[bugfix] fix pass bug: pass really rope dim for npu_rotary_embedding (#6880)
### What this PR does / why we need it?
pass really rope dim for npu_rotary_embedding
**before:**
q_rope, k_rope = torch.ops.vllm.npu_rotary_embedding(
positions, q_flat, k_flat, cos_sin_cache, self.head_dim,
**self.head_dim,** True
)
**after:**
q_rope, k_rope = torch.ops.vllm.npu_rotary_embedding(
positions, q_flat, k_flat, cos_sin_cache, self.head_dim,
**self.rope_dim,** True
)
### Does this PR introduce _any_ user-facing change?
no
### How was this patch tested?
- vLLM version: v0.16.0
- vLLM main:
https://github.com/vllm-project/vllm/commit/15d76f74e2fdb12a95ea00f0ca283acf6219a2b7
---------
Signed-off-by: zjks98 <zhangjiakang4@huawei.com>
Signed-off-by: aipaes <82140963+aipaes@users.noreply.github.com>
Co-authored-by: zjks98 <zhangjiakang4@huawei.com>