ddc309f82089d0ddb1b7ac20342bb26b

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

• Loss: 1.2294
• Data Size: 1.0
• Epoch Runtime: 12.8357
• Bleu: 12.6588

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	17.7144	0	2.3173	0.0290
No log	1	35	17.5179	0.0078	2.4144	0.0269
No log	2	70	16.6155	0.0156	2.1030	0.0310
No log	3	105	14.8986	0.0312	3.0620	0.0402
No log	4	140	13.8295	0.0625	3.4856	0.0462
No log	5	175	12.6466	0.125	4.0377	0.0540
No log	6	210	11.3404	0.25	4.8257	0.0478
No log	7	245	9.5996	0.5	6.8137	0.0323
2.4972	8.0	280	7.4873	1.0	11.7508	0.0232
10.4601	9.0	315	6.9812	1.0	11.1159	0.0205
8.0928	10.0	350	5.1085	1.0	12.2190	0.0134
8.0928	11.0	385	3.5469	1.0	9.9811	0.4195
5.5166	12.0	420	2.2664	1.0	10.3937	3.8922
3.466	13.0	455	1.7359	1.0	11.4021	6.2499
3.466	14.0	490	1.5843	1.0	10.7715	7.9447
2.713	15.0	525	1.4818	1.0	10.8165	8.7921
2.3289	16.0	560	1.4245	1.0	11.1957	9.8322
2.3289	17.0	595	1.3659	1.0	12.0886	10.6717
2.0457	18.0	630	1.3351	1.0	10.0443	11.4403
1.8524	19.0	665	1.3033	1.0	9.8577	12.0174
1.748	20.0	700	1.2855	1.0	9.8277	11.9246
1.748	21.0	735	1.2733	1.0	10.4312	12.0354
1.6095	22.0	770	1.2555	1.0	11.2081	12.3019
1.5372	23.0	805	1.2470	1.0	11.7935	12.1520
1.5372	24.0	840	1.2412	1.0	12.4973	12.0524
1.4536	25.0	875	1.2403	1.0	10.3195	11.8488
1.3963	26.0	910	1.2322	1.0	10.5973	11.8377
1.3963	27.0	945	1.2298	1.0	11.0937	12.1627
1.3322	28.0	980	1.2258	1.0	11.9552	12.3770
1.2762	29.0	1015	1.2218	1.0	12.0660	12.2784
1.2329	30.0	1050	1.2256	1.0	12.6226	12.5182
1.2329	31.0	1085	1.2199	1.0	12.5873	12.4929
1.1652	32.0	1120	1.2150	1.0	10.1448	12.5183
1.1327	33.0	1155	1.2212	1.0	11.4643	12.6969
1.1327	34.0	1190	1.2190	1.0	11.8735	12.6848
1.0749	35.0	1225	1.2171	1.0	12.4821	12.9429
1.0425	36.0	1260	1.2294	1.0	12.8357	12.6588

Framework versions

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