distilbart-cnn-6-6-finetuned-summscreen-10-epochs

This model is a fine-tuned version of sshleifer/distilbart-cnn-6-6 on the SummScreen dataset. It achieves the following results on the evaluation set:

• Loss: 3.4962
• Rouge1: 26.3499
• Rouge2: 7.3999
• Rougel: 18.6087
• Rougelsum: 23.17
• Gen Len: 49.8609

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: 1e-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: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
3.1229	1.0	3673	3.1271	26.6959	7.4401	18.8303	23.7132	49.9763
2.8872	2.0	7346	3.0482	26.6447	7.5599	18.5921	23.2786	49.8195
2.5733	3.0	11019	3.0292	27.425	7.9963	19.3544	24.1281	49.8757
2.3886	4.0	14692	3.0625	27.1291	7.5541	18.9375	23.8729	49.8905
2.215	5.0	18365	3.1118	27.1773	7.551	19.0524	24.1015	49.9142
2.0377	6.0	22038	3.2086	27.2237	7.8821	19.2136	24.0477	49.784
1.9358	7.0	25711	3.3405	26.7555	7.6628	18.8609	23.5264	49.8343
1.8292	8.0	29384	3.4124	26.7741	7.4529	18.9276	23.5827	49.8757
1.7702	9.0	33057	3.4457	26.6281	7.4415	18.7932	23.4608	49.8639
1.7443	10.0	36730	3.4962	26.3499	7.3999	18.6087	23.17	49.8609

Framework versions

• Transformers 4.26.0
• Pytorch 1.13.1
• Datasets 2.9.0
• Tokenizers 0.13.2