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

[Bugfix][Hardware][CPU] Fix mm_limits initialization for CPU backend (#7735)

Browse Source
This commit is contained in:
Isotr0py 2024-08-22 00:23:03 +08:00 committed by GitHub
parent 53328d7536
commit 6925cdbeea
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 3 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
vllm/worker/cpu_model_runner.py
View File

@ -113,8 +113,10 @@ class CPUModelRunner(ModelRunnerBase[CPUModelInput]):
) )
# Multi-modal data support # Multi-modal data support
self.multi_modal_input_mapper = MULTIMODAL_REGISTRY \ self.mm_registry = MULTIMODAL_REGISTRY
self.multi_modal_input_mapper = self.mm_registry \
.create_input_mapper(self.model_config) .create_input_mapper(self.model_config)
self.mm_registry.init_mm_limits_per_prompt(self.model_config)
# Lazy initialization. # Lazy initialization.
self.model: nn.Module # Set after init_Model self.model: nn.Module # Set after init_Model