internetguy

ayy cool! i spoke to you at yc sus this year :)

this is pretty cool! it feels very good for getting a brief summary/overview of something and getting a beginner's level grasp.

this is rich coming from the company that shreds books for training data

will nixpkgs still get updates? how will maintenance work (i am not familiar at all with the nix ecosystem so maybe this is a dumb q)

It is a little ironic that this article reads so much like an AI-generated one.

we've reached another level of dopamine-hit chasing....

I'm fine with it seeing as I can use my student email and get free usage

While I can see this being a very useful skill for many, we must also remember sometimes reading slow is desired, especially when a deep understanding and analysis of a text is required (e.g reading and analysing literature)

not too great, i submitted a photo taken of me at graduation. it got the location totally wrong and was around 50%-70% accurate on my hobbies and interests. it was able to correctly guess my sexuality and ethnicity, which is rather unsurprising.

> MegaTrain stores parameters and optimizer states in host memory (CPU memory) and treats GPUs as transient compute engines. For each layer, we stream parameters in and compute gradients out, minimizing persistent device state

This is pretty awesome. The only compute I have at home is an RTX 3080 with 10 GB of VRAM, so I struggle with training larger models (>40M, 50M params). I get OOM errors and have to optimize a lot.

I have a lot more CPU RAM in my PC, and this would likely increase the size of models I can train locally.

Insane. Not even an apology. And they ask why we should respect the police.

how elitist

nixos or arch for me.

their in-house Workbench distro seems really cool, looks super light and minimalistic. IMHO it's also priced reasonably enough.

might be a cool way to visualize emergent properties of automata

LMAO How ironic...

all non-ivy league researchers? that seems a little harsh IMO. i've read some amazing papers from T50 or even some T100 universities.

This should honestly have been implemented a long time ago. Much of academia is pressured to churn out papers month after month as academia is prioritizing volume over quality or impact.

this is pretty cool, it seems they reuse every image by translating it a little when your cursor is within a range

wow - this is really well made! i've been doing research w/ Transformer-based audio/speech models and this is made with incredible detail. Attention as a concept itself is already quite unintuitive for beginners due to is non-linearity, so this also explains it very well