# ExLlamaV2

An inference library with a flexible mixed-precision quantization format and optimized kernels.

## What it is used for

Running quantized models quickly on a single GPU.