Text-to-Image Generation on 100 prompts (test)
0.2822HPSv2VMPO-Diff
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| VMPO-DiffBase Model=Stable Diffusion v1.5, Fine-tuning Reward Function=HPSv2, Adaptation Method=LoRA, Images per Prompt=102026.02 | 0.2822 | 0.2622 | 0.4973 | 0.2916 | |
| VMPO-R2GBase Model=Stable Diffusion v1.5, Fine-tuning Reward Function=HPSv2, Adaptation Method=LoRA, Images per Prompt=102026.02 | 0.2723 | 0.2713 | 0.3427 | 0.3673 | |
| GRPOBase Model=Stable Diffusion v1.5, Fine-tuning Reward Function=HPSv2, Adaptation Method=LoRA, Images per Prompt=102026.02 | 0.2684 | 0.2653 | 0.3449 | 0.322 | |
| SD1.5 (Base)Base Model=Stable Diffusion v1.5, Images per Prompt=102026.02 | 0.2368 | 0.2717 | 0.0331 | 0.4389 |