t5-arabic-base-finetuned-xlsum-ar

This model is a fine-tuned version of bakrianoo/t5-arabic-base on the xlsum dataset. It achieves the following results on the evaluation set:

• Loss: 3.0328
• Rouge-1: 23.72
• Rouge-2: 10.95
• Rouge-l: 21.59
• Gen Len: 19.0
• Bertscore: 71.81

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: 0.0005
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 250
• num_epochs: 10
• label_smoothing_factor: 0.1

Training results

Framework versions

• Transformers 4.19.4
• Pytorch 1.11.0+cu113
• Datasets 2.2.2
• Tokenizers 0.12.1