e9da85655b71cfdf5572eb654469cad7

This model is a fine-tuned version of facebook/mbart-large-en-ro on the Helsinki-NLP/opus_books [de-fr] dataset. It achieves the following results on the evaluation set:

• Loss: 1.9846
• Data Size: 1.0
• Epoch Runtime: 232.5093
• Bleu: 8.8526

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	6.7515	0	19.8400	0.4308
No log	1	872	3.7312	0.0078	21.5766	1.1052
No log	2	1744	2.9432	0.0156	24.6077	2.5523
0.0583	3	2616	2.5704	0.0312	29.6079	3.7818
0.1724	4	3488	2.3565	0.0625	36.2208	4.6509
2.4368	5	4360	2.1626	0.125	49.8644	5.2806
2.1778	6	5232	2.0266	0.25	76.9490	6.2943
1.8892	7	6104	1.8864	0.5	130.7073	7.0564
1.7003	8.0	6976	1.7479	1.0	234.1093	7.8791
1.4497	9.0	7848	1.7299	1.0	232.2629	8.3945
1.26	10.0	8720	1.7513	1.0	230.9264	8.8920
1.0865	11.0	9592	1.8023	1.0	235.2139	8.6731
0.8866	12.0	10464	1.8794	1.0	232.5389	8.6166
0.7549	13.0	11336	1.9846	1.0	232.5093	8.8526

Framework versions

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