91cd397d79fc9dd01b7b158224820016

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

• Loss: 3.2187
• Data Size: 1.0
• Epoch Runtime: 13.8113
• Bleu: 1.6158

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	27.2746	0	1.8161	0.0015
No log	1	83	27.1766	0.0078	1.8968	0.0014
No log	2	166	27.2891	0.0156	2.2723	0.0008
No log	3	249	26.8685	0.0312	2.9650	0.0008
0.7951	4	332	24.2712	0.0625	3.3803	0.0015
0.7951	5	415	20.1067	0.125	4.1026	0.0021
0.7951	6	498	16.3462	0.25	5.2247	0.0021
3.5672	7	581	12.2835	0.5	8.0506	0.0034
11.3735	8.0	664	7.0681	1.0	13.7821	0.0060
8.6391	9.0	747	5.0629	1.0	12.7269	0.0142
6.0938	10.0	830	4.2062	1.0	13.8398	0.0641
5.2694	11.0	913	3.9666	1.0	13.2515	0.2946
5.081	12.0	996	3.8471	1.0	13.4729	0.4175
4.8027	13.0	1079	3.7648	1.0	13.2391	0.5484
4.6622	14.0	1162	3.7027	1.0	13.1385	0.5798
4.5894	15.0	1245	3.6644	1.0	12.9530	0.6406
4.4537	16.0	1328	3.6289	1.0	13.8658	0.6754
4.3994	17.0	1411	3.5828	1.0	12.9598	0.7474
4.3299	18.0	1494	3.5628	1.0	13.2880	0.7859
4.2342	19.0	1577	3.5258	1.0	13.2278	0.8391
4.1848	20.0	1660	3.5074	1.0	13.1333	0.8469
4.1149	21.0	1743	3.4861	1.0	13.8052	0.8978
4.0798	22.0	1826	3.4677	1.0	14.3385	0.9568
4.0288	23.0	1909	3.4474	1.0	14.1256	1.0051
4.0162	24.0	1992	3.4280	1.0	12.8350	1.0470
3.96	25.0	2075	3.4157	1.0	13.0228	1.1022
3.9173	26.0	2158	3.3987	1.0	12.8699	1.1388
3.8811	27.0	2241	3.3852	1.0	13.3422	1.1356
3.8416	28.0	2324	3.3750	1.0	13.8469	1.1721
3.8165	29.0	2407	3.3596	1.0	14.1080	1.1948
3.7862	30.0	2490	3.3485	1.0	14.3882	1.2216
3.7645	31.0	2573	3.3389	1.0	14.6558	1.3029
3.7457	32.0	2656	3.3284	1.0	13.4624	1.3212
3.6619	33.0	2739	3.3179	1.0	13.9517	1.3264
3.6506	34.0	2822	3.3056	1.0	13.8116	1.3758
3.6494	35.0	2905	3.3010	1.0	13.9967	1.3931
3.6346	36.0	2988	3.2928	1.0	14.3535	1.3951
3.5929	37.0	3071	3.2905	1.0	14.0629	1.4419
3.5822	38.0	3154	3.2777	1.0	14.1572	1.4070
3.5353	39.0	3237	3.2721	1.0	14.3544	1.4220
3.5122	40.0	3320	3.2702	1.0	13.0288	1.4485
3.4808	41.0	3403	3.2638	1.0	13.3382	1.4391
3.4642	42.0	3486	3.2557	1.0	13.1424	1.4439
3.4377	43.0	3569	3.2478	1.0	13.2877	1.5041
3.4228	44.0	3652	3.2471	1.0	13.6846	1.5051
3.3832	45.0	3735	3.2394	1.0	14.1308	1.5161
3.3663	46.0	3818	3.2337	1.0	14.2943	1.5433
3.3804	47.0	3901	3.2342	1.0	14.6711	1.5757
3.3225	48.0	3984	3.2254	1.0	12.9744	1.5517
3.3143	49.0	4067	3.2257	1.0	13.9205	1.5577
3.3178	50.0	4150	3.2187	1.0	13.8113	1.6158

Framework versions

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