List your company

Question Answering on PIQA (zero-shot, Normalized Accuracy)

60.8Normalized Accuracy (zero-shot)RNN PPT
58.40859.02959.6560.271May 11, 2026

Evaluation results

MethodLinks
RNN PPTscale=160M, zero-shot=true, noise level=0%2026.0560.8
Dyck PPTscale=160M, zero-shot=true, noise level=0%2026.0560.7
Baselinescale=160M, zero-shot=true, noise level=0%2026.0560.5
RNN PPTscale=160M, zero-shot=true, noise level=10%2026.0560.5
Dyck PPTscale=160M, zero-shot=true, noise level=10%2026.0560
Baselinescale=160M, zero-shot=true, noise level=10%2026.0559.9
RNN PPTscale=160M, zero-shot=true, noise level=30%2026.0559.7
Baselinescale=160M, zero-shot=true, noise level=30%2026.0559.6
Dyck PPTscale=160M, zero-shot=true, noise level=30%2026.0559.6
RNN PPTscale=160M, zero-shot=true, noise level=50%2026.0559.2
Dyck PPTscale=160M, zero-shot=true, noise level=50%2026.0558.7
Baselinescale=160M, zero-shot=true, noise level=50%2026.0558.5