File size: 5,120 Bytes
0e68ad2 5abf52f 0e68ad2 5abf52f 0e68ad2 5abf52f 0e68ad2 5abf52f 0e68ad2 5abf52f 0e68ad2 5abf52f 0e68ad2 |
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 |
---
language:
- ga
- en
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- ymoslem/IWSLT2023-GA-EN
- ymoslem/FLEURS-GA-EN
- ymoslem/BitesizeIrish-GA-EN
- ymoslem/SpokenWords-GA-EN-MTed
metrics:
- bleu
- wer
model-index:
- name: Whisper Small GA-EN Speech Translation
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: IWSLT-2023, FLEURS, BiteSize, and SpokenWords
type: ymoslem/IWSLT2023-GA-EN
metrics:
- name: Bleu
type: bleu
value: 29.51
- name: Wer
type: wer
value: 67.08689779378658
---
<!-- 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. -->
# Whisper Small GA-EN Speech Translation
This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the IWSLT-2023, FLEURS, BiteSize, and SpokenWords dataset.
It achieves the following results on the evaluation set:
- Loss: 1.8231
- Bleu: 29.51
- Chrf: 44.29
- Wer: 67.0869
## 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: 0.0001
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- training_steps: 4000
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Bleu | Chrf | Validation Loss | Wer |
|:-------------:|:------:|:----:|:-----:|:-----:|:---------------:|:--------:|
| 1.9416 | 0.2155 | 100 | 13.09 | 26.48 | 1.7899 | 104.4575 |
| 1.5186 | 0.4310 | 200 | 18.6 | 35.75 | 1.5696 | 87.5732 |
| 1.2884 | 0.6466 | 300 | 17.57 | 37.19 | 1.4751 | 87.2580 |
| 1.0729 | 0.8621 | 400 | 17.92 | 38.23 | 1.4345 | 99.2346 |
| 0.4574 | 1.0776 | 500 | 22.48 | 39.17 | 1.5585 | 83.1607 |
| 0.4517 | 1.2931 | 600 | 22.53 | 38.38 | 1.5763 | 81.7650 |
| 0.4385 | 1.5086 | 700 | 20.05 | 39.46 | 1.5852 | 96.8483 |
| 0.3934 | 1.7241 | 800 | 26.89 | 42.67 | 1.5332 | 70.6889 |
| 0.3587 | 1.9397 | 900 | 28.95 | 44.16 | 1.5025 | 64.9707 |
| 0.1528 | 2.1552 | 1000 | 28.32 | 42.36 | 1.5882 | 65.8712 |
| 0.1425 | 2.3707 | 1100 | 25.5 | 42.42 | 1.6056 | 75.0113 |
| 0.1389 | 2.5862 | 1200 | 26.52 | 42.11 | 1.6236 | 70.6439 |
| 0.1532 | 2.8017 | 1300 | 25.78 | 41.61 | 1.6196 | 75.9118 |
| 0.1138 | 3.0172 | 1400 | 26.01 | 40.88 | 1.7185 | 69.6983 |
| 0.0661 | 3.2328 | 1500 | 28.74 | 43.16 | 1.6626 | 71.2292 |
| 0.0625 | 3.4483 | 1600 | 29.16 | 43.6 | 1.6835 | 66.3215 |
| 0.0615 | 3.6638 | 1700 | 28.93 | 44.08 | 1.6756 | 68.3476 |
| 0.0611 | 3.8793 | 1800 | 27.77 | 43.67 | 1.6648 | 72.1747 |
| 0.0344 | 4.0948 | 1900 | 28.33 | 44.18 | 1.7351 | 68.1225 |
| 0.0339 | 4.3103 | 2000 | 28.9 | 42.98 | 1.7715 | 67.0869 |
| 0.0369 | 4.5259 | 2100 | 29.83 | 44.87 | 1.7200 | 64.8807 |
| 0.0326 | 4.7414 | 2200 | 28.23 | 43.75 | 1.7232 | 69.3832 |
| 0.0346 | 4.9569 | 2300 | 27.72 | 43.1 | 1.7688 | 72.8050 |
| 0.0167 | 5.1724 | 2400 | 28.73 | 43.26 | 1.8072 | 67.4471 |
| 0.0146 | 5.3879 | 2500 | 29.91 | 44.24 | 1.7801 | 66.4566 |
| 0.0165 | 5.6034 | 2600 | 29.34 | 44.33 | 1.7782 | 68.2125 |
| 0.0143 | 5.8190 | 2700 | 27.78 | 43.07 | 1.7675 | 72.5799 |
| 0.0106 | 6.0345 | 2800 | 29.45 | 43.31 | 1.7660 | 67.5371 |
| 0.0098 | 6.25 | 2900 | 27.89 | 42.67 | 1.7803 | 71.6344 |
| 0.0087 | 6.4655 | 3000 | 27.66 | 43.04 | 1.7786 | 72.0396 |
| 0.0089 | 6.6810 | 3100 | 1.7661| 29.81 | 44.65 | 67.3120 |
| 0.0081 | 6.8966 | 3200 | 1.7744| 29.48 | 44.3 | 68.0324 |
| 0.0095 | 7.1121 | 3300 | 1.8197| 29.55 | 44.2 | 67.5371 |
| 0.0112 | 7.3276 | 3400 | 1.8102| 29.34 | 43.9 | 66.2765 |
| 0.0075 | 7.5431 | 3500 | 1.8004| 29.57 | 44.43 | 67.3570 |
| 0.0111 | 7.7586 | 3600 | 1.8015| 29.56 | 44.57 | 66.4566 |
| 0.009 | 7.9741 | 3700 | 1.8001| 29.7 | 45.24 | 66.6817 |
| 0.005 | 8.1897 | 3800 | 1.8184| 29.21 | 44.4 | 67.4471 |
| 0.0055 | 8.4052 | 3900 | 1.8222| 29.67 | 44.35 | 67.1319 |
| 0.0042 | 8.6207 | 4000 | 1.8231| 29.51 | 44.29 | 67.0869 |
### Framework versions
- Transformers 4.40.2
- Pytorch 2.2.0+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1
|