List your company

Factual Question Answering on LSQA OOD

79.56PrecisionUALIGN
64.074468.094772.11576.1353Dec 16, 2024

Evaluation results

MethodLinks
UALIGNBackbone=Llama-3-8B2024.1279.5655.88
ICL-COTBackbone=Mistral-7B2024.1279.2452.59
ICLBackbone=Llama-3-8B2024.1277.3552.98
UALIGNBackbone=Mistral-7B2024.1276.2952.89
UALIGN (w/o Entro.)Backbone=Llama-3-8B2024.1276.1454.43
UALIGN (w/o Entro.)Backbone=Mistral-7B2024.1275.3450.15
RLKFBackbone=Llama-3-8B2024.1274.9552.46
RL-PPOBackbone=Mistral-7B2024.1274.4949.67
UALIGN (w/o Conf.)Backbone=Llama-3-8B2024.1274.2552.06
RLKFBackbone=Mistral-7B2024.1274.1949.23
ICLBackbone=Mistral-7B2024.1274.0947.71
ICL-COTBackbone=Llama-3-8B2024.1273.9649.37
UALIGN (w/o Conf.)Backbone=Mistral-7B2024.1273.1547.06
ICL-IDKBackbone=Mistral-7B2024.1272.2747.32
RL-PPOBackbone=Llama-3-8B2024.1272.1848.43
R-TuningBackbone=Llama-3-8B2024.1271.5452.15
RL-DPOBackbone=Llama-3-8B2024.1271.0948.88
ITIBackbone=Llama-3-8B2024.1268.5246.99
SFTBackbone=Mistral-7B2024.1268.3344
SFTBackbone=Llama-3-8B2024.1268.0946.63
RL-DPOBackbone=Mistral-7B2024.1267.8243.77
ITIBackbone=Mistral-7B2024.1267.7346.2
ICL-IDKBackbone=Llama-3-8B2024.1266.6750.24
R-TuningBackbone=Mistral-7B2024.1264.6744.05