List your company

Document-level Information Extraction on MUC English 4

58.32MUC ScoreTHINKTWICE
23.729632.709841.6950.6702Jan 26, 2026

Evaluation results

MethodLinks
THINKTWICEBackbone=Qwen 3, Selector=oracle, Protocol=Supervised2026.0158.32
THINKTWICEBackbone=Llama R1, Selector=oracle, Protocol=Supervised2026.0156.6
THINKTWICEBackbone=Qwen 3, Selector=Reward, Protocol=Supervised2026.0142.13
THINKTWICEBackbone=Llama R1, Selector=Reward, Protocol=Supervised2026.0141.99
THINKTWICEBackbone=Qwen 3, Selector=F1 Voting, Protocol=Supervised2026.0141.4
THINKTWICEBackbone=Llama R1, Selector=F1 Voting, Protocol=Supervised2026.0137.24
THINKTWICEBackbone=Qwen 3, Selector=Majority, Protocol=Supervised2026.0136.6
GreedyBackbone=Qwen 3, Selector=Greedy decoding, Protocol=Supervised2026.0135.97
ITERXBackbone=T5-enc large, Selector=None, Protocol=Supervised2026.0135.2
GTTBackbone=BERT base, Selector=None, Protocol=Supervised2026.0132.3
GreedyBackbone=Llama R1, Selector=Greedy decoding, Protocol=Supervised2026.0128.52
TEMPGENBackbone=BART large, Selector=None, Protocol=Supervised2026.0128.3
THINKTWICEBackbone=Llama R1, Selector=Majority, Protocol=Supervised2026.0125.06