xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-14 00:45:36 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Misc] Use RoPE cache for MRoPE (#8396)

Browse Source
This commit is contained in:
Woosuk Kwon 2024-09-11 23:13:14 -07:00 committed by GitHub
parent 295c4730a8
commit 42ffba11ad
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/model_executor/layers/rotary_embedding.py
View File

@ -990,7 +990,7 @@ def get_rope(
base, is_neox_style, dtype, short_factor, long_factor, base, is_neox_style, dtype, short_factor, long_factor,
**extra_kwargs) **extra_kwargs)
elif scaling_type == "mrope": elif scaling_type == "mrope":
return MRotaryEmbedding( rotary_emb = MRotaryEmbedding(
head_size, head_size,
rotary_dim, rotary_dim,
max_position, max_position,