7a583d8577e5f79bc54befed4a12623e

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

• Loss: 4.0743
• Data Size: 1.0
• Epoch Runtime: 24.5907
• Bleu: 6.7658

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	10.7855	0	2.4020	0.2633
No log	1	85	9.5226	0.0078	3.2265	0.3460
No log	2	170	9.0365	0.0156	3.6326	0.3080
No log	3	255	8.4422	0.0312	4.7386	0.4078
No log	4	340	7.8162	0.0625	6.2939	0.5854
0.5131	5	425	6.1052	0.125	8.1055	1.3545
0.5131	6	510	5.1733	0.25	10.6596	0.6543
1.5794	7	595	4.0760	0.5	15.0997	2.4706
3.6159	8.0	680	3.5109	1.0	26.4927	7.2870
2.6892	9.0	765	3.3684	1.0	26.2263	8.6876
2.112	10.0	850	3.4185	1.0	24.1976	9.1893
1.5701	11.0	935	3.6304	1.0	25.2792	7.8953
1.114	12.0	1020	3.8229	1.0	25.4259	8.5545
0.824	13.0	1105	4.0743	1.0	24.5907	6.7658

Framework versions

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