cc75383112b7cfbe11e9c93cac065f84

This model is a fine-tuned version of google-t5/t5-3b on the Helsinki-NLP/opus_books [fi-fr] dataset. It achieves the following results on the evaluation set:

• Loss: 2.1304
• Data Size: 1.0
• Epoch Runtime: 52.5337
• Bleu: 5.5717

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	2.8900	0	3.5486	0.9393
No log	1	88	2.7646	0.0078	4.2820	1.1895
No log	2	176	2.5943	0.0156	8.0258	1.7195
No log	3	264	2.4788	0.0312	14.0628	2.0155
No log	4	352	2.3867	0.0625	19.8327	3.1790
No log	5	440	2.3110	0.125	24.9854	4.8331
0.1874	6	528	2.2165	0.25	28.5716	5.4614
0.7813	7	616	2.1371	0.5	35.0426	5.2105
2.0822	8.0	704	2.0641	1.0	57.5174	5.1753
1.8839	9.0	792	2.0351	1.0	51.0989	5.3004
1.6744	10.0	880	2.0121	1.0	55.9947	5.6737
1.5244	11.0	968	2.0323	1.0	56.1812	5.4096
1.3833	12.0	1056	2.0502	1.0	54.2524	5.6860
1.2452	13.0	1144	2.0869	1.0	56.5898	5.5314
1.112	14.0	1232	2.1304	1.0	52.5337	5.5717

Framework versions

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