Downloads · 30 days
20
39% of all-time downloads
Hubert-noisy-cv-kakeiken-J
utakumi/Hubert-noisy-cv-kakeiken-J
Hubert-noisy-cv-kakeiken-J is a automatic speech recognition model from utakumi. Use it when you need speech turned into text. It is set up for transformers. The card lists the license as apache-2.0.
should probably proofread and complete it, then remove this comment. --
All-time downloads
51
Public
Parameters
94.4M
31 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors378 MB · 100%
At a glance
- Task
- Automatic Speech Recognition
- Library
- transformers
- License
- apache-2.0
- Model type
- hubert
- Access
- Public
- Created
- Jan 13, 2025
- Updated
- Jan 13, 2025
- SHA
- 646be422
Base models
Model card
From the Hugging Face model README
<!-- 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. -->
Hubert-noisy-cv-kakeiken-J
This model is a fine-tuned version of rinna/japanese-hubert-base on the ORIGINAL_NOISY_KAKEIKEN_W - JA dataset. It achieves the following results on the evaluation set:
- Loss: 3.3017
- Wer: 1.0
- Cer: 1.1284
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: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 12500
- num_epochs: 20.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 3.076 | 1.0 | 1639 | 2.2157 | 1.0 | 1.1284 |
| 0.5136 | 2.0 | 3278 | 0.2265 | 0.9990 | 1.0553 |
| 0.4033 | 3.0 | 4917 | 0.2809 | 0.9988 | 1.0661 |
| 0.416 | 4.0 | 6556 | 0.2755 | 0.9994 | 1.0706 |
| 0.4323 | 5.0 | 8195 | 0.3099 | 0.9993 | 1.1033 |
| 3.3425 | 6.0 | 9834 | 3.3130 | 1.0 | 1.1284 |
| 3.3313 | 7.0 | 11473 | 3.3158 | 1.0 | 1.1284 |
| 3.3297 | 8.0 | 13112 | 3.3195 | 1.0 | 1.1283 |
| 3.3269 | 9.0 | 14751 | 3.3184 | 1.0 | 1.1284 |
| 3.3291 | 10.0 | 16390 | 3.3212 | 1.0 | 1.1284 |
| 3.329 | 11.0 | 18029 | 3.3157 | 1.0 | 1.1284 |
| 3.3261 | 12.0 | 19668 | 3.3104 | 1.0 | 1.1284 |
| 3.3215 | 13.0 | 21307 | 3.3075 | 1.0 | 1.1284 |
| 3.3193 | 14.0 | 22946 | 3.3067 | 1.0 | 1.1284 |
| 3.3153 | 15.0 | 24585 | 3.3036 | 1.0 | 1.1284 |
| 3.3121 | 16.0 | 26224 | 3.3033 | 1.0 | 1.1283 |
| 3.3074 | 17.0 | 27863 | 3.3021 | 1.0 | 1.1284 |
| 3.3064 | 18.0 | 29502 | 3.3022 | 1.0 | 1.1283 |
| 3.3079 | 19.0 | 31141 | 3.3020 | 1.0 | 1.1284 |
| 3.3061 | 20.0 | 32780 | 3.3018 | 1.0 | 1.1284 |
Framework versions
- Transformers 4.48.0
- Pytorch 2.5.1+cu124
- Datasets 3.1.0
- Tokenizers 0.21.0
- Task
- Automatic Speech Recognition
- Library
- transformers
- Type
- hubert
- License
- apache-2.0
- Created
- Jan 13, 2025
- Updated
- Jan 13, 2025