EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

[CI]Add performance CI for VLM (#6038)

Browse Source 
Signed-off-by: Xinyuan Tong <justinning0323@outlook.com>
This commit is contained in:
XinyuanTong
2025-05-07 19:20:03 -07:00
committed by GitHub
parent 73600673bb
commit e88dd482ed
5 changed files with 219 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
python/sglang/test/test_utils.py
View File

@@ -79,7 +79,8 @@ DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_FP8_TP1 = "neuralmagic/Meta-Llama-3.1-8B-Ins
DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_FP8_TP2 = "neuralmagic/Meta-Llama-3.1-70B-Instruct-FP8,neuralmagic/Mixtral-8x7B-Instruct-v0.1-FP8,neuralmagic/Qwen2-72B-Instruct-FP8,neuralmagic/Qwen2-57B-A14B-Instruct-FP8,neuralmagic/DeepSeek-Coder-V2-Lite-Instruct-FP8"
DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_QUANT_TP1 = "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4,hugging-quants/Meta-Llama-3.1-8B-Instruct-GPTQ-INT4,hugging-quants/Mixtral-8x7B-Instruct-v0.1-AWQ-INT4"
DEFAULT_SMALL_MODEL_NAME_FOR_TEST_QWEN = "Qwen/Qwen2.5-1.5B-Instruct"
DEFAULT_SMALL_VLM_MODEL_NAME = "Qwen/Qwen2-VL-2B"
DEFAULT_SMALL_VLM_MODEL_NAME_FOR_TEST = "Qwen/Qwen2.5-VL-3B-Instruct"
DEFAULT_VLM_CHAT_TEMPLATE_FOR_TEST = "qwen2-vl"
DEFAULT_IMAGE_URL = "https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true"
DEFAULT_VIDEO_URL = "https://raw.githubusercontent.com/EvolvingLMMs-Lab/sglang/dev/onevision_local/assets/jobs.mp4"