Support page size > 1 + eagle (#4908)

This commit is contained in:
Lianmin Zheng
2025-03-30 00:46:23 -07:00
committed by GitHub
parent 5ec5eaf760
commit b26bc86b36
16 changed files with 374 additions and 71 deletions

View File

@@ -1110,7 +1110,7 @@ class Scheduler(
)
if memory_leak:
msg = (
"KV cache pool leak detected! "
"token_to_kv_pool_allocator memory leak detected! "
f"{available_size=}, {protected_size=}, {self.max_total_num_tokens=}\n"
f"{self.token_to_kv_pool_allocator.available_size()=}\n"
f"{self.tree_cache.evictable_size()=}\n"
@@ -1121,7 +1121,7 @@ class Scheduler(
if len(self.req_to_token_pool.free_slots) != self.req_to_token_pool.size:
msg = (
"Memory pool leak detected!"
"req_to_token_pool memory leak detected!"
f"available_size={len(self.req_to_token_pool.free_slots)}, "
f"total_size={self.req_to_token_pool.size}\n"
)