List your company

Multi-turn role-play on ICLR

96.2Success Rate (SR)EmoLLM
82.05685.72889.493.072Mar 17, 2026

Evaluation results

MethodLinks
EmoLLMBackbone=Qwen3-8B, Method Category=Ours2026.0396.21.214.213.95
PERMBackbone=Qwen3-8B, Method Category=RL Methods2026.0388.31.264.053.88
RLVERBackbone=Qwen3-8B, Method Category=RL Methods2026.0386.61.334.023.81
ICL-AIFBackbone=Qwen3-8B, Method Category=Prompting Methods2026.0386.41.353.993.86
ARG-PromptBackbone=Qwen3-8B, Method Category=Prompting Methods2026.03861.393.953.82
RLFF-ESCBackbone=Qwen3-8B, Method Category=RL Methods2026.03861.393.953.78
PPDPBackbone=Qwen3-8B, Method Category=SFT Methods2026.0385.31.423.93.79
Best-of-NBackbone=Qwen3-8B, Method Category=Prompting Methods2026.0384.91.713.773.64
Ask-an-ExpertBackbone=Qwen3-8B, Method Category=Prompting Methods2026.0384.21.583.853.66
ESCoTBackbone=Qwen3-8B, Method Category=SFT Methods2026.0384.11.473.863.73
Qwen3-8BBackbone=Qwen3-8B, Method Category=Base Model2026.0383.41.643.793.83
SFTBackbone=Qwen3-8B, Method Category=SFT Methods2026.0382.61.363.733.76