jzombie

Non-RAG LLMs don't search the internet, nor even have the capability to do so.

Of course they can. Either you haven't used a good LLM, or you're in denial.

The "here's what people have been saying about ctypes" section is hilarious!

I can't honestly recall what it was doing except I couldn't echo the vars. Can't remember if they were encrypted or just empty.

Related:

https://1password.community/discussion/141554/use-op-run-to-...

https://developer.1password.com/docs/cli/secrets-environment...

Edit: I could be mistaken as well. It has been about a year since I was working on a related project.

Are the secrets encrypted once they become env vars? I feel like this is a silly question without some context, but somehow 1Password is able to keep them private even if echoing them locally, and I haven't yet figured out how.

The Python it writes runs in a sandbox environment with its own containerized filesyatem. That's been obvious to me for a while as I have seen it generate temporary files when uploading things like images to scan.

This does not imply that any large language model executes the code it writes... I get it, they don't. I build models myself and have a fairly good understanding of how they work. But ChatGPT-4 does provide the ability to generate and run code in its own sandbox environment, outside of the model itself.

I didn't see any, just some keys that I assume were temporary. My session ran short and I didn't bother looking into it further.

For instance, if I were using someone else's custom GPT, I wonder if it would be using my particular container system, or someone else's.

For some reason they locked this thread, and I'm having the same issue now.

Where are the data centers located?

I'd rather pay a little more money to a reputable vendor, honestly. If I depend on the backup, I want a reasonable guarantee they will deliver on their promise.

Numerous studies in psychology suggest that the ability to delay gratification is indeed linked to better outcomes in life, including higher academic achievement, better health, and more profound personal satisfaction.

Instant gratification is largely simple decisions by simple people.

If you don't care about UI and want keyboard only, perhaps a terminal browser such as Lynx would be of interest.

Well, I thought I did but I guess I didn't.

Figured out what happened, though; Still not quite satisfied, but I can let it ride.

--------

Edit: 2FA is indeed enabled now. Thanks.

Another solution, that would most definitely work, would be using something like NoMachine or Parsec to just stream the desktop to your other computer. Parsec would probably give you really good quality as well.

My guess is that sharing from Linux would give you more options in terms of how to work around the copy protection.

I don't know if you use Docker, but a seriously weird sharing workaround could be to use a Samba share to mount in Finder, pointed to your drive on Linux... I made a Docker repo that could probably be modified to fit this use case: https://github.com/jzombie/sshfs-mac-docker

Depending on your setup, you might could stream the DVDs over your network. Surely you could somehow share the drive.

Have you contemplated ripping the DVDs and/or using a streaming service?

I wouldn't limit my options to just having a DVD if that means compromising everything else, especially if the computer is for work.

Press Shift + F10 to launch the command prompt then type in 'OOBE\BYPASSNRO' and press Enter. The system will reboot and you can proceed without a MS account.

I did this a couple of times lately on a bare metal install as well as in a VM.

I used MS Teams via a web browser using Linux for a couple of years. Should support video calling via WebRTC.

Why not take a snapshot image of the base install and just revert to that?

I originally was a Poetry user, but after having issues related to installing TensorFlow with it on Mac Silicon, as it wanted to resolve Intel dependencies, I got to thinking that venturing far from Pip may have unintended consequences.

So I made this little project which takes the best features for my own use case from Poetry, and wrapped Pip using simple Bash.

This doesn't replace Pip. It is Pip. There's no dependencies other than Bash, and Pip.

> We just need some core work horse libraries.

Do you mean like PyTorch, TensorFlow, or the others?

Appears to only be useful for inference. During training, it seems like it is simply just broken.

I haven't upgraded yet, but I'm curious if others have experienced what these people are experiencing. One commenter said his Core ML training is 7x slower.

Poetry install tensorflow does not work for me. There's lots of other folks with this issue as well.

It won't install tensorfow on a Mac Mx series without hacking a custom config. I gave up on Poetry after that and try to stick with pip now.

This article does not mention leveraging AI at all, to my knowledge.

Interesting you mentioned using X-plane 12 on your M1. I had considered doing the same. Currently play MS Flight Simulator sometimes via XBox Cloud Gaming, over a VPN to Nicaragua. It isn't bad, but X-plane seems like it could be better for me.

Your custom instruction is good, I mean.