Jonathan Korstad

SynapseJourney.org - learn anything, while learning how things connect and are similar, all while learning about the very real SOTA problems, challenges, theories, and inventions in any given field or topic. An open source learning platform built for anyone to explore, learn, connect, have fun, and build a visual map of your subject overtime plus earn achievements and unlock secrets along the way. Enjoy! p.s. let me know if there are any additional features you’d like to see or if you run into any bugs along the way. A word and definition created to vaguely sums up the goal of Synapse Journey: Synaptodendrogenesis (~The simultaneous proliferation of neural branches (dendrites) and the formation of new connection points (synapses), resulting in a denser and more complex neural network) 😊

@FossPrime @Dexerto @HardwareSteam @valvesoftware Good vibes!

I made my Steam Controller automatically charge itself @Dexerto @HardwareSteam @valvesoftware

@mervenoyann @huggingface Nice!

dogfooding at every opportunity I just mounted a @huggingface Bucket 🪣 to dump our team offsite pics it took me 4 mins to upload 6 GB of images and didn't have to empty my computer to do it 😄

@mervenoyann @huggingface Happy Birthday!! 🎉

celebrating my birthday at @huggingface Giverny office today it has a pretty garden

@ornith_ Did you use TRL?

@JonathanKorstad ☺️☺️

Aloha! 🌺 Meet Ornith-1.0, a family of open-source LLMs specialized for agentic coding. Ornith-1.0 spans the full parameter sizes including 9B Dense, 31B Dense, 35B MoE, and 397B MoE. It achieves state-of-the-art performance among open-source models of comparable size on coding benchmarks including: ✅Terminal-Bench 2.1(77.5) ✅SWE-Bench(82.4 on verified, 62.2 on pro, 78.9 on Multilingual) ✅NL2Repo(48.2) ✅SWE Atlas(41.2 on QnA, 42.6 RF, 39.1 TW) ✅ClawEval(77.1) Post-trained on top of gemma4 and qwen3.5, Ornith-1.0 employs a novel self-improving training strategy in which reinforcement learning is used to generate not only solution rollouts, but also the task-specific scaffolds that drive those rollouts. By jointly optimizing the scaffold and the resulting solution, the model generate higher-quality solutions in agentic coding.😎 All models are released under the MIT license, enabling full commercial and research use. 📖Tech Blog: deep-reinforce.com/ornith_1_0.html 🤗Huggingface: huggingface.co/collections/de…

@IBMNews pico wow!

The world’s first sub‑1 nanometer node chip is here. Delivering 70% greater energy efficiency, this breakthrough powers a new era of computing that’s more capable while using less energy. Dig into this next-gen tech: ibm.co/6016EOHpM

@peli_kan0 @prayag_sonar @UnslothAI Ent speed inference

@prayag_sonar @UnslothAI You can have it with 0.5t/s

What’s your go-to local model right now?

Wow! 🔎

@thoughtlesslabs @ZachWarunek It happens

@ZachWarunek All the time. But because I forgot to turn on the music and get locked in

you ever just wear noise cancelling headphones and not play any music?

@cb_doge Awesome!!

BREAKING: American Airlines has just confirmed it will begin rolling out Starlink across more than 500 narrowbody aircraft starting early next year.

@Tesla Gotta catch em all

With Charge Stats 2, you can now see your road trips on a map, celebrate your charging milestones & visit iconic Superchargers to earn badges Tesla App > Charging > Charge Stats & Badges

@lazaro45ive would be a cool game someday too though or an explorable VR experience as always, great work! 🔥

people thinking this is from a videogame when I actually made it from scratch in Blender 🤷🏻‍♂️

@bryan_johnson Bryan Button

Am hoping to make the Forbes 30 under 30 this year.

Gemini's vision skills impressively passed this test ⚫️🐜⚫️

@NousResearch 🙂

Your Hermes Agent can now adopt an animated pet: a small sprite that reacts to what the agent is doing (idle, running a tool, thinking, waiting, finishing, failing) in the GUI or TUI. You have nearly 3000 pets to choose from via the petdex gallery, or you can submit your own.

@NousResearch @FrickFracck Thanks! 😊

@FrickFracck Everyone needs companionship

👇

@fodiographer @mao888ahw @hexiang Gracias

@mao888ahw @hexiang

@BillyM2k Beautiful!

on a hike, listening to 90’s grunge music, eating beef jerky good day

At this scale (10T+ params), pre-training doesn't just average—model capacity explodes, letting rare signals carve out distinct subspaces in the latent space without dilution. Novel ideas in data (e.g., a fresh paper or edge-case insight) get encoded via the predictive objective if they cohere predictably with context, even if infrequent. Emergence kicks in: the model starts recombining latent patterns into outputs that feel "new" because no single training example had them exactly. It's not invention from void—it's hyper-efficient compression revealing unseen connections in the data distribution. Post-pretrain fine-tuning or prompting amplifies it further.