How imdigitalashish/decentralized-mesh-model-training works
Project type and purpose
• This is a full-stack, decentralized mesh model training system.
• The goal is to coordinate many worker nodes in a peer-to-peer or master/worker mesh, distributing ML training tasks, and provide a web UI for monitoring/control.
File structure
Top level:
– README.md
– backend/
– frontend/
Backend folder:
– master_server.py, worker_node.py, model.py, utils.py
– templates/main.html for server-side views
– experiment/ subfolder containing Go and Python experiments (main.go, main.py, worker.py, master.py)
Frontend folder:
– Next.js app (src/app) in TypeScript with globals.css, layout.tsx, page.tsx
– config files (package.json, tsconfig.json, eslint, postcss)
README
• No textual architecture description beyond that image link.
Detail
• Python script or Go binary as a micro-component if it has distinct responsibilities.
• Template rendering as part of the API layer.
• Mesh links if worker nodes talk to each other directly.