73e890e2245d36eb87096cf4ddb9e7a4

This model is a fine-tuned version of google-t5/t5-large on the Helsinki-NLP/opus_books [fr-no] dataset. It achieves the following results on the evaluation set:

• Loss: 1.9787
• Data Size: 1.0
• Epoch Runtime: 40.3407
• Bleu: 3.0800

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	3.4607	0	3.3462	0.2870
No log	1	86	3.3678	0.0078	3.7920	0.2582
No log	2	172	3.1563	0.0156	5.3835	0.3723
No log	3	258	3.0253	0.0312	6.9070	0.5199
No log	4	344	2.8886	0.0625	8.9859	0.7965
0.1674	5	430	2.7530	0.125	12.7754	0.8146
0.7258	6	516	2.6189	0.25	16.8867	1.0797
0.9771	7	602	2.4721	0.5	26.5765	1.1704
1.4666	8.0	688	2.3142	1.0	42.6459	1.5458
2.3382	9.0	774	2.2111	1.0	40.5502	1.8273
2.215	10.0	860	2.1461	1.0	41.5116	2.0062
2.1348	11.0	946	2.0952	1.0	40.8749	2.2755
2.0025	12.0	1032	2.0570	1.0	39.4688	2.3422
1.9285	13.0	1118	2.0222	1.0	40.7465	2.4164
1.8471	14.0	1204	1.9951	1.0	40.5677	2.5546
1.7777	15.0	1290	1.9772	1.0	45.3299	2.7309
1.7093	16.0	1376	1.9671	1.0	42.0990	2.9486
1.6552	17.0	1462	1.9636	1.0	42.9682	2.8516
1.5849	18.0	1548	1.9511	1.0	41.7736	2.9707
1.5264	19.0	1634	1.9608	1.0	41.9025	2.9817
1.476	20.0	1720	1.9507	1.0	46.6141	2.9891
1.4362	21.0	1806	1.9568	1.0	41.0776	3.1204
1.3905	22.0	1892	1.9743	1.0	42.8900	3.0876
1.3344	23.0	1978	1.9803	1.0	41.8377	3.1262
1.2953	24.0	2064	1.9787	1.0	40.3407	3.0800

Framework versions

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