From 349b3f272a89307aa37b8646c6dc3a7220f29fd1 Mon Sep 17 00:00:00 2001 From: Flavia Beo Date: Wed, 23 Oct 2024 12:12:42 -0300 Subject: [PATCH] Reverts paged_attn changes Signed-off-by: Flavia Beo --- vllm/attention/ops/paged_attn.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/vllm/attention/ops/paged_attn.py b/vllm/attention/ops/paged_attn.py index 076f151ffcb61..92023d5b75f5a 100644 --- a/vllm/attention/ops/paged_attn.py +++ b/vllm/attention/ops/paged_attn.py @@ -34,7 +34,7 @@ class PagedAttention: @staticmethod def get_supported_head_sizes() -> List[int]: - return [32, 64, 80, 96, 112, 120, 128, 192, 256] + return [64, 80, 96, 112, 120, 128, 192, 256] @staticmethod def get_kv_cache_shape(