HyperGAN

Stabilizing GANs by replacing known priors with learnable particle clouds.

Experimental MiniMax H3 source: historical UNI and guarded shared-session candidate; CPU software checks only

Experimental Ripple NVFP4 shared ParticleGAN source; full-model training and GPU qualification remain blocked

Native GLiDE decision-readout particle sliders, shared ParticleGAN integration, and reproducible preference experiments

Source-only P-H3-Image Particle Sliders integration using the shared HyperGAN core; native operations pending license and qualification

Source-only Z-Image shared-game CPU preview extracted for review from HyperGAN/particle-sliders

Supra2-IMG concept sliders: 1,600 steps in 134.4s on RTX A6000, with 6.8 MB LoRAs and 3.4 MB rank-8 distills.

Composable GAN framework with api and user interface

Source-only MiniMax-H3-Image Particle Sliders planning scaffold; native operations unavailable

Muse Glimmer particle-slider integration using upstream ParticleGAN and ridge LoRA distillation