Letting Codex+Astra agents reproduce and fine-tune an Edge Impulse bird sound classifier
dansitu · x · 2026-11-03
Developer dansitu is testing AI coding agents on a bird sound classifier project (sparrow vs parakeet vs background noise): having Codex+Astra fetch the float32 TFLite model from Edge Impulse, reproduce its results locally, then edit the weights for improvement.
Slow airport Wi-Fi is slowing the dataset download, but on a sample of the data the agents already claim some improvement. An in-progress look at autonomous agent experimentation on a real ML task.
More from coding & agent
- Agent-Built 'Deimos' Encyclopedia Profiles 40 Designers Via Multi-Model Drafting — evilrabbit_ · 2026-11-03
- Sentry ships MCP server replay fix hours after a developer's agent feedback — zeeg · 2026-11-03
- A one-night AI-built project nets 333 points and ~22k visitors — Aizkmusic · 2026-11-03
- Five manual checks before trusting an agent actually finished its job — Kodenstudio · 2026-11-03
- Open-source AI pit crew for iRacing runs 5 agents on Kimi, checks every number against telemetry — JeremyCMorgan · 2026-11-03
- Agent control plane benchmark: Smalltalk-80 cascades hit 100% compliance vs 2.5% for Gherkin — RandalSchwartz · 2026-11-03