List your company

Multi-hop Question Answering on HotpotQA (val) (F1)

80F1 ScoreGTBP
75.16476.419577.67578.9305Jun 12, 2026

Evaluation results

MethodLinks
GTBPBackbone=GPT-4.1-mini, Number of intermediate modules K=tuned over {1, 2, 4}, Mini-batch size=15, Number of iterations=152026.0680
GEPABackbone=GPT-4.1-mini, Implementation=DSPy, Reflection mini-batch size=3, Metric-call budget=10002026.0676.97
Few-shotBackbone=GPT-4.1-mini, Shot mode=few-shot, Number of examples=tuned over {1, 3, 5, 10}2026.0676.59
Zero-shotBackbone=GPT-4.1-mini, Shot mode=zero-shot2026.0675.35