b2ab8e49dcb71e09f756f41b49f65d23

This model is a fine-tuned version of facebook/mbart-large-cc25 on the Helsinki-NLP/opus_books [it-pt] dataset. It achieves the following results on the evaluation set:

• Loss: 5.9639
• Data Size: 1.0
• Epoch Runtime: 12.6170
• Bleu: 0.3199

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	15.5090	0	1.6414	0.2043
No log	1	29	8.7051	0.0078	2.0649	1.0218
No log	2	58	7.8696	0.0156	3.3083	0.4334
No log	3	87	7.5790	0.0312	4.3296	0.5795
No log	4	116	8.7429	0.0625	5.2492	0.3622
No log	5	145	6.5636	0.125	6.5791	0.9823
0.7358	6	174	5.7163	0.25	7.9295	2.0386
0.7358	7	203	4.8836	0.5	9.6100	3.2410
0.7358	8.0	232	4.0254	1.0	13.7327	4.0043
2.8873	9.0	261	3.6651	1.0	13.2528	4.9644
2.8873	10.0	290	18.5718	1.0	13.2408	0.0463
6.6421	11.0	319	12.7992	1.0	13.9846	0.0
6.6421	12.0	348	6.8240	1.0	13.0073	0.0532
11.4042	13.0	377	5.9639	1.0	12.6170	0.3199

Framework versions

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