List your company

Symbolic Reasoning on Coin Flip (Accuracy, Average Accuracy)

80.75AccuracyRM-Primed
58.1364.002569.87575.7475Mar 20, 2026

Evaluation results

MethodLinks
RM-PrimedModel=Llama3-8B2026.0380.7568.23
RM-PrimedModel=GPT-3.52026.0379.7566.01
RM-Primed (R+)Model=Llama3-8B, R+ selection=only correct entries used2026.0378.7565.94
RM-Primed (R+)Model=GPT-3.5, R+ selection=only correct entries used2026.037864.99
Few-shot CoTModel=GPT-3.52026.0374.2561.15
Contrastive CoTModel=GPT-3.5, Reflection=with2026.0369.2560.33
Contrastive CoTModel=Llama3-8B, Reflection=with2026.0366.2561.13
Few-shot CoTModel=Llama3-8B2026.0365.7564.19
Contrastive CoTModel=Llama3-8B, Reflection=without2026.0360.7559.56
Contrastive CoTModel=GPT-3.5, Reflection=without2026.035957.01