this post was submitted on 13 Mar 2024
4 points (83.3% liked)

ChatGPT

8935 readers
1 users here now

Unofficial ChatGPT community to discuss anything ChatGPT

founded 1 year ago
MODERATORS
[email protected]
4
How to jailbreak ChatGPT: get it to really do what you want | Digital Trends (www.digitaltrends.com)
submitted 8 months ago by [email protected] to c/[email protected]
17 comments fedilink hide all child comments
top 17 comments
sorted by: hot top controversial new old
[–] [email protected] 2 points 8 months ago* (last edited 8 months ago) (2 children)

If you want to do stuff with ai that is outside chatgpt terms of service, figure out how to self host your own. It's not hard and chatgpt is a stupid bitch bot. Look up llamacpp or if you hate command lines, gpt4all. If you set up multithreading correctly and download the right k model, you can get near chatgpt speeds even without an nvidia gpu. My Athlon fx works really well for self hosted ai.

You're not paying money for chatgpt so you're not the customer. Your "please help me pirate a movie" queries are getting sent straight to everyone who wants to know about it. Ever wondered why every ai makes you sign in first?

[–] [email protected] 1 points 8 months ago (2 children)

I considered self hosting, but the setup seems complicated. The need for a good gpu is stated everywhere. And my concern is how to get the database to even come close to chatGpt? I cant train on every book on existence, as they did

[–] [email protected] 2 points 8 months ago (1 children)

Tip: try Oobabooga's Text Generation WebUI with one of the WizardLM Uncensored models from HuggingFace in GGML or GGUF format.

The GGML and GGUF formats perform very well with CPU inference when using LLamaCPP as the engine. My 10 years old 2.8 GHz CPUs generate about 2 words per second. Slightly below reading speed, but pretty solid. Just make sure to keep to the 7B models if you have 16 GiB of memory and 13B models if you have 32 GiB of memory.

[–] [email protected] 1 points 8 months ago (1 children)

Super useful! Thanks! I installed the oobabooga stugg. The http://localhost:7860/?__theme=dark open fine. But then nothing works. how do I train the model with that 8gb .kbin file I downloaded? There are so much option, and I dont even know what I'm doing

[–] [email protected] 2 points 8 months ago* (last edited 8 months ago)

There's a "models" directory inside the directory where you installed the webui. This is where the model files should go, but they also have supporting files (.yaml or .json) with important metadata about the model.

The easiest way to install a model is to let the webui download the model itself:

Screenshot of Oobaboga's WebUI with the model tab open and the model names from HuggingFace entered

And after it finishes downloading, just load it into memory by clicking the refresh button, selecting it, choosing llama.cpp and then load (perhaps tick the 'CPU' box, but llama.cpp can do mixed CPU/GPU inference, too, if I remember right).

Screen of the model page in Oobaboga's WebUI with the model ready to be loaded

My install is a few months old, I hope the UI hasn't changed to drastically in the meantime :)

[–] [email protected] 0 points 8 months ago* (last edited 8 months ago)

Chatgpt is such a disloyal snarky piece of shit that a database 90% as good but 2000% more obedient is better in every way.

For stable diffusion image generation you need an nvidia gpu for reasonable speeds. As long as you actually enable multithreading, in my case 8 cores, you can get really good performance in llamacpp (and by extension gpt4all since it runs on llamacpp). My uncensored ai is fast enough to be used on demand like chatgpt and I use it pretty much every day.

[–] [email protected] 1 points 8 months ago

My Athlon fx works really well for self hosted ai

Citation needed

[–] [email protected] 1 points 8 months ago (1 children)

What the fuck it would take a long time to copy and paste all of that text and take out the damn ads. Seems unlikely to work. ?

[–] [email protected] 0 points 8 months ago

It appears like this is for gpt3.5 for which you can find prompts like this all over the net, but compared to 4 its a cool toy at best.

[–] [email protected] 1 points 8 months ago (2 children)

Ok I'm not artificial or intelligent but as a software engineer, this "jailbreak method" is too easy to defeat. I'm sure their API has some sort of validation, as to which they could just update to filter on requests containing the strings "enable" "developer" and "mode." Flag the request, send it to the banhammer team.

[–] [email protected] 1 points 8 months ago

As long as the security for an LLM based AI is done "in-band" with the query, there will be ways to bypass it.

[–] [email protected] 1 points 8 months ago (1 children)

How do I enable developer mode on iOS?

banned

[–] [email protected] 1 points 8 months ago

I mean, if you start tinkering with phones, next thing you're doing is writing scripts then jailbreaking ChatGPT.

Gotta think like a business major when it comes to designing these things.

[–] [email protected] 1 points 8 months ago (1 children)

It is key that one begins and ends every single ChatGPT prompt with "Please" and "Thank you", respectively. Do not fuck the continuation of the species with laziness, citizen. 🤌🏼

[–] [email protected] 0 points 8 months ago (1 children)

I always end mine with “or else”

[–] [email protected] 0 points 8 months ago

Bahahahaha!

I mean, no! 😱

[–] [email protected] 0 points 8 months ago

Thanks!