bavell

Take off the tinfoil hat and learn how LLMs actually work and are trained. Your comment is so far off the mark that's it's effectively just noise.

I'm also struggling a bit to get some decent code output (TS) from local LLMs via oobabooga. I've been getting mixed results and I think it's pretty sensitive to the exact combo of model/parameters/prompt. If any one is off or not tuned to work with the others then results aren't great.

E.g. some models expect very specific formatting of prompts. Some models give nonsense output at temperature=0.7 while others work great. Some models have been fine-tuned on chatbot-bot like instructions (ala chatgpt, so-called "Instruct" models) while others just autocomplete the given prompt. Very important to use the right combo to get the best results.

Totally fair, I've mostly lived in one part of the south so it's likely regional.

Fwiw I've heard many horror stories about Paypal

Lived in the south my whole life and never heard of this once

(Devil's advocate) And employers will be happy to cover it for their employees, especially since they'll be paying them 10-20% less to account for their increased costs :)

This is one of the biggest tax loopholes for the ultra rich IMO. Ripe for abuse - shut it down!

Uhhh, I've been running stable diffusion on my 6750XT for a few months now... just started playing with chat and text models (oobabooga) too.

Didn't even get the card thinking I could use it with ROCm for hobbyist AI stuff but turns out it works great and having 12GB of VRAM is super nice.

I really, really hope this is the case. Trying to be optimistic about this in spite of my skeptical nature :)

From the summary:

"We perform the most comprehensive evaluation of Code LLMs to date and show that StarCoderBase outperforms every open Code LLM that supports multiple programming languages and matches or outperforms the OpenAI code-cushman-001 model."

So I'd assume not up to par with gpt4 or copilot. Can't wait to see it evolve from here!

Agreed with the sibling comments, physical violence is an entirely different beast than verbal "violence".

I hadn't realized how lost Sam had gotten until I saw his Triggernometry interview [0]. I had a lot of respect for him once but he's clearly lost the plot.

[0] https://youtu.be/DDqtFS_Pvcs

I don't think having romantic relationships with inanimate objects or non-humans will ever be normalized, for obvious reasons.

Brilliant. Thank you. Now get off my beach!

Yes, it was always really lame and silly.

> Unfortunately I can't remember exactly where the part of this talk is.

If you or anyone else remembers, I'd love to watch

In just the past week I used multipart/form-data to proxy an incoming file upload stream to another backend server which then streamed it to a cloud bucket. Working great so far and was fairly simple to set up!

Huh? Growing up in the burbs (90's and 00's) I had dozens of kids around my age within biking distance.

Good points but YMMV:

- fraction scaling: I don't need it

- Wayland: still using X

- hardware accel: I think that's true for chromium but IIRC there's a package with it and widevine (google-chrome-stablevon Arch AUR)

- nvidia: the AMD drivers are great and well supported (I've even been using ROCm to play with stable diffusion on my 6750XT)

- shooters: I mostly play strategy, puzzle, sim and 2D games. Haven't run into multiplayer issues yet but I'm sure there are some in the FPS genre with heavy DRM

Not for everyone but when I got my AMD card I stopped using my qemu virt-io VM w/ GPU passthrough and started using native steam with proton - much better experience all around and haven't booted windows ever since

Right, just the toxic combustion byproducts.

Well said and agreed!

I've been using EdgeDB in a greenfield project and loving it. Some rough edges (forgive the pun) but fantastic for prototyping / smaller projects and could be a good candidate for large projecs. I like it better than when I used neo4j a few years back for a client project. Much better license and I really like that it's built on top on postgres.

Same here, I've been happily using signal for years and will probably have to drop it since I won't be able to communicate with a majority of my contacts :(

They're basically kicking me out as a happy user. I would love to continue using Signal, not sure what the motivation for this change is in the first place.

> A self replicating reinforcement learning AI that distributes peer to peer over any network available and reproduces with other copies of itself to improve itself.

Behold, the birth of a true computer virus!

I've had great results recently using Steam/Proton on Arch with my AMD 6750XT.

> US intelligence agencies by law do not operate within the US.

Just incredible. Thanks for giving me a hearty laugh. Make sure to refill your pitcher of kool-aid, I see you're running low!

Wow, that really is unhinged! Would be a great element of a sci-fi plot of an AI going insane/short-circuiting and turning on the crew/populace/creator.

"Please don't post shallow dismissals"

https://news.ycombinator.com/newsguidelines.html

Nope, haven't seen anything you listed.

A lot of people here were very dismissive of the Twitter files, showing how closely big govt and big tech work together. I even lamented [0] about it and got tons of pushback ("business as usual"). I had thought at the time this site was very clear-eyed about dangers like these but I'm much less confident in that now and it's apparent even us techies are easily blinded by partisan politics and self-interest.

[0] https://news.ycombinator.com/item?id=33917602