List your company

Commonsense Reasoning on Commonsense Reasoning Suite (Zero-shot)

82.81PIQA AccuracyLlama2-70B
48.17857.16966.1675.151May 11, 2026May 18, 2026May 26, 2026Jun 2, 2026Jun 10, 2026Jun 17, 2026Jun 25, 2026

Evaluation results

MethodLinks
Llama2-70B#Bits=W16A16, Zero-shot evaluation protocol=true, Backbone=Llama2-70B2026.0682.81—83.8680.8557.5977.58—76.53
CAT-Q#Bits=W1.58A16, Zero-shot evaluation protocol=true, Backbone=Llama2-70B2026.0680.34—74.5680.1353.2475.34—72.72
EfficientQAT#Bits=W2A16, Zero-shot evaluation protocol=true, Backbone=Llama2-70B2026.0680.2—61.5880.0149.2373.64—68.93
DB-LLM#Bits=W1*A16, Zero-shot evaluation protocol=true, Backbone=Llama2-70B2026.0679.27—76.1655.9344.4573.32—65.82
Llama2-7B#Bits=W16A16, Zero-shot evaluation protocol=true, Backbone=Llama2-7B2026.0678.84—75.974.6246.4269.46—69.04
SliderQuant#Bits=W2A16, Zero-shot evaluation protocol=true, Backbone=Llama2-70B2026.0678.79—73.0277.1452.7173.75—71.08
OmniQuant#Bits=W2A16, Zero-shot evaluation protocol=true, Backbone=Llama2-70B2026.0674.1—35.4567.2133.2864.33—54.87
DB-LLM#Bits=W1*A16, Zero-shot evaluation protocol=true, Backbone=Llama2-7B2026.0673.18—61.9845.233.5361.72—55.12
CAT-Q#Bits=W1.58A16, Zero-shot evaluation protocol=true, Backbone=Llama2-7B2026.0672.91—60.9560.0633.6260.93—57.69
SliderQuant#Bits=W2A16, Zero-shot evaluation protocol=true, Backbone=Llama2-7B2026.0670.78—57.1557.7931.0660.14—55.38
OmniQuant#Bits=W2A16, Zero-shot evaluation protocol=true, Backbone=Llama2-7B2026.0665.13—40.2850.1323.4655.88—46.98
DECO (NG)Model Scale=Medium, Routing Strategy=NG2026.0564.3636.1829.4751.9820.7352.0919.4839.18
BlockFFNModel Scale=Medium2026.0563.3835.8229.6148.8220.9950.5119.9538.44
DeepSeek-V3 (NG)Model Scale=Medium, Routing Strategy=NG2026.0563.2837.6729.6449.0720.6549.2519.7638.47
ReMoE (NG)Model Scale=Medium, Routing Strategy=NG2026.0563.2236.6429.449.7119.7150.8320.4538.57
DenseModel Scale=Medium, Model Architecture=Dense2026.056336.929.5751.4722.0150.5119.6239.01
DeepSeek-V3 (GA)Model Scale=Medium, Routing Strategy=GA2026.0562.7937.8729.6148.5320.952.1719.8438.82
ReMoE (GA)Model Scale=Medium, Routing Strategy=GA2026.0562.6237.6729.148.6520.0550.6718.2838.15
DECO (GA)Model Scale=Medium, Routing Strategy=GA2026.0561.8637.128.4447.1819.3749.6417.4337.29
TopPModel Scale=Medium2026.0561.3736.4928.5745.7519.0350.5112.6336.34
BiLLM#Bits=W1*A16, Zero-shot evaluation protocol=true, Backbone=Llama2-7B2026.0660.6—34.836.224.452.4—41.68
GPTQ#Bits=W2A16, Zero-shot evaluation protocol=true, Backbone=Llama2-7B2026.0658.32—32.5940.4521.2555.17—41.55
PB-LLM#Bits=W1*A16, Zero-shot evaluation protocol=true, Backbone=Llama2-7B2026.0655.22—30.4929.8822.0150.36—37.59
AWQ#Bits=W2A16, Zero-shot evaluation protocol=true, Backbone=Llama2-7B2026.0650—26.1426.5226.7949.64—35.82
GPTQ#Bits=W2A16, Zero-shot evaluation protocol=true, Backbone=Llama2-70B2026.0649.51—25.0425.0822.749.57—34.38