typed

40 stories and discussions about typed, aggregated from every source we track.

1.

Evaluates typed decisions (choice, score, noul) over 100+ languages in a single forward pass with calibrated probabilities. Outperforms TypeSafe Jev.

1315 points•nandakishor_ml•12 days ago•312 comments•
2.

Run typed option logits and autoregressive JSON locally with open models and WebGPU.

706 points•ilreb•13 days ago•287 comments•
3.

52 Jev-class systems tested on 534 decisions. Jev 1.13.0 leads JevBench v1.3.0 with 74.4; compare open-source, self-hostable and hosted options.

149 points•florianstandhar•9 days ago•39 comments•
4.

Agent evals and guardrails in one request. Built on Jev, Kev and Laya. - openlayer-ai/jevals

47 points•gbayomi•10 days ago•6 comments•
5.

How Jev's typed probabilistic decisions, LangGraph workflows, and Tenuo task-scoped warrants combine in a dependency upgrade agent.

22 points•niyikiza•8 days ago•4 comments•
6.

PHP 7.4 adds typed properties to classes

7 points•magikid•over 7 years ago•1 comment
7.

Strongly typed Elasticsearch DSL written in Rust. Contribute to vinted/elasticsearch-dsl-rs development by creating an account on GitHub.

6 points•rsdbdr203•about 4 years ago•1 comment
8.

Well-Typed are happy to announce Eventlog Live, a program that streams real-time telemetry from any Haskell application to any observability platform that supports the OpenTelemetry protocol.

5 points•abhin4v•6 days ago•0 comments
9.

typesleeve w/ keyboard · 1913-1918, likely later — from the Dynamically Typed collection.

4 points•Bluestein•7 days ago•0 comments•
10.

Single-forward-pass semantic routing in Rust: turn dense LLMs (Llama, Qwen, Mistral, Gemma) into a typed decision API with LoRA/QLoRA training and FP8/FP4 quantization, built on Candle. - neurono-m...

3 points•andrelgcclaudin•5 days ago•1 comment•
11.

Choose your model, tune the research workflow, and take the session with you. The Grainulator research playground.

3 points•aid-ninja•11 days ago•1 comment•
12.
2 points•fanf2•1 day ago•0 comments•
13.

A workspace and framework to build domain-specific systems for agentic knowledge work. - arsumbris/arsumbris

2 points•handfuloflight•3 days ago•0 comments•
14.

The agent reports through a fixed set of typed blocks instead of prose, and the UI draws them.

2 points•skolos•4 days ago•0 comments•
15.

Peek at an image, get booleans back: typed visual questions (choice, score, yes/no) with calibrated probabilities from a 500M VLM, ~400 ms on a MacBook - bykof/peekaboolean

2 points•bykof•4 days ago•0 comments•
16.

An atmospheric 3D maze where JEV's route decisions come alive.

2 points•baigumair•6 days ago•0 comments•
17.

Jev is TypeSafe AI's decision model. Learn what state and typed questions mean, how its answers work, and where type safety ends.

2 points•flashbrew•10 days ago•0 comments•
18.

Our own benchmark of typed-decision models: Jev, its open rebuilds and instruction models on the JevBench Score — Intelligence, Calibration, Speed and Cost, 25 % each, geometric mean.

2 points•oscarfr•10 days ago•0 comments•
19.

Native MLX runtime for Laya typed decision models — 7–14 ms short decisions on M3 Max. No text generation, PyTorch, or cloud API. - mizorewww/laya-mlx

2 points•blueaquilae•11 days ago•1 comment•
20.

DuckDB extension: typed Jev answers as real SQL types - colliber/duckdb-jev

2 points•jmrothweiler•11 days ago•0 comments•
21.

Under 120 MB. One forward pass returns typed, calibrated answers. By toolchain.studio.

2 points•namanvyas•11 days ago•0 comments•
22.

Spreadsheets made for people in, typed, row-identified, versioned, governed tables to your agent, Excel in its original formatting back out. MCP, CLI, REST and SDKs.

1 points•keikubo•about 2 hours ago•0 comments•
23.

The agent reports through a fixed set of typed blocks instead of prose, and the UI draws them.

1 points•skolos•1 day ago•0 comments•
24.

The agent reports through a fixed set of typed blocks instead of prose, and the UI draws them.

1 points•skolos•2 days ago•0 comments•
25.

The agent reports through a fixed set of typed blocks instead of prose, and the UI draws them.

1 points•skolos•3 days ago•0 comments•
26.

Portable local probabilistic inference runtime. Contribute to bulyaki/Credence development by creating an account on GitHub.

1 points•naughtiusmax•4 days ago•0 comments•
27.

Jev is the first in a new class of models that allow us to build LLM applications more efficiently and at lower cost. Typed questions in, typed answers with probabilities out.

1 points•mooreds•6 days ago•0 comments•
28.

An open-source, high-performance System One Model for one-pass typed decisions, with an embeddable C runtime and WebAssembly support. - sqliteai/blink

1 points•marcobambini•7 days ago•0 comments•
29.

Moggi Compiler. Contribute to moggi-lang/moggi development by creating an account on GitHub.

1 points•prolic•7 days ago•1 comment•
30.

The reference for building an app and its cloud as one typed program, with Effect, Alchemy, and a fence that raises the floor.

1 points•handfuloflight•7 days ago•0 comments•
31.

A strongly-typed, polyglot compiler. Contribute to morloc-project/morloc development by creating an account on GitHub.

1 points•arendsee•8 days ago•0 comments•
32.

Jev (TypeSafe System One) inner loop for ego-browser — one ~0.4s typed decision per DOM step instead of an LLM turn. Agent skill for ego lite. - ZephyrDeng/ego-jev

1 points•ZephyrDeng•8 days ago•0 comments•
33.

Gemma 3 4B as a Jev-style System One decision function: one forward pass, choice logits, softmax. Native Rust + Candle. No generate(). - zozo123/gemma-to-jev

1 points•zozo123-IB-IL2•9 days ago•0 comments•
34.

Typed semantic decisions for pandas and Polars with TypeSafe Jev

1 points•jonbaer•9 days ago•0 comments•
35.

Try the Jev AI Playground online, then learn how typed decisions, confidence, and structured AI outputs work.

1 points•lovelyxyz520•10 days ago•0 comments•
36.

Jev walks a real Pokémon FireRed ROM. Contribute to daniel4x/JevEmon development by creating an account on GitHub.

1 points•alfasiii•10 days ago•0 comments•
37.

Your last chat with an AI holds the wreckage of a scorecard. Here’s how to recover it.

1 points•grrrtttt•10 days ago•0 comments•
38.

Local typed decisions, contrastive data curation, and model evaluation. - bespokelabsai/nimble

1 points•Bluestein•10 days ago•0 comments•
39.

Build type-safe, real-time TypeScript applications from one server contract - leonardoventurini/typeferry

1 points•leonardovb•10 days ago•0 comments•
40.

mamori is an open-source Go library that loads config and secrets from 30+ sources into typed, validated structs and keeps them reconciled at runtime.

0 points•xavidop•1 day ago•0 comments

Related topics