| Pretraining KoLD Dataset with pretrained "koelectra-v3" model. | |
| dataset : https://github.com/boychaboy/KOLD | |
| pretrained_model : https://huggingface.co/monologg/koelectra-base-v3-discriminator | |
| So you should use tokenizer with "koelectra-base-v3-discriminator". | |
| label maps are like | |
| > | |
| {0: "not_hate_speech", 1: "hate_speech"} |