albertzeyer

11 ms·

91.
by albertzeyer 2y ago
Somewhat similar: My PyCPython project ( https://github.com/albertz/PyCPython ), where I wrote a C interpreter in Python, and then tried to interpret CPython. I never really completed this project though.
92.
by albertzeyer 2y ago
In Python, some of the stated use cases, like extracting coverage, extracting the call graph, etc, you can get via tracing already (sys.settrace, https://docs.python.org/3/library/sys.html#sys.settrace ). And I wou
93.
by albertzeyer 2y ago
I find it a bit weird that Gimp does not use the latest GTK (i.e. GTK4, which was considered stable since 2020), even though GTK originates in the Gimp project itself. It actually seems to be quite a bit behind: This is now the first releas
94.
by albertzeyer 2y ago
A similar tool for this user-space bind-mount is https://github.com/fritzw/ld-preload-open , which relies on LD_PRELOAD to overwrite common libc functions. Thus this is less reliable as the presented tool which uses ptr
95.
by albertzeyer 2y ago
Probably just "regular" LMs, not large LMs, I assume. I assume some LM with 10-100M params or so, which is cheap to use (and very standard for ASR).
96.
by albertzeyer 2y ago
K2/Kaldi is using more traditional ASR technology. It's probably more difficult to set up but you will more reliable outputs (no hallucinations or so).
97.
AI-generated poetry is indistinguishable from human-written and more favorably(nature.com)
15 points by albertzeyer 2y ago|10 comments
98.
by albertzeyer 2y ago
I remember the first time, we bought some 10BASE2 ethernet cards and BNC connector cables, and spend hours to figure out why it does not work, only then to learn the next day that we also need cable end terminators (if I remember that corre
99.
by albertzeyer 2y ago
It might be a way to reverse the biodiversity loss caused by humans. Biodiversity is important for various reasons ( https://en.wikipedia.org/wiki/Biodiversity#Role_and_benefits... ).
100.
by albertzeyer 2y ago
(Small) neural networks have been discussed to be used in computer chips for branch prediction. I don't really find a good source whether that really landed in production though. Here was some discussion: https://news.ycombi
101.
by albertzeyer 2y ago
I guess we are all in our own bubble, but in my circle of friends (all in Germany), it is rather the opposite: I and many of my friends feel quite European, and don't identify at all as being a German. I would consider myself even quit
102.
by albertzeyer 2y ago
When someone asks "where are you from", you are usually trying to guess what you can tell them that is most exact which they still know. Usually that's the country. Or if it is a big city (Paris, Berlin or so), you might also
103.
by albertzeyer 2y ago
That's not at all my experience. I contributed often to CPython by filling many issues and also one PR, and the whole process was very reasonable.
104.
by albertzeyer 2y ago
There is an advantage of having this as syntax, which is a difference in semantics, which you couldn't get if this would just be an expression, namely: x = func1() if something else func2() In this example, only func1() or fun
105.
by albertzeyer 2y ago
MPD was too restricted in functionality. It doesn't really provide such infinite play list concept. So I felt that I would fight with the limitations of the mpd API more than I really get something out of it. Here are some comparisons
106.
by albertzeyer 2y ago
I did not test this yet, but Dynamic "Mix" Feature and Audio Analysis and Recommendations, those are features I always wanted to have for a local music player. Amarok had some simple variant of this. Spotify of course has as well,
107.
Windows Agent Arena – Evaluating Multi-Modal OS Agents at Scale(microsoft.github.io)
1 points by albertzeyer 2y ago|0 comments
108.
by albertzeyer 2y ago
There is also RPython (used by PyPy) ( https://rpython.readthedocs.io/ ), which is a strict subset of Python, allowing for static analysis, specifically for the translation logic needed by PyPy. Thus, I was told that RPython
109.
by albertzeyer 2y ago
Don't overthink it. Don't search for the perfect project. There never will be one. Just start sth, also multiple potentially multiple things, and keep active on those which are fun. It's good to do small projects (timescale
110.
by albertzeyer 2y ago
I wonder about Keras 3. It's now backend independent again, like in the early days, and supports JAX, TensorFlow, or PyTorch. It's a nice thing if you defined your model and can then easily switch between the frameworks, right? Or
111.
by albertzeyer 2y ago
Not at all. I forked before I used anything with CUDA. I didn't need it but I guessed this could cause all kind of weird problems.
112.
by albertzeyer 2y ago
We considered to use sth like this to cache some Python program state to speed up the startup time, as the startup time was quite long for some of our scripts (due to slow NFS, but also importing lots of libs, like PyTorch or TensorFlow). W
113.
by albertzeyer 2y ago
Oh: Automatic differentiation of Java code using Code Reflection: https://openjdk.org/projects/babylon/articles/auto-diff This sounds interesting. I remember that Swift for TensorFlow was similar. Maybe you c
114.
by albertzeyer 2y ago
So, what's the current state? Can I already run this and get a usable desktop? Or some terminal with POSIX toolchain (shell etc)? SSHd? What kind of desktop would this run actually? The Windows shell, the ReactOS shell, some Unix DE (K
115.
by albertzeyer 2y ago
The blog post seems to contain more details and the core ideas: https://medium.com/yandex/yafsdp-a-tool-for-faster-llm-train...
116.
by albertzeyer 2y ago
It's not just a problem when you are an amateur. This is sth that every project should provide. But there are also many projects which do. Sometimes you need to search a bit for it. Actually I would expect that most big projects have s
117.
by albertzeyer 2y ago
I assume it's just some neural network doing the heavy lifting, or not?
118.
by albertzeyer 2y ago
To me, this actually sounds like a nice feature, if you have full control over it, i.e. you are sure all data stays on your device, it's a local AI, etc. And also, you can easily enable/disable it, and you can easily see its curre
119.
by albertzeyer 2y ago
I have a similar document, much shorter, but contains a few other useful tools which are not mentioned here: https://github.com/albertz/wiki/blob/master/debugging.md
120.
by albertzeyer 2y ago
There have been many. But none of them really successful, in the sense that they were more proof-of-concept, never fully feature-complete, or just not adopted by the community. See the list here: https://github.com/hoeck&#x2

More ›