d866be3915f999906084ee146a64d4a6

This model is a fine-tuned version of facebook/mbart-large-cc25 on the Helsinki-NLP/opus_books [fr-sv] dataset. It achieves the following results on the evaluation set:

• Loss: 4.7249
• Data Size: 1.0
• Epoch Runtime: 23.4469
• Bleu: 0.6315

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	14.5099	0	2.5722	0.0425
No log	1	75	9.1157	0.0078	3.1838	0.1670
No log	2	150	7.9709	0.0156	4.5702	0.1770
No log	3	225	7.4838	0.0312	5.7484	0.1404
No log	4	300	7.4019	0.0625	7.8269	0.1518
No log	5	375	6.6843	0.125	10.1069	0.2183
No log	6	450	6.4293	0.25	12.3869	0.2556
No log	7	525	5.6304	0.5	16.0891	0.5882
5.1956	8.0	600	5.0341	1.0	26.0756	0.9583
4.7375	9.0	675	4.6402	1.0	25.1907	1.3361
13.1269	10.0	750	8.0745	1.0	23.3981	0.0107
6.9412	11.0	825	4.5933	1.0	24.2348	1.6506
4.9768	12.0	900	4.9942	1.0	24.4450	0.4189
4.4863	13.0	975	4.8060	1.0	25.1328	0.3593
4.1397	14.0	1050	4.7346	1.0	23.4493	0.5793
3.7855	15.0	1125	4.7249	1.0	23.4469	0.6315

Framework versions

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