biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[V1] V1 Enablement Oracle (#13726)

Browse Source 
Signed-off-by: rshaw@neuralmagic.com <rshaw@neuralmagic.com>
Co-authored-by: rshaw@neuralmagic.com <rshaw@neuralmagic.com>
Co-authored-by: Nicolò Lucchesi <nlucches@redhat.com>
Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>
Co-authored-by: Michael Goin <michael@neuralmagic.com>
This commit is contained in:
Robert Shaw
2025-03-15 01:02:20 -04:00
committed by GitHub
parent 8c0d15d5c5
commit d4d93db2c5
96 changed files with 1537 additions and 512 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
vllm/v1/attention/backends/flash_attn.py
View File

@@ -196,7 +196,8 @@ class FlashAttentionImpl(AttentionImpl):
if head_size not in support_head_sizes:
raise ValueError(
f"Head size {head_size} is not supported by FlashAttention. "
f"Supported head sizes are: {support_head_sizes}.")
f"Supported head sizes are: {support_head_sizes}. "
"Set VLLM_USE_V1=0 to use another attention backend.")
if attn_type != AttentionType.DECODER:
raise NotImplementedError("Encoder self-attention and "