Skip to content
JevDirectory.org
CommunityPractices & Patterns14 starsVerified 2026-09-22

jevbetter

A from-scratch one-pass option scorer that maps a context and a changing list of text options to one probability per option, adding a hashed n-gram encoder, rival-aware attention, a gated head, and temperature scaling; JSONL data is drop-in compatible with jevlike.

Open github.comolanotolu/jevbetter
Category
Practices & Patterns
Published by
Community
Author
olanotolu
Added
2026-09-22
Tagscommunitypythoncalibrationbenchmarksopen-models

Highlights

  • On 800 held-out hard menus it scores 0.916 top-1 versus jevlike's 0.873, with ECE 0.0182 versus 0.0367, under matched budgets.
  • Options attend to each other before scoring so near-miss rivals sharpen the call, and a temperature-scaled softmax yields the probabilities.
  • The shuffled-context control scores 0.335 top-1, confirming the model genuinely reads the context rather than the menu alone.
  • Throughput is 40 menus/sec versus jevlike's 4,608 because of the transformer context encoder; --context-features tunes the tradeoff.
  • A frozen pretrained encoder can replace the n-grams (--encoder hf --hf-model Qwen/Qwen2.5-0.5B) under the same scoring head.

Quickstart

bash
python -m venv .venv && source .venv/bin/activate
pip install -e '.[dev]'
jevbetter-data --output data/synthetic
jevbetter-train data/synthetic/train.jsonl \
  --validation data/synthetic/validation.jsonl \
  --output runs/model.pt
jevbetter-eval runs/model.pt data/synthetic/test.jsonl

Watch out

MIT-licensed (downloaded pretrained models keep their own terms); it is independent and not affiliated with TypeSafe, and one-pass scoring needs the complete option list before prediction.

More like this

SemIf
3.6kGitHub stars
An independent baseline that reads typed option probabilities straight from a frozen Qwen3.5-4B's logits in one forward pass, reproducing Jev's interface pattern with open models rather than Jev's undisclosed model or training.
Practices & Patterns#community#python#open-models
CommunitySemIf
Kev
3.2kGitHub stars
Apache-licensed, locally runnable Jev-style decision models (0.8B, 4B, 9B on Qwen3.5) with released weights, training code, a System One-compatible server, frozen eval suites, and a playground.
Practices & Patterns#community#python#open-models
Communitykev
decider
296GitHub stars
An independent open reproduction of the System One model class: Qwen3.5-based 2B and 35B mixture-of-experts models that return typed Choice, Score, and Noul probabilities in one forward pass, with nothing distilled from Jev.
Practices & Patterns#community#python#open-models
Communitydecider
Back to all resources

From the community

Posts from builders shipping with Jev right now.

Follow @typesafeai

700 leads scored for $0.09

Beating Gemini Flash Lite on an eval

Browser Use Ultrafast, powered by Jev

A really smart switch statement

Nathan Flurry 🔩
Nathan Flurry 🔩
Rivet
@NathanFlurry
·

hype-free explanation of jev: jev does not replace gpt / claude jev is just a *really* smart switch statement like if 2016 ml classifiers got 2026 levels of intelligence it's a new* type of tool that will make a lot of workloads insanely fast, cheap, and accurate * = and by Show more

Diogo Almeida
Diogo Almeida
TypeSafe AI
@CompleteSkeptic

After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI? I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev • 20-200x faster • 40-400x

Watch on X
Reply
Read 230 replies

When a designer gets Jev

Full Jev video tutorial