RDBT [Anima]

See this page for more info. This page for update log.

Personal finetuned + distilled model. Based on pretrained anima (hf link).

Sharing merges using this model is not allowed.

Random experiment, random quality. New version != better version. Feel free to leave feedback.

It's highly recommended that you use a "trained" base model, not "merged".

• This is a heavily trained LoRA. It needs to be loaded on a base model close to the pretrained model. "merged" models usually have very messy weights. E.g. merged all anima preview versions. Sounds cool, but is nonsense.

• Most (75+%) "merged" base models have merged "turbo" LoRA. Stacking distilled models is not a good idea. The model will break very easily.