Originally posted by: Helena.

IBM Spectrum Conductor Deep Learning Impact version 1.1 delivers an enterprise-class solution for deep learning. It is a new add-on to IBM Spectrum Conductor with Spark version 2.2.1 that enables data scientists to run distributed deep learning workloads. Data scientists can prepare and import data, and run training and inference using industry standard open source and IBM PowerAI deep learning frameworks, like Caffe and TensorFlow. It is available for both x86 and IBM Power servers.

IBM Spectrum Conductor Deep Learning Impact leverages the power of GPUs to provide an engine that supports distributed deep learning across multiple GPUs and multiple hosts; efficiently scaling out on an IBM Spectrum Conductor with Spark cluster.

It also allows for automatic scaling of resource management which makes it possible for compute resources to be automatically added to a training job as the demand increases.

When using IBM Spectrum Conductor Deep Learning Impact, data scientists can use deep learning insights during training and inference. Deep learning insights provides training visualizations and runtime monitoring, enabling the data scientist to judge the model accuracy during training and to stop, adjust and restart training runs as needed.

IBM Spectrum Conductor Deep Learning Impact also includes hyperparameter tuning which uses multiple iterations and parallel processes running on a distributed Spark architecture to help data scientists find better hyperparameters and quickly improve accuracy.

IBM Spectrum Conductor Deep Learning Impact version 1.1 is available on IBM Passport Advantage or try it today and download the free trial now!

For more information about IBM Spectrum Conductor Deep Learning Impact and these deep learning capabilities, go to: www.ibm.biz/DeepLearningImpact