Yea except DeepSeek released a combined Multimodal/generation model that has similar performance to contemporaries and a similar level of reduced training cost ~20 hours ago:

https://huggingface.co/deepseek-ai/Janus-Pro-7B

This seems cool but also a gateway for RSI

You don’t need to run the LLM on the device. It supports 13 different protocols for both text completion and chat completion.

There’s also RisuAI which has some nicer features like better integration with function calling. Sillytavern is working on Function calling more but it’s not quite there yet.

The nice thing about Sillytavern is that it also has plugins for Alltalk for TTS and ComfyUi/A1111 for image generation directly from the roleplay interface.

It also has support for RAG through upload of documents and web scraping and a shitload of other features it would take awhile to list here.

Somebody hit the token chain jackpot

What’s great about lawsuits like this is you really only have to prove intent and they have a record of them asking for similar imagery.

Easiest? Tailscale., set it up on the server and each client you want to access it and it creates auto-resolving P2P VPN tunnels between them all.

I used to work for an algorithmic advertising company.

The gist is that if you get one big spender it offsets the cost of losing a thousand or more other people because those large contracts usually last past the official sale

Isn’t fusion power not as clean as people say it is?

The Practicalities of actual fusion reactors make this seem a lot less appealing than I think I grew up hearing.

I’m happy to see china continue to pump resources into their clean energy mix, but at the same time it feels like this entire concept might end up being more of a meme than we think.

I2FS?

They can’t form new trade unions because everyone already has to join a registered trade union.

Do image previews work over SSH? I admit I’ve never actually tried it…

I wouldn’t bother unless you find yourself doing more through the terminal than through GUIs.

I don’t have a built-in file browser (not using a DE, just i3 window manager), so I use ranger and pure GNU coreutils commands mostly but I still find myself missing the drag-and-drop features that FreeDesktop integration provides for stuff like nautilus.

Facebook is trying to burn the forest around OpenAI and other closed models by removing the market for “models” by themselves, by releasing their own freely to the community. A lot of money is already pivoting away towards companies trying to find products that use the AI instead of the AI itself. Unless OpenAI pivots to something more substantial than just providing multimodal prompt completion they’re gonna find themselves without a lot of runway left.

This is because all LLMs function primarily based on the token context you feed it.

The best way to use any LLM is to completely fill up it’s history with relevant context, then ask your question.

Doesn’t this just do what gets done through convolution anyway?

What’s the point of this.

I thought it was atomic age and information age…

Or was that just empire earth…

There’s a weird implicit conservancy in tech circles around the dictatorial nature of corporate leadership.

It stems from this weird externalization of corporate decision making that just turns everything that happens at large companies into the machinations of the unknowable machine of capital.

“Of course they were fired, they protested in a way that disrupted the business, if the business is disrupted the machine must correct itself, and it did so by releasing the corporate anti-bodies of leadership to fire the disruptive element. Thus the machine is corrected. This is all logically sound, and thus impervious to moral inquisition.”

I’m assuming with as small as a tin foil hat as possible that the CIA and NSA purposefully do this in major closed and open source software

I agree with this wholeheartedly,

but if you feel about this methodology strongly you’re going to get hit with nay-sayers that use the same argument anti-VAT people use, as it’s ostensibly the same mechanism: that the developers farthest downstream would have to take the full amount of the percents piled up in their pricing scheme.

Are all these wacky licenses because the OSI doesn’t have an AGPL (ostensibly anti-cloud) equivalent BSD style permissive license?