PixArt-Σ XL 2 1024-MS
Parameters
610M
Modality
Image
Auxiliary Parameters
4.78B
Architecture
Diffusion Transformer
License
Apache 2.0
Release Date
10 Apr 2024
API Pricing (per 1M)
Self-hosted only
Hardware Sizing
Precision
Workload Memory: 2.91 GB
INT4 · 1024x1024
DiT Backbone Weights: 0.28 GB
VAE Decode Workspace: 0.12 GB
Runtime & Graphs: 2.50 GB
Minimum Required Hardware
1x NVIDIA L4 24GB
2.9 / 21.6 GB usable
1x RTX 3060 12GB
2.9 / 10.8 GB usable
Apple M4 24GB
2.9 / 18.0 GB usable
Determine requirements for your specific workload:
Open in VRAM Calculator
Architecture Diagram
About PixArt-Σ XL 2 1024-MS
PixArt-Σ is an enhanced 610M diffusion transformer introducing weak-to-strong training and KV-compression DiT blocks to support native 4K image generation at low compute cost.
Technical Specifications
Attention
Attention Heads
16
Attention Head Dimension
72
Dimensions
Auxiliary Parameters
4.78B
Hidden Dimension Size
1,152
Number of Layers
28