[Bugfix] Pass quant_config to DiffusionGemma's ParallelLMHead (#48521)
Signed-off-by: Aaron Kang <aaron.h.kang@icloud.com> Co-authored-by: Misha Goin <mgoin64@gmail.com>
Aaron Kang committed
Sep 24, 2026 at 16:17 UTC
0908116dd9e27766efdd46f37a108c5a6b7ade53
1 file changed
+2
vllm/model_executor/models/diffusion_gemma.py
+2
@@ -250,6 +250,8 @@ class DiffusionGemmaForConditionalGeneration(
250
self.lm_head = ParallelLMHead(
251
num_embeddings=text_config.vocab_size,
252
embedding_dim=text_config.hidden_size,
253
+ quant_config=vllm_config.quant_config,
254
+ prefix=maybe_prefix(prefix, "lm_head"),
255
)
256
257
if text_config.tie_word_embeddings: