Qwen3-ASR-1.7B-JA MLX

Unquantized MLX-format conversion of neosophie/Qwen3-ASR-1.7B-JA for mlx-qwen3-asr.

Converted locally with mlx-qwen3-asr 0.3.5 by loading the source checkpoint, remapping Qwen3-ASR weights into MLX module keys, and saving bfloat16 weights without quantization.

Usage

pip install -U mlx-qwen3-asr
mlx-qwen3-asr sample.wav --model whnf/Qwen3-ASR-1.7B-JA-MLX --language Japanese -f vtt -o out
Downloads last month
37
Safetensors
Model size
2B params
Tensor type
BF16
·
MLX
Hardware compatibility
Log In to add your hardware

Quantized

Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for whnf/Qwen3-ASR-1.7B-JA-MLX

Finetuned
(1)
this model