Revert "Fix memory leak when doing chunked prefill" (#1797)

This commit is contained in:
Lianmin Zheng
2024-10-25 10:24:44 -07:00
committed by GitHub
parent 40900baea7
commit c555ce2ca2
6 changed files with 69 additions and 183 deletions

View File

@@ -663,7 +663,6 @@ def run_mmlu_test(
chunked_prefill_size=32,
):
other_args = ["--chunked-prefill-size", str(chunked_prefill_size)]
other_args += ["--mem-fraction-static", "0.85"]
if disable_radix_cache:
other_args += ["--disable-radix-cache"]
if enable_mixed_chunk: