It also really depends on VRAM IMO, I have a 4090 and these days I don't tend to touch anything under 30B (Wizard Uncensored is really good here) if I had dual 3090s I would likely be running a 65B model.
LocalLLaMA
Welcome to LocalLLaMA! Here we discuss running and developing machine learning models at home. Lets explore cutting edge open source neural network technology together.
Get support from the community! Ask questions, share prompts, discuss benchmarks, get hyped at the latest and greatest model releases! Enjoy talking about our awesome hobby.
As ambassadors of the self-hosting machine learning community, we strive to support each other and share our enthusiasm in a positive constructive way.
Rules:
Rule 1 - No harassment or personal character attacks of community members. I.E no namecalling, no generalizing entire groups of people that make up our community, no baseless personal insults.
Rule 2 - No comparing artificial intelligence/machine learning models to cryptocurrency. I.E no comparing the usefulness of models to that of NFTs, no comparing the resource usage required to train a model is anything close to maintaining a blockchain/ mining for crypto, no implying its just a fad/bubble that will leave people with nothing of value when it burst.
Rule 3 - No comparing artificial intelligence/machine learning to simple text prediction algorithms. I.E statements such as "llms are basically just simple text predictions like what your phone keyboard autocorrect uses, and they're still using the same algorithms since <over 10 years ago>.
Rule 4 - No implying that models are devoid of purpose or potential for enriching peoples lives.
If you’re looking for chatbots -
SillyTavern with turbo 3.5 is pretty good but you have to pay for the OAI API.
My experience with pygmalion wasn’t great but some really like it.
Character AI is unquestionably the best model and the easiest to use however it has a strict NSFW policy.
I’ve never tried llama but people did rave about it when it first released. I don’t know how easy it is to get up and running though.
You can use 3.5turbo via poe.com through sillytavern for free
If you can run 13B, I would recommend the following models. The links are for the GGML models. Keep in mind that if you can run a 30B or larger model, there are other LLMs that will work better.
TheBloke/manticore-13b-chat-pyg-GGML
This one is my personal current favorite. I think that it's better than Chronos for short messages. For me it usually sticks to using asterisks for actions, and quotes for speech, which is what I prefer.
TheBloke/chronos-wizardlm-uc-scot-st-13B-GGML
This one feels more "clever" to me all around, and is currently very popular for roleplay. It produces results that are usually longer, so I feel like its better suited for longer dialogue. It also feels to me like it understands the scenarios better, and I usually get slightly more creative results from it.