distill-pegasus-cnn-arxiv-pubmed-v3-e4

This model is a fine-tuned version of theojolliffe/distill-pegasus-cnn-arxiv-pubmed on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.8962
• Rouge1: 49.5676
• Rouge2: 30.7141
• Rougel: 34.191
• Rougelsum: 45.0269
• Gen Len: 125.8333

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

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
2.7729	1.0	795	2.1332	48.4776	29.8247	33.8775	44.0771	126.2407
2.3362	2.0	1590	1.9953	48.7574	30.0148	33.8955	44.3967	126.2407
2.2766	3.0	2385	1.9159	49.3004	30.5548	34.5702	44.8082	125.5
2.1815	4.0	3180	1.8962	49.5676	30.7141	34.191	45.0269	125.8333

Framework versions

• Transformers 4.18.0
• Pytorch 1.11.0+cu113
• Datasets 2.1.0
• Tokenizers 0.12.1