ResearchBenchmarksTask-oriented Dialogue on FramesFollow50.57Success Rate (SR)VLK-RL25.204431.789738.37544.9603Apr 25, 2026Evaluation ResultsMethodMethodLinksSuccess Rate (SR)Hit Rate (HR)VLK-RLbackbone=Qwen-14Bbackbone=Qwen-14B2026.0450.573.32VLK-RLbackbone=Qwen-7Bbackbone=Qwen-7B2026.0448.233.16VLK-RLbackbone=GPT-4o-minibackbone=GPT-4o-mini2026.0445.893.11CAPID2026.0439.272.51GDP-Zero2026.0438.962.49TransferTOD2026.0437.412.45GALAXY2026.0436.122.34ACGOS2026.0431.852.22PPO2026.0426.182.08