Warn users when release_memory_occupation is called without memory saver enabled (#4566)

2025-03-26 15:18:14 +08:00
parent 34e07a65f1
commit 26f07294f1
10 changed files with 50 additions and 12 deletions
--- a/test/srt/test_mla_int8_deepseek_v3.py
+++ b/test/srt/test_mla_int8_deepseek_v3.py
@@ -46,7 +46,7 @@ class TestMLADeepseekV3ChannelInt8(CustomTestCase):
        metrics = run_eval_few_shot_gsm8k(args)
        print(metrics)

-        self.assertGreater(metrics["accuracy"], 0.62)
+        self.assertGreaterEqual(metrics["accuracy"], 0.61)


 class TestDeepseekV3MTPChannelInt8(CustomTestCase):
--- a/test/srt/test_vision_openai_server.py
+++ b/test/srt/test_vision_openai_server.py
@@ -624,7 +624,6 @@ class TestMinicpmoServer(TestOpenAIVisionServer):
                "minicpmo",
                "--mem-fraction-static",
                "0.7",
-                "--tp=2",
            ],
        )
        cls.base_url += "/v1"