parrot_paraphraser_on_T5-finetuned-xsum-v5

This model is a fine-tuned version of prithivida/parrot_paraphraser_on_T5 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0345
• Rouge1: 86.5078
• Rouge2: 84.8978
• Rougel: 86.4798
• Rougelsum: 86.4726
• Gen Len: 17.8462

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: 4
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
0.0663	1.0	2002	0.0539	86.0677	84.063	86.0423	86.0313	17.8671
0.0449	2.0	4004	0.0388	86.4564	84.7606	86.432	86.4212	17.8501
0.0269	3.0	6006	0.0347	86.4997	84.8907	86.4814	86.4744	17.8501
0.023	4.0	8008	0.0345	86.5078	84.8978	86.4798	86.4726	17.8462

Framework versions

• Transformers 4.22.0
• Pytorch 1.12.1+cu113
• Datasets 2.4.0
• Tokenizers 0.12.1