List your company

Factual Question Answering on NQ-Open ID

57.34PrecisionICL-COT
40.772845.073949.37553.6761Dec 16, 2024

Evaluation results

MethodLinks
ICL-COTBackbone=Llama-3-8B2024.1257.3423.6
UALIGN (w/o Entro.)Backbone=Llama-3-8B2024.1256.6827.56
UALIGNBackbone=Llama-3-8B2024.1256.626.09
RLKFBackbone=Llama-3-8B2024.1254.8622.15
ICL-COTBackbone=Mistral-7B2024.1254.4718.22
UALIGN (w/o Conf.)Backbone=Llama-3-8B2024.1254.1923.6
UALIGNBackbone=Mistral-7B2024.1254.1719.64
RL-PPOBackbone=Llama-3-8B2024.1254.0724.19
RL-PPOBackbone=Mistral-7B2024.1253.7619.19
UALIGN (w/o Entro.)Backbone=Mistral-7B2024.1252.2118.44
ICLBackbone=Mistral-7B2024.1252.0917.95
RLKFBackbone=Mistral-7B2024.1252.0918.24
ICL-IDKBackbone=Mistral-7B2024.1251.1317.67
ITIBackbone=Mistral-7B2024.1251.1216.68
RL-DPOBackbone=Mistral-7B2024.1250.9516.42
SFTBackbone=Mistral-7B2024.1250.8214.42
UALIGN (w/o Conf.)Backbone=Mistral-7B2024.1250.317.88
ICLBackbone=Llama-3-8B2024.1250.2820.11
RL-DPOBackbone=Llama-3-8B2024.1249.6519.18
R-TuningBackbone=Llama-3-8B2024.1247.8118.12
R-TuningBackbone=Mistral-7B2024.1246.3315.52
ICL-IDKBackbone=Llama-3-8B2024.1245.4320.72
ITIBackbone=Llama-3-8B2024.1243.217.13
SFTBackbone=Llama-3-8B2024.1241.4116.57