mT5-fine-tune

This model is a fine-tuned version of google/mt5-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.5256
• Rouge1: 0.0822
• Rouge2: 0.0244
• Rougel: 0.0813
• Rougelsum: 0.0814
• Gen Len: 18.9803
• Chrf Score: 20.301
• Chrf Char Order: 6
• Chrf Word Order: 0
• Chrf Beta: 2

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: 0.0001
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 7

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len	Chrf Score	Chrf Char Order	Chrf Word Order	Chrf Beta
3.5479	1.0	1951	2.7435	0.0672	0.021	0.0666	0.0667	18.9323	19.2495	6	0	2
3.1717	2.0	3902	2.6452	0.0746	0.0207	0.0738	0.0737	18.9814	20.1079	6	0	2
3.0151	3.0	5853	2.6014	0.0834	0.0243	0.0826	0.0823	18.9891	20.2875	6	0	2
2.95	4.0	7804	2.5647	0.0765	0.0218	0.0757	0.0757	18.981	20.2327	6	0	2
2.8592	5.0	9755	2.5480	0.0822	0.0242	0.0814	0.0813	18.9819	20.3982	6	0	2
2.8214	6.0	11706	2.5317	0.0841	0.0255	0.0831	0.083	18.9764	20.3935	6	0	2
2.789	7.0	13657	2.5256	0.0822	0.0244	0.0813	0.0814	18.9803	20.301	6	0	2

Framework versions

• Transformers 4.33.0
• Pytorch 2.0.0
• Datasets 2.1.0
• Tokenizers 0.13.3