mt5-small

This model is a fine-tuned version of google/mt5-small on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.3524
• Rouge1: 21.18
• Rouge2: 6.37
• Rougel: 20.84

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: 9
• eval_batch_size: 9
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel
4.6211	1.45	500	2.5968	16.96	4.86	16.73
3.1269	2.9	1000	2.4790	17.62	5.0	17.58
2.884	4.35	1500	2.4077	17.67	5.06	17.4
2.7627	5.8	2000	2.4003	18.67	5.42	18.26
2.638	7.25	2500	2.3953	18.76	5.49	18.44
2.5427	8.7	3000	2.3837	18.97	6.04	18.62
2.4846	10.14	3500	2.3957	20.17	6.23	19.88
2.3867	11.59	4000	2.3558	19.5	6.24	19.1
2.3651	13.04	4500	2.3225	19.6	6.18	19.2
2.2846	14.49	5000	2.3385	19.34	6.3	18.9
2.2351	15.94	5500	2.3413	20.42	6.44	19.93
2.1862	17.39	6000	2.3418	20.04	6.35	19.51
2.1375	18.84	6500	2.3438	21.02	6.56	20.45
2.0961	20.29	7000	2.3451	20.82	6.81	20.6
2.0686	21.74	7500	2.3571	20.46	6.57	20.03
2.0253	23.19	8000	2.3672	20.49	6.21	20.16
1.9997	24.64	8500	2.3524	21.18	6.37	20.84
1.9627	26.09	9000	2.3780	20.9	5.96	20.4
1.9561	27.54	9500	2.3808	21.06	6.59	20.76
1.902	28.99	10000	2.3739	20.73	6.09	20.41
1.8837	30.43	10500	2.3786	20.65	6.27	20.35
1.8587	31.88	11000	2.3853	20.44	6.23	20.0

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1+cu116
• Datasets 2.10.1
• Tokenizers 0.13.2