e6f8d099c2a80d45144d3a01936b488e

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

• Loss: 2.7399
• Data Size: 1.0
• Epoch Runtime: 29.4615
• Bleu: 4.9891

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	12.9722	0	2.9103	0.1568
No log	1	88	5.1781	0.0078	3.5080	0.8185
No log	2	176	4.0243	0.0156	4.6759	1.7675
No log	3	264	3.6400	0.0312	6.5857	1.9102
No log	4	352	3.3374	0.0625	8.3241	3.2994
No log	5	440	3.0319	0.125	11.1517	3.7263
0.2839	6	528	2.7076	0.25	13.5217	4.3892
0.9231	7	616	2.3614	0.5	16.9778	4.7927
2.0869	8.0	704	2.2113	1.0	29.0496	5.1338
1.5903	9.0	792	2.3178	1.0	29.2749	5.6954
1.1879	10.0	880	2.4009	1.0	26.1508	5.0705
0.9312	11.0	968	2.6025	1.0	27.1503	5.2318
0.7235	12.0	1056	2.7399	1.0	29.4615	4.9891

Framework versions

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