SOTA Commonsense Question Answering on Abductive NLI (aNLI) (val) and PapersWithCode

0.812Accuracy

DeBERTa-v3-L (CANDLE Distilled)

Updated 4d ago

Evaluation Results

Method	Links
DeBERTa-v3-L (CANDLE Distilled) 2024.01		0.812
CAR-DeBERTa-v3-L 2024.01		0.796
CAR-DeBERTa-v3-L 2024.01		0.789
DeBERTa-v3-L (MR) 2024.01		0.76
DeBERTa-v3-L (MR) 2024.01		0.751
GPT-4 2024.01		0.75
VERA-T5-xxl (CANDLE Distilled) 2024.01		0.738
ChatGPT + Self-consistent chain-of-thought 2024.01		0.732
VERA-T5-xxl 2024.01		0.732
STL-Adapter 2024.01		0.713
VERA-T5-xxl 2024.01		0.712
ChatGPT + Chain-of-thought 2024.01		0.705
VERA-T5-xxl 2024.01		0.703
ChatGPT 2024.01		0.693
ROBERTa-L 2024.01		0.655
SMLM 2024.01		0.653
GPT 3.5 2024.01		0.618
DeBERTa-v3-L 2024.01		0.599
LLAMA2 2024.01		0.575
LLAMA2 2024.01		0.559
Mistral-v0.1 2024.01		0.51