dialogue-samsum-pro-spacy

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

• Loss: 0.3250
• Rouge1: 48.0216
• Rouge2: 24.8336
• Rougel: 40.4736
• Rougelsum: 40.4829
• Gen Len: 18.1944

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
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
0.3887	1.0	1841	0.3393	46.6582	23.7253	39.6291	39.5901	17.7531
0.3486	2.0	3682	0.3261	48.0581	24.6841	40.3964	40.4169	18.187
0.3249	3.0	5523	0.3250	48.0216	24.8336	40.4736	40.4829	18.1944

Framework versions

• Transformers 4.20.1
• Pytorch 1.12.1+cu102
• Datasets 2.10.1
• Tokenizers 0.12.1