f3c96362ac0d54ef92daf623dcd555ef

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

• Loss: 2.0634
• Data Size: 1.0
• Epoch Runtime: 33.3974
• Bleu: 0.8831

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	3.0878	0	2.8774	0.1701
No log	1	70	3.0142	0.0078	3.3484	0.1612
No log	2	140	2.8223	0.0156	5.5043	0.2565
No log	3	210	2.6577	0.0312	7.0274	0.3782
No log	4	280	2.5478	0.0625	9.6439	0.4202
No log	5	350	2.4582	0.125	12.0514	0.6013
No log	6	420	2.3558	0.25	16.0349	0.3410
0.3948	7	490	2.2673	0.5	22.0499	0.3497
2.3506	8.0	560	2.1732	1.0	38.9860	0.4199
2.2634	9.0	630	2.1298	1.0	34.5599	0.4587
2.1392	10.0	700	2.0938	1.0	39.4260	0.5695
2.0813	11.0	770	2.0692	1.0	36.4187	0.6320
2.019	12.0	840	2.0553	1.0	35.7192	0.6972
1.9366	13.0	910	2.0344	1.0	35.0084	0.7801
1.8906	14.0	980	2.0345	1.0	36.6243	0.8230
1.8295	15.0	1050	2.0268	1.0	35.0934	0.7277
1.7829	16.0	1120	2.0329	1.0	37.0892	0.8031
1.7519	17.0	1190	2.0293	1.0	36.6521	0.8046
1.6902	18.0	1260	2.0258	1.0	36.4588	0.8535
1.6587	19.0	1330	2.0445	1.0	35.4485	0.8125
1.6054	20.0	1400	2.0402	1.0	36.2016	0.8714
1.5666	21.0	1470	2.0473	1.0	36.4630	0.8767
1.556	22.0	1540	2.0634	1.0	33.3974	0.8831

Framework versions

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