t5-finetuned-ar-to-arsl

This model is a fine-tuned version of PRAli22/arat5-arabic-dialects-translation on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4184
• Rouge1: 0.0
• Rouge2: 0.0
• Rougel: 0.0
• Rougelsum: 0.0
• Gen Len: 5.2837

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: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
1.7177	1.0	631	0.5319	0.0	0.0	0.0	0.0	5.1458
0.648	2.0	1262	0.4079	0.0	0.0	0.0	0.0	5.1918
0.4575	3.0	1893	0.3779	0.0	0.0	0.0	0.0	5.2567
0.3147	4.0	2524	0.3666	0.0	0.0	0.0	0.0	5.2662
0.2603	5.0	3155	0.3682	0.0	0.0	0.0	0.0	5.2726
0.2376	6.0	3786	0.3755	0.0	0.0	0.0	0.0	5.2758
0.2157	7.0	4417	0.3767	0.0	0.0	0.0	0.0	5.2821
0.182	8.0	5048	0.3994	0.0	0.0	0.0	0.0	5.2821
0.1631	9.0	5679	0.3910	0.0	0.0	0.0	0.0	5.2868
0.1526	10.0	6310	0.3991	0.0	0.0	0.0	0.0	5.2853
0.1463	11.0	6941	0.4110	0.0	0.0	0.0	0.0	5.2805
0.1301	12.0	7572	0.4094	0.0	0.0	0.0	0.0	5.2805
0.1278	13.0	8203	0.4126	0.0	0.0	0.0	0.0	5.2821
0.1253	14.0	8834	0.4184	0.0	0.0	0.0	0.0	5.2837
0.1165	15.0	9465	0.4184	0.0	0.0	0.0	0.0	5.2837

Framework versions

• Transformers 4.39.3
• Pytorch 2.1.2
• Datasets 2.18.0
• Tokenizers 0.15.2