58fceacaa752d5e3474d761054b94d40

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

• Loss: 2.0810
• Data Size: 1.0
• Epoch Runtime: 264.9675
• Bleu: 10.1440

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	10.7814	0	23.0063	0.1624
No log	1	1000	3.7832	0.0078	24.3671	2.8235
No log	2	2000	3.4683	0.0156	26.5974	3.6917
No log	3	3000	2.9326	0.0312	31.0007	4.4876
0.1208	4	4000	2.5241	0.0625	39.1177	5.4169
2.5342	5	5000	2.2655	0.125	54.3441	6.1478
0.1364	6	6000	2.0353	0.25	83.4401	7.2585
0.171	7	7000	1.8527	0.5	142.1893	8.4845
1.657	8.0	8000	1.7065	1.0	262.8612	16.5878
1.424	9.0	9000	1.6926	1.0	259.6856	13.0326
1.2043	10.0	10000	1.6815	1.0	261.0464	10.3316
1.0234	11.0	11000	1.7398	1.0	260.1502	10.6781
0.8361	12.0	12000	1.8461	1.0	260.9129	11.2870
0.6943	13.0	13000	1.9749	1.0	260.6425	11.4679
0.5505	14.0	14000	2.0810	1.0	264.9675	10.1440

Framework versions

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