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

4
vllm/model_executor/models/zamba2.py
View File

@@ -128,7 +128,6 @@ class Zamba2Attention(nn.Module):
tp_size = get_tensor_model_parallel_world_size()
self.config = config
self.num_hybrid_layers = num_hybrid_layers
self.rope_theta = config.rope_theta
self.attention_hidden_size = config.attention_hidden_size
self.total_num_attention_heads = config.num_attention_heads
@@ -233,8 +232,7 @@ class Zamba2Attention(nn.Module):
head_size=self.attention_head_dim,
rotary_dim=self.attention_head_dim,
max_position=config.max_position_embeddings,
base=self.rope_theta,
rope_scaling=None,
rope_parameters=config.rope_parameters,
is_neox_style=True,
)