[V1] Fix multimodal profiling for Molmo (#11325)

Signed-off-by: ywang96 <ywang@example.com> Co-authored-by: ywang96 <ywang@example.com>
2024-12-19 08:27:22 -08:00
parent 6c7f881541
commit 7379b3d4b2
4 changed files with 24 additions and 4 deletions
--- a/vllm/v1/worker/gpu_model_runner.py
+++ b/vllm/v1/worker/gpu_model_runner.py
@@ -638,7 +638,7 @@ class GPUModelRunner:
            # Compute MM hashes (if enabled)
            mm_hashes = None
            if self.use_hash:
-                mm_hashes = self.mm_hasher.hash_mm_data(dummy_mm_data)
+                mm_hashes = self.mm_hasher.hash_dummy_mm_data(dummy_mm_data)

            dummy_mm_kwargs = self.mm_input_mapper_client.process_inputs(
                mm_data=dummy_mm_data,