c7604cf43f2ead5903e79dc76fa69bb7

This model is a fine-tuned version of google/mt5-large on the Helsinki-NLP/opus_books [fi-no] dataset. It achieves the following results on the evaluation set:

• Loss: 2.1777
• Data Size: 1.0
• Epoch Runtime: 39.5704
• Bleu: 6.2517

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	21.3467	0	3.5736	0.0079
No log	1	85	22.9127	0.0078	4.4974	0.0082
No log	2	170	24.4305	0.0156	5.5673	0.0044
No log	3	255	23.3816	0.0312	8.5154	0.0039
No log	4	340	22.7459	0.0625	11.5559	0.0040
1.5516	5	425	21.5469	0.125	13.8482	0.0033
1.5516	6	510	20.2011	0.25	17.5870	0.0043
6.0479	7	595	11.4255	0.5	26.3378	0.0045
9.8735	8.0	680	3.5418	1.0	42.3483	0.1807
3.6569	9.0	765	2.5037	1.0	39.9547	4.0870
2.9705	10.0	850	2.3234	1.0	38.7879	4.5832
2.7775	11.0	935	2.2432	1.0	39.1521	4.9747
2.5337	12.0	1020	2.1994	1.0	40.3694	5.1626
2.4171	13.0	1105	2.1733	1.0	39.9761	5.4123
2.2872	14.0	1190	2.1624	1.0	41.8520	5.4914
2.1794	15.0	1275	2.1449	1.0	39.4007	5.7806
2.0549	16.0	1360	2.1618	1.0	40.6027	5.9617
1.9814	17.0	1445	2.1540	1.0	39.2313	5.9820
1.8869	18.0	1530	2.1517	1.0	40.7726	6.0788
1.8155	19.0	1615	2.1777	1.0	39.5704	6.2517

Framework versions

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