List your company

Zero-shot Common Sense Reasoning Suite (PIQA, ARC-e, ARC-c, BoolQ, HellaSwag, Winogrande) (test dev)

79.35PIQAKirin
59.465264.627669.7974.9524Feb 9, 2026

Evaluation results

MethodLinks
KirinBackbone=Llama2-13B, Evaluation Protocol=Zero-shot2026.0279.3556.443.9466.2175.9468.9865.14
SpikeQuantBackbone=Llama2-13B, Evaluation Protocol=Zero-shot2026.0279.3356.4844.0366.1575.5368.1164.94
Fp16Backbone=Llama2-13B, Evaluation Protocol=Zero-shot2026.0279.0559.8544.6268.5376.2270.0966.39
AtomBackbone=Llama2-13B, Evaluation Protocol=Zero-shot2026.0277.6957.5842.9267.4673.7768.5164.66
Fp16Backbone=Llama2-7B, Evaluation Protocol=Zero-shot2026.0277.3752.5341.3873.1272.9966.8564.04
SpikeQuantBackbone=Llama2-7B, Evaluation Protocol=Zero-shot2026.0277.0955.3940.3673.4972.0965.2763.95
KirinBackbone=Llama2-7B, Evaluation Protocol=Zero-shot2026.0276.7754.5540.5373.9172.166.2264.01
QuantizationBackbone=Llama2-13B, Evaluation Protocol=Zero-shot2026.0276.7155.1341.4767.0373.0965.1963.1
AtomBackbone=Llama2-7B, Evaluation Protocol=Zero-shot2026.0276.2852.138.9969.7969.8163.6961.78
QuantizationBackbone=Llama2-7B, Evaluation Protocol=Zero-shot2026.0275.4653.9640.1968.0769.2161.5661.41
Fp16Backbone=Opt-2.7B, Evaluation Protocol=Zero-shot2026.0274.8154.3431.3160.2860.5960.9357.04
KirinBackbone=Opt-2.7B, Evaluation Protocol=Zero-shot2026.0274.8153.9131.4861.6958.8359.9156.77
SpikeQuantBackbone=Opt-2.7B, Evaluation Protocol=Zero-shot2026.0274.2153.7931.6661.1959.1159.7556.62
Fp16Backbone=Opt-1.3B, Evaluation Protocol=Zero-shot2026.0272.4250.829.6957.6853.7359.6754
KirinBackbone=Opt-1.3B, Evaluation Protocol=Zero-shot2026.0271.8648.7828.6753.6652.2557.7752.17
SpikeQuantBackbone=Opt-1.3B, Evaluation Protocol=Zero-shot2026.0271.4949.6627.2253.2751.4658.5651.94
OPTQBackbone=Opt-2.7B, Evaluation Protocol=Zero-shot2026.0271.3848.1927.8255.4349.8654.8251.25
QuantizationBackbone=Opt-2.7B, Evaluation Protocol=Zero-shot2026.0271.1650.8428.7559.8555.5760.4654.44
OmniQuantBackbone=Llama2-13B, Evaluation Protocol=Zero-shot2026.0269.6947.3933.162.8458.9655.854.63
OPTQBackbone=Opt-1.3B, Evaluation Protocol=Zero-shot2026.0268.3446.1727.6551.5746.5355.2649.25
QuantizationBackbone=Opt-1.3B, Evaluation Protocol=Zero-shot2026.0267.7942.9325.2649.5145.9454.9347.73
SpikeLLMBackbone=Llama2-13B, Evaluation Protocol=Zero-shot2026.0266.4955.330.1264.1647.4351.5452.51
OmniQuantBackbone=Llama2-7B, Evaluation Protocol=Zero-shot2026.0266.1545.231.1463.5156.4453.4352.65
SpikeLLMBackbone=Llama2-7B, Evaluation Protocol=Zero-shot2026.0264.4748.7427.363.2743.2956.8350.65
SmoothQuantBackbone=Llama2-13B, Evaluation Protocol=Zero-shot2026.0264.4741.7530.8962.2946.6851.749.63
LRQuantBackbone=Opt-2.7B, Evaluation Protocol=Zero-shot2026.0264.0447.8120.6554.133.5953.5145.62
OneBitBackbone=Opt-2.7B, Evaluation Protocol=Zero-shot2026.0263.8743.3924.454.2838.1851.6745.97
SmoothQuantBackbone=Llama2-7B, Evaluation Protocol=Zero-shot2026.0263.1140.0331.5758.4743.3852.848.23
OneBitBackbone=Opt-1.3B, Evaluation Protocol=Zero-shot2026.0262.5741.2524.0659.4534.2651.1445.46
LRQuantBackbone=Opt-1.3B, Evaluation Protocol=Zero-shot2026.0260.2343.2219.857.9830.7151.7843.95