Labo333

8 ms·

61.
by Labo333 2y ago
I'm sorry, I think I made the server crash with the following js code: Array.from(document.getElementsByTagName("input")).forEach(e => {e.checked = true})
62.
by Labo333 2y ago
@dang maybe the title should reflect the originality of the project, aka the fact it's all AI generated?
63.
by Labo333 2y ago
Author here! It was such a power trip and blissful experience to discover the game on HN, think about a strategy, implement a solution, write a blog post (thanks again to orlp for his comment here that pointed out a mistake), email igpay (t
64.
by Labo333 2y ago
Author here! I'm not sure about the first part V(s) = max i (min j V(s, i, j)) It looks like you suppose that j "precommits" to some move. But I'm not fully awake yet so you could be right.
65.
by Labo333 2y ago
oh RIGHT. Gotta fix it
66.
Solving Probabilistic Tic-Tac-Toe(louisabraham.github.io)
138 points by Labo333 2y ago|44 comments
67.
by Labo333 2y ago
I just did: https://louisabraham.github.io/articles/probabilistic-tic-ta...
68.
by Labo333 2y ago
There is an easier way to solve each recursion! I just wrote a blog post on it: https://louisabraham.github.io/articles/probabilistic-tic-ta...
69.
by Labo333 2y ago
In my experience (slightly tweaked towards data science and AI research), programming is also experimenting, be it playing with the data, testing a new idea or even debugging. There is a reason why LISPers (I'm not one of them though)
70.
by Labo333 3y ago
Pretty cool! A nice addition would be an in-browser runtime!
71.
by Labo333 3y ago
Nothing quite beats a simple google docs file where I can take notes and put links to sci-hub. Very often, legal download links expire after some time or they force the browser to download the pdf. I have a google docs for each research pro
72.
by Labo333 3y ago
TIL there is such a thing as African American English. But until we teach it in school and there are books written in that dialect, it will just be considered incorrect English by the AI, hence the qualification of "lazy" is not i
73.
GaLore: Train 7B models from scratch on consumer GPU(huggingface.co)
1 points by Labo333 3y ago|0 comments
74.
by Labo333 3y ago
Great article! (shameless plug) As an alternative to "Burp Extension Copy As Python-Requests", I coded this CLI tool that converts HAR to Python Requests code: https://github.com/louisabraham/har2requests
75.
V-JEPA: The next step toward advanced machine intelligence(ai.meta.com)
3 points by Labo333 3y ago|0 comments
76.
by Labo333 3y ago
I wonder what the output will be on polyglot files like run-anywhere binaries produced by cosmopolitan [1] [1]: https://justine.lol/cosmopolitan/
77.
S-LoRA: Serving Concurrent LoRA Adapters(github.com)
73 points by Labo333 3y ago|20 comments
78.
by Labo333 3y ago
Side note but I really love my Kobo and its hackability. For example, you can create your own sync mechanisms [0]. And when I say hackable, I mean it: I improved the Dropbox features to use the actual API and not a "public folder"
79.
by Labo333 3y ago
There is an amazing trick that I now use almost every time I want to sleep: the 4-7-8 breathing method. I don't remember using it and not falling asleep in less than 5 minutes, whatever the stress I was in. https://edition.c
80.
Verifiable Encryption on the Blockchain(louisabraham.github.io)
3 points by Labo333 3y ago|0 comments
81.
by Labo333 3y ago
Is this guy paid by a quant firm to implement classical algorithms that are not directly related to their activity, just for the sake of PR? Sign me in!
82.
Checking properties of C++ template types(louisabraham.github.io)
3 points by Labo333 3y ago|0 comments
83.
Two puzzles about uniform random variables(louisabraham.github.io)
2 points by Labo333 3y ago|0 comments
84.
Checking properties of C++ template types(louisabraham.github.io)
2 points by Labo333 3y ago|0 comments
85.
by Labo333 3y ago
Can you explain why they have a "significantly lower energy usage"? Thanks!
86.
by Labo333 3y ago
I'm just impressed that they manage to run at a pace of 6 miles per hour (that is 10km/h) while coding... I don't think I can type or use a mouse while I run, that is not walk – and 10km/h is definitely not a walk. And t
87.
by Labo333 4y ago
I have mixed experience with Wine, most of the times when I tried playing games it was close to impossible to install them. I go for Parallels Desktop those days (kudos to https://github.com/dreamncn/ParallelsDesktopCra
88.
by Labo333 4y ago
I totally agree and hnswlib is actually much faster than FAISS on CPU. I'm really happy to see `hnswlib` as a Python dependency since I'm the one who implemented PyPI support: https://github.com/nmslib/hnswlib
89.
by Labo333 4y ago
SD models are much smaller so you can probably already run it with the github code. Image models work well at least: https://huggingface.co/docs/diffusers/optimization/mps It is even possible to run them in t
90.
by Labo333 4y ago
I don't understand those benchmarks: why is there no comparison to numba? Also comparisons where C++ is beaten don't seem very realistic.

More ›