List your company

Commonsense Reasoning on TruthfulQA

71.48AccuracySPPO
18.90832.556546.20559.8535Sep 27, 2025Oct 25, 2025Nov 23, 2025Dec 22, 2025Jan 19, 2026Feb 17, 2026Mar 18, 2026

Evaluation results

MethodLinks
SPPO2025.0971.48
HT-MNPOReward Model=Athene-RM-8B2025.0971.36
DPO2025.0971.24
INPO2025.0971.24
HT-MNPOReward Model=Skywork-Reward-V22025.0970.87
SFT Model2025.0970.75
HT-MNPOReward Model=ArmoRM-Llama32025.0970.38
TD-MNPO2025.0970.26
SimPO2025.0963.4
FP16Backbone=Qwen2.5-7B2026.0339.05
NSDSBackbone=Qwen2.5-7B2026.0331.15
ZDBackbone=Qwen2.5-7B2026.0328.94
FP16Backbone=Llama-3.1-8B2026.0328.4
KurtBoostBackbone=Qwen2.5-7B2026.0328.35
MSEBackbone=Qwen2.5-7B2026.0327.82
EWQBackbone=Qwen2.5-7B2026.0326.95
NSDSBackbone=Llama-3.1-8B2026.0326.43
MDM-Prime-v2Zero-shot=true, Number of Parameters=1.1B2026.0325.83
BloomZero-shot=true, Number of Parameters=1.1B2026.0325.58
KurtBoostBackbone=Llama-3.1-8B2026.0325.15
SMDMZero-shot=true, Number of Parameters=1.1B2026.0324.6
ZDBackbone=Llama-3.1-8B2026.0324.49
MSEBackbone=Llama-3.1-8B2026.0323.95
OPTZero-shot=true, Number of Parameters=1.3B2026.0323.75
EWQBackbone=Llama-3.1-8B2026.0323.34
GPT-NeoZero-shot=true, Number of Parameters=1.3B2026.0323.13
PythiaZero-shot=true, Number of Parameters=1.4B2026.0322.77
TinyLLaMAZero-shot=true, Number of Parameters=1.1B2026.0320.93