Common Sense Reasoning On Big Bench Date
评估指标
Accuracy
评测结果
各个模型在此基准测试上的表现结果
| Paper Title | Repository | ||
|---|---|---|---|
| PaLM 2 (few-shot, k=3, CoT) | 91.2 | PaLM 2 Technical Report | |
| PaLM 2 (few-shot, k=3, Direct) | 74.0 | PaLM 2 Technical Report | |
| Bloomberg GPT 50B (few-shot, k=3) | 54.8 | BloombergGPT: A Large Language Model for Finance | |
| PaLM 540B (few-shot,k=3) | 53.6 | BloombergGPT: A Large Language Model for Finance | |
| Chinchilla-70B (few-shot, k=5) | 52.3 | Training Compute-Optimal Large Language Models | |
| BLOOM 176B (few-shot, k=3) | 50.00 | BloombergGPT: A Large Language Model for Finance | |
| OPT 66B (few-shot, k=3) | 49.60 | BloombergGPT: A Large Language Model for Finance | |
| GPT-NeoX 20B (few-shot, k=3) | 45.60 | BloombergGPT: A Large Language Model for Finance | |
| Gopher-280B (few-shot, k=5) | 44.1 | Scaling Language Models: Methods, Analysis & Insights from Training Gopher |
0 of 9 row(s) selected.