efc2f0f3ad38fba941d2cf7e3b89b04d

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

• Loss: 3.5978
• Data Size: 1.0
• Epoch Runtime: 25.0390
• Bleu: 4.9265

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	8.6600	0	2.2778	0.0794
No log	1	75	6.5269	0.0078	2.8663	0.7332
No log	2	150	5.3531	0.0156	4.6225	1.7839
No log	3	225	5.0769	0.0312	5.8577	2.4407
No log	4	300	4.8408	0.0625	6.9479	2.8221
No log	5	375	4.5226	0.125	9.9208	3.0897
No log	6	450	3.8524	0.25	11.6724	3.3505
No log	7	525	3.3828	0.5	14.8770	2.4880
2.5376	8.0	600	2.6955	1.0	24.8158	4.6150
1.9259	9.0	675	2.6926	1.0	24.3749	4.4653
1.4294	10.0	750	2.8201	1.0	23.1688	4.3863
0.9685	11.0	825	3.0722	1.0	22.9417	4.7071
0.6825	12.0	900	3.3397	1.0	24.1269	4.4333
0.6853	13.0	975	3.5978	1.0	25.0390	4.9265

Framework versions

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