Long-form Question Answering on FetaQA (test)
32.61BLEUChain-of-Table
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Chain-of-TableBackbone=PaLM 22025.01 | 32.61 | 66 | 44 | 56 | |
| Fine-TuningBackbone=T5-large2025.01 | 30.54 | 63 | 41 | 53 | |
| DaterBackbone=PaLM 22025.01 | 29.47 | 63 | 41 | 53 | |
| TableMasterBackbone=gpt-4o2025.01 | 28.94 | 66.06 | 45.29 | 54.56 | |
| End-to-End QABackbone=PaLM 22025.01 | 28.37 | 63 | 41 | 53 | |
| End-to-End QABackbone=Codex2025.01 | 27.96 | 62 | 40 | 52 | |
| End-to-End QABackbone=gpt-4o2025.01 | 24.91 | 62.05 | 41.29 | 50.36 |