IBM has just announced a new "Deep Learning as a Service" project for AI developers, referred to as DLaaS. With this service, users can train neural networks such as TensorFlow, PyTorch, and Caffe with a popular framework without the need to purchase and maintain costly hardware. For example, data scientists can use only the resources they need to train the model, paying for GPU time.
Each cloud processing unit is set up in accordance with the principle of ease of use, and is ready for deep learning network programming without the user having to manage the infrastructure.
IBM researchers mentioned in a white paper:
Users can select based on a set of supported deep learning frameworks, neural network models, training data, cost constraints, etc., and then the DLAaS service will help complete the rest and provide them with an interactive, iterable training experience.
In short, to use this service, users only need to prepare their data, upload to DLaaS, start training, and then download training results. This kind of operation is quite easy, allowing developers to save training for a few days or even weeks.
Clearly, IBM wants to work hard to overcome the challenges of training neural networks, or at least to lower the barriers to entry in this industry.
It is reported that the current DLaaS is running on IBM's excellent Watson platform, which means it has tested the world's most advanced artificial intelligence system.
8.4V Charger,8.4V Li-Ion Battery Charger,Battery Charger For Drone,Li Ion Battery Charger
ShenZhen Yinghuiyuan Electronics Co.,Ltd , https://www.yhypoweradapter.com