PyTorch - For the TensorFlow developers and others DL enthusiasts
Abdul Muneer (~abdulmuneer) |
PyTorch is the new and upcoming deep learning framework that is shaking up the industry after Google launched TensorFlow. For a long time Tensorflow was the de-facto leader in deep learning frameworks and was far more sophisticated than alternatives like Caffe and Torch. However, its leadership was also because of the lack of competing alternatives.
When Facebook stealthily developed PyTorch along with a few participating organisations and open sourced it, the research community embraced it for its simplicity and the natural flow of coding style. Pytorch also liberates developers from lot of restrictions that other existing frameworks enforced. In this talk I intend to explore what makes PyTorch so special and appealing. The talk explains the programming model in comparison with TensorFlow which will shed more insights into both frameworks.
Introductory knowledge of Machine Learning concepts. Familiarity with any Deep Learning framework.
I am a Neural Networks developer at minds.ai. I work on adding AI to automobiles for driver-assistance and vehicle stability controls. I have always been a passionate about Python with career long dedication to mastering it. My recent interests are in advancements in Artificial Intelligence especially the re-inforcement and attention learning networks.