BART-Firefox-Simplification-Secondary

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: 1.2009

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: 269
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss
2.492	1.0	180	2.4670
2.1175	2.0	360	2.1910
1.8209	3.0	540	1.9628
1.5473	4.0	720	1.8677
1.3788	5.0	900	1.7496
1.2256	6.0	1080	1.6135
1.0851	7.0	1260	1.5506
1.0034	8.0	1440	1.4427
0.9083	9.0	1620	1.3785
0.839	10.0	1800	1.3249
0.7747	11.0	1980	1.2845
0.7332	12.0	2160	1.2503
0.7062	13.0	2340	1.2189
0.6824	14.0	2520	1.2087
0.6568	15.0	2700	1.2009

Framework versions

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