Hacker News Reader

fishingboyon Google has the most big pre-trained models while Alibaba has the largest

Yes, it is the sum of all models. But M6-10T (MoE) of Alibaba is a 10 trillion one.

fishingboyon BMList – A list of big pre-trained models (GPT-3, DALL-E2...)

As far as I am concerned, there are many ways to compress a model such as quantization, pruning, and knowledge distillation.

By the way, I found a package called BMCook when I browsed the OpenBMB repo, which implements several algorithms and also compares it with other model compression packages. Hope this can help you.