CTO & co-founder @v7labs. Reliable AI agents at work, strange small models and weird interfaces for fun. I build everything I write about.

London, England
Joined October 2010
Include

Only show posts containing:

Exclude

Hide posts containing:

Time range
-
Minimum likes
Replying to @askmaddyy
Glad you liked it!
1
12
Replying to @superdoccimo
Very cool to see you trying to reproduce it. Will share full weights in a few days, instead of just int4.
6
Replying to @_figarroh_
Sorry about that, if you dm me your email I’ll delete it tonight and fix the bug! How did you find it?
2
I missed that one, thanks for sharing.
1
17
Replying to @ChadNauseam
Thank you!
1
14
Replying to @vitobotta
You can actually try it in the browser! But yes. I tried `open index.html` and it spat out hex code. When I decoded it, it was actually html, not sure how that happened.
1
8
Me: Mum, can we have a world model? Mum: We have a world model at home. World model at home:
2
177
Replying to @maxbittker
Thank you, Max! Really appreciate it. I did spend a few months on this, on and off, and then wrote up the whole thing over the course of ten days. There were a lot of failed attempts along the way 😅
1
1
24
Replying to @yibie
谢谢!
12
Trained a 360M model to pretend to be a terminal. There's no real shell behind it, but it runs vim, tmux and snake, and mostly remembers your files 😅
6
4
6
43
3,119
Thanks for posting. I see so many parallels to software engineering and can only assume that, as time goes on, more and more groups will have to face very similar challenges in their industries
1
392
Very well thought through outline of the future of maths. I see so many parallels to software engineering and can only assume that, as time goes on, more and more groups will have to face very similar challenges
I’m very opinionated. So it’s nearly impossible for me to hear someone’s arguments and *not* find something I disagree with. (This includes reading things I myself wrote years - or weeks - ago.) And yet the amazing Grant Sanderson @3blue1brown has put out a long, detailed video on the future of mathematics, in which I cannot find a single flaw. I’ll be sending this to every PhD student and postdoc I know, and strongly recommend everyone watch it. youtube.com/watch?v=LKrwhjOI… (Of course I tried to say many of the same things in my conversation with Thomas Koberda. But Grant’s version, unsurprisingly, does an infinitely better job at communicating those ideas!)
71
Great to finally see a promising diffusion LLM
1/4 Today, we’re releasing Mercury 3. Mercury 3 is by far our most powerful model and the most intelligent diffusion LLM ever released. It outperforms GPT-6 Luna and Haiku 5.5 on quality while running at 1000+ tok/sec.
5
186
Replying to @dremnik
Right now I’m experimenting with building my own from scratch… what are you using? The choice of graph engine only seem to matter at a pretty large scale in my experience
1
1
136
Replying to @dremnik
A bad graph is worse than no graph. But with the right structure they can save so much compute.
1
4
459
I'm backporting some of the algorithmic changes to the Erlang version. Stay tuned!
24
Replying to @Mascobot
Even a small difference in the order of floating-point operations can lead to minute but potentially detectable differencesf
2
70