[Model] Migrate Glm, Arcee, CWM and Mellum to the Transformers modeling backend (#59679)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
co-authored by
Claude Opus 5.5
parent
ea106db806
commit
8faacd707f
@@ -572,7 +572,6 @@ class GteNewModel(BertWithRope):
|
||||
super().__init__(vllm_config=vllm_config, prefix=prefix, **kwargs)
|
||||
|
||||
# GteNewModel only gate_up_proj does not have bias.
|
||||
# Hack method learned from vllm/model_executor/models/glm.py
|
||||
for layer in self.encoder.layers:
|
||||
layer.mlp.gate_up_proj.bias = None
|
||||
layer.mlp.gate_up_proj.skip_bias_add = True
|
||||
|
||||
Reference in New Issue
Block a user