Mindscend Image 0.2
A small class-conditional image generator trained from scratch by Mindscend. It makes 64×64 pictures of ten subjects and runs in the browser (ONNX Runtime Web, WebGPU) in the Mindscend AI Lab.
| Architecture | Rectified-flow U-Net (levels ×1, ×2, ×2, ×2; attention at the 2nd level), class-conditional with classifier-free guidance |
| Parameters | 38.3M |
| Resolution | 64×64 RGB |
| Classes | flower, bird, insect, boat, train, mountain, building, water, bridge, sculpture (index 10 = no class) |
| Training data | 60,000 images (6,000 per class) picked from PD12M by caption; photos only, no people |
| Training | 80,000 steps, batch 64, AdamW 2e-4, EMA weights, one RTX 3080 Ti |
| Format | ONNX fp32, opset 17 (browser); Core ML FP16 (Apple) |
Quality
FID against 60,000 training images, 30 Euler steps, only 200 generated samples (20 per class). FID on so few samples reads high; compare these only with our own runs measured the same way:
| guidance | FID | Inception score |
|---|---|---|
| 1.0 | 121.2 | 5.09 |
| 1.5 | 115.3 | 5.23 |
| 2.0 | 111.1 | 5.44 |
| 3.0 | 111.0 | 5.63 |
It is a research model: subjects are recognisable, details are not. Many PD12M photos are old, so many results look like old photographs.
Use
mindscend-image-0.2.onnx is one velocity call: inputs x (b×3×64×64, float32), t (b, float32, 1 = noise,
0 = image), y (b, int64 class); output v. Sampling with N Euler steps and guidance g, conditional and
unconditional (y = 10) in one batch of 2:
x = randn(3, 64, 64)
for i = N .. 1: vc, vu = model([x, x], [i/N, i/N], [class, 10]); x -= (vu + g * (vc - vu)) / N
image = clamp(x * 0.5 + 0.5, 0, 1)
coreml/MindscendImage.mlpackage is the same model as a Core ML ML program (FP16, fixed batch of 2: the class and
the no-class row), used by the Mindscend AI Lab app on iPhone, iPad and Mac.
config.json holds the class names and the Lab's defaults (20 steps, guidance 2.0).
License
Weights: Apache-2.0. Training images: PD12M, public domain / CC0 (metadata CDLA-Permissive-2.0).