From a39c428d3f01b8ca2eaaef2aa9dd4aed626009ee Mon Sep 17 00:00:00 2001 From: Zhangheng Date: Sun, 7 Jun 2026 17:12:42 +0800 Subject: [PATCH] [UnifiedTree][CI]: Reduce HiCache PP KL test concurrency to avoid decode OOM (#27483) --- .../test_unified_radix_cache_hicache_pp_kl.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/test/registered/radix_cache/unified_radix_tree/test_unified_radix_cache_hicache_pp_kl.py b/test/registered/radix_cache/unified_radix_tree/test_unified_radix_cache_hicache_pp_kl.py index edfb56dd2..6e6068404 100644 --- a/test/registered/radix_cache/unified_radix_tree/test_unified_radix_cache_hicache_pp_kl.py +++ b/test/registered/radix_cache/unified_radix_tree/test_unified_radix_cache_hicache_pp_kl.py @@ -32,7 +32,7 @@ class TestUnifiedQwen3HiCachePP(UnifiedRadixTreeTestMixin, CustomTestCase): hicache_io_backend = "direct" hicache_mem_layout = "page_first_direct" - max_running_requests = 4 + max_running_requests = 2 kl_threshold = 0.012 gsm8k_threshold = 0.7 num_gsm8k_questions = 50