Large Language Models (LLMs) from the GPT family have become extremely popular, leading to a race towards reducing their inference costs to allow for efficient local computation. Yet, the vast majority of existing work focuses on weight-only quantization, which can reduce runtime costs in the memory-bound one-token-at-a-time generative setting, but does not address them in compute-bound scenarios, such as batched inference or prompt processing. In this paper, we address the general quantization problem, where both weights and activations should be quantized. We show, for the first time, that the majority of inference computations for large generative models such as LLaMA, OPT, and Falcon can be performed with both weights and activations being cast to 4 bits, in a way that leads to practical speedups, while at the same time maintaining good accuracy. We achieve this via a hybrid quantization strategy called QUIK, which compresses most of the weights and activations to 4-bit, while keeping some outlier weights and activations in higher-precision. The key feature of our scheme is that it is designed with computational efficiency in mind: we provide GPU kernels matching the QUIK format with highly-efficient layer-wise runtimes, which lead to practical end-to-end throughput improvements of up to 3.4x relative to FP16 execution. We provide detailed studies for models from the OPT, LLaMA-2 and Falcon families, as well as a first instance of accurate inference using quantization plus 2:4 sparsity. Large language models (LLMs) from the Generative Pretrained Joint weight-activation quantization methods, which can Transformer (GPT) family (Radford et al., 2019) provide computational improvements, but either focus exclusively are massively popular.

Whether you're looking for love or ways to celebrate your loved one, technology is playing an increasingly important role – especially during a pandemic. After all, many of us are forced to remain socially distant for the time being. Valentine's Day might be celebrated at home this year, as opposed to dining in a restaurant, and florists may sell more bouquets to online customers instead of in-store shoppers. As the expression goes, there's an app for that. Interestingly, even online dating apps aren't just used to find a mate over the internet, but quite literally to date online – until it's safe to meet in person.