smallsuper/

Mistral-7B-v0.1-4bit-64rank

This is a bare model without any output layer or classification head. It has been quantized to be used for feature extraction tasks.

This model is intended to be used as a base for training on downstream tasks. In order to use it for predictions and inference, it should be fine-tuned on a specific task with an appropriate output layer or classification head added.

The model has been quantized using the following parameters:

Lora alpha: 16

Lora rank: 64

Lora target modules: all-linear

bits: 4

LoftQ iterations: 5