0a6c3a66f7a7fcd9b06549f82220ee04

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

• Loss: 1.7937
• Data Size: 1.0
• Epoch Runtime: 279.1149
• Bleu: 6.6173

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	22.1300	0	21.0753	0.0120
No log	1	688	25.4380	0.0078	23.7346	0.0107
No log	2	1376	15.5069	0.0156	26.9697	0.0157
No log	3	2064	6.2432	0.0312	33.1115	0.0229
0.4503	4	2752	3.1230	0.0625	42.4554	0.2406
0.298	5	3440	2.4048	0.125	59.4517	0.5248
2.7599	6	4128	2.1351	0.25	91.0323	4.9698
2.485	7	4816	2.0095	0.5	156.4995	6.0790
2.2912	8.0	5504	1.9112	1.0	285.9636	5.7336
2.1686	9.0	6192	1.8658	1.0	281.1763	6.2217
2.0627	10.0	6880	1.8358	1.0	282.2311	6.2266
1.9688	11.0	7568	1.8129	1.0	280.5447	6.3950
1.8904	12.0	8256	1.8009	1.0	281.0382	6.7348
1.8266	13.0	8944	1.7919	1.0	282.6544	6.5020
1.789	14.0	9632	1.7805	1.0	281.1056	6.6004
1.7135	15.0	10320	1.7826	1.0	276.3046	6.5719
1.6412	16.0	11008	1.7733	1.0	275.8294	6.7821
1.6268	17.0	11696	1.7812	1.0	279.2836	6.6139
1.535	18.0	12384	1.7881	1.0	278.9885	6.6738
1.5292	19.0	13072	1.7836	1.0	279.3306	6.6685
1.4805	20.0	13760	1.7937	1.0	279.1149	6.6173

Framework versions

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