File size: 3,196 Bytes
3388588 d0036f3 3388588 d0036f3 3388588 d0036f3 3388588 d0036f3 3388588 d0036f3 3388588 d0036f3 3388588 |
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 |
---
library_name: transformers
license: apache-2.0
base_model: openai/whisper-large-v3
tags:
- whisper-event
- generated_from_trainer
datasets:
- OUTCOMESAI/medical_n_common_speech_corpus_50_50
metrics:
- wer
model-index:
- name: Whisper Large V3 Common n Medical 50 50
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: OUTCOMESAI/medical_n_common_speech_corpus_50_50 en
type: OUTCOMESAI/medical_n_common_speech_corpus_50_50
metrics:
- name: Wer
type: wer
value: 5.218643517767322
---
<!-- 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 Large V3 Common n Medical 50 50
This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the OUTCOMESAI/medical_n_common_speech_corpus_50_50 en dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3196
- Wer: 5.2186
## 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: 5e-07
- train_batch_size: 64
- eval_batch_size: 32
- seed: 42
- distributed_type: multi-GPU
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 250
- training_steps: 5000
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|:-------------:|:------:|:----:|:---------------:|:------:|
| 5.126 | 0.0969 | 250 | 0.3694 | 5.6601 |
| 4.367 | 0.1938 | 500 | 0.3586 | 5.8156 |
| 4.1514 | 0.2907 | 750 | 0.3511 | 5.8839 |
| 3.962 | 0.3876 | 1000 | 0.3450 | 5.7805 |
| 3.9038 | 0.4845 | 1250 | 0.3403 | 6.1746 |
| 3.8313 | 0.5814 | 1500 | 0.3359 | 5.9738 |
| 3.7778 | 0.6783 | 1750 | 0.3333 | 5.9218 |
| 3.7421 | 0.7752 | 2000 | 0.3306 | 6.1327 |
| 3.7367 | 0.8721 | 2250 | 0.3281 | 5.6561 |
| 3.6878 | 0.9690 | 2500 | 0.3257 | 5.5154 |
| 3.6769 | 1.0659 | 2750 | 0.3242 | 5.4803 |
| 3.6508 | 1.1628 | 3000 | 0.3235 | 5.4634 |
| 3.6292 | 1.2597 | 3250 | 0.3220 | 5.3512 |
| 3.6179 | 1.3566 | 3500 | 0.3210 | 5.2254 |
| 3.6032 | 1.4535 | 3750 | 0.3206 | 5.2207 |
| 3.5922 | 1.5504 | 4000 | 0.3201 | 5.3038 |
| 3.5743 | 1.6473 | 4250 | 0.3198 | 5.2633 |
| 3.5882 | 1.7442 | 4500 | 0.3198 | 5.2254 |
| 3.6021 | 1.8411 | 4750 | 0.3196 | 5.2186 |
| 3.5865 | 1.9380 | 5000 | 0.3193 | 5.2213 |
### Framework versions
- Transformers 4.48.0.dev0
- Pytorch 2.1.0+cu118
- Datasets 3.2.1.dev0
- Tokenizers 0.21.0
|