From 564050766de85422e13518043470042e90f5b972 Mon Sep 17 00:00:00 2001 From: Yineng Zhang Date: Wed, 17 Sep 2025 14:34:58 -0700 Subject: [PATCH] fix: update dsv3 fp4 ut (#10584) --- test/srt/test_deepseek_v3_fp4_4gpu.py | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/test/srt/test_deepseek_v3_fp4_4gpu.py b/test/srt/test_deepseek_v3_fp4_4gpu.py index bb9e13554..657c0cf9c 100644 --- a/test/srt/test_deepseek_v3_fp4_4gpu.py +++ b/test/srt/test_deepseek_v3_fp4_4gpu.py @@ -141,8 +141,8 @@ class TestDeepseekV3FP4MTP(CustomTestCase): f'{metrics["accuracy"]=:.3f}\n' f"{avg_spec_accept_length=:.2f}\n" ) - self.assertGreater(metrics["accuracy"], 0.935) - self.assertGreater(avg_spec_accept_length, 2.9) + self.assertGreater(metrics["accuracy"], 0.94) + self.assertGreater(avg_spec_accept_length, 2.04) def test_bs_1_speed(self): args = BenchArgs(port=int(self.base_url.split(":")[-1]), max_new_tokens=2048) @@ -156,8 +156,8 @@ class TestDeepseekV3FP4MTP(CustomTestCase): f"{acc_length=:.2f}\n" f"{speed=:.2f} token/s\n" ) - self.assertGreater(acc_length, 2.9) - self.assertGreater(speed, 130) + self.assertGreater(acc_length, 2.04) + self.assertGreater(speed, 150) class TestDeepseekV3FP4CutlassMoE(CustomTestCase):