Question Answering on HotpotQA (test) (Tool Use Metrics)
28.4EMUALA-M+Backoff
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| UALA-M+BackoffBackbone=LLaMA2-13B2024.01 | 28.4 | 1,525 | |
| UALA-S+BackoffBackbone=LLaMA2-13B2024.01 | 27.3 | 539 | |
| UALA-MBackbone=LLaMA2-13B2024.01 | 26.5 | 1,525 | |
| SCBackbone=LLaMA2-13B2024.01 | 26.4 | 0 | |
| UALA-SBackbone=LLaMA2-13B2024.01 | 25.6 | 539 | |
| ReAct+BackoffBackbone=LLaMA2-13B2024.01 | 25 | 2,371 | |
| CoTBackbone=LLaMA2-13B2024.01 | 24.8 | 0 | |
| UALA-M+BackoffBackbone=LLaMA2-7B2024.01 | 24.6 | 1,386 | |
| UALA-S+BackoffBackbone=LLaMA2-7B2024.01 | 24.2 | 479 | |
| ReAct+BackoffBackbone=LLaMA2-7B2024.01 | 23.8 | 2,324 | |
| StandardBackbone=LLaMA2-13B2024.01 | 23.4 | 0 | |
| UALA-SBackbone=LLaMA2-7B2024.01 | 22.8 | 479 | |
| UALA-MBackbone=LLaMA2-7B2024.01 | 22.6 | 1,386 | |
| SCBackbone=LLaMA2-7B2024.01 | 22 | 0 | |
| CoTBackbone=LLaMA2-7B2024.01 | 21.8 | 0 | |
| StandardBackbone=LLaMA2-7B2024.01 | 21.2 | 0 | |
| ReActBackbone=LLaMA2-13B2024.01 | 19.8 | 2,371 | |
| ReActBackbone=LLaMA2-7B2024.01 | 16.8 | 2,324 |