Stable Diffusion v1-4
| Property | Value |
|---|---|
| License | CreativeML OpenRAIL-M |
| Authors | Robin Rombach, Patrick Esser |
| Training Data | LAION-2B(en) and improved aesthetics datasets |
| Primary Use | Text-to-Image Generation |
What is diffusers-generation-text-box?
Stable Diffusion v1-4 is a sophisticated latent diffusion model designed for high-quality text-to-image generation. It represents a significant advancement in AI image synthesis, combining an autoencoder with a diffusion model trained in latent space. The model was fine-tuned for 225k steps at 512x512 resolution on the "laion-aesthetics v2 5+" dataset.
Implementation Details
The model utilizes a complex architecture that includes a ViT-L/14 text encoder and a UNet backbone for latent diffusion. It operates with a relative downsampling factor of 8, transforming images from HxWx3 to latents of H/f x W/f x 4. The training process involved 32 A100 GPUs with AdamW optimizer and a learning rate of 0.0001.
- Supports both PyTorch and JAX/Flax implementations
- Includes built-in safety modules for content filtering
- Optimized for 512x512 image generation
- Supports different scheduler options including PNDM and Euler
Core Capabilities
- High-quality image generation from text descriptions
- Supports classifier-free guidance sampling
- Memory-efficient with options for float16 precision
- Handles complex compositional prompts
Frequently Asked Questions
Q: What makes this model unique?
The model combines state-of-the-art latent diffusion techniques with improved aesthetics training and classifier-free guidance, resulting in higher quality image generation compared to previous versions.
Q: What are the recommended use cases?
The model is intended for research purposes, including safe deployment studies, artistic applications, educational tools, and research on generative models. It should not be used for creating harmful or misleading content.
