2ffdcffd70eb8f3c41bfbccd14df9209

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

• Loss: 2.3441
• Data Size: 1.0
• Epoch Runtime: 333.7994
• Bleu: 9.4162

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	9.3893	0	28.9578	0.2080
No log	1	1286	3.2857	0.0078	32.6526	8.1477
0.0727	2	2572	2.8567	0.0156	34.7107	7.3171
0.0807	3	3858	2.5293	0.0312	39.7063	5.9600
0.1113	4	5144	2.3319	0.0625	50.3564	6.8175
2.3098	5	6430	2.2014	0.125	69.8308	7.2238
2.1357	6	7716	2.0828	0.25	107.2597	8.8328
2.5511	7	9002	2.3429	0.5	184.3453	13.5245
1.872	8.0	10288	1.9355	1.0	334.1931	8.7933
1.6041	9.0	11574	1.8910	1.0	337.7265	9.7002
1.3249	10.0	12860	1.9350	1.0	334.5037	9.3442
1.0926	11.0	14146	2.0589	1.0	337.4164	8.8583
0.9308	12.0	15432	2.1886	1.0	334.9627	9.1417
0.7153	13.0	16718	2.3441	1.0	333.7994	9.4162

Framework versions

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