A public repo where I'll put my latest KTO results for the v0.2.3 version of my Humanize model. Working towards a final version, but I figure I'll throw this out there. Similar to the first, the model is unusable without KTO, though it slightly more coherent at base. This model gives longer responses than the previous, and isn't great at basic chatting. I'll try to update it a bit less spastically but no promises.
Basic Sampler Settings
Temp: 0.8 TopK: 80 TopP: 0.9 RepPen: 1.1
Formatting Settings for Silly Tavern. It should end up as ChatML with username/character names.
- Downloads last month
- 22
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support
HF Inference deployability: The model has no library tag.