2d921873baea05afbcdcf67debdf7580

This model is a fine-tuned version of facebook/mbart-large-50 on the Helsinki-NLP/opus_books [fr-no] dataset. It achieves the following results on the evaluation set:

• Loss: 4.3274
• Data Size: 1.0
• Epoch Runtime: 24.4286
• Bleu: 4.7146

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	11.4340	0	2.4022	0.1060
No log	1	86	9.7941	0.0078	3.1293	0.3876
No log	2	172	8.8955	0.0156	3.5843	0.2696
No log	3	258	8.4337	0.0312	4.7256	0.6412
No log	4	344	7.6835	0.0625	6.2131	0.7732
0.4301	5	430	6.6806	0.125	8.3339	1.3186
1.4997	6	516	6.4004	0.25	10.6593	1.6043
1.6665	7	602	18.3758	0.5	14.8248	0.0
7.4167	8.0	688	3.9553	1.0	25.7836	2.6216
3.6414	9.0	774	3.5492	1.0	26.0319	3.9327
2.653	10.0	860	3.3694	1.0	24.5324	4.8542
2.1175	11.0	946	3.4610	1.0	25.0039	5.1236
1.4093	12.0	1032	3.7954	1.0	25.6803	4.7693
1.0473	13.0	1118	4.0693	1.0	24.5941	4.6989
0.7443	14.0	1204	4.3274	1.0	24.4286	4.7146

Framework versions

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