frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Show HN: Lar-JEPA – A Testbed for Orchestrating Predictive World Models

https://github.com/snath-ai/Lar-JEPA
2•axdithya•1h ago
Hey HN,

The current paradigm of agentic frameworks (LangChain, AutoGPT) relies on prompting LLMs and parsing conversational text strings to decide the next action. This works for simple tasks but breaks down for complex reasoning because it treats the agent's mind like a scrolling text document.

As research shifts toward Joint Embedding Predictive Architectures (JEPAs) and World Models, we hit an orchestration bottleneck. JEPAs don't output text; they output abstract mathematical tensors representing a predicted environmental state. Traditional text-based frameworks crash if you try to route a NumPy array.

We built Lar-JEPA as a conceptual testbed to solve this. It uses the Lár Engine,a deterministic, topological DAG ("PyTorch for Agents") to act as the execution spine. Key Features for Researchers: Mathematical Routing (No Prompting): You write deterministic Python RouterNodes that evaluate the latent tensors directly (e.g., if collision_probability > 0.85: return "REPLAN"). Native Tensor Logging: We custom-patched our AuditLogger with a TensorSafeEncoder. You can pass massive PyTorch/NumPy tensors natively through the execution graph, and it gracefully serializes them into metadata ({ "__type__": "Tensor", "shape": [1, 768] }) without crashing JSON stringifiers. System 1 / System 2 Testing: Formally measure fast-reflex execution vs. deep-simulation planning. Continuous Learning: Includes a Default Mode Network (DMN) architecture for "Sleep Cycle" memory consolidation.

We've included a standalone simulation where a Lár System 2 Router analyzes a mock JEPA's numerical state prediction, mathematically detects an impending collision, vetoes the action, and replans—all without generating a single word of English text. Repo: https://github.com/snath-ai/Lar-JEPA Would love to hear your thoughts on orchestration for non-autoregressive models.

OpenClaw Hype: Analysis of Chatter from the Dark Web (BleepingComputer)

https://www.bleepingcomputer.com/news/security/the-openclaw-hype-analysis-of-chatter-from-open-so...
1•ildar•57s ago•0 comments

Any Website Can Take Full Control of Your OpenClaw Agent (Oasis)

https://www.oasis.security/blog/openclaw-vulnerability
1•ildar•2m ago•0 comments

Show HN: Cecil – open-source memory and identity protocol for AI

https://github.com/johnkf5-ops/cecil-protocol
1•JohnKnopf•5m ago•1 comments

New AirSnitch attack breaks Wi-Fi encryption in homes, offices, and enterprises

https://arstechnica.com/security/2026/02/new-airsnitch-attack-breaks-wi-fi-encryption-in-homes-of...
1•brie22•6m ago•0 comments

Show HN: HeadshotAI – Professional AI headshots in 30 seconds for £4.99

https://headshotai-mvp-jb.netlify.app
1•jackbventures•14m ago•0 comments

China piles pressure on Japan after Takaichi Sanae's triumph

https://www.economist.com/asia/2026/02/26/china-piles-pressure-on-japan-after-takaichi-sanaes-tri...
2•petethomas•20m ago•0 comments

The Cron King

http://thecronking.com/
1•thebuilderjr•25m ago•0 comments

Can LLMs reason about math? The Subtraction Trick Test

https://haversine.substack.com/p/can-llms-reason-about-math-the-subtraction
3•MakeAJiraTicket•25m ago•0 comments

Show HN: Toggl lock‑screen widget for real‑time time tracking

https://apps.apple.com/ca/app/toggl-lockscreen-widget/id6756894639
1•dakoder36•30m ago•0 comments

How Anthropic Teams use Claude Code [pdf]

https://www-cdn.anthropic.com/58284b19e702b49db9302d5b6f135ad8871e7658.pdf
2•carlsborg•30m ago•0 comments

User Interview #3

https://twitter.com/AviSchiffmann/status/2026798365489725742
1•d_silin•31m ago•0 comments

Show HN: CodeLeash: framework for quality agent development, NOT an orchestrator

https://codeleash.dev/
2•cadamsdotcom•32m ago•0 comments

Principia Mathematica, Vol. 1

https://www.gutenberg.org/cache/epub/78050/pg78050-images.html
2•petethomas•32m ago•0 comments

Show HN: VeryBot – Self-hosted AI assistant for work

https://github.com/charlie0077/verybot
1•verybot•33m ago•0 comments

The Hunt for Dark Breakfast – Can we derive breakfasts we have never observed?

https://moultano.wordpress.com/2026/02/22/the-hunt-for-dark-breakfast/
32•moultano•37m ago•13 comments

Hong Kong jailed my dad to stop me speaking out, says activist

https://www.bbc.com/news/articles/cly897pv4j9o
2•tartoran•37m ago•0 comments

Parakeet.cpp – Parakeet ASR inference in pure C++ with Metal GPU acceleration

https://github.com/Frikallo/parakeet.cpp
7•noahkay13•38m ago•1 comments

Preserving the Web Is Not the Problem. Losing It Is

https://www.techdirt.com/2026/02/17/preserving-the-web-is-not-the-problem-losing-it-is/
3•the-mitr•43m ago•0 comments

Frankensqlite a Rust reimplementation of SQLite with concurrent writers

https://frankensqlite.com/
1•rahimnathwani•43m ago•0 comments

Show HN: Selling my painting to fund a Steam Cleaning startup in Libya

https://juicebox.money/v5/base:139
1•Ali_ll•44m ago•1 comments

Flexport Atlas: Browse the world's container shipping network

https://atlas.flexport.com
1•thedogeye•45m ago•0 comments

How Predator iPhone Spyware Can Defeat iOS Orange and Green Recording Indicators

https://www.jamf.com/blog/predator-spyware-ios-recording-indicator-bypass-analysis/
3•transpute•47m ago•0 comments

A Nationwide Book Ban Bill Has Been Introduced in the House of Representatives

https://bookriot.com/hr7661-book-ban-legislation/
19•LostMyLogin•52m ago•1 comments

The Compact Disc Story – Kees Immink

https://www.researchgate.net/publication/294484774_Compact_disc_story
2•pipeline_peak•53m ago•0 comments

The Web Weaver's Magical Loom

https://sidwashere.substack.com/p/the-web-weavers-magical-loom
1•SidsCreativTech•53m ago•0 comments

GitHub code view fails to load when content contains non-ASCII characters

https://www.githubstatus.com/incidents/kv1lzpgzr9yp
1•csmantle•55m ago•1 comments

Brain on "Not X but Y"

https://hollisrobbinsanecdotal.substack.com/p/metannoying
1•HR01•58m ago•0 comments

Exploring the Gleam FFI (2024)

https://www.jonashietala.se/blog/2024/01/11/exploring_the_gleam_ffi/
1•notagoodidea•58m ago•0 comments

Faster Than Light Travel, Free from Paradoxes

https://nirmalyakajuri.substack.com/p/faster-than-light-travel-free-from
1•virgildotcodes•59m ago•0 comments

Astronomers Wake Up to 800k Notifications from Observatory

https://gizmodo.com/astronomers-wake-up-to-800000-notifications-from-observatory-watching-the-nig...
1•wglb•59m ago•0 comments