0ad8c2ecf6815815030b4e207bb2af1b

This model is a fine-tuned version of google/umt5-base on the Helsinki-NLP/opus_books [fr-pt] dataset. It achieves the following results on the evaluation set:

• Loss: 2.3568
• Data Size: 1.0
• Epoch Runtime: 12.8602
• Bleu: 7.8708

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	11.7090	0	1.3744	0.1954
No log	1	31	11.8378	0.0078	1.5109	0.1584
No log	2	62	11.5957	0.0156	2.1794	0.2111
No log	3	93	11.5673	0.0312	2.8158	0.1973
No log	4	124	11.5255	0.0625	3.8381	0.1563
No log	5	155	11.1080	0.125	4.1333	0.1654
No log	6	186	10.3154	0.25	5.2091	0.2149
2.5995	7	217	9.4188	0.5	7.4415	0.1732
2.5995	8.0	248	8.2699	1.0	11.7606	0.2447
9.2674	9.0	279	7.8717	1.0	11.9712	0.3183
10.3468	10.0	310	6.3612	1.0	12.8705	0.4939
10.3468	11.0	341	5.1569	1.0	9.7600	1.3232
7.6294	12.0	372	4.0364	1.0	10.9309	3.9073
5.7404	13.0	403	3.5976	1.0	10.8226	6.0704
5.7404	14.0	434	3.2598	1.0	11.1467	7.2694
4.7752	15.0	465	2.9971	1.0	12.2766	4.2953
4.7752	16.0	496	2.8488	1.0	12.6613	4.7700
4.213	17.0	527	2.7522	1.0	13.0987	5.1268
3.7812	18.0	558	2.6686	1.0	10.4024	5.3641
3.7812	19.0	589	2.6141	1.0	10.4839	5.6942
3.4767	20.0	620	2.5515	1.0	10.5228	6.3556
3.2547	21.0	651	2.5141	1.0	11.6514	6.3945
3.2547	22.0	682	2.4885	1.0	12.0017	6.6382
3.0838	23.0	713	2.4586	1.0	11.9956	6.8956
3.0838	24.0	744	2.4497	1.0	12.6230	6.8351
2.8882	25.0	775	2.4209	1.0	10.2937	6.8204
2.7929	26.0	806	2.4110	1.0	10.1908	6.9615
2.7929	27.0	837	2.4048	1.0	10.9968	7.0479
2.6492	28.0	868	2.3790	1.0	11.5217	7.0460
2.6492	29.0	899	2.3759	1.0	11.4738	7.1835
2.5644	30.0	930	2.3645	1.0	12.3540	7.2969
2.481	31.0	961	2.3603	1.0	12.4761	7.3014
2.481	32.0	992	2.3567	1.0	9.9603	7.4073
2.3755	33.0	1023	2.3446	1.0	10.7345	7.4959
2.2968	34.0	1054	2.3546	1.0	11.8768	7.4700
2.2968	35.0	1085	2.3649	1.0	12.0682	7.5651
2.2337	36.0	1116	2.3586	1.0	12.8577	7.7334
2.2337	37.0	1147	2.3568	1.0	12.8602	7.8708

Framework versions

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