3fa651ae4098ff158c41963dd45ea26f

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

• Loss: 2.9247
• Data Size: 1.0
• Epoch Runtime: 99.9175
• Bleu: 7.7503

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	8.2932	0	8.8124	0.2396
No log	1	367	4.7360	0.0078	10.3623	3.1137
No log	2	734	4.3401	0.0156	10.6132	4.0427
No log	3	1101	4.0037	0.0312	13.5390	7.0928
No log	4	1468	3.0994	0.0625	17.1962	10.1187
0.1871	5	1835	2.5217	0.125	23.5135	8.2708
9.6921	6	2202	2.5881	0.25	35.3885	10.0388
7.1376	7	2569	3.2690	0.5	54.3256	6.4547
2.3887	8.0	2936	2.3186	1.0	100.0643	12.9116
1.8443	9.0	3303	2.2327	1.0	98.6656	13.7467
1.5588	10.0	3670	2.3916	1.0	100.6215	8.3050
1.161	11.0	4037	2.5093	1.0	100.2174	8.9844
0.9576	12.0	4404	2.6915	1.0	100.1221	9.4485
0.7418	13.0	4771	2.9247	1.0	99.9175	7.7503

Framework versions

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