Bart-CNN-dataset

This model is a fine-tuned version of facebook/bart-large-cnn on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.2222
• Rouge1: 0.4398
• Rouge2: 0.1996
• Rougel: 0.2964
• Rougelsum: 0.4096
• Gen Len: 95.364

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: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• 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	1.4136	0.4361	0.2058	0.2957	0.4075	99.678
1.3139	2.0	500	1.4521	0.444	0.2085	0.3035	0.4138	90.808
1.3139	3.0	750	1.5573	0.4409	0.2046	0.2945	0.4102	100.502
0.7471	4.0	1000	1.6873	0.4429	0.205	0.2985	0.4119	96.34
0.7471	5.0	1250	1.8544	0.4395	0.2016	0.2964	0.409	100.1
0.4392	6.0	1500	2.0239	0.4407	0.2012	0.2946	0.4085	97.476
0.4392	7.0	1750	2.1492	0.4409	0.199	0.2947	0.4101	94.41
0.2886	8.0	2000	2.2222	0.4398	0.1996	0.2964	0.4096	95.364

Framework versions

• Transformers 4.46.3
• Pytorch 2.5.1+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3