The Sad State of Gaza (youtube.com)

1 points by bazalia 10s ago 0 comments

Zen vs. Tribalism (amandaknox.substack.com)

1 points by rossant 1m ago 0 comments

DeLLMa: Decision Making Under Uncertainty with Large Language Models (dellma.github.io)

1 points by giuliomagnifico 3m ago 0 comments

WTF? Is GPT Channeling the Bible? (scribd.com)

2 points by ocbcordoba 3m ago 1 comments

The Decade of Deep Learning (2019) (bmk.sh)

1 points by serviette 8m ago 0 comments

MetaMask extension bug causes 100s of GBs of extraneous data to be written (tomshardware.com)

2 points by fortran77 8m ago 0 comments

AMD CEO Sees Chips from TSMC's US Plant Costing 5%-20% More (bloomberg.com)

3 points by mfiguiere 9m ago 0 comments

Developing Our Position on AI (recurse.com)

1 points by jakelazaroff 9m ago 0 comments

How do you manage your taxes? (taxhero.vc)

1 points by salleisha 12m ago 1 comments

I co-authored books with GPT. It became more than a mirror

1 points by KindLight 12m ago 0 comments

AI industry's size obsession is killing ROI, engineer argues (theregister.com)

2 points by rntn 13m ago 0 comments

Nintendo Switch 2 physical game price differences (sethmlarson.dev)

3 points by SethMLarson 15m ago 0 comments

ICE block founder's wife fired by DOJ in retaliation for the app (newsweek.com)

3 points by arunabha 15m ago 1 comments

LLMs Transmit Traits via Hidden Signals to Other Models (twitter.com)

1 points by Amezarak 16m ago 0 comments

AWS merges malicious PR into Amazon Q (lastweekinaws.com)

2 points by QuinnyPig 16m ago 0 comments

Evidence of death row inmate Robert Roberson's innocence stronger (expressnews.com)

1 points by YeGoblynQueenne 17m ago 0 comments

AccuWeather to discontinue free access to Core Weather API (developer.accuweather.com)

26 points by TerribleTurnout 17m ago 9 comments

Ordering Without Barriers: Engineering Our Websites for Accessibility (thesidedish.flipdish.com)

2 points by marc_omorain 17m ago 0 comments

Solid gold superheated to 14 times its melting temperature (nature.com)

2 points by Bluestein 18m ago 0 comments

Within 5 Years, All Engineers Will Be Systems Integrators (quicktea.ai)

2 points by bhaktasn 19m ago 0 comments

iOS and Android are heading in opposite design directions (birchtree.me)

1 points by speckx 22m ago 0 comments

Helm vs. Kustomize: Differences Explained and How to Combine Them (substack.com)

1 points by kirshiyin 22m ago 0 comments

Show HN: UGMM-NN – A FF neural network using univariate Gaussian mixture neurons (github.com)

1 points by zakeria 23m ago 0 comments

Justice Dept. Told Trump in May That His Name Is Among Many in the Epstein Files (wsj.com)

18 points by impish9208 25m ago 3 comments

Ask HN: Best Service for Blog Newsletter?

1 points by fxtentacle 26m ago 0 comments

Medical Hypotheses: Could body piercing be a cause of rheumatoid arthritis? (sciencedirect.com)

1 points by nateb2022 28m ago 0 comments

The world is run by old men in a hurry (archive.is)

2 points by akyuu 34m ago 0 comments

Julian LeFay, Father of the Elder Scrolls and demoscene pioneer has passed away (twitter.com)

1 points by g4v75o 34m ago 1 comments

AppleCare One launches as a single plan to cover multiple Apple devices (appleinsider.com)

1 points by busymom0 35m ago 0 comments

For those who run Fedora as a server (versus CentOS/Alma/Rocky), why? (old.reddit.com)

2 points by sipofwater 36m ago 0 comments

Explore Kinabalu Park and More UNESCO World Heritage Sites (blog.google)

2 points by gnabgib 36m ago 0 comments

Major Rule About Cooking Meat Turns Out to Be Wrong (seriouseats.com)

21 points by voxadam 41m ago 9 comments

FCC proposal aims to Nix long-term gigabit internet speed goals (tomshardware.com)

4 points by NN88 42m ago 1 comments

Silicon Valley AI Startups Are Embracing China's Controversial 996 Work Schedule (wired.com)

5 points by pranshum 44m ago 6 comments

Notable People (Map) (tjukanovt.github.io)

2 points by rmason 44m ago 1 comments

How to increase your surface area for luck (usefulfictions.substack.com)

33 points by jger15 48m ago 9 comments

Show HN: I built a place where product makers can anonymously confess their sins (confessionsrepo.com)

3 points by gigamick 49m ago 0 comments

Ask HN: What NAS (Synology) do you use in your home lab as a developer?

6 points by lavren1974 51m ago 6 comments

PulseTimer – AI-Powered Work Timer with Smart Presets and Privacy-Focused Design (timer.toxi360.org)

2 points by toxi360 51m ago 1 comments

Graph-based vector indices explained through the "FES theorem" (blog.kuzudb.com)

5 points by semihsalihoglu 51m ago 1 comments

Why Do Victims of Massacres Go to Their Deaths? (benlandautaylor.com)

19 points by jger15 52m ago 23 comments

It's a Beelink, Baby (world.hey.com)

4 points by 0xedb 54m ago 0 comments

Wayback 0.1 Released (wayback.freedesktop.org)

2 points by wicket 55m ago 0 comments

Trump administration ramps up its campaign against colleges (cnbc.com)

8 points by standardUser 57m ago 1 comments

Chris Marker's 'Dialector' Foreshadowed Machine Intelligence (frieze.com)

2 points by kosmavision 58m ago 0 comments

Building low-feature apps is good source of calm (indiehackers.com)

3 points by Norcim133 1h ago 0 comments

The Case for Sabotage (collectiveactionintech.substack.com)

4 points by thunderbong 1h ago 0 comments

Microsoft SharePoint victim count hits 400 orgs in ongoing attacks (theregister.com)

7 points by rntn 1h ago 0 comments

The soft body physics of JellyCar (gamedeveloper.com)

3 points by jcalx 1h ago 0 comments

Employee – CEO pay gap historically wide (cnn.com)

43 points by e12e 1h ago 25 comments

End-to-end long-horizon reasoning with one Transformer model

5 hyluo 3 7/23/2025, 1:15:39 PM subconscious.dev ↗

Comments (3)

hyluo · 6h ago

- We build the Thread Inference Model (TIM) based on the transformer architecture, and its dedicated runtime TIMRUN.

- TIM + TIMRUN = Intelligent workflow generation, context engineering, and multi-hop tool use happens at the runtime level

- TIM + TIMRUN supports virtually unlimited reasoning enabled by context pruning, significantly improves the efficiency for long-horizon reasoning tasks

- Inference API is live at https://subconscious.dev/

- More details: https://github.com/subconscious-systems/TIMRUN

datapalo · 4h ago

I’m more than ready to stop my workflow being derailed by opening new chats—looking forward to LLM assistants that actually remember our conversation.

kevin8704 · 5h ago

Really cool work! I saw the “Selective Working Memory” section, are there hooks to swap in a custom retrieval store or memory layer, or is that all internal to TIM?

Thanks for sharing!