bart-cnn-science-v3-e5

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

• Loss: 0.8090
• Rouge1: 54.0053
• Rouge2: 35.5018
• Rougel: 37.3204
• Rougelsum: 51.5456
• Gen Len: 142.0

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: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	398	0.9935	51.9669	31.8139	34.4748	49.5311	141.7407
1.1747	2.0	796	0.8565	51.7344	31.7341	34.3917	49.2488	141.7222
0.7125	3.0	1194	0.8252	52.829	33.2332	35.8865	50.1883	141.5556
0.4991	4.0	1592	0.8222	53.582	33.4906	35.7232	50.589	142.0
0.4991	5.0	1990	0.8090	54.0053	35.5018	37.3204	51.5456	142.0

Framework versions

• Transformers 4.19.2
• Pytorch 1.11.0+cu113
• Datasets 2.2.2
• Tokenizers 0.12.1