0b99f2c59efa339bc669ee1ad6404947

This model is a fine-tuned version of google/umt5-small on the Helsinki-NLP/opus_books [en-no] dataset. It achieves the following results on the evaluation set:

• Loss: 2.8015
• Data Size: 1.0
• Epoch Runtime: 16.1127
• Bleu: 6.5099

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	19.5040	0	1.8164	0.0551
No log	1	87	19.7877	0.0078	2.3806	0.0613
No log	2	174	19.4318	0.0156	2.5573	0.0522
No log	3	261	19.1174	0.0312	2.8387	0.0869
No log	4	348	18.1931	0.0625	3.3181	0.0857
0.8606	5	435	16.8773	0.125	4.0564	0.0827
4.411	6	522	13.8636	0.25	6.1705	0.1425
5.3619	7	609	9.9907	0.5	9.3401	0.1464
6.7979	8.0	696	6.9065	1.0	15.9693	0.2216
8.0446	9.0	783	5.1407	1.0	14.3013	0.7074
6.6046	10.0	870	4.5308	1.0	15.0072	2.4172
5.8458	11.0	957	4.2031	1.0	15.0180	2.5097
5.5477	12.0	1044	3.9590	1.0	15.3057	2.6652
5.1566	13.0	1131	3.7686	1.0	15.2963	3.0976
4.83	14.0	1218	3.5775	1.0	15.1575	3.5723
4.5322	15.0	1305	3.4101	1.0	15.9932	3.9312
4.445	16.0	1392	3.2813	1.0	14.4618	4.3110
4.2768	17.0	1479	3.2138	1.0	15.2863	4.5780
4.1393	18.0	1566	3.1576	1.0	15.1322	4.8100
4.0361	19.0	1653	3.1251	1.0	15.1464	4.9201
3.9382	20.0	1740	3.0968	1.0	15.1369	5.0012
3.84	21.0	1827	3.0731	1.0	15.4416	5.1079
3.8089	22.0	1914	3.0324	1.0	16.1804	5.2143
3.7134	23.0	2001	3.0245	1.0	14.2684	5.4818
3.6262	24.0	2088	2.9995	1.0	14.8129	5.4222
3.6048	25.0	2175	2.9825	1.0	14.7220	5.4213
3.5626	26.0	2262	2.9654	1.0	14.9187	5.5030
3.5273	27.0	2349	2.9595	1.0	15.1866	5.5954
3.4709	28.0	2436	2.9402	1.0	15.7237	5.6550
3.4198	29.0	2523	2.9172	1.0	15.7879	5.7443
3.4119	30.0	2610	2.9093	1.0	14.3025	5.7331
3.3313	31.0	2697	2.8949	1.0	14.2849	5.8333
3.3101	32.0	2784	2.8836	1.0	14.6021	5.9586
3.2747	33.0	2871	2.8879	1.0	14.7945	5.9238
3.2538	34.0	2958	2.8790	1.0	15.1808	6.0059
3.2457	35.0	3045	2.8797	1.0	14.7276	6.0394
3.1796	36.0	3132	2.8599	1.0	14.9706	6.0593
3.1589	37.0	3219	2.8587	1.0	15.5686	6.1553
3.1516	38.0	3306	2.8529	1.0	14.5410	6.1577
3.1018	39.0	3393	2.8384	1.0	14.8202	6.2180
3.089	40.0	3480	2.8378	1.0	15.0126	6.2073
3.0987	41.0	3567	2.8307	1.0	15.2370	6.2604
3.0259	42.0	3654	2.8352	1.0	15.0191	6.2415
3.0104	43.0	3741	2.8241	1.0	15.6302	6.2377
2.9799	44.0	3828	2.8255	1.0	15.0524	6.2596
2.9478	45.0	3915	2.8068	1.0	15.0966	6.2095
2.9045	46.0	4002	2.8273	1.0	14.4436	6.3158
2.8668	47.0	4089	2.8058	1.0	14.7387	6.3670
2.8903	48.0	4176	2.8119	1.0	15.5057	6.5110
2.8865	49.0	4263	2.7904	1.0	15.5379	6.5642
2.8611	50.0	4350	2.8015	1.0	16.1127	6.5099

Framework versions

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