List your company

Multi-turn Strategic Gameplay on Multi-turn Game Suite

54.9SimpleNegotiation Win RateMEMO
22.55630.95339.3547.747Mar 9, 2026

Evaluation results

MethodLinks
MEMOBackbone=GPT-4o-mini, Type=OURS2026.0354.952.455.642.741.849.56.4
MEMOBackbone=Qwen2.5-7B-Instruct, Type=OURS2026.034848.46031.13444.36.1
SPIRALBackbone=Qwen2.5-7B-Instruct, Type=RL2026.0345.7—56.7—32.7——
MIPROBackbone=Qwen2.5-7B-Instruct, Type=PROMPT2026.0342.447.553.82.220.933.47.3
TextGradBackbone=GPT-4o-mini, Type=PROMPT2026.034244.655.67.123.634.618.4
UnstableBaselineBackbone=Qwen2.5-7B-Instruct, Type=RL2026.0341.130.452.753.347.34543.3
MIPROBackbone=GPT-4o-mini, Type=PROMPT2026.0338.450.955.119.719.136.712.4
TextGradBackbone=Qwen2.5-7B-Instruct, Type=PROMPT2026.0337.129.352.87.122.429.921.7
GEPABackbone=GPT-4o-mini, Type=PROMPT2026.0336.840.452.23.326.93211.3
GEPABackbone=Qwen2.5-7B-Instruct, Type=PROMPT2026.0334.431.755.83.319.328.814.8
baselineBackbone=GPT-4o-mini, Type=STATIC2026.0331.332.239.10.321.425.144.9
CoTBackbone=GPT-4o-mini, Type=STATIC2026.0327.825.746.530.424.831.128.7
ToTBackbone=Qwen2.5-7B-Instruct, Type=STATIC2026.0327.120.742.222.715.125.640.2
ToTBackbone=GPT-4o-mini, Type=STATIC2026.0326.32751.745.123.834.836.5
baselineBackbone=Qwen2.5-7B-Instruct, Type=STATIC2026.032417.145.32.815.120.930.1
CoTBackbone=Qwen2.5-7B-Instruct, Type=STATIC2026.0323.818.74225.813.624.843.4