[Feature] Support mtp overlap schedule (#7001)

2026-04-23 00:17:25 +08:00 · 2026-04-01 14:24:26 +08:00
parent c6f0c5c3a6
commit c29e86fc9d
23 changed files with 215 additions and 138 deletions
@@ -573,7 +573,7 @@ class EngineArgs:
            self.enable_prefix_caching = False
        if (
            not current_platform.is_cuda()
-            or self.speculative_config is not None
+            or (self.speculative_config is not None and self.enable_logprob)
            or self.splitwise_role == "prefill"
            or self.dynamic_load_weight
        ):