3B model?

#1
by KatyTheCutie - opened

Will you experiment with 3B models anytime soon? I think StableLM's 3B model has potential, the Zephyr model is outstanding for a 3B model but its not that good for Roleplay.

Will you experiment with 3B models anytime soon? I think StableLM's 3B model has potential, the Zephyr model is outstanding for a 3B model but its not that good for Roleplay.

I will try to experiment with 3B models today. I will definitely try with StableLM's model for you.

Thank you! You're the best ☺️

HuggingFaceH4/no_robots: for more human-like chats.
Teknium/openhermes: for increased knowledge and intelligence.
Airoboros 3.1: for better writing ability.
Unalignment/toxic-dpo-v0.1: for uncensored responses.

Just some suggestions.

Would you want me to experiment with a MoE or Finetune for 3B?

A 3B finetune would be nice but anything would be good!

A 3B finetune would be nice but anything would be good!

All files should be uploaded soon with information. https://huggingface.co/Walmart-the-bag/zephyr-quiklang-3b

Thank you! Can you perhaps make a Q6 GGUF of it? I don't want to bother @TheBloke with it, Thanks again!

Thank you! Can you perhaps make a Q6 GGUF of it? I don't want to bother @TheBloke with it, Thanks again!

I am attempting to, but keep failing. I will keep trying but I would suggest TheBloke to do it since I am unexperienced with llama.cpp and gguf quants.

TheBloke has published GGUF version.

Walmart-the-bag changed discussion status to closed

Sign up or log in to comment