bert-small2bert-small-finetuned-cnn_daily_mail-summarization-finetuned-multi_news

This model is a fine-tuned version of mrm8488/bert-small2bert-small-finetuned-cnn_daily_mail-summarization on the multi_news dataset. It achieves the following results on the evaluation set:

• Loss: 4.3760
• Rouge1: 38.5318
• Rouge2: 12.7285
• Rougel: 21.4358
• Rougelsum: 33.4565
• Gen Len: 128.985

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: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 5
• mixed_precision_training: Native AMP
• label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
4.6946	0.89	400	4.5393	37.164	11.5191	20.2519	32.1568	126.415
4.5128	1.78	800	4.4185	38.2345	12.2053	20.954	33.0667	128.975
4.2926	2.67	1200	4.3866	38.4475	12.6488	21.3046	33.2768	129.0
4.231	3.56	1600	4.3808	38.7008	12.6323	21.307	33.3693	128.955
4.125	4.44	2000	4.3760	38.5318	12.7285	21.4358	33.4565	128.985

Framework versions

• Transformers 4.20.1
• Pytorch 1.11.0
• Datasets 2.1.0
• Tokenizers 0.12.1