bart-large-cnn-xsum

This model is a fine-tuned version of facebook/bart-large-cnn on the xsum dataset. It achieves the following results on the evaluation set:

• Loss: 2.0314

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: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 500
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.9458	0.3921	500	1.8663
1.7833	0.7842	1000	1.9308
1.3364	1.1762	1500	1.9378
1.3562	1.5683	2000	1.9538
1.3173	1.9604	2500	1.8672
0.9227	2.3525	3000	2.0590
0.8619	2.7446	3500	2.0314

Framework versions

• Transformers 4.46.3
• Pytorch 2.5.1+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3