RDF-to-text generation on GEM2 Factual (test)
98.4GrammaticalityLlama 3.3
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Llama 3.3Architecture=Neural model2025.12 | 98.4 | 2.7 | 7.6 | |
| GPT-4.1Architecture=Rule-based NLG2025.12 | 73 | 3.4 | 10.8 | |
| Qwen 3 235BArchitecture=Rule-based NLG2025.12 | 73 | 2.6 | 4 | |
| BARTArchitecture=Neural model2025.12 | 68.9 | 52.7 | 51.2 | |
| Qwen 2.5 72BArchitecture=Rule-based NLG2025.12 | 66 | 2.1 | 6.6 | |
| Llama 3.3 70BArchitecture=Rule-based NLG2025.12 | 63.3 | 5 | 12.7 |