[Model] Mamba2 causal conv1d Refactor to Split Prefill and Decode Requests for Corresponding Kernels (#17146)

Signed-off-by: Chih-Chieh-Yang <7364402+cyang49@users.noreply.github.com>
2025-05-06 20:59:30 -04:00
parent 6de3e13413
commit 18dd5e01f2
8 changed files with 151 additions and 123 deletions
--- a/vllm/model_executor/models/granitemoehybrid.py
+++ b/vllm/model_executor/models/granitemoehybrid.py
@@ -338,7 +338,6 @@ class GraniteMoeHybridModel(nn.Module):
        attn_metadata = get_forward_context().attn_metadata
        mamba2_metadata = prepare_mamba2_metadata(
            chunk_size=self.config.mamba_chunk_size,
-            input_ids=input_ids,
            attn_metadata=attn_metadata,
        )