Model Zoo – Practicing with Pre-trained Models

In order to make it easier for you to experience Deci and to practice using it, this page lists models that have been pre-trained and pre-packaged for you. They are ready to be served for inference with INFERY (Deci's Runtime Inference Python Package).

The Deci Model Hub is displayed by clicking the Model Hub tab on the navigation bar.


Simply click on a Clone to Lab button and the model will be benchmarked and waiting for you in the Lab page of the platform. You can then optimize this model for the production environment of your choice and then deploy it in RTIC or INFERY.

See the Quickstart for a description of this flow.

Note – Because you are using a ready-made model to practice with Deci, these models have already been benchmarked. Therefore, the benchmark results of a model will be automatically fetched upon cloning the model.

Each row in this page represents another model that is owned by DECI and made public.

  • Name – Specifies the name assigned by Deci to this model.
  • Domain/Task – Specifies the type of data and the purpose of the data set on which this model was trained, such as Computer Vision/Classification.
  • Dataset – Specifies the name of a public dataset on which this model was trained.
  • Accuracy – Specifies the accuracy to which this model was trained. Accuracy is the primary method for evaluating a model. It specifies the proportion of predictions that the model gets right.
  • Framework – Specifies the framework (programming language) used to develop this model.
  • Input Dimensions – Specifies the input dimensions expected by the model. This is the syntax that the model is configured to handle and which will be used to measure the model’s performance.
  • Batch Size – Specifies the production inference batch size for which Deci optimized the model.
  • Deci Score – Deci assigns a standardized/normalized score to describe the efficiency of a model’s runtime performance on a specific production environment, without compromising accuracy. A model has two types of performance – accuracy and efficiency. A Deci Score is a normalized grade of between one and 10 that evaluates a model’s runtime efficiency on a specific hardware and batch size set up. It is comprised of the performance metrics and measures for Accuracy, Throughput, Latency, Memory Footprint and Model Size.
  • Owner – Deci Public. The models listed on this page are intended to be used by the public with the Deci platform.
  • Downloads – Specifies the number of times this model was downloaded by Deci users to be used in Deci throughout the world.