Dataset used: https://huggingface.co/datasets/dair-ai/emotion (Transformed in such a way that label is word not int, can be found compressed in the repo)
Order of operations:
- split_dataset.py DATASET_PATH TRAIN_TEST_RATIO
- (Optional) balance_dataset.py TRAIN_DATASET_PATH
- create_transformers_dataset.py
- train.py
- serve.py
- evaluate.py
- generate.py
- python -m wandb sweep sweep.yml
- have a great classifier :)
Some of these python files need to be edited for your use. This code is a supplement of this workshop: https://www.youtube.com/watch?v=fyydvBcJTn8 (In slovak language)