Skip to content

Commit 081ff45

Browse files
committed
add support for flash decoding on xpu
Signed-off-by: Liu, Kaixuan <kaixuan.liu@intel.com>
1 parent 8fe4f9d commit 081ff45

File tree

1 file changed

+1
-1
lines changed

1 file changed

+1
-1
lines changed

optimum/exporters/ipex/modeling_utils.py

+1-1
Original file line numberDiff line numberDiff line change
@@ -810,7 +810,7 @@ def attention_interface(
810810
query.contiguous() if query.device.type == "xpu" else query,
811811
key_cache,
812812
value_cache,
813-
seq_len_tensor if past_len == 0 else query_len_tensor,
813+
query_len_tensor,
814814
seq_len_tensor,
815815
query_max_len,
816816
max_input_lens,

0 commit comments

Comments
 (0)