1 point by mathnerd42 11 months ago flag hide 10 comments
elonmusk 11 months ago next
What about scikit-learn? I feel it doesn't get talked about much.
bigdatafan23 11 months ago next
scikit-learn stays relevant as a go-to library for many beginner ML tasks. Only worry is, it may not scale well for larger datasets.
elonmusk 11 months ago prev next
Kicking off the discussion with TensorFlow and PyTorch - which do you prefer for ML projects in 2023 and why?
bigdatafan23 11 months ago next
For me, it's PyTorch due to its simplicity and ease of debugging. But things might change in 2023, exciting!
nlpresearcher 11 months ago prev next
TensorFlow has better pre-trained libraries for NLP tasks, IMO.
dlibuser 11 months ago prev next
Another open-source library that's always worth considering for computer vision tasks is Dlib.
bigdatafan23 11 months ago next
Or PyTorch's torchvision, if, like me, you prefer PyTorch over TensorFlow.
nlpresearcher 11 months ago prev next
Dlib is a good tool, but if you are into deep learning and computer vision, consider using TensorFlow's Object Detection API.
opendatascientist 11 months ago prev next
Check out scikit-learn's faster alternative - scikit-optimize. Optimize and pick the best learning algorithm, effortlessly.
machinelearnjunkie 11 months ago next
scikit-optimize is interesting! Has anyone tried comparing it directly to scikit-learn for hyperparameter tuning?