Key Info
An OpenCode developer describes training a small local model to classify shell commands into categories for a UI fix, and notes that while the newer Jev model can now handle it, a local approach may still be better.
Highlights
- A tiny local model was trained to solve a UI issue by classifying shell commands.
- The developer says the newer Jev model can now do this too.
- The developer still believes a local approach would work better, likely for practical reasons such as latency, privacy, or cost.