a4965a7391a742ddf48e5c43e9bc5fc7

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

• Loss: 1.8703
• Data Size: 1.0
• Epoch Runtime: 43.0252
• Bleu: 1.3379

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	2.7522	0	2.6413	0.2646
No log	1	70	2.6963	0.0078	3.5144	0.1729
No log	2	140	2.3937	0.0156	8.3206	0.4878
No log	3	210	2.2819	0.0312	14.2673	0.6709
No log	4	280	2.2235	0.0625	19.9946	1.0777
No log	5	350	2.1060	0.125	25.6098	1.1408
No log	6	420	2.0098	0.25	24.5880	0.9596
0.3341	7	490	1.9177	0.5	30.0931	1.0734
1.913	8.0	560	1.8336	1.0	48.5925	1.2052
1.7698	9.0	630	1.8165	1.0	40.1015	1.1431
1.5953	10.0	700	1.7876	1.0	42.6063	1.1467
1.4939	11.0	770	1.7920	1.0	43.4989	1.2180
1.4147	12.0	840	1.8323	1.0	47.7017	1.2892
1.2901	13.0	910	1.8472	1.0	40.3420	1.3578
1.2123	14.0	980	1.8703	1.0	43.0252	1.3379

Framework versions

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