Commonsense Reasoning on QASC (dev)
84.02AccuracyGenerated Knowledge Prompting
Evaluation results
| Method | Links | |
|---|---|---|
| Generated Knowledge PromptingInference Model=UQA-11b-ft2021.10 | 84.02 | |
| Context sentencesInference Model=UQA-11b-ft2021.10 | 82.61 | |
| Random sentencesInference Model=UQA-11b-ft2021.10 | 82.18 | |
| Vanilla baselineInference Model=UQA-11b-ft2021.10 | 81.75 | |
| UQA-11b-ft (prev. SOTA)2021.10 | 81.75 | |
| AnswersInference Model=UQA-11b-ft2021.10 | 81.53 | |
| Retrieval-based knowledgeInference Model=T5-11b2021.10 | 76.89 | |
| Few-shot GPT-3 Infer. (finetuned config)Inference Model=UQA-11b-ft2021.10 | 66.09 | |
| Generated Knowledge PromptingInference Model=T5-11b2021.10 | 58.32 | |
| Context sentencesInference Model=T5-11b2021.10 | 55.83 | |
| Few-shot GPT-3 Infer.2021.10 | 53.8 | |
| AnswersInference Model=T5-11b2021.10 | 52.48 | |
| Random sentencesInference Model=T5-11b2021.10 | 49.35 | |
| Vanilla baselineInference Model=T5-11b2021.10 | 48.16 |