Instructions to use iljab/omniASR-CTC-7B-v2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use iljab/omniASR-CTC-7B-v2 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="iljab/omniASR-CTC-7B-v2")# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("iljab/omniASR-CTC-7B-v2", device_map="auto") - Notebooks
- Google Colab
- Kaggle
omniASR-CTC-7B-v2 (HuggingFace)
Converted from Meta's official fairseq2 checkpoint
omniASR-CTC-7B-v2.pt
(card omniASR_CTC_7B_v2). Tokenizer:
omniASR_tokenizer_written_v2.model.
| Property | Value |
|---|---|
| HF class | Wav2Vec2ForCTC |
| Encoder layers | 128 |
| Hidden size | 2048 |
| Attention heads | 16 |
| FFN intermediate | 8192 |
| Vocabulary size | 10288 |
| CTC blank | id 0 (<s>) |
| Source | Meta Omnilingual ASR v2, Apache-2.0 |
from transformers import Wav2Vec2ForCTC, AutoProcessor
processor = AutoProcessor.from_pretrained("iljab/omniASR-CTC-7B-v2")
model = Wav2Vec2ForCTC.from_pretrained("iljab/omniASR-CTC-7B-v2")
- Downloads last month
- 6