cd5ac0396c8baf2a64916b0bdad9b4ed

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

• Loss: 3.6903
• Data Size: 1.0
• Epoch Runtime: 13.9298
• Bleu: 9.2473

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	9.2186	0	1.5789	0.2249
No log	1	31	7.8248	0.0078	2.1816	0.2741
No log	2	62	7.5447	0.0156	2.3733	0.3527
No log	3	93	6.9544	0.0312	3.2070	0.3803
No log	4	124	6.4377	0.0625	4.3919	1.1416
No log	5	155	5.7863	0.125	6.0806	1.2125
No log	6	186	5.0049	0.25	8.1399	1.9876
0.8774	7	217	4.1997	0.5	10.4376	3.2581
0.8774	8.0	248	3.4824	1.0	14.0748	4.6720
2.5924	9.0	279	3.2444	1.0	13.3294	6.0135
2.613	10.0	310	3.2265	1.0	14.0302	7.5174
2.613	11.0	341	3.2647	1.0	14.8535	7.5746
1.869	12.0	372	3.3601	1.0	13.0517	7.0081
1.2897	13.0	403	3.5095	1.0	12.9070	10.4037
1.2897	14.0	434	3.6903	1.0	13.9298	9.2473

Framework versions

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