Viewing profile — bigdict

But... they are equivalent?

Was it a controlled study or just correlation?

Huh? Triton inference server and Triton the kernel language are two distinct, very different things… Is this AI-generated?

> What actually happens: random Internet users spend two seconds skimming, then click their favorite. > They're not reading carefully. They're not fact-checking, or even trying. Uh…

cuz area and power

There's antimony, arsenic, aluminum, selenium…

What's the point of the relu in the loss function? Its inputs are nonnegative anyway.

Exactly, it's a natural way to write a college essay . I've never not cringed reading an article/blog post that is structured that way, it comes across very contrived. I've also no…

"delves" dashes an explicit "conclusion" section at the end

Lisp In Small Pieces

Pairs great with the tale of the Death Valley Germans: https://www.otherhand.org/home-page/search-and-rescue/the-hu... .

Once you realize that elisp is a better shell programming language than bash, it ceases to appear insane.

Has this been used widely since?

Thank you so much for continuing to support Gemma 3 with these updates.

Amazing, I've been wishing for this! Do you have any estimates on how much accuracy is first lost then recovered compared to the original bf16 and the naively quantized models?

Sure, you can get better model performance by throwing more compute at the problem in different places. Does is it improve perf on an isoflop basis?

Gemma 3 is.

* the code is at https://github.com/google-deepmind/gemma * you download the weights at https://www.kaggle.com/models/google/gemma-3/

"undo their disgusting propaganda, apply our beautiful correct opinions" This is so cringe.

Why is it timely?

Have you seen the RFK confirmation hearing?

You are quoting Wikipedia, not US foreign policy :)

Does it have an entry for what we don't know we don't know?

> What gave you the impression I was assuming bad faith? You said "I would guess you're not asking a serious question here"