bc82c2bd7429994a0bb5b27adffa4f33

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

• Loss: 2.1278
• Data Size: 1.0
• Epoch Runtime: 21.6920
• Bleu: 4.2545

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	15.4150	0	2.5465	0.0111
No log	1	88	15.2606	0.0078	2.5777	0.0121
No log	2	176	14.7890	0.0156	3.5348	0.0133
No log	3	264	14.0644	0.0312	4.5839	0.0131
No log	4	352	13.9049	0.0625	6.1717	0.0133
No log	5	440	11.0301	0.125	7.8851	0.0149
1.081	6	528	8.9344	0.25	10.6653	0.0150
3.7493	7	616	5.9982	0.5	14.5224	0.0147
4.5302	8.0	704	2.8173	1.0	24.9038	1.2415
3.7076	9.0	792	2.4542	1.0	21.9678	1.8482
3.2077	10.0	880	2.3253	1.0	21.8970	2.0676
2.9692	11.0	968	2.2765	1.0	22.6091	2.3553
2.8193	12.0	1056	2.2384	1.0	21.4813	2.6851
2.7487	13.0	1144	2.2033	1.0	22.5017	2.9447
2.6309	14.0	1232	2.1828	1.0	21.9026	3.2756
2.5517	15.0	1320	2.1783	1.0	23.0023	3.2794
2.4967	16.0	1408	2.1661	1.0	20.9630	3.5091
2.4337	17.0	1496	2.1488	1.0	21.0490	3.5234
2.3646	18.0	1584	2.1418	1.0	21.8422	3.5040
2.353	19.0	1672	2.1434	1.0	21.4715	3.8377
2.2909	20.0	1760	2.1283	1.0	22.0806	4.0097
2.2462	21.0	1848	2.1233	1.0	21.5011	3.9880
2.2333	22.0	1936	2.1252	1.0	22.1982	4.1439
2.1529	23.0	2024	2.1174	1.0	22.3929	3.9746
2.1306	24.0	2112	2.1253	1.0	22.8676	3.9714
2.0824	25.0	2200	2.1269	1.0	20.9700	4.0298
2.0616	26.0	2288	2.1224	1.0	21.8319	4.0111
2.0048	27.0	2376	2.1101	1.0	22.4927	3.9453
1.9662	28.0	2464	2.1191	1.0	23.3229	4.1390
1.9527	29.0	2552	2.1229	1.0	20.8483	4.3526
1.9034	30.0	2640	2.1229	1.0	20.7968	4.3342
1.871	31.0	2728	2.1278	1.0	21.6920	4.2545

Framework versions

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