9559809085d81ea0918f6cc09adc8a11

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

• Loss: 3.6486
• Data Size: 1.0
• Epoch Runtime: 20.9637
• Bleu: 2.9522

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	9.6954	0	2.0643	0.2991
No log	1	70	7.3586	0.0078	2.9052	0.4269
No log	2	140	6.3621	0.0156	4.2957	0.7367
No log	3	210	6.0442	0.0312	6.2398	1.0680
No log	4	280	5.7339	0.0625	7.5422	1.3191
No log	5	350	5.3398	0.125	9.5736	1.4760
No log	6	420	4.3855	0.25	10.7838	1.6985
0.7223	7	490	3.1550	0.5	13.4291	1.7767
2.7579	8.0	560	2.9552	1.0	21.8813	2.0613
2.1118	9.0	630	2.9647	1.0	22.5246	2.1925
1.4601	10.0	700	3.1970	1.0	20.9238	2.3621
1.0066	11.0	770	3.4365	1.0	20.6748	2.7413
0.8526	12.0	840	3.6486	1.0	20.9637	2.9522

Framework versions

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