Skip to content

Commit 572e8fb

Browse files
committed
update config
Signed-off-by: Amog Kamsetty <amogkamsetty@gmail.com>
1 parent 0a45661 commit 572e8fb

File tree

2 files changed

+4
-4
lines changed

2 files changed

+4
-4
lines changed

vllm/config.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -573,7 +573,7 @@ def __post_init__(self) -> None:
573573
sliding_window_pattern = getattr(self.hf_text_config,
574574
"sliding_window_pattern", None)
575575
has_interleaved_attention = sliding_window_pattern is not None or (
576-
isinstance(sliding_window, list))
576+
isinstance(sliding_window, list)) or hasattr(self.hf_text_config, "global_attention_layers")
577577

578578
if not self.disable_sliding_window and has_interleaved_attention:
579579
if (backend :=

vllm/model_executor/models/transformers.py

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -310,9 +310,9 @@ def create_attention_instances(self) -> dict[int, Attention]:
310310

311311
for i in range(start, end):
312312
sliding_window = None
313-
if i in global_attention_layers:
314-
assert self.config.sliding_window is not None
315-
sliding_window = self.config.sliding_window
313+
if i not in global_attention_layers:
314+
assert self.config.interleaved_sliding_window is not None
315+
sliding_window = self.config.interleaved_sliding_window
316316
attention_instances[i] = Attention(
317317
num_heads=num_heads,
318318
head_size=head_size,

0 commit comments

Comments
 (0)