From 7d2d6624b1ee1c857320f36429758f421ebb4067 Mon Sep 17 00:00:00 2001 From: EdwardXuy Date: Tue, 8 Sep 2026 10:39:58 +0800 Subject: [PATCH] [NPU] Enable L1 prefix cache for Kimi-K3 W4A8 accuracy test (gpqa) (#38125) Co-authored-by: EdwardXuy --- .../npu/accuracy/kimi_k3/test_npu_kimi_k3_w4a8_32p_gpqa.py | 1 - 1 file changed, 1 deletion(-) diff --git a/test/registered/npu/accuracy/kimi_k3/test_npu_kimi_k3_w4a8_32p_gpqa.py b/test/registered/npu/accuracy/kimi_k3/test_npu_kimi_k3_w4a8_32p_gpqa.py index b3d8ae60d..319537dec 100644 --- a/test/registered/npu/accuracy/kimi_k3/test_npu_kimi_k3_w4a8_32p_gpqa.py +++ b/test/registered/npu/accuracy/kimi_k3/test_npu_kimi_k3_w4a8_32p_gpqa.py @@ -96,7 +96,6 @@ KIMI_K3_W4A8_32P_OTHER_ARGS = [ "triton", "--speculative-eagle-topk", 1, - "--disable-radix-cache", "--disable-custom-all-reduce", "--watchdog-timeout", 9000,