d6b6ceb27d1777d3fe9645a25748b672

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

• Loss: 1.6071
• Data Size: 1.0
• Epoch Runtime: 28.5217
• Bleu: 11.7345

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	7.2550	0	1.6023	0.0123
No log	1	27	7.2385	0.0078	2.2670	0.0177
No log	2	54	7.0206	0.0156	9.2015	0.0102
No log	3	81	6.6734	0.0312	16.8072	0.0157
No log	4	108	6.0831	0.0625	23.7250	0.0228
No log	5	135	4.0899	0.125	27.9795	0.0824
No log	6	162	2.9214	0.25	33.1854	0.5155
No log	7	189	2.2261	0.5	29.7139	0.9866
0.9472	8.0	216	1.7074	1.0	33.8068	6.6796
0.9472	9.0	243	1.6046	1.0	37.2319	7.4012
1.9242	10.0	270	1.5585	1.0	40.7400	8.0748
1.9242	11.0	297	1.5488	1.0	25.4026	8.4004
1.4718	12.0	324	1.5618	1.0	28.5771	8.9630
1.1963	13.0	351	1.5645	1.0	31.5454	9.3696
1.1963	14.0	378	1.6022	1.0	36.4729	9.2487
0.9982	15.0	405	1.6071	1.0	28.5217	11.7345

Framework versions

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