132ae70869a32d91269ad4e768243915

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

• Loss: 2.3422
• Data Size: 1.0
• Epoch Runtime: 40.5573
• Bleu: 3.3316

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	22.0997	0	3.4174	0.0117
No log	1	83	22.3477	0.0078	3.8759	0.0082
No log	2	166	22.5143	0.0156	6.5727	0.0090
No log	3	249	21.6966	0.0312	8.7333	0.0082
0.7123	4	332	21.4725	0.0625	11.3150	0.0081
0.7123	5	415	17.5101	0.125	14.0681	0.0079
0.7123	6	498	14.3276	0.25	18.8458	0.0081
3.1732	7	581	12.1104	0.5	26.6615	0.0040
11.804	8.0	664	6.1639	1.0	43.8036	0.0232
7.7827	9.0	747	3.0087	1.0	40.9826	0.2163
3.6345	10.0	830	2.5683	1.0	40.8644	2.0186
3.0921	11.0	913	2.4533	1.0	41.8374	2.2874
2.9184	12.0	996	2.3901	1.0	43.5541	2.6282
2.6759	13.0	1079	2.3524	1.0	41.0899	2.8204
2.5621	14.0	1162	2.3239	1.0	41.4385	2.7636
2.4805	15.0	1245	2.3187	1.0	42.4047	2.9852
2.3343	16.0	1328	2.3088	1.0	41.4409	3.0188
2.2424	17.0	1411	2.3066	1.0	42.5654	3.0232
2.1763	18.0	1494	2.3048	1.0	40.6371	3.1532
2.0676	19.0	1577	2.3112	1.0	40.7142	3.2457
2.0232	20.0	1660	2.3241	1.0	41.5369	3.1587
1.9303	21.0	1743	2.3243	1.0	43.1921	3.3780
1.8476	22.0	1826	2.3422	1.0	40.5573	3.3316

Framework versions

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