FuturisticVibes/Luca-MN-bf16-8.0bpw-h8-exl2

I have no idea what I’m doing… if this causes the apocalypse someone please let me know.

Luca-MN-bf16 8.0bpw h8 EXL2

Includes measurement.json file for further quantization

I’ll be taking a break until mid-2025 to recoup some funds, might still do small models occasionally, but anything big will have to wait. See you all next year! 😊

Original Model: https://huggingface.co/rAIfle/Luca-MN-bf16

Original Model Card

Luca-MN-bf16

This thing was just intended as an experiment but it turned out quite good. I had it both name and prompt imagegen for itself.

Created by running a high-r LoRA-pass over Nemo-Base with 2 epochs of some RP data, then a low-r pass with 0.5 epochs of the c2-data, then 3 epochs of DPO using jondurbin/gutenberg-dpo-v0.1.

Prompting

Use the Mistral V3-Tekken context- and instruct-templates. Temperature at about 1.25 seems to be the sweet spot, with either MinP at 0.05 or TopP at 0.9. DRY/Smoothing etc depending on your preference.

Quantized versions

• iMat GGUFs, courtesy of the Quant-Cartel