bart-base-cnn-YT-transcript-sum

This model is a fine-tuned version of ainize/bart-base-cnn on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.4969
• Rouge1: 27.1516
• Rouge2: 14.6227
• Rougel: 23.3968
• Rougelsum: 25.4786
• Gen Len: 19.9954

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	216	1.5374	24.7307	11.5124	20.6823	22.9189	19.9630
No log	2.0	432	1.4976	26.825	14.0512	23.2078	25.2044	19.9583
1.5449	3.0	648	1.4969	27.1516	14.6227	23.3968	25.4786	19.9954
1.5449	4.0	864	1.5345	27.2526	15.0873	23.8556	25.7798	19.9861
0.9	5.0	1080	1.5962	26.8267	14.7267	23.2263	25.2149	19.9676
0.9	6.0	1296	1.6378	26.8444	14.8753	23.254	25.2943	19.9815
0.5749	7.0	1512	1.6819	27.1776	14.898	23.2454	25.4298	19.9583
0.5749	8.0	1728	1.7360	26.9518	15.308	23.6574	25.2991	19.9769
0.5749	9.0	1944	1.7796	27.9253	15.7998	24.4827	26.4424	19.9769
0.3668	10.0	2160	1.8078	26.9211	15.0903	23.4484	25.4369	19.9815
0.3668	11.0	2376	1.8405	27.4434	15.3608	23.903	25.8117	19.9861
0.255	12.0	2592	1.8447	27.7175	15.7173	24.2096	26.0946	19.9815
0.255	13.0	2808	1.8834	27.2409	15.3865	23.7314	25.7682	19.9815
0.192	14.0	3024	1.8796	27.2939	15.5502	23.8294	25.7409	19.9815
0.192	15.0	3240	1.8851	27.6741	15.771	24.1976	26.1196	19.9722

Framework versions

• Transformers 4.33.2
• Pytorch 2.0.1+cu117
• Datasets 2.14.5
• Tokenizers 0.13.3