blessu

Thanks for your question. As indicated in the paper, we implemented similar solutions as in [ECCV'20 paper](https://arxiv.org/abs/2007.09654), and focused on those only based on `binary_cross_entropy`. Thus, the implementation of other...

Thanks for your interest. Yes, it's a one-dimensional list of label frequency in the training set as prepared in the [dataset_prep step](https://github.com/Roche/BalancedLossNLP/blob/main/PubMed/dataset_prep.ipynb)

Hi @zzzmm1 , thanks for your interest. Both datasets have their licences, thus we cannot provide them directly. Please check the dataset section of README https://github.com/Roche/BalancedLossNLP#datasets IMHO, it's helpful to...

Thanks for your interest, the following snippet should work: ``` # after downloading dataset from kaggle and unzipping it import json training_data = [] test_data = [] for line in...

> We need to git clone https://huggingface.co/bert-base-uncased and put it to the ./Reuters/berts ? > > thanks Yes, you may also need to confirm the files are accessible from `./berts/bert-base-uncased/`...

Thanks for your interest. I have searched the first error message (`packaging.version.InvalidVersion: Invalid version: '0.10.1,

torch.fx was added in PyTorch 1.8.0, while the environment required for the experiments is transformers==4.4.2 torch==1.7.0 While you may get it through with a new pytorch version, it can be...

Thanks for your interest. Yes, exactly, it's the number of training instances as prepared in the [dataset_prep step](https://github.com/Roche/BalancedLossNLP/blob/main/PubMed/dataset_prep.ipynb)

Hi, @zqudm, thanks for your questions. Please find my answers below (Q1 and Q2 are quite related so they are answered together). > The first question is about the way...

Hi @YangJenHao, thanks for your question. If you are referring to Table 3 in Appendix, the description of evaluation is mainly in the section "A.2 Additional Effectiveness Check", especially: >...