egyptian_modernbert_fineweb2

Modern EgyBert is a 🤗 Modern Bert transformers model trained on the Egyptian Arabic Split of the FineWeb2 dataset.

Language(s) (NLP): Egyptian Arabic

Finetuned from model: Modern Bert Base

It achieves the following results on the evaluation set:

• eval_loss: 2.5126
• eval_runtime: 134.9987
• eval_samples_per_second: 68.267
• eval_steps_per_second: 8.533
• epoch: 3
• step: 120000

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 32
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 3

Framework versions

• Transformers 4.49.0.dev0
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0