[CI] upgrade to newest pta (#187)

Upgrade to newest torch-npu Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com> Co-authored-by: angazenn <zengyanjia@huawei.com>
2025-02-27 16:40:23 +08:00
parent fd18ae6494
commit 6042c210bc
4 changed files with 27 additions and 56 deletions
--- a/vllm_ascend/ops/rotary_embedding.py
+++ b/vllm_ascend/ops/rotary_embedding.py
@@ -42,7 +42,7 @@ def rope_forward_oot(
        # TODO: Remove the contiguous in the future.
        query = query.contiguous()
        key = key.contiguous()
-        torch_npu.npu_rope(
+        torch_npu._npu_rotary_embedding(
            positions,
            query,
            key,