[Bugfix] Fix unsupported FA version check for Turing GPU (#12828)

8108ac84 · Isotr0py · GitHub · afe74f7a · 8108ac84
Unverified Commit 8108ac84 authored Feb 07, 2025 by Isotr0py Committed by GitHub Feb 06, 2025
Show whitespace changes
Inline Side-by-side

Showing with 1 addition and 1 deletion

vllm/attention/backends/utils.py vllm/attention/backends/utils.py +1 -1

No files found.
--- a/vllm/attention/backends/utils.py
+++ b/vllm/attention/backends/utils.py
@@ -612,5 +612,5 @@ try:
        return fa_version

    VLLM_FLASH_ATTN_VERSION = flash_attn_version()
-except ImportError:
+except (ImportError, AssertionError):
    VLLM_FLASH_ATTN_VERSION = None