mt5-small-synthetic-data-plus-translated-bs32

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: 0.8369
• Rouge1: 0.6206
• Rouge2: 0.4859
• Rougel: 0.5972
• Rougelsum: 0.5979

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: 32
• eval_batch_size: 32
• 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: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
19.4785	1.0	38	11.5404	0.0055	0.0008	0.0051	0.0051
11.9977	2.0	76	6.4079	0.0101	0.0015	0.0089	0.0094
7.5027	3.0	114	3.0626	0.0542	0.0093	0.0482	0.0487
4.8939	4.0	152	2.2496	0.0492	0.0182	0.0429	0.0437
3.64	5.0	190	1.7984	0.1870	0.0826	0.1598	0.1601
2.8662	6.0	228	1.4518	0.1852	0.0916	0.1653	0.1659
2.4493	7.0	266	1.3124	0.4183	0.2586	0.4014	0.4026
2.1362	8.0	304	1.2444	0.4386	0.2716	0.4176	0.4196
1.9923	9.0	342	1.1876	0.4587	0.3034	0.4387	0.4404
1.8438	10.0	380	1.1486	0.5198	0.3637	0.4979	0.4988
1.7212	11.0	418	1.1031	0.5402	0.3848	0.5160	0.5169
1.6315	12.0	456	1.0707	0.5556	0.3999	0.5325	0.5341
1.5623	13.0	494	1.0437	0.5808	0.4309	0.5583	0.5593
1.5269	14.0	532	1.0188	0.5986	0.4540	0.5773	0.5772
1.4668	15.0	570	0.9982	0.5922	0.4511	0.5731	0.5737
1.4357	16.0	608	0.9777	0.5965	0.4549	0.5768	0.5773
1.3684	17.0	646	0.9623	0.6123	0.4722	0.5901	0.5907
1.3675	18.0	684	0.9461	0.6135	0.4771	0.5915	0.5919
1.3285	19.0	722	0.9324	0.6150	0.4754	0.5916	0.5918
1.288	20.0	760	0.9271	0.6179	0.4803	0.5964	0.5968
1.2529	21.0	798	0.9129	0.6156	0.4789	0.5939	0.5940
1.2216	22.0	836	0.9017	0.6163	0.4817	0.5941	0.5941
1.2322	23.0	874	0.8948	0.6208	0.4839	0.5985	0.5986
1.2062	24.0	912	0.8838	0.6139	0.4778	0.5904	0.5912
1.1642	25.0	950	0.8761	0.6150	0.4818	0.5939	0.5951
1.1699	26.0	988	0.8759	0.6152	0.4794	0.5929	0.5932
1.1428	27.0	1026	0.8662	0.6158	0.4806	0.5935	0.5946
1.195	28.0	1064	0.8609	0.6126	0.4758	0.5898	0.5908
1.1619	29.0	1102	0.8568	0.6152	0.4776	0.5924	0.5936
1.1172	30.0	1140	0.8548	0.6181	0.4788	0.5951	0.5964
1.1141	31.0	1178	0.8526	0.6148	0.4766	0.5904	0.5914
1.1176	32.0	1216	0.8488	0.6201	0.4834	0.5963	0.5972
1.0959	33.0	1254	0.8475	0.6225	0.4847	0.5983	0.5993
1.0954	34.0	1292	0.8437	0.6220	0.4859	0.5987	0.5986
1.0844	35.0	1330	0.8420	0.6206	0.4851	0.5969	0.5974
1.1041	36.0	1368	0.8398	0.6222	0.4865	0.5991	0.5992
1.0736	37.0	1406	0.8386	0.6225	0.4867	0.5991	0.6001
1.0816	38.0	1444	0.8376	0.6229	0.4871	0.5994	0.6001
1.0537	39.0	1482	0.8372	0.6242	0.4876	0.6004	0.6013
1.092	40.0	1520	0.8369	0.6206	0.4859	0.5972	0.5979

Framework versions

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