A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 01:00 UTC
Remove Qwen Omni workaround that's no longer necessary (#21057)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Harry Mellor committed
Jul 16, 2025 at 17:25 UTC
a931b4cdcf70c45d7da6945e0ffca3c372808e20
1 file changed
-7
vllm/transformers_utils/config.py
-7
@@ -733,13 +733,6 @@ def get_hf_text_config(config: PretrainedConfig):
733
"""Get the "sub" config relevant to llm for multi modal models.
734
No op for pure text models.
735
"""
736
- # This block should be unnecessary after https://github.com/huggingface/transformers/pull/37517
737
- if hasattr(config, "thinker_config"):
738
- # TODO(suyang.fy): Refactor code.
739
- # For Qwen2.5-Omni, change hf_text_config to
740
- # thinker_config.text_config.
741
- return config.thinker_config.text_config
742
-
736
text_config = config.get_text_config()
737
738
if text_config is not config: