mt5-small-synthetic-data-plus-translated-bs64-ep20

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: 1.4566
• Rouge1: 0.2300
• Rouge2: 0.1163
• Rougel: 0.2056
• Rougelsum: 0.2055

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: 5.6e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
20.7702	1.0	19	11.1481	0.0054	0.0019	0.0053	0.0057
14.5169	2.0	38	6.9805	0.0043	0.0008	0.0043	0.0043
10.9637	3.0	57	5.0126	0.0043	0.0	0.0042	0.0038
8.5736	4.0	76	3.4889	0.0183	0.0036	0.0164	0.0164
6.9785	5.0	95	3.1837	0.0532	0.0068	0.0492	0.0499
5.7159	6.0	114	2.7398	0.0134	0.0009	0.0124	0.0127
4.7695	7.0	133	2.3907	0.0672	0.0248	0.0563	0.0562
4.1002	8.0	152	2.1993	0.1233	0.0514	0.1036	0.1038
3.6707	9.0	171	2.0524	0.1737	0.0689	0.1399	0.1411
3.3974	10.0	190	1.9375	0.2102	0.0854	0.1730	0.1733
3.1294	11.0	209	1.8367	0.2270	0.0974	0.1917	0.1919
3.0166	12.0	228	1.7489	0.2263	0.1061	0.1947	0.1956
2.7434	13.0	247	1.6770	0.2334	0.1073	0.1991	0.1999
2.7288	14.0	266	1.6183	0.2258	0.1057	0.1958	0.1967
2.6084	15.0	285	1.5695	0.2264	0.1089	0.1982	0.1986
2.5373	16.0	304	1.5259	0.2217	0.1086	0.1960	0.1961
2.4698	17.0	323	1.4961	0.2281	0.1124	0.2017	0.2014
2.3932	18.0	342	1.4754	0.2316	0.1141	0.2051	0.2048
2.375	19.0	361	1.4614	0.2281	0.1140	0.2032	0.2030
2.3621	20.0	380	1.4566	0.2300	0.1163	0.2056	0.2055

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0