bart-base-finetuned-xsum

This model is a fine-tuned version of facebook/bart-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3082
• Rouge1: 13.998
• Rouge2: 8.2042
• Rougel: 13.2682
• Rougelsum: 13.6101
• Gen Len: 20.0

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: 0.002
• train_batch_size: 10
• eval_batch_size: 10
• seed: 42
• gradient_accumulation_steps: 5
• total_train_batch_size: 50
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 8
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	180	1.5657	13.284	6.5156	12.5098	12.8132	20.0
No log	2.0	360	1.0051	13.3398	6.426	12.4508	12.9061	20.0
1.9306	3.0	540	0.7440	12.8324	6.2848	11.9985	12.4138	20.0
1.9306	4.0	720	0.5412	13.2678	6.7085	12.4631	12.9015	20.0
1.9306	5.0	900	0.4288	13.4092	6.9196	12.4719	12.9804	20.0
0.625	6.0	1080	0.3653	13.8777	7.53	13.0145	13.4782	20.0
0.625	7.0	1260	0.3255	14.585	8.5087	13.8215	14.1797	20.0
0.625	8.0	1440	0.3082	13.998	8.2042	13.2682	13.6101	20.0

Framework versions

• Transformers 4.27.0.dev0
• Pytorch 1.13.1+cu116
• Datasets 2.9.0
• Tokenizers 0.13.2