mt5-base-finetuned-english-finetuned-english-arabic

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

• Loss: 3.4788
• Rouge-1: 22.55
• Rouge-2: 9.84
• Rouge-l: 20.5
• Gen Len: 19.0
• Bertscore: 71.39

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: 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

Training Loss	Epoch	Step	Validation Loss	Rouge-1	Rouge-2	Rouge-l	Gen Len	Bertscore
4.999	1.0	1172	3.9343	17.67	5.93	15.86	19.0	69.69
4.008	2.0	2344	3.6655	19.48	7.67	17.67	19.0	70.49
3.7463	3.0	3516	3.5503	20.47	8.24	18.6	19.0	70.86
3.5924	4.0	4688	3.4942	20.95	8.45	19.05	19.0	71.0
3.4979	5.0	5860	3.4788	21.34	8.75	19.39	19.0	71.11

Framework versions

• Transformers 4.19.0
• Pytorch 1.11.0+cu113
• Datasets 2.2.1
• Tokenizers 0.12.1