Commonsense Question Answering on SocialIQA (SIQA) (val)

70.7Accuracy

ChatGPT + Chain-of-thought

Updated 1mo ago

Evaluation Results

Method	Links
ChatGPT + Chain-of-thought 2024.01		70.7
ChatGPT + Self-consistent chain-of-thought 2024.01		69.7
ChatGPT 2024.01		69.5
GPT 3.5 2024.01		68
DeBERTa-v3-L (CANDLE Distilled) 2024.01		65.9
STL-Adapter 2024.01		64.4
CAR-DeBERTa-v3-L 2024.01		64
CAR-DeBERTa-v3-L 2024.01		63.8
DeBERTa-v3-L (MR) 2024.01		62.1
DeBERTa-v3-L (MR) 2024.01		59.7
VERA-T5-xxl (CANDLE Distilled) 2024.01		59.4
VERA-T5-xxl 2024.01		58.2
VERA-T5-xxl 2024.01		58.1
VERA-T5-xxl 2024.01		57.7
GPT-4 2024.01		57
MICO 2024.01		56
LLAMA2 2024.01		50.3
COMET-DynGen 2024.01		50.1
SMLM 2024.01		48.5
LLAMA2 2024.01		48.3
DeBERTa-v3-L 2024.01		47.8
ROBERTa-L 2024.01		47.3
Self-talk 2024.01		46.2
Mistral-v0.1 2024.01		42.9