43f238e5bc548f3fc7b03ee9ceeb50a9

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

• Loss: 1.6133
• Data Size: 1.0
• Epoch Runtime: 828.7915
• Bleu: 14.8256

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	8.8008	0	67.7606	0.2274
No log	1	3177	2.5848	0.0078	74.0987	6.6989
0.047	2	6354	2.2470	0.0156	81.3440	7.8151
2.1442	3	9531	1.9308	0.0312	92.9631	9.3910
1.9569	4	12708	1.7963	0.0625	117.9589	9.4802
1.7647	5	15885	1.6629	0.125	165.7951	14.1034
1.6266	6	19062	1.5573	0.25	257.0280	14.4013
1.4565	7	22239	1.4512	0.5	444.5387	11.6849
1.2971	8.0	25416	1.3477	1.0	820.7993	16.0099
1.0941	9.0	28593	1.3408	1.0	822.7588	28.2093
0.9725	10.0	31770	1.3666	1.0	824.6350	15.3033
0.8212	11.0	34947	1.4266	1.0	819.1045	18.0941
0.6712	12.0	38124	1.5069	1.0	817.2837	17.6470
0.5536	13.0	41301	1.6133	1.0	828.7915	14.8256

Framework versions

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