Why, of all names possible, you thought TILF was the best one?

You need to find an abliterated finetune, where someone sends prompts that would hit the guardrails, traces the activated neurons, finds the pathway that leads to refusal, and deletes it.

I don't understand this move. A frontend desktop application is the opposite of what I and anyone else I know uses Ollama for. It's a local LLM backend. It's been around long enough now that any long term users have found, created and/or adjusted to their own front end interface.

I'm comfy, but some of the cutting edge local LLMs have been a little bit slow to be available recently, maybe this frontend focus is why.

I will now go and look at other options like Ollama that have either been fully UI integrated since the start, or that is committed to just being a headless backend. If any of them seem better, I'll consider switching, I probably should have done this sooner.

I hope this isn't the first step in Ollama dropping the local CLI focus, offering a subscription and becoming a generic LLM interface like so many of these tools seem to converge on.

There is a petition to repeal the Online Safety Act[0].

The initial government response can be read as “lol, no”.

[0] https://petition.parliament.uk/petitions/722903

If you're writing server stuff, at the coarse-grained level of isolation that Deno provides you're better off using just about anything else and restricting access to network/disks/etc through systemd. Unlike Deno, it can restrict access to specific filesystem paths and network addresses (whitelist/blacklist, your choice), and you're not locked into using just Deno and not forced to write JS/TS.

See `man systemd.exec`, `systemd-analyze security`, https://wiki.archlinux.org/title/Systemd/Sandboxing

Suppose I'd like to use models like this one to perform web searches. Is there anything available in the open-source world that would let me do that without much tinkering needed?

I think it’s something that even Google should consider: publishing open-source models with the possibility of grounding their replies in Google Search.

16 points and it returns a 429 (Too many requests). I wonder what the avg. factor between HN points and site visits is approximately.