search

udacity / deep-learning-v2-pytorch

Projects and exercises for the latest Deep Learning ND program https://www.udacity.com/course/deep-learning-nanodegree--nd101
MIT License
5.26k stars 5.32k forks source link

deep-learning-v1 keras/tf student here, are the videos mostly the same? #414

Open nyck33 opened 1 year ago

nyck33 commented 1 year ago

I took it a long time ago and we used Keras and TF but I want to learn Pytorch now.

Are the videos and curriculum mostly the same, just implementation is in Pytorch?

Siddharth-Latthe-07 commented 1 month ago

When transitioning from Keras and TensorFlow to PyTorch, the fundamental concepts of deep learning and neural networks remain the same, but the implementation details and some aspects of the workflow will differ.

key differences:-

  1. Syntax and API: Keras/TF: Keras is high-level and designed to be user-friendly, often allowing for quick model building with less code. PyTorch: PyTorch provides more control and flexibility, with a more explicit approach to building and training models.

  2. Dynamic vs. Static Graphs: Keras/TF: TensorFlow (especially in its early versions) used static computation graphs, which required defining the entire computation graph before running it. PyTorch: Uses dynamic computation graphs, allowing you to modify the graph on the fly. This can make debugging easier and more intuitive.

Hope, this helps Thanks