9719bc8ccf2cf39c858026f6c15e898e

This model is a fine-tuned version of facebook/mbart-large-50-many-to-many-mmt on the Helsinki-NLP/opus_books [fr-pt] dataset. It achieves the following results on the evaluation set:

• Loss: 2.6595
• Data Size: 1.0
• Epoch Runtime: 16.2787
• Bleu: 9.9842

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	7.1296	0	1.2755	0.4828
No log	1	31	4.8102	0.0078	1.7038	1.2952
No log	2	62	3.5218	0.0156	2.2359	4.0128
No log	3	93	3.1881	0.0312	3.4010	4.9513
No log	4	124	2.8720	0.0625	4.1919	5.0907
No log	5	155	2.6261	0.125	5.7186	6.0193
No log	6	186	2.4622	0.25	7.4086	7.3636
0.4311	7	217	2.3007	0.5	9.2420	7.9751
0.4311	8.0	248	2.1624	1.0	13.4369	10.8006
1.2742	9.0	279	2.2148	1.0	13.3917	10.6475
1.1394	10.0	310	2.3417	1.0	13.4760	15.4858
1.1394	11.0	341	2.5018	1.0	13.7842	9.3105
0.6495	12.0	372	2.6595	1.0	16.2787	9.9842

Framework versions

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