bert-base-parsbert-uncased-finetuned-khorshid-accelerate

This model is a fine-tuned version of HooshvareLab/bert-base-parsbert-uncased on a sentiment analysis dataset of Hala-Khorshid T.V program dataset. We used 🤗 Accelerate to train the model.

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3.0

Training results

More information needed

Framework versions

• Transformers 4.25.1
• Pytorch 1.13.1+cu116
• Datasets 2.8.0
• Tokenizers 0.13.2