Share your thoughts, 1 month free Claude Pro on usSee more

Zero-shot Commonsense Reasoning on PIQA zero-shot

76.93Accuracy

FP16

Updated 29d ago

Evaluation Results

Method	Links
FP16 2026.04		76.93
TriLM-3.9B 2026.06		74.44
BitNetV1-7B 2026.06		74.37
TequilaLLM-3B 2026.06		73.9
TernaryLLM-8B+KD 2026.06		73.7
Qwen3-8B + CAT-Q 2026.06		73.45
BitNetV1-3.9B 2026.06		73.2
FBI-LLM 2026.04		72.6
Qwen3-4B + CAT-Q 2026.06		71.55
LBLLM 2026.04		70.97
Original Endpoint 2026.06		70.95
Qwen3-4B + CAT-Q 2026.06		70.62
TriLM-1.5B 2026.06		70.3
BitNetV1-1.3B 2026.06		68.8
Qwen3-1.7B + CAT-Q 2026.06		68.4
Onebit 2026.04		68.01
Original Pythia 2026.06		66.43
Yat (pn+α) 2026.05		62.44
Yat (sb+α) 2026.05		62.37
GELU 2026.05		61.77
Yat (sb+ca) 2026.05		61.5
Yat (pn+ca) 2026.05		61.26
Dual Learned Matching 2026.06		56.42
Learned Matching 2026.06		55.98
Dual Learned Matching 2026.06		55.71
Learned Matching 2026.06		55.44
Raw Interpolation 2026.06		55.39
Raw Interpolation 2026.06		54.35