xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-01-26 16:14:32 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Bugfix] use blockmanagerv1 for encoder-decoder (#9084)

Browse Source 
Co-authored-by: Roger Wang <ywang@roblox.com>
This commit is contained in:
Chen Zhang 2024-10-04 21:45:38 -07:00 committed by GitHub
parent a95354a36e
commit dac914b0d6
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 5 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/engine/arg_utils.py
View File

@ -903,6 +903,11 @@ class EngineArgs:
"--enable-prefix-caching is currently not "
"supported for multimodal models and has been disabled.")
self.enable_prefix_caching = False
if model_config.is_encoder_decoder_model:
logger.warning(
"Block Manager v2 does not support encoder-decoder models"
" currently. Using Block Manager v1 as fallback.")
self.use_v2_block_manager = False
cache_config = CacheConfig(
block_size=self.block_size if self.device != "neuron" else