dialogue-samsum

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

• Loss: 0.3249
• Rouge1: 48.0133
• Rouge2: 24.9057
• Rougel: 40.6842
• Rougelsum: 40.6602
• Gen Len: 18.2384

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: 2e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
0.3968	0.9997	1841	0.3374	47.4452	24.2213	40.0832	40.024	18.3875
0.3432	2.0	3683	0.3270	47.721	24.8189	40.4846	40.4736	18.143
0.324	2.9992	5523	0.3249	48.0133	24.9057	40.6842	40.6602	18.2384

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1