Large page size aligned hierarchical caching (#4581)

2025-04-01 22:38:15 -07:00
parent 9eb49e878b
commit e119f04215
8 changed files with 242 additions and 71 deletions
--- a/python/sglang/srt/managers/scheduler.py
+++ b/python/sglang/srt/managers/scheduler.py
@@ -1282,7 +1282,7 @@ class Scheduler(
        ]

        if self.enable_hierarchical_cache:
-            self.tree_cache.read_to_load_cache()
+            self.tree_cache.ready_to_load_cache()

        if adder.new_chunked_req is not None:
            assert self.chunked_req is None