GPT-6B-J example based on the official [Huggingface Transformers] library.
Credit to https://github.com/paulcjh/gpt-j-6b/blob/main/gpt-j-t4.ipynb
python3 -m pip install einops transformersYou can run the following command with an AWS EC2 g4dn.xlarge. It will generate the continuation of the sentence "Why AutoGluon is great?".
bash run_generation.sh
***output_context: Why AutoGluon is great?
Autogluon is the best platform for AI - Analytics. This is simply because of the fact that Autogluon makes it possible for everyone to build, test and distribute their own AI-based solutions. This has been made possible by its contribution to further and seamless machine learning in the industry.
However, let's start by discussing what can Autogluon do for you?
What Autogluon does
Provides cutting edge AI solutions for any business that wants to make them smarter and faster.
One could argue that today, this is limited only to those in the industry. However, Autogluon is currently working on the provision of such products to those outside of the industry. The product is still not fully operational, however.
For any business that is already using Autogluon, some things can be already done with the cloud product. But, how about something totally different? We also have a development-ready product, that you could useastical workspaces. However, these workspaces are still being worked on and are probably a few months from public release. But, what if you could start coding your own intelligent applications?
Well, now you can! Autogluon comes with one such example of a project that has been built on the Autogluon platform. The project will be announced in the weeks ahead.
For the full picture, what is Autogluon?onial AI platform is a machine learning development platform.
It provides pre-built, open source solutions for building an AI model, deploying it on-premise, and hosting it on the cloud. You can also develop your own custom solutions in Autogluon. This is what makes the AI model development so interesting.
Autogluon also contains a machine learning library that provides useful APIs for model deployment and training as well as for making predictions.
For the last point, the Prediction Engine provides data scientists with built-in models that can be used to make predictions. There is also a model factory to build models, as well as REST APIs and a web client for integration with third-party applications.
If you want a project to be hosted, Autogluon can set up a service for you. You can also import and run AI projects from GitHub as well as share code using GitLab.
Currently, there are two products available in Autogluon. One is a cloud-based product, and the other is a development-ready product. We will talk about the cloud-based product first.
Our cloud-based product, Autogluon Cloud is easy to use, and enables anyone to build intelligent solutions. You can also deploy your project on the cloud in a couple of minutes.
In fact, Autogluon Cloud is free to use until your project is ready to go live. Once you want to put your model to work, you need to pay a monthly fee.
However, if you don't want to spend any money, then you could deploy your model locally. This is the basis of our development-ready product, Autogluon Dev.
The autogluon dev
Autogluon Dev is a free project that can be used to deploy a project on-premise. That means that the project can be run anywhere with an internet connection.
This also means that you can run an AI model in your home, your office, or wherever is most convenient.
For developers, Autogluon Dev provides an ideal environment to work on. It has the ability to import and run project from Github.
Additionally, Autogluon Dev will allow you to take an AI project that you have developed on-premise and put it to work on the cloud.
We will