Fix pipeline parallel with multimodal models with the Transformers modelling backend (#37057)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Harry Mellor
2026-03-16 10:20:37 +00:00
committed by GitHub
parent d8f8a7aad2
commit 122f75d939
2 changed files with 31 additions and 8 deletions
@@ -38,7 +38,7 @@ class CausalMixin(VllmModelForTextGeneration):
# Tell `Base.load_weights` to skip
# `lm_head` if the model has tied word embeddings
tie_word_embeddings = getattr(self.text_config, "tie_word_embeddings", False)
tie_word_embeddings = self._get_tie_word_embeddings()
if tie_word_embeddings:
self.skip_prefixes.append("lm_head.")