Entropia Riko v0.2.0 — AI Workflow Framework

From node-editor prototype to AI workflow framework:

• IDE-style project system (project.riko manifest, 5 templates, browser APIs)
• strong-typed port system with connection validation
• professional nodes: file_input / dataset / checkpoint_save / checkpoint_load
• bake/cache + reproducible experiment records
• multi-workflow metadata + dependency graph
• real AI pipeline examples (CV / diffusion / audio / video)

Entropia Riko v0.1.5

Blender-style default tabs: Train / Inference / Layout, each loading its own root graph.

Entropia Riko v0.1.4

Default tab auto-loads the MNIST root workflow; single practical layout.

Entropia Riko v0.1.3

Fully decoupled UI: frontend mounts the entropia-template-ui npm package.

Entropia Riko v0.1.2

Engineering: modular server routers, ARCHITECTURE/CONTRIBUTING docs, ruff lint + CI, rmsnorm dim fix, repo cleanup.

Entropia Riko v0.1.1

train-to-model / load-from-model nodes (safetensors + Houdini-style file picker), /api/train save_path, train+infer example pairs.

Entropia Riko v0.1.0

Initial release: node-graph deep-learning editor (PyTorch + TensorFlow/Keras).