This is a pruned version of Gemma 4 26B A4B down from 128 to 96 experts based on a Heretic abliterated version.

Safetensors

Model size

20B params

Tensor type

BF16

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for blascotobasco/Gemma-4-96E-A4B-Heretic

Base model

Finetuned

Finetuned

(4)

this model

Quantizations