Natural Language Inference on ANLI R2 (Accuracy)

81.14Accuracy

In-Squeeze

Updated 1mo ago

Evaluation Results

Method	Links
In-Squeeze 2026.02		81.14
In-Squeeze 2026.02		79.63
Cont-Squeeze 2026.02		79.52
Direct Fine-tuning 2026.02		79.3
Direct Fine-tuning 2026.02		78.57
Cont-Squeeze 2026.02		78.46
Direct Fine-tuning 2026.02		77.23
Direct Fine-tuning 2026.02		75.11
Direct Fine-tuning 2026.02		74.72
In-Squeeze 2026.02		74.33
In-Squeeze 2026.02		74.16
Cont-Squeeze 2026.02		73.66
Cont-Squeeze 2026.02		72.88
Direct Fine-tuning 2026.02		72.38
Self-consistency 2022.03		64.5
PaLM 2-L 2023.05		63.4
CoT-prompting 2022.03		58.9
Standard-prompting (no-rationale) 2022.03		55.8
PaLM 2-M 2023.05		49.5
PaLM 2-S 2023.05		48.8
PaLM 2023.05		48.7
Cont-Squeeze 2026.02		29.3
Cont-Squeeze 2026.02		21.26
Zero-shot 2026.02		19.9