Advertisement
If you have a new account but are having problems posting or verifying your account, please email us on hello@boards.ie for help. Thanks :)
Hello all! Please ensure that you are posting a new thread or question in the appropriate forum. The Feedback forum is overwhelmed with questions that are having to be moved elsewhere. If you need help to verify your account contact hello@boards.ie

Pytorch V Keras

Options
  • 30-03-2021 10:17am
    #1
    aidanki
    Registered Users Posts: 1,262 ✭✭✭


    I have recently completed a beginner course in Python so I now can read and semi understand what is going on in python code, v proud of that considering I knew zilch a few weeks ago.

    I would like to do some expts with Machine Learning, using either Pytorch or Keras as these seem to be the popular ones.

    Anyone care to comment on which is the better one to use

    ResNet50 also seems to be a v popular Neual Network, are there any better ones available


Comments

  • Options
    #2
    Talisman
    Registered Users Posts: 6,007 ✭✭✭Talisman


    If you know nothing of machine learning it doesn't really matter which one you choose as there is a learning curve involved in each.

    PyTorch operates at a lower level than Keras. Keras runs on top of TensorFlow which operates at the same level as PyTorch. What this means is that Keras is simpler to code, but PyTorch will feel more Pythonic and also executes faster.

    As for neural networks, it's horses for courses the network is trained to solve the problem. Resnet50 is just an image classifier, it won't be of much use to you for doing sentiment analysis or sales predictions.

    If you are interested in the field of machine learning then start with the simpler stuff in the scikit-learn library before moving on to Deep Learning. You'll need to have a good understanding of the fundamentals to know what you are doing with neural networks.

    You should also check out Kaggle, the site has easy to follow courses which will teach you the basics. https://www.kaggle.com/learn/overview


Advertisement