mbart-large-50-many-to-many-mmt-finetuned-pt-to-en

This model is a fine-tuned version of facebook/mbart-large-50-many-to-many-mmt on the None dataset. It achieves the following results on the evaluation set:

• Loss: 3.8212
• Bleu: 5.1667
• Gen Len: 53.5411

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: 0.0002
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
No log	1.0	154	2.9955	1.7894	44.0952
No log	2.0	308	2.8970	2.326	46.8317
No log	3.0	462	2.9511	3.0767	55.8505
2.5276	4.0	616	3.0830	3.3555	48.737
2.5276	5.0	770	3.2703	3.5714	47.0602
2.5276	6.0	924	3.4241	4.3465	52.7744
1.0364	7.0	1078	3.5665	4.4909	52.705
1.0364	8.0	1232	3.6942	4.6399	50.4717
1.0364	9.0	1386	3.7852	5.2082	54.6205
0.3319	10.0	1540	3.8212	5.1667	53.5411

Framework versions

• Transformers 4.40.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1