Explanation Generation on CFMS
10.6BLEU-4GPT-4o
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GPT-4oPrompting Strategy=RAG 1-shot2026.03 | 10.6 | 73.84 | |
| LoRA FTModel=InternVL2.5-8B2026.03 | 9.35 | 72.36 | |
| InternVL2.5-8BPrompting Strategy=PGDS2026.03 | 8.76 | 72.3 | |
| InternVL2.5-8BPrompting Strategy=RAG 1-shot2026.03 | 8.58 | 71.69 | |
| GPT-4oPrompting Strategy=Random 1-shot2026.03 | 8.39 | 72.59 | |
| InternVL2.5-8BPrompting Strategy=Random 1-shot2026.03 | 8.33 | 72.31 | |
| Qwen2.5-VL-7B-InstructPrompting Strategy=Random 1-shot2026.03 | 8.24 | 71.83 | |
| Qwen2.5-VL-7B-InstructPrompting Strategy=RAG 1-shot2026.03 | 8.18 | 71.92 | |
| Qwen2.5-VL-7B-InstructPrompting Strategy=PGDS2026.03 | 8.1 | 71.52 | |
| GPT-4oPrompting Strategy=Zero-shot2026.03 | 8.07 | 72.81 | |
| LoRA FTModel=Qwen2.5-VL-7B-Instruct2026.03 | 7.51 | 71.24 | |
| Gemini-2.5-FlashPrompting Strategy=RAG 1-shot2026.03 | 7.34 | 73.03 | |
| Qwen2.5-VL-7B-InstructPrompting Strategy=Zero-shot2026.03 | 7.29 | 72.04 | |
| Gemini-2.5-FlashPrompting Strategy=Random 1-shot2026.03 | 6.88 | 72.55 | |
| InternVL2.5-8BPrompting Strategy=Zero-shot2026.03 | 6.77 | 70.72 | |
| Gemini-2.5-FlashPrompting Strategy=Zero-shot2026.03 | 6.3 | 72.41 | |
| LoRA FTModel=LLaVA-1.5-7B2026.03 | 5.71 | 69.32 |