BART-Firefox-Simplification-Elementary

This model is a fine-tuned version of facebook/bart-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4478

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: 2.5e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 404
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss
2.1568	1.0	270	1.6097
1.7893	2.0	540	1.2922
1.4104	3.0	810	1.0986
1.192	4.0	1080	0.9639
1.0033	5.0	1350	0.8523
0.8717	6.0	1620	0.7346
0.7336	7.0	1890	0.6595
0.6545	8.0	2160	0.6199
0.5909	9.0	2430	0.5550
0.5096	10.0	2700	0.5255
0.4778	11.0	2970	0.5069
0.4307	12.0	3240	0.4759
0.404	13.0	3510	0.4557
0.3798	14.0	3780	0.4505
0.384	15.0	4050	0.4478

Framework versions

• Transformers 4.48.3
• Pytorch 2.6.0+cu126
• Datasets 3.0.1
• Tokenizers 0.21.0