DeepSpark unveiled, no-cost RHEL developer subscription, and Google’s Machine Learning…

Databricks has unveiled DeepSpark, a new milestone in Apache Spark. DeepSpark uses neural networks to automate software development, such as writing test cases, fixing bugs, implementing features according to specs, and reviewing pull requests for their correctness, simplicity and style. Databricks wrote in a blog post that scaling Spark’s development has been top priority. As a multifaceted program trained to examine diffs against a Spark codebase, DeepSpark automatically write its own patches for Spark. By reviewing pull requests, this AI can both enforce a high and consistent standard for code quality as well as make constructive suggestions, wrote the company. It consists of three 15-layer convolutional neural networks on a 12,000-node Spark cluster using 1.2PB of memory. By using these three networks in synchronicity, DeepSpark is able to effectively review…


Link to Full Article: DeepSpark unveiled, no-cost RHEL developer subscription, and Google’s Machine Learning…