Qwen3-ASR-0.6B-JA-Anime-Galgame-hf

This is a Transformers-native -hf layout conversion of jaykwok/Qwen3-ASR-0.6B-JA-Anime-Galgame, which was fine-tuned from the non--hf Qwen3-ASR checkpoint.

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)

Changelog

• 2026-09-02 — chat_template.jinja gained an assistant-role rendering block, copied verbatim from the base Qwen/Qwen3-ASR-1.7B-hf / Qwen/Qwen3-ASR-0.6B-hf template (both are byte-identical). transformers 5.15.0 changed Qwen3ASRProcessor.apply_transcription_request()'s language forcing from a system-prompt string + add_generation_prompt=True to an assistant-turn prefill ("language <NAME><asr_text>") + continue_final_message=True (upstream commit c7f9c881). This repository's template previously had no branch for assistant-role messages at all, so continue_final_message raised ValueError("continue_final_message is set but the final message does not appear in the chat...") on every transcription request under transformers>=5.15.0. The fix is purely additive: the existing system/user rendering and the add_generation_prompt branch are unchanged, so callers still on transformers<5.15.0 see no behavior change.

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.