bart-base-cnndm

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

• Loss: 1.5802
• Rouge1: 25.0336
• Rouge2: 12.5344
• Rougel: 20.8721
• Rougelsum: 23.5806
• Gen Len: 19.9998

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: 3e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 4

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
1.845	1.0	8972	1.6461	24.8325	12.327	20.6952	23.3653	19.9998
1.7427	2.0	17945	1.6098	24.9118	12.4577	20.786	23.4624	19.9998
1.6727	3.0	26917	1.5881	24.9723	12.4738	20.8317	23.5195	19.9994
1.6288	4.0	35888	1.5802	25.0336	12.5344	20.8721	23.5806	19.9998

Framework versions

• Transformers 4.27.1
• Pytorch 2.0.1+cu118
• Datasets 2.9.0
• Tokenizers 0.13.3