<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

hu-covid-fake-news-classifier

This model is a finetuned version of the hubert-base-cc and was finetuned for covid fake news detection in hungarian for the BME MIGT2022 competition.

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3.0

Framework versions

• Transformers 4.18.0
• Pytorch 1.11.0
• Datasets 2.0.0
• Tokenizers 0.11.0