init src 0.9.2

vllm/zero_overhead/v1/outputs.py

@@ -0,0 +1,14 @@
+
+
+
+from dataclasses import dataclass
+from vllm.v1.outputs import ModelRunnerOutput
+
+@dataclass
+class ZeroV1ModelRunnerOutput(ModelRunnerOutput):
+    # [num_reqs]
+    fix_req_ids: list[str] = None
+    fix_sampled_token_ids:list[list[int]] = None
+    fix_draft_req_ids:list[str] = None
+    fix_draft_tokens_ids:list[list[int]] = None
+    is_output_valid:bool = True