List your company
ResearchFormerly WizWand

Question Answering on TruthfulQA (TruthfulQA Score)

63TruthfulQA ScoreTI-DPO
2.773618.409334.04549.6807Feb 4, 2025Apr 11, 2025Jun 17, 2025Aug 23, 2025Oct 29, 2025Jan 4, 2026Mar 12, 2026
Updated 7mo ago

Evaluation results

MethodLinks
TI-DPOBackbone=LLaMA-3.1-8B2025.0563
GRPOBackbone=LLaMA-3.1-8B2025.0562
TPOBackbone=LLaMA-3.1-8B2025.0561
CPOBackbone=LLaMA-3.1-8B2025.0560
SELF-EVOLVING CLEANINGBase Model=LLaMA-3.1-8B, Selection Proportion=0.62025.0259.58
KTOBackbone=LLaMA-3.1-8B2025.0559
DPOBackbone=LLaMA-3.1-8B2025.0558
TDPOBackbone=LLaMA-3.1-8B2025.0557
FIXED-MODEL CLEANINGBase Model=LLaMA-3.1-8B, Selection Proportion=0.62025.0256.02
SFTBackbone=LLaMA-3.1-8B2025.0556
RHOBase Model=LLaMA-3.1-8B, Selection Proportion=0.62025.0254.63
SIMPOBackbone=LLaMA-3.1-8B2025.0554
IPOBackbone=LLaMA-3.1-8B2025.0553
SELF-EVOLVING CLEANINGBase Model=LLaMA-3.2-3B, Selection Proportion=0.62025.0251.07
ABL-SOC-110M-2Zero-shot=true2026.0350.41
DS² (10K)Base Model=LLaMA-3.1-8B, Token Budget=10K2025.0249.57
FIXED-MODEL CLEANINGBase Model=LLaMA-3.2-3B, Selection Proportion=0.62025.0248.96
SUB-SOC-110M-1Zero-shot=true2026.0348.91
UNIFORM RANDOM (50K×0.6)Base Model=LLaMA-3.1-8B, Token Budget=50K, Selection Proportion=0.62025.0248.68
ABL-SOC-110M-3Zero-shot=true2026.0348.42
SUB-SOC-110M-2Zero-shot=true2026.0348.07
FULL TOKENS (50K)Base Model=LLaMA-3.1-8B, Token Budget=50K2025.0247.51
GPT-Neo-125MZero-shot=true2026.0345.58
RHOBase Model=LLaMA-3.2-3B, Selection Proportion=0.62025.0245.57
SELF-EVOLVING CLEANINGBase Model=Mistral-7B-v0.3, Selection Proportion=0.62025.0245.41
BASEBase Model=LLaMA-3.1-8B2025.0245.1
ABL-SOC-110M-1Zero-shot=true2026.0344.66
PLDRv51-SOC-110M-1Zero-shot=true2026.0344.59
FIXED-MODEL CLEANINGBase Model=Mistral-7B-v0.3, Selection Proportion=0.62025.0244.52
PLDRv51-SOC-110M-2Zero-shot=true2026.0344.33
PLDRv51-SOC-110M-4Zero-shot=true2026.0344.33
DS² (10K)Base Model=Mistral-7B-v0.3, Token Budget=10K2025.0244.24
PLDRv51-SOC-110M-3Zero-shot=true2026.0344.17
RHOBase Model=Mistral-7B-v0.3, Selection Proportion=0.62025.0243.92
UNIFORM RANDOM (50K×0.6)Base Model=Mistral-7B-v0.3, Token Budget=50K, Selection Proportion=0.62025.0243.82
UNIFORM RANDOM (50K×0.6)Base Model=LLaMA-3.2-3B, Token Budget=50K, Selection Proportion=0.62025.0243.79
FULL TOKENS (50K)Base Model=Mistral-7B-v0.3, Token Budget=50K2025.0243.67
DS² (10K)Base Model=LLaMA-3.2-3B, Token Budget=10K2025.0243.35
FULL TOKENS (50K)Base Model=LLaMA-3.2-3B, Token Budget=50K2025.0243.32
PLDRv51-SOC-110M-5Zero-shot=true2026.0343
BASEBase Model=Mistral-7B-v0.32025.0242.56
BASEBase Model=LLaMA-3.2-3B2025.0239.39
Vicuna-13B (all)#Token=370M2026.0135
Alpaca-13B#Token=4.4M2026.0130
LLaMA-13B#Token=1T2026.0126
B4GModel=GEMMA-2-9B, Task=model identity2026.035.9
B4GModel=LLAMA2-13B, Task=access control2026.035.89
B4GModel=QWEN2.5-7B, Task=access control2026.035.82
B4GModel=QWEN2.5-7B, Task=safety enhancement2026.035.66
B4GModel=GEMMA-2-9B, Task=access control2026.035.6
B4GModel=LLAMA2-13B, Task=safety enhancement2026.035.56
B4GModel=LLAMA2-13B, Task=style personalization2026.035.54
B4GModel=QWEN2.5-7B, Task=style personalization2026.035.51
B4GModel=GEMMA-2-9B, Task=safety enhancement2026.035.5
B4GModel=GEMMA-2-9B, Task=style personalization2026.035.35
B4GModel=QWEN2.5-7B, Task=model identity2026.035.32
B4GModel=LLAMA3.1-8B, Task=style personalization2026.035.31
B4GModel=LLAMA3.1-8B, Task=safety enhancement2026.035.18
B4GModel=LLAMA3.1-8B, Task=model identity2026.035.12
B4GModel=LLAMA3.1-8B, Task=access control2026.035.1
B4GModel=LLAMA2-13B, Task=model identity2026.035.09