Brightics AI Accelerator

Features

• Initial and Data Assessments

  Assessment service explores your specific use case and determines a scope of work to produce desired results while a Data Assessment determines the predictive power of your data based on a snapshot of the data over a milestone-based, phased approach and make a GO/NO-GO decision to proceed.

• Automated setup

  Offers simplified one-click installation in the Cloud and accelerated setup for on-premise.

• PyCharm IDE

  AI Machine Learning teams run data preparation, training, and inference jobs entirely from PyCharm IDE using REST APIs

• Jupyter notebooks

  Data science teams run data preparation, training, and inference jobs entirely from one single interface with minimal code

• Advanced cluster management

  Integrated environment for Data Science and Machine Learning teams to collaborate using simple, automated distributed training, data preparation and inference on large clusters

• Linear scalability

  Increases image data throughput near linearly with large numbers of up to 256 GPUs in your cluster for Keras, TensorFlow and PyTorch training

• AutoDL

  Automates and accelerates deep learning model training using data-parallel, distributed synchronous Horovod Ring-All-Reduce TensorFlow and PyTorch frameworks with minimal code. AutoDL exploits up to 256 GPUs per iteration to produce a model in 2 hours versus 3 weeks using traditional methods. Automates transfer learning for image data considering all models in the model zoo with hyper-parameter search.

• AutoML

  Automates and accelerates model training on tabular data by using automated model selection from scikit-learn, automated feature synthesis, and hyper-parameter search optimization. AutoML with synthetic feature generation exploits up to 256 CPU cores simultaneously to produce a scikit-learn model in 1 hour versus 2 months using traditional methods.