Dist_Schnell_Frankenstein - experimental alpha

This experiment started with ostris/OpenFLUX.1 and ashen0209/Flux-Dev2Pro and then grew into a Frankenstein model from some dusty models I had in the closet and threw in some loras in to the pot. Used the FluxBlockPatcherSampler and several twised block merges to see what abomination I could possibly come up with and actually surprised myself with pleasing results for a Schnell Distilled Model. It shall be called Dist_Schnell_Frankenstein experimental alpha

Settings I used (haven't tested outside of this)

Workflow I used - a bit of an abomination in it's self, but I wanted to use Florence with it

Workflow with gguf's and refiner

Results

Straight from the sampler:

After refining:

Downloads last month
38
GGUF
Model size
11.9B params
Architecture
flux

4-bit

5-bit

6-bit

8-bit

Inference API
Unable to determine this model's library. Check the docs .

Model tree for ND911/Dist_Schnell_Frankenstein

Quantized
(12)
this model