5e24ce48e8861558fe55400f88651223

This model is a fine-tuned version of facebook/mbart-large-50-many-to-one-mmt on the Helsinki-NLP/opus_books [en-it] dataset. It achieves the following results on the evaluation set:

• Loss: 2.5671
• Data Size: 1.0
• Epoch Runtime: 205.7961
• Bleu: 6.1673

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	6.3353	0	17.2871	0.4116
No log	1	808	4.9832	0.0078	18.9028	0.6862
No log	2	1616	4.4294	0.0156	21.0383	1.1176
No log	3	2424	3.9470	0.0312	24.6304	1.7129
0.1279	4	3232	3.5461	0.0625	31.6600	2.4148
3.4448	5	4040	3.1592	0.125	44.0322	3.2246
2.9074	6	4848	2.8128	0.25	67.9403	4.1556
2.4207	7	5656	2.4849	0.5	113.5253	4.9888
2.1489	8.0	6464	2.2324	1.0	207.3360	5.9003
1.7566	9.0	7272	2.1711	1.0	204.4574	6.2117
1.4621	10.0	8080	2.2058	1.0	206.1582	6.3089
1.2369	11.0	8888	2.2919	1.0	205.0531	6.3776
0.995	12.0	9696	2.3953	1.0	204.7778	6.2355
0.8289	13.0	10504	2.5671	1.0	205.7961	6.1673

Framework versions

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