mlx-qwen3-asr-ja-en-speech-translation-4bit
MLX 4-bit quantization of voiceping-ai/qwen3-asr-ja-en-speech-translation.
- Source model:
voiceping-ai/qwen3-asr-ja-en-speech-translation
- Quantization: 4-bit
- Group size: 64
- Format: MLX safetensors
- Intended use: drop-in Bee ASR model dir (
config.json, model.safetensors, tokenizer.json)
Note: this repo was produced from the original fine-tune with a local converter patch for tied embeddings (lm_head.weight tied to model.embed_tokens.weight).
