Flowchart-Oriented Dialogue on PFDial (in-domain)
100INGALlama-3.1-8B
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Llama-3.1-8BTraining protocol=Fine-tuned on PFDial2026.02 | 100 | 92.63 | 75.35 | 0 | 77 | |
| Qwen2.5-7BTraining protocol=Fine-tuned on PFDial2026.02 | 100 | 95.18 | 75.63 | 0 | 80 | |
| LLAMA3.1 8B-instrctFramework=FloCA, Base Model=LLAMA3.1-8B-instruct2026.02 | 100 | 92.91 | 83.56 | 0 | 1,066 | |
| Qwen2.5-7B-instructFramework=FloCA, Base Model=Qwen2.5-7B-instruct2026.02 | 100 | 94.05 | 79.6 | 0 | 833 | |
| Llama-3.3-70BFramework=FloCA, Base Model=Llama-3.3-70B2026.02 | 100 | 97.73 | 90.65 | 0 | 271 | |
| Qwen2.5-72BFramework=FloCA, Base Model=Qwen2.5-72B2026.02 | 100 | 94.33 | 85.54 | 0 | 271 | |
| Deepseek-R1Framework=FloCA, Base Model=Deepseek-R12026.02 | 100 | 98.3 | 94.33 | 0 | 60 |