List your company
ResearchFormerly WizWand

Physical Commonsense Reasoning on PIQA (test)

90.7AccuracyUL20B
48.63259.553570.47581.3965May 10, 2022Jan 9, 2023Sep 10, 2023May 12, 2024Jan 11, 2025Sep 12, 2025May 15, 2026
Updated 5mo ago

Evaluation results

MethodLinks
UL20Bmode=supervised finetuning, leaderboard submission=true2022.0590.7
UNICORN 11BModel=UNICORN 11B2022.1090.1
Lourie et al.status=previous state-of-the-art2022.0590.1
UnifiedQA 11BModel=UnifiedQA 11B2022.1089.5
ScoreBackbone=DeBERTa Large2022.1087.41
Few-shot*Backbone=Qwen2.5-7B, Few-shot settings=5-shot ID2025.0986
TEAMBackbone=DeBERTa Large2022.1085.9
FP16Model=LLAMA3-70B, Evaluation Protocol=Zero-shot, Bit-width=Full-precision2024.0384.66
Llama3.1-8B-Instruct + SFTBase Model=Llama3.1-8B-Instruct, Training=SFT2026.0584.6
Qwen3-8B + SFT + WeMask(TF)Base Model=Qwen3-8B, Training=SFT, Variant=TF2026.0584.44
Llama3.1-8B-Instruct + SFT + WeMask(TF)Base Model=Llama3.1-8B-Instruct, Training=SFT, Variant=TF2026.0584.22
Llama3.1-8B-Instruct + WeMask(SFT)Base Model=Llama3.1-8B-Instruct, Training=WeMask(SFT)2026.0584.22
Qwen3-8B + WeMask(SFT)Base Model=Qwen3-8B, Training=WeMask(SFT)2026.0584.11
Qwen3-8B + SFTBase Model=Qwen3-8B, Training=SFT2026.0584.06
LLAMA 1Size=65B2023.0782.8
LLAMA 2Size=70B2023.0782.8
ICRBackbone=Qwen2.5-7B2025.0982.6
M²IVBackbone=Qwen2.5-7B2025.0982.5
FalconSize=40B2023.0782.4
LLAMA 1Size=33B2023.0782.3
LIVEBackbone=Qwen2.5-7B2025.0982
MPTSize=30B2023.0781.9
LLAMA 2Size=34B2023.0781.9
I2CLBackbone=Qwen2.5-7B2025.0981.2
FP16Model=LLAMA3-8B, Evaluation Protocol=Zero-shot, Bit-width=Full-precision2024.0380.74
MPTSize=7B2023.0780.6
LLAMA 2Size=13B2023.0780.5
LLAMA 1Size=13B2023.0780.1
LLAMA 1Size=7B2023.0779.8
ScoreBackbone=RoBERTa Large2022.1079.4
Interaction-Aware Influence Function (Ours)Base Model=Llama-3.1-8B, Evaluation Protocol=Fine-tuned, Number of Seeds=52026.0579.39
LLAMA 2Size=7B2023.0778.8
QuaRotModel=LLAMA3-70B, Evaluation Protocol=Zero-shot, Bit-width=4-bit2024.0378.07
OracleModel Size=Large, Expertise Distribution=Dist. 32026.0477.8
REALMModel Size=Large, Noise Type=Systematic, Expertise Distribution=Dist. 32026.0477.67
RandomBase Model=Llama-3.1-8B, Evaluation Protocol=Fine-tuned, Number of Seeds=52026.0577.5
LESSBase Model=Llama-3.1-8B, Evaluation Protocol=Fine-tuned, Number of Seeds=52026.0577.4
Additive IFBase Model=Llama-3.1-8B, Evaluation Protocol=Fine-tuned, Number of Seeds=52026.0577.38
FalconSize=7B2023.0776.7
REALMModel Size=Large, Noise Type=Uniform, Expertise Distribution=Dist. 32026.0476.5
Zero-shotBackbone=Qwen2.5-7B2025.0976.2
REALMModel Size=Large, Noise Type=Asymmetric, Expertise Distribution=Dist. 32026.0476.17
NOISYModel Size=Large, Noise Type=Systematic, Expertise Distribution=Dist. 32026.0475.6
QuaRotModel=LLAMA3-8B, Evaluation Protocol=Zero-shot, Bit-width=4-bit2024.0375.14
Llama3.1-8B-InstructBase Model=Llama3.1-8B-Instruct2026.0575.14
RDS+Base Model=Llama-3.1-8B, Evaluation Protocol=Fine-tuned, Number of Seeds=52026.0575.14
NV-EmbedBase Model=Llama-3.1-8B, Evaluation Protocol=Fine-tuned, Number of Seeds=52026.0575.13
TEAMBackbone=RoBERTa Large2022.1074.55
OracleModel Size=Base, Expertise Distribution=Dist. 32026.0467.68
REALMModel Size=Base, Noise Type=Systematic, Expertise Distribution=Dist. 32026.0466.92
REALMModel Size=Base, Noise Type=Uniform, Expertise Distribution=Dist. 32026.0465.79
REALMModel Size=Base, Noise Type=Asymmetric, Expertise Distribution=Dist. 32026.0465.77
NOISYModel Size=Base, Noise Type=Systematic, Expertise Distribution=Dist. 32026.0464.87
Qwen3-8BBase Model=Qwen3-8B2026.0563.11
Few-shot*Backbone=Llama2-7B, Few-shot settings=5-shot ID2025.0959.8
ICRBackbone=Llama2-7B2025.0957
M²IVBackbone=Llama2-7B2025.0956.8
LIVEBackbone=Llama2-7B2025.0956.4
I2CLBackbone=Llama2-7B2025.0955.6
NOISYModel Size=Large, Noise Type=Asymmetric, Expertise Distribution=Dist. 32026.0453.47
OracleModel Size=Small, Expertise Distribution=Dist. 32026.0452.88
REALMModel Size=Small, Noise Type=Uniform, Expertise Distribution=Dist. 32026.0452.88
Zero-shotBackbone=Llama2-7B2025.0952.2
NOISYModel Size=Base, Noise Type=Asymmetric, Expertise Distribution=Dist. 32026.0452.1
NOISYModel Size=Large, Noise Type=Uniform, Expertise Distribution=Dist. 32026.0451.99
NOISYModel Size=Base, Noise Type=Uniform, Expertise Distribution=Dist. 32026.0451.47
REALMModel Size=Small, Noise Type=Systematic, Expertise Distribution=Dist. 32026.0451.32
NOISYModel Size=Small, Noise Type=Systematic, Expertise Distribution=Dist. 32026.0450.84
NOISYModel Size=Small, Noise Type=Asymmetric, Expertise Distribution=Dist. 32026.0450.58
NOISYModel Size=Small, Noise Type=Uniform, Expertise Distribution=Dist. 32026.0450.36
REALMModel Size=Small, Noise Type=Asymmetric, Expertise Distribution=Dist. 32026.0450.25