EpochGo: local-first multi-agent AutoML on Ollama, $0 API costs and 100% offline
Grouchy-Excitement28 · reddit · 2026-10-13
EpochGo is an open-source, local-first AutoML engine running three agents on Ollama (qwen2.5-coder:7b, 4.7GB VRAM at 4-bit): a Feature Architect (profiling, IQR outlier clipping, serialized Scikit-Learn pipelines), a Model Tournament agent (unbiased eval of 7 baselines incl. XGBoost/CatBoost, top-3 hyperparameter search), and a Serving agent that generates a standalone FastAPI microservice with an in-browser playground. Engineering: asyncio.tothread for CPU-bound sweeps, AST-parsed LLM code with deterministic fallbacks. Empirical note: on small tabular datasets (<5k samples), regularized linear baselines often beat boosted trees. MIT licensed, Docker one-command deploy.
More from coding & agent
- LangChain's deepagents agent harness hits 30k GitHub stars — LangChain · 2026-10-13
- AI instantly adds cross-screen local multiplayer to Halo decompilation — _sholtodouglas · 2026-10-13
- diginomica: context and harnesses make agents smarter, but humans must stay in the loop — jonerp · 2026-10-13
- UCSD's Mental-Models framework gives agents explicit Theory-of-Mind for multi-agent systems — UCSanDiego · 2026-10-13
- Dev pushes back on 'slop' discourse: AI also supercharges debugging and refactoring — jonmarkgo · 2026-10-13
- LocalLM Lab SDK adds MCP Apps support, rendering tool UIs inline in chat — AdventurousKeys · 2026-10-13