Hugging Face's logo

KaraKaraWitch
/
Llama-MiraiFanfare-3.3-70B

Text Generation
Transformers
Safetensors
llama
mergekit
Merge
conversational
text-generation-inference
Inference Endpoints
Model card Files Community
2

Settings

#2
by Evoc - opened
Discussion
Evoc
•
edited 9 days ago

What are the recommended settings for this model? ChatML context/instruct? Or Alpaca?

KaraKaraWitch
 Owner
•
edited 9 days ago

ChatML appears to work fine. I think Alpaca is also fine based on the discussion mentioned but I would personally stick to chat formats if possible :)

Evoc
•
edited 9 days ago

ChatML appears to work fine. I think Alpaca is also fine based on the discussion mentioned but I would personally stick to chat formats if possible :)

Thanks!

Evoc changed discussion status to closed
Nohobby

Konnect's Llam@ception works really well with this.
Here it is w/ the samplers I use: https://files.catbox.moe/aa1sfk.json

KaraKaraWitch
 Owner

Konnect's Llam@ception works really well with this.
Here it is w/ the samplers I use: https://files.catbox.moe/aa1sfk.json

Whatever boats your float. Though I have to say that is an overly long system prompt for any model.

Sign up or log in to comment