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 16, 00:45 UTC
[Bugfix] Fix unsupported FA version check for Turing GPU (#12828)
Isotr0py committed
Feb 7, 2025 at 01:18 UTC
8108ac841d66515b58252edc26ba63da6cf980e5
1 file changed
+1
-1
vllm/attention/backends/utils.py
+1
-1
@@ -612,5 +612,5 @@ try:
612
return fa_version
613
614
VLLM_FLASH_ATTN_VERSION = flash_attn_version()
615
-except ImportError:
615
+except (ImportError, AssertionError):
616
VLLM_FLASH_ATTN_VERSION = None