From db11fe1f9547a6a29a1c8b13483658fc6080067e Mon Sep 17 00:00:00 2001 From: "Wang, Yi A" Date: Sun, 29 Jun 2025 23:02:10 -0700 Subject: [PATCH] minor fix Signed-off-by: Wang, Yi A --- .../text_generation_server/models/custom_modeling/qwen2_5_vl.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/server/text_generation_server/models/custom_modeling/qwen2_5_vl.py b/server/text_generation_server/models/custom_modeling/qwen2_5_vl.py index a9cfc065..231d02b5 100644 --- a/server/text_generation_server/models/custom_modeling/qwen2_5_vl.py +++ b/server/text_generation_server/models/custom_modeling/qwen2_5_vl.py @@ -460,7 +460,7 @@ class Qwen2_5VLAttention(nn.Module): # execute flash attention if SYSTEM == "ipex": attn_output = torch.empty_like(query) - if query.device.dtype == "xpu": + if query.device.type == "xpu": ipex.llm.functional.varlen_attention( query.contiguous(), key.contiguous(),