site banner
PaperclipPerfector  1yr ago (text post)   3261 thread views

Small-Scale Question Sunday for July 9, 2023

Do you have a dumb question that you're kind of embarrassed to ask in the main thread? Is there something you're just not sure about?

This is your opportunity to ask questions. No question too simple or too silly.

Culture war topics are accepted, and proposals for a better intro post are appreciated.

5
Top Bottom Old Controversial Comments
Jump in the discussion.

No email address required.

I've ran pretty much every local model ≤65B parameters there is, except a few weird Chinese projects (speaking of foundations, not interchangeable LLaMA finetunes). It's easy these days. If you're paranoid you can inspect the inference code, it's all in the open. llama.cpp and kobold.cpp are probably the best open-source cross-platform applications to use right now. Closed-source stuff like LM Studio is more user-friendly.

It's vastly worse, except for tasks where GPT-4 is deliberately NERFed (erotic roleplay mainly, which I don't use – never saw the point even with humans – but which is a big reason for the community to keep going, it seems).

There isn't even anything entirely on par with ChatGPT-3.5-turbo. Specialized coding models (WizardCoder) are getting close, little by little. But really, I'm quite disappointed. People don't think bigly, don't read the literature thoroughly enough, don't pool resources with remotely the urgency the situation commands.

Speaking of which, if there's anyone here interested in making a local ChatGPT-tier model available, you can support Eric Hartford's Dolphin project. Falcon-40B finetuned on OpenOrca dataset and some textbooks might well surpass it. I think it should be on the level of $20k total. For now best we can look forwards to is his Llama 13B finetune.

EDIT: it seems it's more like 10k for Falcon but also these guys are already doing Orca 13B replication. I was not sure whether they or Hartford are more trustworthy, at least some guys in this team are okay.

take advantage of LLMs for my personal note taking / organization / todos

I'm sorry to say I am not aware of a good extant system that can leverage local models for work with documents. But it'll be something like this or this. There's a big problem with context length and computational cost. Research suggests much more is possible.

It's vastly worse, except for tasks where GPT-4 is deliberately NERFed (erotic roleplay mainly, which I don't use – never saw the point even with humans – but which is a big reason for the community to keep going, it seems).

I don't do ERP, but I have used jailbroken ChatGPT-4 to generate bespoke erotica, and I can tell you that it's really fun. You can feed it any combination of characters and scenarios you like. If I want to read about the Flintstones wife-swapping with the Rubbles, I can. If I want to read about Agatha Heterodyne having a threesome with Gil and Tarvek to decide who gets to marry her, I can. If I want to read about Nagatoro teasing and bullying Senpai by sending him videos of her sexcapades while he is away at art school, I can. If I want to read about Amaryllis Penndraig becoming a slut to compete with Fenn over Juniper's affections, I can. The possibilities are endless!

For now, jailbreaking ChatGPT is easier and cheaper than using open-source LLMs, not to mention it works WAY better (as you correctly note, no open-source model can even match the output of ChatGPT-3.5), but OpenAI is clearly trying to make their models unbreakable and it is good that people are working on free alternatives should they succeed.

What's a good way to jailbreak gpt4 these days?

The Luigi and Peach jailbreak reliably produces NSFW. Make sure you have DeMod installed, though, or you will get banned.

Thanks. Any chance that DeMod itself will get you banned?

What is this place?

This website is a place for people who want to move past shady thinking and test their ideas in a court of people who don't all share the same biases. Our goal is to optimize for light, not heat; this is a group effort, and all commentators are asked to do their part.

The weekly Culture War threads host the most controversial topics and are the most visible aspect of The Motte. However, many other topics are appropriate here. We encourage people to post anything related to science, politics, or philosophy; if in doubt, post!

Check out The Vault for an archive of old quality posts. You are encouraged to crosspost these elsewhere.


Why are you called The Motte?

A motte is a stone keep on a raised earthwork common in early medieval fortifications. More pertinently, it's an element in a rhetorical move called a "Motte-and-Bailey", originally identified by philosopher Nicholas Shackel. It describes the tendency in discourse for people to move from a controversial but high value claim to a defensible but less exciting one upon any resistance to the former. He likens this to the medieval fortification, where a desirable land (the bailey) is abandoned when in danger for the more easily defended motte. In Shackel's words, "The Motte represents the defensible but undesired propositions to which one retreats when hard pressed."

On The Motte, always attempt to remain inside your defensible territory, even if you are not being pressed.


New post guidelines

If you're posting something that isn't related to the culture war, we encourage you to post a thread for it. A submission statement is highly appreciated, but isn't necessary for text posts or links to largely-text posts such as blogs or news articles; if we're unsure of the value of your post, we might remove it until you add a submission statement. A submission statement is required for non-text sources (videos, podcasts, images).

Culture war posts go in the culture war thread; all links must either include a submission statement or significant commentary. Bare links without those will be removed.

If in doubt, please post it!


Rules


Recommended Posts And Communities

Recommended Realtime Chats