List your company

Factual Question Answering on SciQ

76.44PrecisionUALIGN
63.23266.66170.0973.519Dec 16, 2024

Evaluation results

MethodLinks
UALIGNBackbone=Llama-3-8B2024.1276.4448
RL-PPOBackbone=Llama-3-8B2024.1275.745.8
UALIGN (w/o Entro.)Backbone=Llama-3-8B2024.1275.3947.5
UALIGN (w/o Conf.)Backbone=Llama-3-8B2024.1274.0545
UALIGNBackbone=Mistral-7B2024.1273.2146.7
UALIGN (w/o Entro.)Backbone=Mistral-7B2024.1272.945.6
RLKFBackbone=Llama-3-8B2024.1272.3644.9
ITIBackbone=Llama-3-8B2024.1272.3543.8
ICL-COTBackbone=Llama-3-8B2024.1272.3445.9
SFTBackbone=Llama-3-8B2024.1272.1845.4
ICL-COTBackbone=Mistral-7B2024.1271.8744.2
R-TuningBackbone=Llama-3-8B2024.1271.3844.9
RL-PPOBackbone=Mistral-7B2024.1271.3544.1
RL-DPOBackbone=Llama-3-8B2024.1271.2344.2
RLKFBackbone=Mistral-7B2024.1270.6643.9
ICLBackbone=Llama-3-8B2024.1270.4344.3
UALIGN (w/o Conf.)Backbone=Mistral-7B2024.1269.8443.5
ICLBackbone=Mistral-7B2024.1268.6242.2
ICL-IDKBackbone=Llama-3-8B2024.1268.3643
ITIBackbone=Mistral-7B2024.1266.944.9
RL-DPOBackbone=Mistral-7B2024.1266.4441.8
SFTBackbone=Mistral-7B2024.1265.8542.5
R-TuningBackbone=Mistral-7B2024.1264.4440.1
ICL-IDKBackbone=Mistral-7B2024.1263.7439.2