Text-to-Visualization on Text2Vis
0.97Code Exec Success (%)RL-Text2Vis
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| RL-Text2VisModel Scale (Parameter Count)=14B, Evaluation Protocol (Zero-Shot/SFT/RL)=RL (Gen-8)2026.01 | 0.97 | 0.35 | 4.1 | 4.03 | 0.29 | |
| RL-Text2VisModel Scale (Parameter Count)=7B, Evaluation Protocol (Zero-Shot/SFT/RL)=RL (Gen-8)2026.01 | 0.91 | 0.31 | 3.84 | 3.86 | 0.22 | |
| Gemini-2.0-FlashEvaluation Protocol (Zero-Shot/SFT/RL)=Zero-Shot2026.01 | 0.9 | 0.35 | 3.73 | 3.66 | 0.26 | |
| GPT-4oEvaluation Protocol (Zero-Shot/SFT/RL)=Zero-Shot2026.01 | 0.87 | 0.39 | 3.32 | 3.3 | 0.3 | |
| Qwen-2.5Model Scale (Parameter Count)=14B, Evaluation Protocol (Zero-Shot/SFT/RL)=SFT2026.01 | 0.87 | 0.36 | 3.42 | 3.28 | 0.18 | |
| Qwen-2.5Model Scale (Parameter Count)=7B, Evaluation Protocol (Zero-Shot/SFT/RL)=SFT2026.01 | 0.85 | 0.3 | 3.34 | 3.32 | 0.16 | |
| Gemini-1.5-FlashEvaluation Protocol (Zero-Shot/SFT/RL)=Zero-Shot2026.01 | 0.82 | 0.32 | 3.26 | 2.95 | 0.16 | |
| Qwen-2.5Model Scale (Parameter Count)=14B, Evaluation Protocol (Zero-Shot/SFT/RL)=Zero-Shot2026.01 | 0.78 | 0.29 | 3.12 | 2.94 | 0.14 | |
| Qwen-2.5Model Scale (Parameter Count)=7B, Evaluation Protocol (Zero-Shot/SFT/RL)=Zero-Shot2026.01 | 0.77 | 0.27 | 2.81 | 2.69 | 0.13 | |
| Llama-3.1Model Scale (Parameter Count)=8B, Evaluation Protocol (Zero-Shot/SFT/RL)=Zero-Shot2026.01 | 0.7 | 0.25 | 1.81 | 1.62 | 0.08 | |
| CodeLlamaModel Scale (Parameter Count)=7B, Evaluation Protocol (Zero-Shot/SFT/RL)=Zero-Shot2026.01 | 0.62 | 0.25 | 2.34 | 1.5 | 0.02 | |
| CodeLlamaModel Scale (Parameter Count)=13B, Evaluation Protocol (Zero-Shot/SFT/RL)=Zero-Shot2026.01 | 0.54 | 0.16 | 1.81 | 1.45 | 0.04 | |
| MistralModel Scale (Parameter Count)=7B, Evaluation Protocol (Zero-Shot/SFT/RL)=Zero-Shot2026.01 | 0.42 | 0.22 | 1.57 | 1.39 | 0.07 | |
| Qwen-2.5-CoderModel Scale (Parameter Count)=7B, Evaluation Protocol (Zero-Shot/SFT/RL)=Zero-Shot2026.01 | 0.32 | 0.24 | 1.37 | 1.23 | 0.04 |