Multilingual Table Question Answering on MultiTEND French
25.58EMMultiLink
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MultiLink2025.02 | 25.58 | 64.71 | 73.22 | 59.86 | 84.64 | 74.42 | |
| SMART2025.02 | 18.38 | 61.23 | 60.97 | 44.69 | 85.67 | 78.99 | |
| Fine-tuned LlamaBackbone=Llama, Evaluation Protocol=Fine-tuned2025.02 | 16.53 | 54.22 | 56.61 | 41.26 | 80.14 | 68.38 | |
| RAG for LLMEvaluation Protocol=RAG2025.02 | 15.62 | 60.51 | 67.03 | 49.28 | 73.88 | 70.4 | |
| Few-shot LLMEvaluation Protocol=Few-shot2025.02 | 10.65 | 56.06 | 59.39 | 36.64 | 62.85 | 63.94 | |
| Zero-shot LLMEvaluation Protocol=Zero-shot2025.02 | 0.61 | 50.36 | 58.92 | 33.86 | 57.36 | 59.93 |