Search results

Qwen3-TTS-12Hz-0.6B-Base fine-tuning fails due to embedding dimension mismatch (2048 vs 1024)

Status: Open.

#198 In QwenLM/Qwen3-TTS;

· shubhamragade opened on Feb 10, 2026

如何实现并发推理？

Status: Open.

#197 In QwenLM/Qwen3-TTS;

· maxin9966 opened on Feb 10, 2026

Incompatible library version : we need a requirements.txt file

Status: Open.

#196 In QwenLM/Qwen3-TTS;

· mrinalinigarg opened on Feb 9, 2026

The training methodology of the base model is inconsistent with the description provided in the paper.

Status: Open.

#194 In QwenLM/Qwen3-TTS;

· foresttigeryao opened on Feb 9, 2026

Setting <code>pad_token_id</code> to <code>eos_token_id</code>:2150 for open-end generation.

Status: Open.

#184 In QwenLM/Qwen3-TTS;

· LIUKAI0815 opened on Feb 6, 2026

The last few words of the audio generated from long text will be swallowed.

Status: Open.

#181 In QwenLM/Qwen3-TTS;

· Deadikun opened on Feb 5, 2026

Finetuning Base results in progressively faster speech with every epoch

Status: Open.

#179 In QwenLM/Qwen3-TTS;

· dragonfyre13 opened on Feb 5, 2026

Qwen/Qwen3-TTS-12Hz-0.6B-Base -&gt; RuntimeError: The size of tensor a (2048) must match the size of tensor b (1024) at non-singleton dimension 2

Status: Open.

#174 In QwenLM/Qwen3-TTS;

· fablevi opened on Feb 4, 2026

incorrect pypi package source target url owner: <code>qwen</code> rather than <code>qwenlm</code>

Status: Open.

#173 In QwenLM/Qwen3-TTS;

· evdcush opened on Feb 4, 2026

Inference is too slow

Status: Open.

#171 In QwenLM/Qwen3-TTS;

· skytodmoon opened on Feb 4, 2026

qwen-tts hard-pins transformers==4.57.3, preventing co-installation with qwen-asr despite verified compatibility

Status: Open.

#156 In QwenLM/Qwen3-TTS;

· joshuasundance-swca opened on Feb 1, 2026

Mixed Chinese-English numeral pronunciation in cross-lingual voice cloning

Status: Open.

#151 In QwenLM/Qwen3-TTS;

· aichase opened on Feb 1, 2026