bart-large-cnn-samsum

This model is a fine-tuned version of facebook/bart-large-cnn on the samsum dataset. It achieves the following results on the evaluation set:

• Loss: 0.3028
• Rouge1: 0.4139
• Rouge2: 0.2105
• Rougel: 0.3191
• Rougelsum: 0.3193
• Gen Len: 60.0134

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
• gradient_accumulation_steps: 16
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 200
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
2.9128	0.4344	100	0.3621	0.3984	0.1999	0.3038	0.3038	60.8888
0.3205	0.8689	200	0.3097	0.4102	0.2138	0.3186	0.3188	60.6345
0.2702	1.3033	300	0.3041	0.4159	0.211	0.3179	0.3179	60.077
0.251	1.7377	400	0.2964	0.4191	0.2154	0.3229	0.3233	59.9022
0.2262	2.1721	500	0.3055	0.4135	0.208	0.3178	0.3179	60.4132
0.1906	2.6066	600	0.3028	0.4139	0.2105	0.3191	0.3193	60.0134

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.19.1