Skip to content

GitLab

Unverified Commit 8085aca7 authored by alanhe151220037's avatar alanhe151220037 Committed by GitHub
Browse files

[Bug fix] Fix ascend mla in aclgraph (#9925)

parent 0096798e
Show whitespace changes
Inline Side-by-side
Showing with 1 addition and 1 deletion
python/sglang/srt/layers/attention/ascend_backend.py
View file @ 8085aca7
...@@ -368,7 +368,7 @@ class AscendAttnBackend(AttentionBackend): ...@@ -368,7 +368,7 @@ class AscendAttnBackend(AttentionBackend):
-1, layer.tp_v_head_num, self.page_size, self.kv_lora_rank -1, layer.tp_v_head_num, self.page_size, self.kv_lora_rank
) )
q_nope = q.view(-1, layer.tp_q_head_num, 1, self.kv_lora_rank) q_nope = q.view(-1, layer.tp_q_head_num, 1, self.kv_lora_rank).contiguous()
q_rope = q_rope.view(-1, layer.tp_q_head_num, 1, self.qk_rope_head_dim) q_rope = q_rope.view(-1, layer.tp_q_head_num, 1, self.qk_rope_head_dim)
if self.forward_metadata.seq_lens_cpu_int is None: if self.forward_metadata.seq_lens_cpu_int is None:
actual_seq_len_kv = self.forward_metadata.seq_lens_cpu_list actual_seq_len_kv = self.forward_metadata.seq_lens_cpu_list
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment