biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

Update rope_scaling to rope_parameters in preparation for Transformers v5 (#28542)

Browse Source 
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Harry Mellor
2025-11-19 18:06:36 +01:00
committed by GitHub
parent d44e9df7d4
commit a8b70304d6
104 changed files with 542 additions and 910 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/model_executor/models/commandr.py
View File

@@ -156,8 +156,6 @@ class CohereAttention(nn.Module):
self.max_position_embeddings = getattr(
config, "model_max_length", None
) or getattr(config, "max_position_embeddings", 8192)
self.rope_theta = config.rope_theta
self.rope_scaling = getattr(config, "rope_scaling", None)
self.use_qk_norm = getattr(config, "use_qk_norm", False)
self.qkv_proj = QKVParallelLinear(
self.hidden_size,
@@ -179,8 +177,7 @@ class CohereAttention(nn.Module):
self.head_dim,
rotary_dim=self.head_dim,
max_position=self.max_position_embeddings,
base=self.rope_theta,
rope_scaling=self.rope_scaling,
rope_parameters=config.rope_parameters,
is_neox_style=False,
)