@OpenAIDevs Imagine a Codex LAN party https://t.co/w9UkQyAsLD

@OpenAIDevs Imagine a Codex LAN party https://t.co/w9UkQyAsLD

Court poster présenté @Institut_Dataia sur les métriques et définitions de la #fairness en #MachineLearning | Revue de littérature des métriques de l'#équité en #IA. 👉https://t.co/jMTdsHdCp8 https://t.co/YVjlzaM8o8

@jxnlco https://t.co/Sdp2FsxF0z

docs: https://t.co/rfQOK0muLR

Seven great design tools with examples from @scuethics https://t.co/5apFa9aMhb

Thoughts? Honestly, it's catching up to Twitter in terms of usefulness to me https://t.co/1KNgrgaE5H

We’re open-sourcing Unlimited OCR — built to read long documents in one pass. With 3B total parameters and only 500M activated, Unlimited OCR sets new end-to-end SOTA results on OmniDocBench v1.5 and v1.6. The key innovation is Reference Sliding Window Attention (R-SWA), inspired by how humans transcribe books: keeping the source, recent context, and next words in focus, while softly forgetting what’s no longer needed. With constant KV Cache size and lower attention cost, Unlimited OCR can transcribe 40+ pages in a single forward pass — without losing context or slowing down. Explore the model👇: --GitHub: https://t.co/5ZJBsEldKd --Hugging Face: https://t.co/4FKFr9EfOu

+1 more

https://t.co/I1x8xQ3g5T

OMG! Fugu Ultra is ridiculously good at these 3D renders. https://t.co/Noi2CCydEG

https://t.co/3zfJqyJuE2

https://t.co/1b4UZDTMFt

https://t.co/zrpic33P60

https://t.co/bNlTKl7PN5

https://t.co/8Ct5biuwp7

https://t.co/MI8mMNDVhb

CharacterAI deck (2021) via garcia v. character technologies, inc. https://t.co/vpxI2Q5gOE

Daniel Cressy has become the first person in the state of Louisiana to be cured of sickle cell disease through gene therapy https://t.co/EsEwQJahKD

Found the dumbest useful optimization in my Hermes Agent setup today: Two hourly maintenance cronjobs were waking the full agent just to run scripts. Switched them to script-only no_agent mode. Savings: ~1.1M tokens/day, ~33M/month. Tiny fix. Huge leak plugged. https://t.co/V4fI2SRwBg

From governance & compliance to community growth & developer engagement, OSPOs are shaping the future of #OpenSource. Speak at OSPOlogy + #OSPOSummit China on September 7 in Shanghai. 🎤 CFP closes July 12. Submit now: https://t.co/SO5xfmgZM2 https://t.co/hclLnOyOOL

What is actually limiting model routers for coding tasks? Most routers treat picking a model as a static, one-off classification. This paper identifies the real bottleneck as information deficit. Simply augmenting a vanilla LLM router with task-dimension-level performance statistics yields a 15.3% relative gain, surpassing a heuristic router built on the same priors. Motivated by that, the authors propose Agent-as-a-Router, which formalizes routing as a Context to Action to Feedback to Context loop. It closes the information gap by accumulating execution-grounded experience during deployment, so the router improves from real outcomes instead of guessing once. Why does it matter? If you orchestrate multiple coding models by cost and skill, treating routing as a learning loop rather than a one-shot classifier is the lever. This means that routing becomes an agentic process that updates its priors as it runs. Paper: https://t.co/717q9Ep45M Learn to build effective AI agents in our academy: https://t.co/LRnpZN7L4c

My wife spent half an hour this morning researching the minimum heights for rides at Disney World and then putting the info in a spreadsheet. Wondering how ChatGPT would do with it, I gave it a few examples and it was able to recreate the information in seconds https://t.co/ntZINZvbuh

@soroushjp @deliprao @abacaj I was working with this one from @huggingface last night - works well https://t.co/ZqW1pjtNAS

Faintly positive on QuickVue yesterday, negative on a Lucira and 2 On/Go (hours apart). Faintly positive on QuickVue again this morning and negative on On/Go. Going to take one more Lucira, but looks like a COVID false positive https://t.co/X9CGpgj861

Thank you all for the overwhelming interest in our new Member of Technical Staff (RSI Lab) role. Just to clarify many questions asked, the role is not remote. We expect candidates to move to Tokyo, Japan. https://t.co/OUnbrC4BNc Please read the job description for more details! https://t.co/oHM6CMlvXl

Something that's been pretty nice in the GitHub Copilot app is how helps me with pixel pushing responsive layouts. I've been working on a design update for my app PocketCal, and I used it to test screen sizes and it gave me screenshots based on all the ranges I asked for! https://t.co/h7P1rKHLP6

I am also open to other suggestions. One goal of this sabbatical is to explore the space of possibilities more. Work at answerai is the baseline to beat: $150k/yr and lots of exploration possible. Would love to hear from others on finding the freedom vs. work Pareto frontier :) https://t.co/uAXAuUX7WD

implicitly or explicitly, that’s got to be the hope of the big players. https://t.co/01rht8Vrgm

LinkedIn messaging is embarrassingly bad. So I built a faster, private, open-source inbox. ⌨️ Superhuman-style shortcuts ⚡ Instant search with local message sync 🌙 Dark mode, emoji, and more 🔒 Runs completely local as a Chrome extension Open source and free. Get it below 👇 https://t.co/ZgFFVAt6t6

https://t.co/J4YHJzccN8

We taught a brand-new mini-series this year at @SCSatCMU on Modern GPU Programming for ML Systems, as part of the ML Systems course, touching on fun questions like what data layout swizzling is, how to use 3D TMA, and state-of-the-art Blackwell programming. We released a curated online book based on the materials: https://t.co/5ZJg2lySNO check it out

Get @glyphengineapp now on the App Store! You a find the perfect emoji or SF Symbol for your app, or generate a custom symbol yourself. Link below: https://t.co/O6j3jKQHQw