Automatic Speech Recognition
Safetensors
MLX
Chinese
English
mlx-audio
qwen3_asr
asr
speech-to-text
taiwan-mandarin
mixed-quantization
4-bit precision
Instructions to use Alkd/TEA-ASR-1.1-MLX-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use Alkd/TEA-ASR-1.1-MLX-4bit with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir TEA-ASR-1.1-MLX-4bit Alkd/TEA-ASR-1.1-MLX-4bit
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
| { | |
| "base": "Qwen/Qwen3-ASR-1.7B", | |
| "adapter": "outputs/tea11_17b_v6_r_s0", | |
| "selected_row_delta": null, | |
| "selected_row_summary": null, | |
| "decode_config": "s2tw", | |
| "tokenizer_stats": { | |
| "clean": 150482, | |
| "marked": 1161, | |
| "merges": [ | |
| 151387, | |
| 138969 | |
| ] | |
| }, | |
| "verification": { | |
| "checked_sequences": 152643, | |
| "decode_mismatches": 0, | |
| "sentinel_leaks": 0, | |
| "examples": [], | |
| "is_fast": true | |
| } | |
| } | |