Constrained-IR-Benchmark

10-domain retrieval test set. Each domain is a separate config with a test split.

Domains

config	samples
business_selection	25
company	25
course_selection	25
drug	25
enterprise_collaboration	25
phone	25
rental_house	25
restaurant	25
resume	25
vector_database	25

Load

from datasets import load_dataset

# 单个 domain
ds = load_dataset("WangSenQWQ/Constrained-IR-Benchmark", "phone", split="test")

# 遍历全部 domain 做实测
domains = ['business_selection', 'company', 'course_selection', 'drug', 'enterprise_collaboration', 'phone', 'rental_house', 'restaurant', 'resume', 'vector_database']
for name in domains:
    test = load_dataset("WangSenQWQ/Constrained-IR-Benchmark", name, split="test")
    # evaluate(test)

Fields

• domain: domain name
• query: query text
• pos: list of positive passages ({"text": ...})
• neg: list of negative passages ({"text": ...})