umbragit/Qwen3-TTS
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

fix: adjust finetune script for stable training.

Browse Source
This commit is contained in:
Xiong Wang
2026-02-04 11:15:24 +08:00
parent 1ab0dd7535
commit 680d4e9d0e
3 changed files with 5 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
finetuning/README.md
View File

@@ -50,9 +50,9 @@ python sft_12hz.py \
--init_model_path Qwen/Qwen3-TTS-12Hz-1.7B-Base \
--output_model_path output \
--train_jsonl train_with_codes.jsonl \
--batch_size 2 \
--lr 2e-5 \
--num_epochs 3 \
--batch_size 32 \
--lr 2e-6 \
--num_epochs 10 \
--speaker_name speaker_test
```

2
finetuning/sft_12hz.py
View File

@@ -110,7 +110,7 @@ def train():
sub_talker_logits, sub_talker_loss = model.talker.forward_sub_talker_finetune(talker_codec_ids, talker_hidden_states)
loss = outputs.loss + sub_talker_loss
loss = outputs.loss + 0.3 * sub_talker_loss
accelerator.backward(loss)

2
pyproject.toml
View File

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
[project]
name = "qwen-tts"
version = "0.0.4"
version = "0.0.6"
description = "Qwen-TTS python package"
readme = "README.md"
requires-python = ">=3.9"