xsum_and_billsum_and_samsum_old

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

• Loss: 3.6572
• Rouge1: 0.0873
• Rouge2: 0.0186
• Rougel: 0.0712
• Rougelsum: 0.0711
• Gen Len: 19.0
• Precision: 0.069
• Recall: -0.2018
• F1: -0.0
• Hashcode: roberta-large_L17_no-idf_version=0.3.12(hug_trans=4.28.1)-rescaled

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: 72
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 6
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len	Precision	Recall	F1	Hashcode
No log	1.0	42	3.7360	0.0789	0.0152	0.0679	0.068	19.0	0.071	-0.2095	-0.0	roberta-large_L17_no-idf_version=0.3.12(hug_trans=4.28.1)-rescaled
No log	2.0	84	3.6936	0.085	0.0193	0.071	0.0712	19.0	0.0898	-0.2007	-0.0	roberta-large_L17_no-idf_version=0.3.12(hug_trans=4.28.1)-rescaled
No log	3.0	126	3.6739	0.0854	0.0202	0.0712	0.0713	19.0	0.0973	-0.1935	-0.0	roberta-large_L17_no-idf_version=0.3.12(hug_trans=4.28.1)-rescaled
No log	4.0	168	3.6643	0.0865	0.0194	0.0709	0.0709	19.0	0.0845	-0.197	-0.0	roberta-large_L17_no-idf_version=0.3.12(hug_trans=4.28.1)-rescaled
No log	5.0	210	3.6589	0.087	0.0186	0.0711	0.0711	19.0	0.0698	-0.2012	-0.0	roberta-large_L17_no-idf_version=0.3.12(hug_trans=4.28.1)-rescaled
No log	6.0	252	3.6572	0.0873	0.0186	0.0712	0.0711	19.0	0.069	-0.2018	-0.0	roberta-large_L17_no-idf_version=0.3.12(hug_trans=4.28.1)-rescaled

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.0+cu118
• Datasets 2.12.0
• Tokenizers 0.13.3