LED-cnn-dataset-summarization

This model is a fine-tuned version of pszemraj/led-base-book-summary on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.0098
• Rouge1: 0.4061
• Rouge2: 0.1676
• Rougel: 0.2695
• Rougelsum: 0.3756
• Gen Len: 79.036

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: 8
• eval_batch_size: 8
• seed: 42
• 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	250	1.8883	0.4074	0.1733	0.2733	0.3741	81.696
1.9196	2.0	500	1.8782	0.4105	0.1738	0.2735	0.3789	85.312
1.9196	3.0	750	1.8763	0.408	0.1734	0.2747	0.3754	84.348
1.4188	4.0	1000	1.9043	0.4086	0.1716	0.273	0.3795	79.842
1.4188	5.0	1250	1.9344	0.4084	0.1686	0.2713	0.377	79.926
1.168	6.0	1500	1.9623	0.4121	0.1733	0.2749	0.3813	77.228
1.168	7.0	1750	2.0004	0.4092	0.1711	0.273	0.3794	77.102
1.0279	8.0	2000	2.0098	0.4061	0.1676	0.2695	0.3756	79.036

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.19.1