faldore

joined 1 year ago
sorted by: new top controversial old
LLM Web-UI recommendations in c/localllama@poweruser.forum
[–] faldore@alien.top 1 points 11 months ago (1 children)

How chat-ui local? Last I tried they still require mongo.

Anyone have a 1B or 3B model that is mostly coherent? in c/localllama@poweruser.forum
[–] faldore@alien.top 1 points 11 months ago

Did you see Samantha-phi-1.5?

https://huggingface.co/ehartford/samantha-phi

Note this is pre-ChatML

I think I'll re-train this with ChatML

Anyone have a 1B or 3B model that is mostly coherent? in c/localllama@poweruser.forum
[–] faldore@alien.top 1 points 11 months ago

Phi-1.5

ORCA 2 Released open source! in c/localllama@poweruser.forum
[–] faldore@alien.top 1 points 11 months ago

Yes. This is real Microsoft Orca that they never released. But version 2.

I'm having... Trouble. in c/localllama@poweruser.forum
[–] faldore@alien.top 1 points 1 year ago

Need to make a Dolphin version

Down to memory lane, 2022 - "Google's LaMDA Ai is sentient, I swear" in c/localllama@poweruser.forum
[–] faldore@alien.top 1 points 1 year ago

No my Samantha model is not sentient.

I want to try to develop that though and see i can get it closer

dolphin-2.2-70b in c/localllama@poweruser.forum
[–] faldore@alien.top 1 points 1 year ago

Try SynthIA

1
dolphin-2.2-70b (alien.top)
 

Today I released Dolphin 2.2 70b (deprecating version 2.1, which had a EOS token bug)

https://huggingface.co/ehartford/dolphin-2.2-70b

This model is based on llama2, so it is suitable for commercial or non-commercial use.

This model is trained on top of the amazing StellarBright base model.

New in 2.2 is conversation and empathy. With an infusion of curated Samantha and WizardLM DNA, Dolphin can now give you personal advice and will care about your feelings, and with extra training in long multi-turn conversation.

This model is uncensored. I have filtered the dataset to remove alignment and bias. This makes the model more compliant. You are advised to implement your own alignment layer before exposing the model as a service. It will be highly compliant to any requests, even unethical ones. Please read my blog post about uncensored models. https://erichartford.com/uncensored-models You are responsible for any content you create using this model. Enjoy responsibly.

Dolphin-2.2-70b's training was sponsored by a16z.

Thanks to Jon Durbin, Wing Lian, WizardLM, and TheBloke.

​

Example output

Google Brain cofounder says Big Tech companies are lying about the risks of AI wiping out humanity because they want to dominate the market in c/localllama@poweruser.forum
[–] faldore@alien.top 1 points 1 year ago (1 children)

Agree it's very strange they didn't use his name