Difference From The Non--hf Repository
The source repository is intended for the qwen-asr wrapper / original Qwen3-ASR
layout. This repository is intended for native Hugging Face Transformers loading.
The conversion keeps the fine-tuned weights unchanged and only rewrites the
repository layout to match Qwen/Qwen3-ASR-0.6B-hf:
- config / processor / tokenizer files come from the official
-hf template.
- safetensors keys are rewritten as:
thinker.audio_tower.* -> model.audio_tower.*thinker.audio_tower.proj1.* -> model.multi_modal_projector.linear_1.*thinker.audio_tower.proj2.* -> model.multi_modal_projector.linear_2.*thinker.model.* -> model.language_model.*
- tensor count after conversion:
611.
- converted tensor bytes:
1564852224.
Requirements
Requires the stable transformers >= 5.13.0 release for native Qwen3-ASR support:
pip install "transformers>=5.13.0"
With uv:
uv pip install "transformers>=5.13.0"
Usage
from transformers import AutoModelForMultimodalLM, AutoProcessor
model_id = "jaykwok/Qwen3-ASR-0.6B-JA-Anime-Galgame-hf"
processor = AutoProcessor.from_pretrained(model_id)
model = AutoModelForMultimodalLM.from_pretrained(
model_id,
torch_dtype="auto",
device_map="auto",
)
inputs = processor.apply_transcription_request(
audio="path/to/audio.wav",
language="Japanese",
).to(model.device, model.dtype)
output_ids = model.generate(**inputs, max_new_tokens=256)
generated_ids = output_ids[:, inputs["input_ids"].shape[1]:]
text = processor.decode(generated_ids, return_format="transcription_only")[0]
print(text)
Conversion
This repository was produced with:
uv run python -m tools.asr.convert_qwen3_asr_to_hf `
--source-model-dir models/jaykwok-Qwen3-ASR-0.6B-JA-Anime-Galgame `
--output-dir agents/temp/20260630_131549_qwen3_asr_0p6b_hf_conversion `
--template-repo Qwen/Qwen3-ASR-0.6B-hf `
--max-shard-size 768MB
Notes
This model is specialized for Japanese anime / galgame style speech. It should
be evaluated on your own data before production use.