Puse a Jev, la nueva IA de @typesafeai, a jugar al Tetris en modo súper difícil. Decidió cada jugada en unos 0,3 segundos. Acomodó 357 piezas e hizo 134 líneas en solo 2 minutos. Increíble.
PatternsBeam search
Parallel beam search
Keep the top-k branches at each taxonomy level instead of only the winner, so a wrong first choice can still be recovered.
- Category
- Patterns
- Also known as
- Beam search
- Related terms
- 3
- Directory entries
- 1
- Docs
- docs.typesafe.ai
- Added
- 2026-09-24
Definition
Beam search over Choice probabilities asks one question per surviving node at each level and keeps the best few paths, pruning the rest. It uses the probability distribution the answer already returns as the ranking signal.
It is the mitigation for greedy hierarchical classification: committing to the single top choice at level one turns a 0.7 into a hard decision, while a beam keeps a runner-up alive until the evidence separates.
Tagspatternsclassification
Related terms
Definitions that connect to this one.
Hierarchical classificationWalk a deep taxonomy level by level, using each Choice answer to pick the next level's options, keeping several paths alive.PatternsProbability distributionThe floats summing to 1 across your options or levels that every Choice and Score answer returns under probabilities.Answers & ConfidenceRe-rankingScore each query-candidate pair with one question and reorder a shortlist; the CLERC example lifts top-1 accuracy from 5% to 18%.Patterns
From the directory
Classify documents through deep patent, retail product, biomedical, and source-code hierarchies using parallel beam search over Choice probabilities.
Cookbooks & DemosDocs#official#cookbook#classification
Official
From the community
Posts from builders shipping with Jev right now.
Jev plays Tetris: 134 lines in two minutes
Support answers in a Mac app
First @typesafeai use case, live in our Mac app: setup and troubleshooting help when no model is loaded. Model downloading, load failed, API returning 503, phone won't pair: the user asks, Jev reads the question with the whole built-in manual as state and decides, with Show more
A local Jev build with room to get faster
ローカルTypesafe Jev、できたんやないか？ もうちょい速度は改善余地ありそう
typesafeのJev、これローカルLLMで割といけるんじゃないか？ ていうかそういう使い方するのもローカルLLMの醍醐味だったわけで！
jev-review: a local-first score loop for coding agents
built `jev-review` @typesafeai it's an experimental, local-first MCP plugin that gives coding agents a score quality feedback loop across different metrics. agents call jev while they work, get scored, make improvements, and repeat the loop try below 👇
@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
The X algorithm, rebuilt with Jev
rebuilt the X algorithm with Jev - uses real weights - simulates virality of your post - has a global feed (you see everyone) it's insanely accurate
Hard-coded rules moved to Jev
Now using @typesafeai Jev in aiseotracker.com, linkdr.com, genppt.com, etc AI ends up vibe coding so much AI regex slop if you don't read the code, so I can finally move all this hard-coding to Jev and it's insanely fast! Also for regular LLM Show more
