a482abe5c0c3df2e55fe77af50223b30

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

• Loss: 2.4162
• Data Size: 1.0
• Epoch Runtime: 12.9575
• Bleu: 5.4267

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	24.1044	0	1.7023	0.0013
No log	1	77	24.1850	0.0078	2.8246	0.0021
No log	2	154	23.4822	0.0156	2.1373	0.0039
No log	3	231	22.1881	0.0312	2.8963	0.0034
No log	4	308	20.5519	0.0625	3.1263	0.0022
No log	5	385	17.3638	0.125	4.1881	0.0008
2.0437	6	462	12.6606	0.25	5.6267	0.0021
6.1418	7	539	8.2146	0.5	8.3646	0.0055
8.7939	8.0	616	5.0185	1.0	14.5166	0.0205
6.8281	9.0	693	3.9108	1.0	14.3469	0.0948
5.1822	10.0	770	3.4798	1.0	13.1189	0.5779
4.8265	11.0	847	3.2817	1.0	12.7046	1.4472
4.4252	12.0	924	3.1466	1.0	12.9419	1.8448
4.1807	13.0	1001	3.0419	1.0	13.3224	2.2019
4.0807	14.0	1078	2.9745	1.0	13.3528	2.4143
3.9248	15.0	1155	2.9171	1.0	13.2870	2.6509
3.8525	16.0	1232	2.8732	1.0	13.4281	2.7693
3.7494	17.0	1309	2.8366	1.0	12.5643	2.9671
3.6911	18.0	1386	2.7954	1.0	12.9707	3.1357
3.5872	19.0	1463	2.7712	1.0	13.8654	3.2620
3.5785	20.0	1540	2.7459	1.0	13.3650	3.4082
3.486	21.0	1617	2.7112	1.0	13.2648	3.5018
3.4522	22.0	1694	2.6877	1.0	13.5224	3.7139
3.395	23.0	1771	2.6623	1.0	13.4110	3.9639
3.3551	24.0	1848	2.6430	1.0	14.2624	4.0686
3.2847	25.0	1925	2.6286	1.0	12.5905	4.1673
3.2428	26.0	2002	2.6152	1.0	13.0139	4.2583
3.2186	27.0	2079	2.5961	1.0	12.8912	4.2933
3.1648	28.0	2156	2.5841	1.0	13.6841	4.3143
3.142	29.0	2233	2.5703	1.0	13.3015	4.3932
3.0906	30.0	2310	2.5622	1.0	13.5234	4.4793
3.0849	31.0	2387	2.5449	1.0	14.0385	4.5656
3.0196	32.0	2464	2.5371	1.0	15.2850	4.6850
3.0039	33.0	2541	2.5238	1.0	12.5780	4.6726
2.9926	34.0	2618	2.5162	1.0	12.7979	4.7165
2.9676	35.0	2695	2.5134	1.0	12.7531	4.7666
2.8917	36.0	2772	2.4989	1.0	13.0001	4.7960
2.9226	37.0	2849	2.4914	1.0	13.8380	4.8517
2.8577	38.0	2926	2.4861	1.0	13.9003	4.9313
2.8482	39.0	3003	2.4732	1.0	14.1276	5.0349
2.8188	40.0	3080	2.4690	1.0	13.7970	5.1179
2.7777	41.0	3157	2.4635	1.0	12.7962	5.0593
2.7564	42.0	3234	2.4576	1.0	13.0918	5.0874
2.7427	43.0	3311	2.4526	1.0	13.1732	5.1692
2.7413	44.0	3388	2.4511	1.0	13.7614	5.2781
2.7009	45.0	3465	2.4456	1.0	13.9087	5.2347
2.6772	46.0	3542	2.4371	1.0	13.8138	5.3230
2.6589	47.0	3619	2.4311	1.0	13.9322	5.3383
2.669	48.0	3696	2.4261	1.0	13.9409	5.4561
2.6068	49.0	3773	2.4189	1.0	14.3229	5.3914
2.6003	50.0	3850	2.4162	1.0	12.9575	5.4267

Framework versions

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