e1df5231ce4cfd3c4bbce30ac3720722

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

• Loss: 2.9626
• Data Size: 1.0
• Epoch Runtime: 97.8937
• Bleu: 6.5780

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	6.7355	0	8.3919	0.3798
No log	1	367	5.6509	0.0078	9.2553	0.6305
No log	2	734	5.0569	0.0156	10.7366	0.9096
No log	3	1101	4.4190	0.0312	13.3452	1.4115
No log	4	1468	3.9710	0.0625	17.2455	2.0960
0.2209	5	1835	3.5723	0.125	23.2385	3.0114
3.4098	6	2202	3.2060	0.25	34.8854	4.0343
2.8946	7	2569	2.8628	0.5	54.4539	4.8588
2.4327	8.0	2936	2.5828	1.0	99.0636	5.7343
2.024	9.0	3303	2.5179	1.0	97.3995	6.1899
1.6556	10.0	3670	2.5736	1.0	98.1031	6.4199
1.3533	11.0	4037	2.6517	1.0	96.5354	6.2619
1.1312	12.0	4404	2.7988	1.0	96.7894	6.3813
0.9315	13.0	4771	2.9626	1.0	97.8937	6.5780

Framework versions

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