[Core] Use individual MM items in P0/P1 cache and model runner (#22570)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
2025-08-13 22:18:07 +08:00
parent 20d65aa755
commit 19b927e52d
24 changed files with 549 additions and 486 deletions
--- a/tests/v1/kv_connector/unit/utils.py
+++ b/tests/v1/kv_connector/unit/utils.py
@@ -154,7 +154,7 @@ def create_request(
        prompt_token_ids=prompt_token_ids,
        sampling_params=sampling_params,
        pooling_params=None,
-        multi_modal_inputs=None,
+        multi_modal_kwargs=None,
        multi_modal_placeholders=None,
        multi_modal_hashes=None,
        eos_token_id=EOS_TOKEN_ID,