minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

Accelerate deepseek fp4 b200 ci (#11993)

Browse Source
This commit is contained in:
Qiaolin Yu
2025-10-25 15:29:14 -07:00
committed by GitHub
parent bcecf27e7c
commit d98b81e260
1 changed files with 5 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
test/srt/test_deepseek_v3_fp4_4gpu.py
+5
View File
@@ -32,6 +32,8 @@ class TestDeepseekV3FP4(CustomTestCase):
"flashinfer_trtllm", "flashinfer_trtllm",
"--quantization", "--quantization",
"modelopt_fp4", "modelopt_fp4",
"--kv-cache-dtype",
"fp8_e4m3",
] ]
cls.process = popen_launch_server( cls.process = popen_launch_server(
cls.model, cls.model,
@@ -100,6 +102,9 @@ class TestDeepseekV3FP4MTP(CustomTestCase):
"1", "1",
"--speculative-num-draft-tokens", "--speculative-num-draft-tokens",
"4", "4",
"--kv-cache-dtype",
"fp8_e4m3",
"--enable-beta-spec",
] ]
cls.process = popen_launch_server( cls.process = popen_launch_server(
cls.model, cls.model,