Graph-to-Sequence Generation on TekGEN (test)
0.253B ScoreDLM4G-2.o
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DLM4G-2.o2026.04 | 0.253 | 0.522 | 0.414 | 0.847 | 0.82 | |
| DLM4G-2.oevaluation paradigm=Pretrain, #P=63M2026.04 | 0.253 | 0.522 | 0.414 | — | — | |
| DLM4G-2.oevaluation paradigm=Finetune, #P=63M2026.04 | 0.253 | 0.522 | 0.414 | — | — | |
| DLM4G-2.oevaluation paradigm=Zero-shot, #P=63M2026.04 | 0.253 | 0.522 | 0.414 | — | — | |
| DLM4G-1.o2026.04 | 0.247 | 0.493 | 0.375 | 0.795 | 0.781 | |
| DLM4G-1.oevaluation paradigm=Pretrain, #P=50M2026.04 | 0.247 | 0.493 | 0.375 | — | — | |
| GPT-2 (B)evaluation paradigm=Finetune, #P=355M2026.04 | 0.228 | 0.366 | 0.211 | — | — | |
| GPT-2 (S)evaluation paradigm=Finetune, #P=124M2026.04 | 0.226 | 0.358 | 0.208 | — | — | |
| ReGen-SCST2026.04 | 0.219 | 0.385 | 0.223 | 0.698 | 0.719 | |
| ReGen-CE2026.04 | 0.199 | 0.372 | 0.214 | 0.612 | 0.701 | |
| T5 (L)evaluation paradigm=Finetune, #P=770M2026.04 | 0.199 | 0.37 | 0.211 | — | — | |
| Rule-based2026.04 | 0.189 | 0.309 | 0.301 | 0.509 | 0.672 | |
| T5 (S)evaluation paradigm=Finetune, #P=60M2026.04 | 0.189 | 0.352 | 0.203 | — | — | |
| Qwen2.5evaluation paradigm=Zero-shot, #P=7B2026.04 | 0.182 | 0.312 | 0.234 | — | — | |
| LLaMa-3evaluation paradigm=Zero-shot, #P=8B2026.04 | 0.176 | 0.341 | 0.251 | — | — | |
| DeepSeekevaluation paradigm=Zero-shot, #P=7B2026.04 | 0.121 | 0.345 | 0.256 | — | — | |
| GPT-o4-minievaluation paradigm=Zero-shot, #P=8B2026.04 | 0.121 | 0.327 | 0.277 | — | — |