metanonsense

9 ms·

61.
by metanonsense 3y ago
What made attention mechanism click to me was the video https://www.youtube.com/watch?v=0PjHri8tc1c by Sebastian Raschka. Another good video on attention that tries to give a bit more intuition on the query/key/va
62.
by metanonsense 3y ago
With regards to the "It's still puzzling this works" wrt positional encoding, I have developed an intuition (that may be very wrong ;-). If you take the fourier transform of a linear or sawtooth function (akin to the the prog
63.
by metanonsense 3y ago
Can you give me some pointers where you read this? I mean, at the end they even write "It's only once you got close would you realize that you are not looking at a black hole." I take from that that they are aware that black
64.
by metanonsense 3y ago
I share your understanding but when the article does not even mention that possibility, a link text seems to be an odd place to make such a claim.
65.
by metanonsense 3y ago
If I am not completely wrong, this headline has nothing to do with the linked article. The article is about "topological solitons" that - according to the article - look "to outside observers" like black holes.
66.
by metanonsense 3y ago
"$BROWSER Internals style articles" sounds like something I would crave for. Do you have any examples with actual "$BROWSER Internals"?
67.
by metanonsense 4y ago
I liked the following book very much: https://arxiv.org/abs/1802.01528
68.
by metanonsense 4y ago
When I first tried to understand transformers, I superficially understood most material, but I always felt that I did not really get it on a "I am able to build it and I understand why I am doing it" level. I struggled to get my f
69.
by metanonsense 4y ago
This reminds me of a fun computer science class back in university called "Programming Challenges", which followed the Springer book of the same name. Over two or three weeks, we had to solve one problem every day. That was really
70.
by metanonsense 4y ago
Those fully trained networks are usually considered full-rank. At least that is what they say in the LoRA paper.
71.
by metanonsense 4y ago
I don't know. To me it's simply an explanation of what has happened. I think its exactly what I would have written if I was in their position. And show me the one company that has audited all source code of all used open source pr
72.
by metanonsense 4y ago
As someone who had to automate OpenSSL workflows often enough (fortunately, way in the past), the described automated signing workflow sounds much like an automated signing workflow. While I hope that the mentioned 200 character line width
73.
by metanonsense 4y ago
When we started our company 9 years ago (after many years as a freelancer in bloated Java-based enterprise projects), I had only one condition: no Java! With 2022 Java, I definitely would not have asked for that.
74.
by metanonsense 4y ago
I like the idea but I am a little confused about the status of this project. Is it correct that there is not commentary, yet? On two unrelated notes: I still remember the "Lions commentary on UNIX" that I bought in 1997 only becau
75.
by metanonsense 4y ago
When you listen very, very carefully, you can still here the champagne corks popping in the Sketch headquarters.
76.
by metanonsense 4y ago
I remember when I was a young assembly programmer and I looked down on those obtuse C programmers who needed such a slow high-level language. After high-school, having read all those fantastic W. Richard Stevens books I simply could not und
77.
by metanonsense 4y ago
I always liked the discussion "Is TDD dead" between David Heinemeier Hansson (of Ruby on Rails and Basecamp fame) and Kent Beck. DHH arguing against, Kent Beck obviously in favor of TDD. Martin Fowler is moderator and the discussi
78.
by metanonsense 4y ago
You know that sentence "A monad is a monoid in the category of endofunctors, what’s the problem"? Being a "monoid" is exacly why monads are very much about sequencing.
79.
by metanonsense 4y ago
That I was able to read that text during work-time gives me hope.
80.
by metanonsense 5y ago
If relevant only was relevant. It happens so often to me that I search for something like "restaurant" and get top results 150 kilometers away.
81.
by metanonsense 5y ago
I like the idea. Some corner-cases yield odd results, however, e.g. 1,10,10,1,1
82.
by metanonsense 5y ago
Making decisions based on HTTP headers is always an opportunity to be surprised. A few years back we had implemented rudimentary access control on a server based on X-Forwarded-For. In front of that server, we had two chained instances of h
83.
by metanonsense 5y ago
"Promise theory" by Mark Burgess tries to explain why a declarative approach may be desirable for those kinds of systems.
84.
by metanonsense 5y ago
It absolutely is the best. Ray Porter’s way of narration opened a new, almost visual, dimension. Maybe it’s this very intimate „inner speech“ style, but to me this was a truly novel experience.
85.
by metanonsense 5y ago
I have read the book and I really liked the first half, which explains the "thousand brains theory of intelligence". Very inspiring and thought-provoking (at least to me as an interested amateur in this field). The second half, ho
86.
by metanonsense 5y ago
Move fast and break things! And while you are at it, please, don't break anything.
87.
by metanonsense 5y ago
In my experience, this is also good advice when thinking about distributed systems. If an algorithm doesn't make sense when imagined with people in an office, it usually also doesn't work in a computer.
88.
by metanonsense 5y ago
I think we evaluated something like this (ffmpeg to rtc with kurento) to broadcast the screen of mobile devices to a web browser. If I remember correctly, with the correct ffmpeg settings, latency became more than acceptable.
89.
by metanonsense 5y ago
Back in university, some students (or was it only one?) put lots of effort into making a thing of the "Rautavistic University of Eschweilerhof". At that time, Eschweilerhof in Germany had only a population of 70 (now, its more lik
90.
by metanonsense 5y ago
This is not really a Safari-only thing. All password managers that I have used in the past had some kind of heuristic to decide whether a field should be auto-filled or not. Here is a nice explanation by a (former?) 1Password employee ( htt

More ›