Feat/support rerank (#6058)

This commit is contained in:
woodx
2025-06-17 01:50:01 +08:00
committed by GitHub
parent 91a066ec6a
commit e30ef368ab
20 changed files with 684 additions and 30 deletions

View File

@@ -41,6 +41,8 @@ DEFAULT_MOE_MODEL_NAME_FOR_TEST = "mistralai/Mixtral-8x7B-Instruct-v0.1"
DEFAULT_SMALL_MOE_MODEL_NAME_FOR_TEST = "Qwen/Qwen1.5-MoE-A2.7B"
# MLA test models
DEFAULT_SMALL_EMBEDDING_MODEL_NAME_FOR_TEST = "Alibaba-NLP/gte-Qwen2-1.5B-instruct"
DEFAULT_SMALL_CROSS_ENCODER_MODEL_NAME_FOR_TEST = "cross-encoder/ms-marco-MiniLM-L6-v2"
DEFAULT_MLA_MODEL_NAME_FOR_TEST = "deepseek-ai/DeepSeek-Coder-V2-Lite-Instruct"
DEFAULT_MLA_FP8_MODEL_NAME_FOR_TEST = "neuralmagic/DeepSeek-Coder-V2-Lite-Instruct-FP8"
DEFAULT_MODEL_NAME_FOR_TEST_MLA = "lmsys/sglang-ci-dsv3-test"