20231088/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

[Bugfix] Revert max_prompt_len validation for decoder-only models. (#16741)

Browse Source 
Signed-off-by: David Heineman <david@davidheineman.com>
This commit is contained in:
David Heineman 2025-04-16 21:33:15 -07:00 committed by GitHub
parent cb072ce93b
commit 607029e515
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/engine/llm_engine.py
View File

@ -2062,7 +2062,7 @@ class LLMEngine:
raise ValueError(f"The {prompt_type} prompt cannot be empty")
max_prompt_len = self.model_config.max_model_len
if len(prompt_ids) >= max_prompt_len:
if len(prompt_ids) > max_prompt_len:
if prompt_type == "encoder" and model_config.is_multimodal_model:
mm_registry = self.input_preprocessor.mm_registry
mm_processor = mm_registry.create_processor(

2
vllm/v1/engine/processor.py
View File

@ -354,7 +354,7 @@ class Processor:
raise ValueError(f"Token id {max_input_id} is out of vocabulary")
max_prompt_len = self.model_config.max_model_len
if len(prompt_ids) >= max_prompt_len:
if len(prompt_ids) > max_prompt_len:
if prompt_type == "encoder" and model_config.is_multimodal_model:
mm_registry = self.input_preprocessor.mm_registry
mm_processor = mm_registry.create_processor(