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 14, 05:30 UTC
[fix] fix correct assertion syntax error in attention utils. (#22154)
Signed-off-by: zitian.zhao <zitian.zhao@tencentmusic.com>
ZiTian.Zhao committed
Aug 4, 2025 at 10:24 UTC
e27d25a0dcbb71a0d2e2a27d7e2b606a8df30320
1 file changed
+3
-1
vllm/v1/attention/backends/utils.py
+3
-1
@@ -97,7 +97,9 @@ def _make_metadata_with_slice(
97
98
query_start_loc = slice_query_start_locs(attn_metadata.query_start_loc,
99
request_slice)
100
- assert len(query_start_loc >= 2)
100
+ assert len(query_start_loc) >= 2, (
101
+ f"query_start_loc must have at least 2 elements, "
102
+ f"got {len(query_start_loc)}")
103
query_start_loc_cpu = slice_query_start_locs(
104
attn_metadata.query_start_loc_cpu, request_slice)
105