Multilingual Table Question Answering on MultiTEND Japanese
23.53EMMultiLink
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MultiLink2025.02 | 23.53 | 63.26 | 72.18 | 55.75 | 85.32 | 72.03 | |
| SMART2025.02 | 18.05 | 58.11 | 62.03 | 43.3 | 84.76 | 74.6 | |
| Fine-tuned LlamaBackbone=Llama, Evaluation Protocol=Fine-tuned2025.02 | 16.4 | 56.22 | 58.7 | 43.32 | 79.5 | 68.43 | |
| RAG for LLMEvaluation Protocol=RAG2025.02 | 12.02 | 57.36 | 65.29 | 45.12 | 75.02 | 68.14 | |
| Few-shot LLMEvaluation Protocol=Few-shot2025.02 | 9.87 | 56.14 | 61.51 | 35.93 | 64.47 | 63.46 | |
| Zero-shot LLMEvaluation Protocol=Zero-shot2025.02 | 0.54 | 47.35 | 58.09 | 30.63 | 57.01 | 60.14 |