7132eac5f8a3d007b76b88db764976c7

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

• Loss: 2.0901
• Data Size: 1.0
• Epoch Runtime: 81.3317
• Bleu: 7.8459

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	15.2536	0	7.2985	0.0211
No log	1	390	15.6662	0.0078	8.9328	0.0211
No log	2	780	13.8073	0.0156	8.9606	0.0190
No log	3	1170	13.4585	0.0312	11.2179	0.0221
No log	4	1560	11.7099	0.0625	14.1549	0.0137
0.7843	5	1950	9.2046	0.125	20.1830	0.0089
1.4287	6	2340	4.9232	0.25	29.3691	0.0543
4.054	7	2730	2.9184	0.5	46.5163	2.9863
3.1519	8.0	3120	2.5208	1.0	85.5254	4.8923
2.9129	9.0	3510	2.4149	1.0	86.1225	5.4005
2.7849	10.0	3900	2.3566	1.0	83.9788	5.6694
2.6958	11.0	4290	2.3041	1.0	82.6600	5.9535
2.5942	12.0	4680	2.2601	1.0	83.0294	6.2272
2.5384	13.0	5070	2.2363	1.0	82.6532	6.3636
2.4667	14.0	5460	2.2089	1.0	82.2912	6.5239
2.3524	15.0	5850	2.1965	1.0	82.5349	6.6398
2.3166	16.0	6240	2.1625	1.0	81.8239	6.7495
2.2953	17.0	6630	2.1555	1.0	81.9901	6.8783
2.2332	18.0	7020	2.1440	1.0	83.2302	7.0288
2.1893	19.0	7410	2.1239	1.0	83.5916	7.1203
2.1559	20.0	7800	2.1173	1.0	81.5685	7.1722
2.1161	21.0	8190	2.1141	1.0	81.6963	7.3281
2.043	22.0	8580	2.1068	1.0	82.8431	7.4121
2.0294	23.0	8970	2.1009	1.0	81.9595	7.4750
1.982	24.0	9360	2.0951	1.0	80.4355	7.5181
1.9439	25.0	9750	2.0982	1.0	81.8319	7.5364
1.9049	26.0	10140	2.1002	1.0	82.3029	7.6096
1.8619	27.0	10530	2.0860	1.0	81.8496	7.6852
1.8604	28.0	10920	2.0850	1.0	81.8561	7.7166
1.8383	29.0	11310	2.0821	1.0	81.0283	7.7468
1.796	30.0	11700	2.0823	1.0	82.0984	7.7893
1.7357	31.0	12090	2.0848	1.0	81.7833	7.8345
1.707	32.0	12480	2.0905	1.0	82.0953	7.8734
1.7016	33.0	12870	2.0901	1.0	81.3317	7.8459

Framework versions

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