Using llama.cpp release b3772 for quantization. Original model: https://huggingface.co/TheDrummer/Cydonia-22B-v1. All quants made using imatrix option with dataset from here. Run them in LM Studio. Full F16 weights. Uses Q8_0 for embed and output weights. Very high quality, near perfect, recommended. Uses …

BeaverAI proudly presents... Cydonia 22B v1 💿. I christen this model, 'Miqu 2 Mini' - @invisietch. You can mix it up and see which works best for you. What's Next? I might release a v1.1...

It's able to write engaging and impactful content, carry out roleplay mostly effectively, and manage to respond well. It shocked me with the first output by introducing a character that is not written anywhere in the card.

Using llama.cpp release b3972 for quantization. Original model: https://huggingface.co/TheDrummer/Cydonia-22B-v1.2. All quants made using imatrix option with dataset from here. Run them in LM Studio. Full F16 weights. Uses Q8_0 for embed and output weights. Very high quality, near perfect, recommended. Uses …

Sep 18, 2024 · So right now, I have managed to get Cydonia 22B V1.1 to work for character description extraction. I switched from silly tavern to the newest update of lm studio 0.3.3, and cut down the text that I was giving it. That helped greatly.

Nov 24, 2024 · Cydonia is a very good all around model, but having only a 3060 gpu makes roleplay a little less authentic especially when Rocinante-12B-v1.1-Q6_K operates as good at 30 tokes/s. Whereas Cydonia 22B runs the prompt at 9 t/s the output is very similar.

I broke the new Cydonia and I'm not sure how. I don't think it's anything to do with your finetune specifically, probably an issue with KCPP. Cydonia or Unslop Nemo have pretty much been my daily drivers since you first released them.

Cydonia 22B v1.1 💿. I christen this model, ‘Miqu 2 Mini’ - @invisietch. What’s New with v1.1? Lightly decensored: Less refusals; Less positivity: Less plot armor, less moralizing; Holds back better on the moist; Smarter! Links. Original: https://huggingface.co/TheDrummer/Cydonia-22B-v1.1; GGUF: https://huggingface.co/TheDrummer/Cydonia ...

Nov 6, 2024 · The Cydonia-22B-v1 model excels at open-ended, interactive storytelling and role-playing. It can engage in detailed character interactions, develop complex plots, and generate vivid descriptions and worldbuilding.

Cydonia 22B v1.1 💿 I christen this model, 'Miqu 2 Mini' - @invisietch. What's New with v1.1? Lightly decensored: Less refusals; Less positivity: Less plot armor, less moralizing; Holds back better on the moist; Smarter! Links Original: https://huggingface.co/TheDrummer/Cydonia-22B-v1.1; GGUF: https://huggingface.co/TheDrummer/Cydonia-22B-v1 ...