arxiv.org
A personalized algorithmic feed based on your likes. It finds people who liked the same posts as you, and shows you what else they've liked recently.
Feed on BlueskyFeeds Stats
- 💙 Liked by 30,215 users
- 📅 Updated about 1 month ago
- ⚙️ Provider linklonk.com
- 📈 In the last 30 days, there were 812 posts about this feed. These posts got a total of 13K likes and had 5.4K reposts.
For You Likes over time
Like count prediction
The feed For You gains approximately 1.7K likes per day.
Feed Preview for For You
this is a very important paper, although not easy to read
it takes hyperconnections (HC), which is basically just a smarter way to do back propagation, and stabilizes it so that it’s actually usable at scale
Alexander Doria@dorialexander.bsky.social
So we already have the first major paper of 2026, DeepSeek mHC: Manifold-Constrained Hyper-Connections. arxiv.org/pdf/2512.24880 Short thread: this is actually an engineering paper, taking as a starting points ideas already exposed in an original Hyper-Connections (HC) paper from ByteDance.
0
3
40
I am so so glad that melatonin is an OTC drug.
2
0
20
Saving lives is cool, but if you write really well about parenthood you can cause lives.
2
2
21
Saving lives is cool, but if you write really well about parenthood you can cause lives.
0
0
5
Boy it sure was nice when Anthropic's limits were doubled.
0
0
13
Happy Gary Marcus was wrong about 2024 and also about 2025 day for those who celebrate.
4
10
37
If you have 10-20 tweets on a theme, you should patch the holes and turn them into an essay on the real internet.
4
3
46
25 years of progress in audio (and IT) technology: I replaced my old bulky Sony class A amplifier (2x 60W) by a new and shiny FOSI class D amplifier ... a tiny tiny box producing 2x 300W. Class D = digital, uses switching to amplify the signal. Efficient, the metal box doesn't even have holes.
2
0
8
“Can’t believe NYC elected a Muslim as mayor.”
In truth—a Black American Christian woman swearing in a Ugandan American Muslim of South Asian ethnicity as his Syrian American wife holds the Qurans he swears in on, all while standing on the steps of a subway station—is the most NYC thing imaginable.
702
4186
25636
Cameras, lenses, framing, and everything in-between have fascinated me for many year.
This morning I read Bartosz Ciechanowski's article on the subject. It's the best explainer I've seen. The interactivity really sells it.
Great article to kick off your year with:
ciechanow.ski/cameras-and-...
0
1
11
Struggling to find a more accurate description of parenthood than my two year old sat on a stool singing me incy wincy spider whilst I'm in the shower
0
0
6
add "ask me a few clarifying questions" to any prompt to get better alignment, especially in early planning mode
also works with people but they get cranky if you ask too often
1
0
13
When clocks strike midnight on the first day each year, a quiet creative revolution starts as hundreds of works move from private control into the public domain, unlocking stories, songs, and images for anyone to reuse.
The world celebrates this big moment with an official Public Domain Day. (1/2)
1
94
263
Happy New Year.
THERE. I said it & I really mean it, even though the odds are it will be every bit as challenging & frustrating as the last one.
media.tenor.com
a man with a beard and a black shirt is sitting in a dark room and says `` not amused '' .
ALT: a man with a beard and a black shirt is sitting in a dark room and says `` not amused '' .
293
518
5418
1995: Never EVER run any code you find on the Internet!
2025: curl -LsSf nobodyeverreadsthis/install.sh | sh
4
0
26
Simon Willison has a great summary of LLM progress in 2025:
• Reasoning models have made LLMs useful for web search.
• AI agents now work well for Deep Research and Coding.
• AI image generation went mainstream.
• Vibe coding became a big business.
• OpenAI lost the lead & Google’s Gemini caught up
simonwillison.net
2025: The year in LLMs
This is the third in my annual series reviewing everything that happened in the LLM space over the past 12 months. For previous years see Stuff we figured out about …
7
16
91
There's this crank on Twitter who spends all day posting about how his theory of consciousness is so obviously correct and everyone who disagrees just needs to be corrected once they find the right way to do it and it's simultaneously the most condescending and ridiculous thing I've read.
2
0
12
So we already have the first major paper of 2026, DeepSeek mHC: Manifold-Constrained Hyper-Connections. arxiv.org/pdf/2512.24880
Short thread: this is actually an engineering paper, taking as a starting points ideas already exposed in an original Hyper-Connections (HC) paper from ByteDance.
arxiv.org
1
8
47
Call me suspicious and cynical, but a new player suddenly drops a 40B model that beats all of the 400B+ frontier models at coding? 🤔
Amazing if true, but I think I’ll wait until real-world performance is confirmed independently.
Sung Kim@sungkim.bsky.social
The best open-weight coding LLM? Another Chinese quant company, like High-Flyer/DeepSeek, decided to enter AI and released the IQuest-Coder-V1, which is the best coding open-weight LLM per them. Blog: iquestlab.github.io Models: huggingface.co/collections/...
0
0
6
Like most human beings, chatbots struggle to make new discoveries. Fortunately there’s a tiny irritating minority of humans who do make discoveries, in part by systematically doubting received explanations. Not clear whether it would be possible or ethical to increase their number.
www.nytimes.com
Opinion | I Asked ChatGPT to Solve an 800-Year-Old Italian Mystery. What Happened Surprised Me.
7
2
32
This post was a massive success so wanna repost it here. Dunno if BlueSky is now more open for these topics? steipete.me/posts/2025/s...
steipete.me
Shipping at Inference-Speed | Peter Steinberger
Why I stopped reading code and started watching it stream by.
6
4
37