Skip to content

[Bug]: PR #669 forces model.talker needs to be present. Is this expected? #728

@divyanshsinghvi

Description

@divyanshsinghvi

Your current environment

The output of python collect_env.py
Your output of `python collect_env.py` here

🐛 Describe the bug

In #669

If a model doesn't intialize a variable self.model.talker = None it will fail.

(EngineCore_DP0 pid=34323) [Stage-0] ERROR 01-10 13:07:58 [core.py:843]   File "/home/divyansh/code/open_source/vllm-omni/vllm_omni/worker/gpu_model_runner.py", line 566, in _dummy_run
(EngineCore_DP0 pid=34323) [Stage-0] ERROR 01-10 13:07:58 [core.py:843]     if self.model.talker is not None and hasattr(self.model, "talker_mtp") and num_tokens_padded == 1:
(EngineCore_DP0 pid=34323) [Stage-0] ERROR 01-10 13:07:58 [core.py:843]        ^^^^^^^^^^^^^^^^^
(EngineCore_DP0 pid=34323) [Stage-0] ERROR 01-10 13:07:58 [core.py:843]   File "/home/divyansh/code/open_source/vllm-omni/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1964, in __getattr__
(EngineCore_DP0 pid=34323) [Stage-0] ERROR 01-10 13:07:58 [core.py:843]     raise AttributeError(
(EngineCore_DP0 pid=34323) [Stage-0] ERROR 01-10 13:07:58 [core.py:843] AttributeError: 'CosyVoice3Model' object has no attribute 'talker'

Possible Fix

Should be checking if hasattr(self.model, "talker") unless it's a design choice and must be present

Before submitting a new issue...

  • Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

Metadata

Metadata

Assignees

Labels

bugSomething isn't working

Type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions