Introduction

Backprop is a serverless model platform that makes it simple for developers to use machine learning models in any application.

Each model is exposed as a private API that deploys with zero downtime, scales automatically, and requires no maintenance, enabling to focus on building great applications.

Using pre-trained models

When you Sign up, you'll have access to pre-trained models like GPT2, BERT, T5, CLIP.
These models are trained to perform various tasks in a zero-shot setting, from classifying images and text to generating stories.

As a criterion, our pre-trained models are trained to solve broad tasks, and not niche ones. It means you will find models that can classify text given any set of labels, but not models that can tell popular tweets apart from unpopular ones.

To solve your niche tasks, check out Deploying models.

Supported tasks

Backprop is organised with a tasks first approach. First you pick a task you want to solve and then you select the appropriate model.

As an example, you may want to do the Text Classification task using a model optimised for English or for 100+ languages.

See the table below for a list of supported tasks. Try them out in Sandbox and view the API docs for more in-depth information.

TaskDescriptionReferences
Q&AAnswers questions given contextAPI Sandbox
Text ClassificationClassifies text given labelsAPI Sandbox
Image ClassificationClassifies images given labelsAPI Sandbox
Text SummarisationSummarises long textAPI Sandbox
Emotion detectionDetects emotions from textAPI Sandbox
Text GenerationGenerates text based on the given input textAPI Sandbox
Text VectorisationTurns variable size text into a fixed size vectorAPI Sandbox
Image VectorisationTurns image into vectorAPI Sandbox
CustomUse your deployed model as you see fitAPI Sandbox