8a00dae926d79c92ddbd0aa2648dc45b

This model is a fine-tuned version of google/mt5-large on the Helsinki-NLP/opus_books [fi-pl] dataset. It achieves the following results on the evaluation set:

• Loss: 2.4164
• Data Size: 1.0
• Epoch Runtime: 34.9631
• Bleu: 2.7051

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	20.3236	0	3.0794	0.0117
No log	1	70	19.2670	0.0078	3.6533	0.0118
No log	2	140	14.5807	0.0156	6.3044	0.0118
No log	3	210	12.3757	0.0312	9.3386	0.0123
No log	4	280	10.6561	0.0625	11.8784	0.0112
No log	5	350	7.6517	0.125	13.8170	0.0103
No log	6	420	8.1701	0.25	18.0152	0.0093
2.1794	7	490	8.4589	0.5	22.9373	0.0130
6.9933	8.0	560	3.3705	1.0	39.1747	0.1134
4.0676	9.0	630	2.7627	1.0	34.0768	0.6739
3.2027	10.0	700	2.5811	1.0	34.5257	1.5314
2.9697	11.0	770	2.4990	1.0	34.3232	1.6353
2.9003	12.0	840	2.4588	1.0	36.6172	1.8702
2.6861	13.0	910	2.4243	1.0	34.8621	1.9516
2.5783	14.0	980	2.4168	1.0	36.5225	1.9847
2.4547	15.0	1050	2.4072	1.0	34.6742	2.0813
2.3639	16.0	1120	2.3923	1.0	36.1785	2.1539
2.327	17.0	1190	2.3997	1.0	33.6013	2.3461
2.21	18.0	1260	2.4097	1.0	35.0281	2.3455
2.1605	19.0	1330	2.4102	1.0	33.5976	2.4446
2.0708	20.0	1400	2.4164	1.0	34.9631	2.7051

Framework versions

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