| 0.0569 | 96.7742 | 3000 | 0.7105 | 0.3964 | 0.0918 |
| 0.0132 | 100.0 | 3100 | 0.7093 | 0.3960 | 0.0915 |
### Framework versions
- Transformers 4.42.0.dev0
- Pytorch 2.3.1+cu121
- Datasets 2.19.2
- Tokenizers 0.19.1
|