4966ac501748d756bc84cb75c8913cda

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

• Loss: 3.1017
• Data Size: 1.0
• Epoch Runtime: 26.2700
• Bleu: 6.0714

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	7.8580	0	2.4247	0.4905
No log	1	91	6.3259	0.0078	3.3552	1.7822
No log	2	182	5.6126	0.0156	4.6390	2.1084
No log	3	273	5.3795	0.0312	5.7078	2.0559
No log	4	364	5.0829	0.0625	7.0377	2.5933
No log	5	455	4.1455	0.125	8.7233	2.9658
No log	6	546	3.3250	0.25	11.3480	3.4999
0.4224	7	637	2.6789	0.5	17.3167	4.2796
2.2001	8.0	728	2.3384	1.0	28.1657	9.4805
1.5345	9.0	819	2.4065	1.0	26.0358	9.3155
1.0357	10.0	910	2.6319	1.0	27.5458	9.0920
0.694	11.0	1001	2.8348	1.0	28.0565	7.8990
0.4021	12.0	1092	3.1017	1.0	26.2700	6.0714

Framework versions

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