f9f41a0f89b33fb2c07d205e1e26813b

This model is a fine-tuned version of google/mt5-small on the Helsinki-NLP/opus_books [en-no] dataset. It achieves the following results on the evaluation set:

• Loss: 2.5256
• Data Size: 1.0
• Epoch Runtime: 14.9447
• Bleu: 5.4189

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	24.7306	0	1.7708	0.0009
No log	1	87	24.4981	0.0078	2.8416	0.0009
No log	2	174	24.0173	0.0156	2.7828	0.0009
No log	3	261	23.9039	0.0312	2.9664	0.0015
No log	4	348	22.2791	0.0625	3.4660	0.0016
1.058	5	435	19.6922	0.125	4.2597	0.0010
5.3662	6	522	14.8597	0.25	5.8722	0.0024
5.83	7	609	9.6132	0.5	8.9930	0.0037
6.8261	8.0	696	5.2840	1.0	14.9620	0.0325
6.6149	9.0	783	3.7924	1.0	13.9719	0.4429
4.9574	10.0	870	3.3892	1.0	14.6994	1.3554
4.4405	11.0	957	3.2125	1.0	14.9218	1.9539
4.2746	12.0	1044	3.0943	1.0	15.5061	2.3989
4.0628	13.0	1131	3.0252	1.0	15.5147	2.6657
3.8933	14.0	1218	2.9535	1.0	16.1197	3.0631
3.7342	15.0	1305	2.9128	1.0	16.2269	3.1671
3.6903	16.0	1392	2.8682	1.0	14.4820	3.5035
3.5971	17.0	1479	2.8382	1.0	14.5036	3.6614
3.5275	18.0	1566	2.8120	1.0	15.0770	3.7921
3.4727	19.0	1653	2.7887	1.0	14.3960	3.8094
3.3867	20.0	1740	2.7564	1.0	14.4340	3.8698
3.3212	21.0	1827	2.7461	1.0	14.4230	4.0295
3.3134	22.0	1914	2.7215	1.0	14.2917	4.1399
3.2497	23.0	2001	2.7105	1.0	14.0975	4.1338
3.1809	24.0	2088	2.6950	1.0	13.9646	4.2330
3.1636	25.0	2175	2.6752	1.0	14.3319	4.2981
3.1244	26.0	2262	2.6631	1.0	15.2468	4.4268
3.1143	27.0	2349	2.6583	1.0	15.7072	4.4021
3.0509	28.0	2436	2.6432	1.0	15.8659	4.4354
3.0285	29.0	2523	2.6296	1.0	15.9876	4.6023
3.0068	30.0	2610	2.6226	1.0	16.1165	4.6507
2.9563	31.0	2697	2.6140	1.0	13.7827	4.6323
2.9334	32.0	2784	2.6070	1.0	14.3865	4.6868
2.88	33.0	2871	2.5981	1.0	14.6992	4.8183
2.8755	34.0	2958	2.5974	1.0	14.4411	4.8294
2.8746	35.0	3045	2.5883	1.0	15.1472	4.8546
2.8177	36.0	3132	2.5754	1.0	14.4105	4.9302
2.7918	37.0	3219	2.5783	1.0	14.4150	4.9763
2.7873	38.0	3306	2.5682	1.0	14.3907	5.0512
2.7449	39.0	3393	2.5583	1.0	13.8708	5.0262
2.7539	40.0	3480	2.5549	1.0	14.0173	5.0774
2.7303	41.0	3567	2.5488	1.0	13.9626	5.1274
2.6798	42.0	3654	2.5520	1.0	14.4316	5.1657
2.6767	43.0	3741	2.5434	1.0	14.1089	5.1690
2.6503	44.0	3828	2.5438	1.0	14.0546	5.1647
2.6252	45.0	3915	2.5322	1.0	13.8464	5.2778
2.5873	46.0	4002	2.5453	1.0	13.9957	5.2616
2.5381	47.0	4089	2.5279	1.0	14.3024	5.2626
2.5765	48.0	4176	2.5283	1.0	13.9604	5.3135
2.5594	49.0	4263	2.5234	1.0	14.3919	5.3666
2.5347	50.0	4350	2.5256	1.0	14.9447	5.4189

Framework versions

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