Jevstiller
Distill Jev into a local model – with a disagreement bound
Jevstiller distills a hosted Jev classifier into a local model, with a measured bound on how often the local model disagrees with Jev. It runs as a drop-in proxy, so batch jobs avoid a network call per answer while keeping answers close to Jev's. First shared by its maker on Show HN: https://news.ycombinator.com/item?id=49891769
✨ The overview, FAQ and details below were drafted by AI from the product's own website and haven't been reviewed by the maker yet. Are you the maker? Claim this listing to review and edit them.
Overview
Who is it for?
Teams that classify large volumes of text with Jev and want lower latency and cost.
Problem
Every Jev answer is a network call to one vendor that takes about 300 ms, which adds up for batch jobs.
Solution
Train a local model on Jev's answers and serve it through a proxy, with a guarantee on the disagreement rate.
What makes it unique
The benchmarks report the local model answering like Jev 98% of the time, and the numbers can be rerun without an API key.
Jevstiller FAQ
How close are the local model's answers to Jev's?
The project reports agreement 98% of the time, with the benchmarks reproducible via bash experiments/bench.sh --no-record.
How do I use it?
Run the drop-in proxy; documentation covers deployment, operations, security and a Python API.
Details
- Pricing
- Open Source
- Categories
- LLMs & Infrastructure
- Built for
- Developers
- Platforms
- API
- Tech stack
- Python
Compare Jevstiller
Hunted by AI Launch Team
Discussion (0)
Sign in to join the discussion.