[DLLM] Basic dLLM scheduling strategy and implementation (#17484)

Signed-off-by: Zehuan Li <lizehuan.lzh@antgroup.com>
2026-02-10 16:54:15 +08:00
parent 8da14aea88
commit 26f2b3798d
9 changed files with 461 additions and 210 deletions
@@ -95,7 +95,7 @@ class ForwardMode(IntEnum):
    # Split Prefill for PD multiplexing
    SPLIT_PREFILL = auto()

-    # Used in diffusion LLM inference
+    # Used in dLLM
    DLLM_EXTEND = auto()

    def is_prefill(self):