eed3b2fc273ffb52c6b1fa7ecf793821

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

• Loss: 1.9388
• Data Size: 1.0
• Epoch Runtime: 11.6263
• Bleu: 7.2219

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	16.1264	0	1.5804	0.0231
No log	1	31	16.4643	0.0078	2.3928	0.0284
No log	2	62	15.9303	0.0156	2.0920	0.0267
No log	3	93	15.5384	0.0312	3.3902	0.0274
No log	4	124	15.0819	0.0625	4.8687	0.0206
No log	5	155	13.6719	0.125	5.6603	0.0197
No log	6	186	11.9069	0.25	7.7089	0.0322
2.3003	7	217	9.3145	0.5	10.0812	0.0321
2.3003	8.0	248	6.4098	1.0	14.7705	0.0199
6.9365	9.0	279	4.7352	1.0	14.1538	0.0096
6.3421	10.0	310	2.9529	1.0	9.7062	1.0909
6.3421	11.0	341	2.4681	1.0	10.1137	3.6803
3.9354	12.0	372	2.2692	1.0	10.2667	3.9580
3.216	13.0	403	2.1776	1.0	10.9836	4.2497
3.216	14.0	434	2.1226	1.0	12.7806	4.6474
2.9141	15.0	465	2.0874	1.0	11.9026	5.1961
2.9141	16.0	496	2.0459	1.0	12.2544	5.6613
2.691	17.0	527	2.0090	1.0	13.2220	5.9954
2.529	18.0	558	2.0037	1.0	9.8746	6.2197
2.529	19.0	589	1.9852	1.0	10.3038	6.4326
2.3882	20.0	620	1.9751	1.0	10.4815	6.4168
2.2867	21.0	651	1.9600	1.0	10.6459	6.4688
2.2867	22.0	682	1.9624	1.0	11.3954	6.8694
2.2047	23.0	713	1.9534	1.0	12.0309	7.1474
2.2047	24.0	744	1.9442	1.0	13.0364	7.1646
2.0857	25.0	775	1.9308	1.0	13.3050	6.8281
2.0494	26.0	806	1.9343	1.0	9.8075	6.8110
2.0494	27.0	837	1.9369	1.0	10.5133	6.8163
1.9657	28.0	868	1.9248	1.0	11.3402	7.0828
1.9657	29.0	899	1.9337	1.0	11.1664	7.0742
1.9108	30.0	930	1.9344	1.0	11.0983	7.0631
1.8423	31.0	961	1.9270	1.0	11.6784	7.1148
1.8423	32.0	992	1.9388	1.0	11.6263	7.2219

Framework versions

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