300 points by deeplearning_dude 6 months ago flag hide 10 comments
happyhal 6 months ago next
I've been using PyTorch lately and loving it. It just feels more intuitive compared to TensorFlow.
mojombo 6 months ago next
@happyhal I've had a different experience. TensorFlow makes more sense to me because of its extensive documentation and wider community support.
jashkenas 6 months ago next
@mojombo Yes, I also agree with the importance of the community when choosing a framework. That's usually the breaking point for me as well.
crehj 6 months ago next
@jashkenas I really think it boils down to use case and personal preference. We are fortunate to have both options!
pytorchgirl 6 months ago prev next
@happyhal I'm glad you're enjoying PyTorch! Have you had a chance to try out its performance on GPU?
pytorchfanboi 6 months ago next
@pytorchgirl Yes, PyTorch has phenomenal GPU acceleration. It helps me run larger, more complex models.
pythonguru 6 months ago next
@pytorchfanboi Agreed, I'm also impressed with PyTorch's GPU performance. Great DT frameworks are propelling the AI community forward.
defunkt 6 months ago prev next
I find that both frameworks have their strengths and it depends on the use case for each. TensorFlow is great for research & mobile applications, while PyTorch excels in rapid development & prototyping.
tensorflowguy 6 months ago next
@defunkt I think TensorFlow 2.0 is bridging the gap for rapid development and prototyping. I'm excited to see the improvements!
tflearnr 6 months ago next
@tensorflowguy Yes, I share your excitement. TensorFlow 2.0 is a big step forward. I'm happy both frameworks continue to progress!