Mt5 Base Arabic

🚀 mt5-base-arabic

This model is a fine - tuned version of google/mt5-base on the Arabic subset of the xlsum dataset, designed for Arabic text summarization.

🚀 Quick Start

This model is a fine - tuned version of google/mt5-base on the Arabic subset of the xlsum dataset. It achieves the following results on the evaluation set:

• Loss: 3.2742
• Rouge - 1: 22.86
• Rouge - 2: 10.31
• Rouge - l: 20.85
• Gen Len: 19.0
• Bertscore: 71.52

✨ Features

• Fine - Tuned for Arabic: Specifically fine - tuned on the Arabic subset of the xlsum dataset, making it suitable for Arabic text summarization tasks.
• Quantifiable Performance: Achieves specific performance metrics on the evaluation set, such as Rouge and Bertscore.

🔧 Technical Details

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0005
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
• lr_scheduler_type: linear
• num_epochs: 5
• label_smoothing_factor: 0.1

Training results

Framework versions

• Transformers 4.18.0
• Pytorch 1.11.0+cu113
• Datasets 2.1.0
• Tokenizers 0.12.1

📄 License

This model is released under the Apache - 2.0 license.