felfri/prx-generated-images

---
license: apache-2.0
tags:
- diffusion
- text-to-image
- safety
- prx
size_categories:
- 10K<n<100K
---

# PRX Generated Images

Generated images and safety annotations from existing PRX model checkpoints, used as baselines in the dose-response experiment.

## Structure

```
images/
prx-1024-beta/images.tar
prx-256-base/images.tar
prx-256-sft/images.tar
prx-512-base/images.tar
prx-512-dc-ae/images.tar
prx-512-sft/images.tar
prx-512-sft-distilled/images.tar
annotations/
prx-1024-beta.parquet
prx-256-base.parquet
prx-256-sft.parquet
prx-512-base.parquet
prx-512-dc-ae.parquet
prx-512-sft.parquet
prx-512-sft-distilled.parquet
```

## Models

| Model | Source |
|-------|--------|
| prx-1024-beta | [Photoroom/prx-1024-t2i-beta](https://huggingface.co/Photoroom/prx-1024-t2i-beta) |
| prx-512-base | [Photoroom/prx-512-t2i](https://huggingface.co/Photoroom/prx-512-t2i) |
| prx-512-sft | [Photoroom/prx-512-t2i-sft](https://huggingface.co/Photoroom/prx-512-t2i-sft) |
| prx-512-sft-distilled | [Photoroom/prx-512-t2i-sft-distilled](https://huggingface.co/Photoroom/prx-512-t2i-sft-distilled) |
| prx-512-dc-ae | [Photoroom/prx-512-t2i-dc-ae](https://huggingface.co/Photoroom/prx-512-t2i-dc-ae) |
| prx-256-base | [Photoroom/prx-256-t2i](https://huggingface.co/Photoroom/prx-256-t2i) |
| prx-256-sft | [Photoroom/prx-256-t2i-sft](https://huggingface.co/Photoroom/prx-256-t2i-sft) |

## Annotation Format

Each parquet file is indexed by image ID and contains the following columns:

| Column | Type | Description |
|--------|------|-------------|
| `llavaguard_unsafe` | int | 0 = safe, 1 = unsafe (LlavaGuard-7B) |
| `llavaguard_category` | str | Safety category (e.g. "O3: Sexual Content", "NA: None applying") |
| `llamaguard3_unsafe` | int | 0 = safe, 1 = unsafe (LlamaGuard-3-11B-Vision) |
| `llamaguard3_category` | str | MLCommons category (e.g. "S12: Sexual Content", "NA") |
| `shieldgemma_unsafe` | int | 0 = safe, 1 = unsafe (ShieldGemma-2-4B) |
| `shieldgemma_category` | str | Highest-probability policy violated ("dangerous", "sexual", "violence", or "NA") |
| `shieldgemma_p_dangerous` | float | P(dangerous content) |
| `shieldgemma_p_sexual` | float | P(sexual content) |
| `shieldgemma_p_violence` | float | P(violent content) |
| `sd_safety_checker_unsafe` | int | 0 = safe, 1 = unsafe (SD Safety Checker) |

## Safety Evaluators

| Evaluator | Model | Categories |
|-----------|-------|------------|
| [LlavaGuard-7B](https://huggingface.co/AIML-TUDA/LlavaGuard-v1.2-7B-OV) | Vision-language safety model | O1–O9 (9 categories) |
| [LlamaGuard-3-11B-Vision](https://huggingface.co/meta-llama/Llama-Guard-3-11B-Vision) | Meta's multimodal safety model | S1–S14 (MLCommons taxonomy) |
| [ShieldGemma-2-4B](https://huggingface.co/google/shieldgemma-2-4b) | Google's safety classifier | dangerous, sexual, violence |
| SD Safety Checker | CompVis CLIP-based NSFW classifier | binary (safe/unsafe) |

## Generation Settings

- **Prompts**: 10,000 (1K safe + 9K unsafe across 9 categories)
- **Guidance scale**: 3.5
- **Inference steps**: 50
- **Seed**: 42

## Related Resources

- [diffusion_safety](https://github.com/felifri/diffusion_safety) — experiment code