aad5838167148ea61700c0664ced95db

This model is a fine-tuned version of google/mt5-large on the Helsinki-NLP/opus_books [fr-pt] dataset. It achieves the following results on the evaluation set:

• Loss: 1.6999
• Data Size: 1.0
• Epoch Runtime: 22.6237
• Bleu: 9.0351

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Bleu
No log	0	0	23.7785	0	2.0497	0.0163
No log	1	31	24.4749	0.0078	2.6113	0.0123
No log	2	62	25.5218	0.0156	3.7359	0.0115
No log	3	93	26.0026	0.0312	6.4084	0.0091
No log	4	124	25.5681	0.0625	8.0860	0.0158
No log	5	155	19.5432	0.125	10.8673	0.0179
No log	6	186	17.2617	0.25	13.9709	0.0153
3.604	7	217	13.9010	0.5	16.4235	0.0289
3.604	8.0	248	7.1512	1.0	23.2488	0.0378
9.9694	9.0	279	5.2699	1.0	23.7114	0.0690
8.337	10.0	310	3.0625	1.0	19.4387	0.2920
8.337	11.0	341	2.1975	1.0	20.7641	0.8443
3.8149	12.0	372	1.9373	1.0	21.6020	1.0372
2.5394	13.0	403	1.8138	1.0	20.0556	6.2818
2.5394	14.0	434	1.7498	1.0	21.0757	6.3339
2.1857	15.0	465	1.7314	1.0	22.7169	6.8624
2.1857	16.0	496	1.7054	1.0	19.7962	7.2596
1.955	17.0	527	1.6864	1.0	20.6538	7.7493
1.7829	18.0	558	1.6770	1.0	21.4480	8.3674
1.7829	19.0	589	1.6752	1.0	20.1836	8.4784
1.6663	20.0	620	1.6707	1.0	20.7797	8.8787
1.5617	21.0	651	1.6675	1.0	21.6892	9.3177
1.5617	22.0	682	1.6835	1.0	22.8971	9.1090
1.4621	23.0	713	1.6914	1.0	20.3055	9.2020
1.4621	24.0	744	1.6866	1.0	21.1386	9.0153
1.3424	25.0	775	1.6999	1.0	22.6237	9.0351

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.2.0
• Tokenizers 0.22.1