cbee14b4723877d024c52e185dfdf57f

This model is a fine-tuned version of facebook/mbart-large-cc25 on the Helsinki-NLP/opus_books [fr-no] dataset. It achieves the following results on the evaluation set:

• Loss: 4.9984
• Data Size: 1.0
• Epoch Runtime: 25.4824
• Bleu: 3.8997

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	14.7153	0	2.7815	0.0201
No log	1	86	10.6747	0.0078	3.3457	0.1602
No log	2	172	9.4438	0.0156	3.9242	0.1664
No log	3	258	7.8104	0.0312	5.6486	0.1782
No log	4	344	7.0974	0.0625	7.1808	0.2943
0.4306	5	430	6.2872	0.125	9.2771	0.6206
1.5608	6	516	5.6884	0.25	12.1334	0.9372
1.9814	7	602	5.0092	0.5	15.7538	1.4266
2.724	8.0	688	4.2757	1.0	26.4025	1.9760
3.821	9.0	774	3.9494	1.0	25.9110	2.8209
3.1582	10.0	860	4.5657	1.0	25.4878	2.6718
10.5101	11.0	946	6.3540	1.0	26.3779	0.1214
5.535	12.0	1032	3.9837	1.0	27.0421	3.6079
2.4842	13.0	1118	3.8890	1.0	24.5501	3.9739
1.9461	14.0	1204	4.1200	1.0	25.8110	3.7675
1.5314	15.0	1290	4.4265	1.0	25.9130	3.8790
1.1194	16.0	1376	4.6593	1.0	24.7992	4.8279
0.8629	17.0	1462	4.9984	1.0	25.4824	3.8997

Framework versions

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