List your company

Geopolitical Event Forecasting on POLECAT-FOR-ME Llama 3.1 knowledge cut-off compliant (test)

22.2ISRGPT-4o
10.65613.65316.6519.647Jan 22, 2026

Evaluation results

MethodLinks
GPT-4oType=Proprietary LLM2026.0122.217.315.337.918.615.620.815.823.524
GPT-4o-miniType=Proprietary LLM2026.0120.321.315.337.217.716.319.214.122.722.5
ThinkTank-ME (Elite Ensemble - Weighted Best-of-N)Backbone=Llama-3.1-8B, Leader Strategy=Elite Ensemble, Variant=Weighted Best-of-N2026.0118.315.714.451.623.426.911.517.524.625
ThinkTank-ME (Wisdom Aggregation - Weighted Best-of-N)Backbone=Llama-3.1-8B, Leader Strategy=Wisdom Aggregation, Variant=Weighted Best-of-N2026.011813.914.450.722.226.610.917.523.924.4
ThinkTank-ME (Elite Ensemble - Majority Voting)Backbone=Llama-3.1-8B, Leader Strategy=Elite Ensemble, Variant=Majority Voting2026.011812.813.550.722.626.911.21723.924.2
ThinkTank-ME (Wisdom Aggregation - Majority Voting)Backbone=Llama-3.1-8B, Leader Strategy=Wisdom Aggregation, Variant=Majority Voting2026.0117.711.913.950.421.626.911.218.123.623.9
ThinkTank-ME (Elite Ensemble - Vanilla Best-of-N)Backbone=Llama-3.1-8B, Leader Strategy=Elite Ensemble, Variant=Vanilla Best-of-N2026.0117.714.111.850.919.224.8816.122.823.9
Baselines (All-data Training)Training=All-data2026.0117.410.716.249.919.325.38.515.822.722.6
ThinkTank-ME (Wisdom Aggregation - Vanilla Best-of-N)Backbone=Llama-3.1-8B, Leader Strategy=Wisdom Aggregation, Variant=Vanilla Best-of-N2026.0116.313.19.648.418.219.6915.521.222.2
ThinkTank-ME (Expert Routing)Backbone=Llama-3.1-8B, Leader Strategy=Expert Routing2026.0113.98.411.43915.623.710.713.518.619.4
Baselines (No Training)Training=None2026.0111.17.89.63417.812.410.18.315.515.7