keyphrase-extractions_bart-large

This model is a fine-tuned version of facebook/bart-large on the kp20k dataset. It achieves the following results on the evaluation set:

• Loss: 1.7257
• Rouge1: 0.4713
• Rouge2: 0.2385
• Rougel: 0.384
• Rougelsum: 0.3841
• Gen Len: 18.3164
• Phrase match: 0.1917

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• 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	Phrase match
2.5104	1.0	730	1.8021	0.464	0.2336	0.3765	0.3766	18.9074	0.1784
1.8436	2.0	1460	1.7473	0.4709	0.2381	0.3834	0.3836	17.8127	0.1891
1.6864	3.0	2190	1.7257	0.4713	0.2385	0.384	0.3841	18.3164	0.1917

Framework versions

• Transformers 4.26.0
• Pytorch 1.13.1+cu116
• Datasets 2.8.0
• Tokenizers 0.13.2