List your company

Search-augmented Multi-turn Tool Calling on HotpotQA (Success Rate)

38.32Success RateENVRL-GiGPO
37.924838.027438.1338.2326Jun 16, 2026

Evaluation results

MethodLinks
ENVRL-GiGPOBackbone=Qwen2.5-3B-Instruct2026.0638.32
GiGPOBackbone=Qwen2.5-3B-Instruct2026.0637.94