Mxbonn

This looks really cool! The cuteness of the demo alone makes me want one. Ideally I would also love for it to have a real use case in my home but right now I really cant think of one.

uv and ruff are one of the best things that happened in the python ecosystem the last years. I hope this acquisition does not put them on a path to doom.

Now that PyTorch is also ending their anaconda package distribution, I think a lot of ml/ds people should give uv a shot.

Used a framework for the last years of my PhD and would very much recommend it! Now got a macbook from work and I have to say that I do prefer it but it also comes with a higher cost.. As I was no longer using my framework I turned it into my homeserver using this server kit: https://frame.work/be/en/products/framework-laptop-13-mainbo.... Any other laptop would have started gathering dust but my framework keeps being useful thanks to its modular design.

What happened to text prompts that were shown as early results in SAM1? I assume they never really got them working well?

I love framework. Have been using as my main device during the last years of my PhD. However now that I got a company mac book I was using it less and less, so I decided to buy the Cooler Master mainboard case and now I'm using it as a home server. Wouldn't have been able to do that with any other laptop.

I've been using the claude model over the past months and I have to say I usually prefer it over ChatGPT.

Some things that impressed me particular are:

* It can often give you working urls of images related to your query. (e.g. give me some images to use with this blog paragraph).

* It can list relevant publications corresponding to paragraphs, chatgpt often halucinates new papers but claude consistenly gives highly cited and relevant suggestions.

* It can work with pdf inputs.

This is largely due to Meta maintaining an internal performance-oriented production version of CPython: https://github.com/facebookincubator/cinder .

Couldn't immediately find it but who sponsors/pays for the compute?

Finishing PhD on efficient deep learning for computer vision.

Looking for a job as research engineer/scientist with anything related to ML efficiency or computer vision.

--

Location: Belgium

Remote: Yes

Willing to relocate: Possibly

Technologies: Deep learning, mainly PyTorch but knowledge of most of the stack.

Résumé/CV: https://maxim.bonnaerens.com/resume.pdf

Email: found in resume.

Alright so nearly all the 'cheap' stars got detected, but did any of the 'premium' stars get detected?

"... a better project might be to read papers until you find something you’re really interested in that comes with clean code, and trying to implement an extension to it."

Doesn't have to be clean but starting from a reproducible codebase at least gives you the guarantees that you have everything. Too often papers omit details that are crucial for a successful reproduction.

My approach is similar! However, my cards are more made for spaced reading than for spaced repetition studying. As such, each card just corresponds to a highlight.

(Also replace Kindle with Google play books)

For me, Midjourney was the killer app which showed that chat apps can be good interfaces for products. It takes away many UI development costs and problems and allows you to focus on the product/backend.

The part where he jumps to repeat a letter is both smart and funny to watch. Indeed, highly recommended watching the full video!

As an ML researcher I think about learning a new language like rust whenever I see articles about it appearing here. However, when I consider which parts of my job could be easily replaced with another language besides Python, the list is nearly empty.

I was kind of hoping that this would be a site that let anyone roast landing pages. This however is landing page for company that roasts landing pages.

Not sure if it's beginner friendly but I found implementing NeRF from scratch a good exercise. Especially since it reveals many details that are not immediately obvious from the paper.