[PR #21] Add compatibility for Transformers 5.x #22

New issue

Open

opened 2026-10-30 21:03:38 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented 2026-10-30 21:03:38 +00:00

Owner

Copy link

📋 Pull Request Information

Original PR: https://github.com/1038lab/ComfyUI-QwenTTS/pull/21
Author: @cvrunmin
Created: 7/18/2026
Status: 🔄 Open

Base: main ← Head: main

---

📝 Commits (3)

• 96acacf transformers 5.x compat
• df7b6c2 better version check
• 7051417 fix some model weight get reinitialized

📊 Changes

12 files changed (+330 additions, -39 deletions)

View changed files

➕ .gitignore (+220 -0)
➖ qwen_tts/core/models/__pycache__/__init__.cpython-312.pyc (+0 -0)
➖ qwen_tts/core/models/__pycache__/configuration_qwen3_tts.cpython-312.pyc (+0 -0)
➖ qwen_tts/core/models/__pycache__/modeling_qwen3_tts.cpython-312.pyc (+0 -0)
➖ qwen_tts/core/models/__pycache__/processing_qwen3_tts.cpython-312.pyc (+0 -0)
📝 qwen_tts/core/models/configuration_qwen3_tts.py (+4 -1)
📝 qwen_tts/core/models/modeling_qwen3_tts.py (+74 -31)
➖ qwen_tts/core/tokenizer_12hz/__pycache__/configuration_qwen3_tts_tokenizer_v2.cpython-312.pyc (+0 -0)
➖ qwen_tts/core/tokenizer_12hz/__pycache__/modeling_qwen3_tts_tokenizer_v2.cpython-312.pyc (+0 -0)
📝 qwen_tts/core/tokenizer_12hz/modeling_qwen3_tts_tokenizer_v2.py (+32 -7)
➖ qwen_tts/inference/__pycache__/qwen3_tts_model.cpython-312.pyc (+0 -0)
➖ qwen_tts/inference/__pycache__/qwen3_tts_tokenizer.cpython-312.pyc (+0 -0)

📄 Description

This PR aims to solve the incompatability between Qwen3-TTS package (inlined at qwen_tts) and Transformers 5.x

Summary

1. resolve ImportError on transformers.utils.generic.check_model_inputs (from transformers.utils.generic import check_model_inputs at qwen_tts/core/tokenizer_12hz/modeling_qwen3_tts_tokenizer_v2.py)
2. change input_embeds into inputs_embeds (note: this change happened in Transformers 5.2.0)
3. remove cache_position if necessary
4. fix KeyError for rope_type = 'default'
5. fix 'Qwen3TTSTalkerConfig' object has no attribute 'pad_token_id': #9, #10

---

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/1038lab/ComfyUI-QwenTTS/pull/21 **Author:** [@cvrunmin](https://github.com/cvrunmin) **Created:** 7/18/2026 **Status:** 🔄 Open **Base:** `main` ← **Head:** `main` --- ### 📝 Commits (3) - [`96acacf`](https://github.com/1038lab/ComfyUI-QwenTTS/commit/96acacf22e1777220a25cd4316e3109f59bcf309) transformers 5.x compat - [`df7b6c2`](https://github.com/1038lab/ComfyUI-QwenTTS/commit/df7b6c2ef8c7e11709566e40816689738b32947c) better version check - [`7051417`](https://github.com/1038lab/ComfyUI-QwenTTS/commit/7051417736c2cbe8dbedf05596971ae4c0206b86) fix some model weight get reinitialized ### 📊 Changes **12 files changed** (+330 additions, -39 deletions) <details> <summary>View changed files</summary> ➕ `.gitignore` (+220 -0) ➖ `qwen_tts/core/models/__pycache__/__init__.cpython-312.pyc` (+0 -0) ➖ `qwen_tts/core/models/__pycache__/configuration_qwen3_tts.cpython-312.pyc` (+0 -0) ➖ `qwen_tts/core/models/__pycache__/modeling_qwen3_tts.cpython-312.pyc` (+0 -0) ➖ `qwen_tts/core/models/__pycache__/processing_qwen3_tts.cpython-312.pyc` (+0 -0) 📝 `qwen_tts/core/models/configuration_qwen3_tts.py` (+4 -1) 📝 `qwen_tts/core/models/modeling_qwen3_tts.py` (+74 -31) ➖ `qwen_tts/core/tokenizer_12hz/__pycache__/configuration_qwen3_tts_tokenizer_v2.cpython-312.pyc` (+0 -0) ➖ `qwen_tts/core/tokenizer_12hz/__pycache__/modeling_qwen3_tts_tokenizer_v2.cpython-312.pyc` (+0 -0) 📝 `qwen_tts/core/tokenizer_12hz/modeling_qwen3_tts_tokenizer_v2.py` (+32 -7) ➖ `qwen_tts/inference/__pycache__/qwen3_tts_model.cpython-312.pyc` (+0 -0) ➖ `qwen_tts/inference/__pycache__/qwen3_tts_tokenizer.cpython-312.pyc` (+0 -0) </details> ### 📄 Description This PR aims to solve the incompatability between Qwen3-TTS package (inlined at `qwen_tts`) and Transformers 5.x ## Summary 1. resolve `ImportError` on `transformers.utils.generic.check_model_inputs` (`from transformers.utils.generic import check_model_inputs` at qwen_tts/core/tokenizer_12hz/modeling_qwen3_tts_tokenizer_v2.py) 2. change `input_embeds` into `inputs_embeds` (note: this change happened in Transformers 5.2.0) 3. remove `cache_position` if necessary 4. fix KeyError for rope_type = 'default' 5. fix 'Qwen3TTSTalkerConfig' object has no attribute 'pad_token_id': #9, #10 --- _{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

bsakaguchi added the

pull-request

label 2026-10-30 21:03:38 +00:00

Sign in to join this conversation.

No Branch/Tag specified

Branches Tags

main

No results found.

Labels

Clear labels   

pull-request

Mirrored from GitHub Pull Request

No labels

pull-request

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

1038lab/ComfyUI-QwenTTS#22

No description provided.