AraBART-finetuned-ar-wikilingua

This model is a fine-tuned version of moussaKam/AraBART on the wiki_lingua dataset. It achieves the following results on the evaluation set:

• Loss: 3.9990
• Rouge-1: 23.82
• Rouge-2: 8.97
• Rouge-l: 21.05
• Gen Len: 19.06
• Bertscore: 72.08

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 250
• num_epochs: 8
• label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge-1	Rouge-2	Rouge-l	Gen Len	Bertscore
4.2331	1.0	5111	4.0713	21.42	7.69	19.08	18.79	71.22
3.9438	2.0	10222	4.0251	23.1	8.63	20.59	18.41	71.86
3.7372	3.0	15333	3.9744	22.98	8.47	20.3	19.2	71.74
3.5782	4.0	20444	3.9680	23.37	8.67	20.79	18.93	71.85
3.4509	5.0	25555	3.9643	23.42	8.85	20.71	19.33	71.88
3.3471	6.0	30666	3.9831	23.41	8.75	20.69	19.18	71.97
3.2673	7.0	35777	3.9917	23.93	9.13	21.16	19.0	72.11
3.214	8.0	40888	3.9990	23.94	9.1	21.21	19.13	72.11

Framework versions

• Transformers 4.18.0
• Pytorch 1.10.0+cu111
• Datasets 2.1.0
• Tokenizers 0.12.1