[XPU] Add gpt-oss model support for Intel GPU (#27786)

Signed-off-by: Kunshang Ji <kunshang.ji@intel.com>
This commit is contained in:
Kunshang Ji
2025-11-05 02:17:23 +00:00
committed by GitHub
parent 4ea62b77f5
commit 18b39828d9
4 changed files with 101 additions and 6 deletions
+2 -1
View File
@@ -27,6 +27,7 @@ from vllm.attention.utils.fa_utils import (
if is_flash_attn_varlen_func_available():
from vllm.attention.utils.fa_utils import (
flash_attn_supports_sinks,
flash_attn_varlen_func,
get_scheduler_metadata,
reshape_and_cache_flash,
@@ -497,7 +498,7 @@ class FlashAttentionImpl(AttentionImpl):
self.sinks = sinks
if self.sinks is not None:
assert self.vllm_flash_attn_version == 3, (
assert flash_attn_supports_sinks(), (
"Sinks are only supported in FlashAttention 3"
)
assert self.sinks.shape[0] == num_heads, (