I have a text classification dataset. The aim is to predict the category of an article based on its title. I have about 100 categories and 10 thousands instances. I've tried models like RNN, LSTM. I've tried pre-trained word embedding models like GloVE. LSTM gave better results than RNN. I also got an improvement with Glove. As for BERT, I expected to have a better performance, but it was the opposite, as I had the worst performance with this model. I've never implemented BERT before. I've only followed tutorials and books I've read (so I don't know if it's because BERT doesn't perform well in this case or if it's because I haven't implemented it properly). So my problem isn't necessarily a bug in my code. But improving the performance of my model. Is this platform designed to help improve a model's performance (The reason I ask is that it's a bit like asking for a free lunch, asking people to improve their own model.) ? If so, I'd like to share the link to my dataset with the code I've written, including how I implemented the architecture of my BERT model. Thank you for your understanding

  • 2
    $\begingroup$ You can get help as long as it is about Data Science. In addition, you should clearly identify your problem and prepare it. You don't have to share the full data set: Most data scientists will only do some of the work for you. Instead, you can simplify with a sample and have a better chance of getting an answer. $\endgroup$ May 28 at 11:54


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Browse other questions tagged or ask your own question.