nateraw/bert-base-uncased-ag-news

The model was finetuned on the AG News dataset using PyTorch Lightning . Sequence length 128, learning rate 2e-5, batch size 32, 4 T4 GPUs, 4 epochs . Data came from HuggingFace's datasets package. The data can be viewed on nlp viewer. The code can be found here. The model is based on the data from HuggedFace's dataset package. It is not the best model, it has limitations and bias. It has been described as 'not the best' by the authors of this article. The training procedure was described as a training procedure with a training sequence of 128 and a training period of 32 . The training sequence was,