583291cd7fd4442d961732f1a812a378

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

• Loss: 3.1004
• Data Size: 1.0
• Epoch Runtime: 15.1206
• Bleu: 2.0107

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	22.8551	0	1.7380	0.0007
No log	1	75	22.6731	0.0078	2.3068	0.0007
No log	2	150	21.9580	0.0156	2.2608	0.0008
No log	3	225	20.9401	0.0312	2.9772	0.0021
No log	4	300	19.1929	0.0625	3.4257	0.0033
No log	5	375	16.7386	0.125	4.2602	0.0008
No log	6	450	13.3689	0.25	5.6277	0.0009
No log	7	525	9.1680	0.5	8.6906	0.0022
8.4373	8.0	600	5.8878	1.0	14.0675	0.0093
6.8746	9.0	675	4.4173	1.0	13.9376	0.0150
5.3129	10.0	750	3.9211	1.0	12.5278	0.0898
5.0202	11.0	825	3.7685	1.0	13.1778	0.3491
4.7453	12.0	900	3.6834	1.0	13.8237	0.4400
4.6122	13.0	975	3.6053	1.0	13.7579	0.5845
4.4648	14.0	1050	3.5586	1.0	13.9024	0.6611
4.3538	15.0	1125	3.5184	1.0	13.9139	0.6382
4.272	16.0	1200	3.4743	1.0	14.0328	0.6863
4.2215	17.0	1275	3.4448	1.0	12.8123	0.7799
4.1271	18.0	1350	3.4189	1.0	12.9035	0.7764
4.082	19.0	1425	3.3954	1.0	13.0466	0.9124
4.0451	20.0	1500	3.3718	1.0	13.5896	0.9273
3.9919	21.0	1575	3.3487	1.0	12.9714	1.0240
3.9567	22.0	1650	3.3299	1.0	13.4183	1.0280
3.9051	23.0	1725	3.3229	1.0	13.9033	1.1077
3.8432	24.0	1800	3.3014	1.0	14.9776	1.1978
3.8327	25.0	1875	3.2812	1.0	13.2021	1.2136
3.7931	26.0	1950	3.2737	1.0	13.1801	1.1906
3.7656	27.0	2025	3.2638	1.0	13.2863	1.2855
3.7105	28.0	2100	3.2472	1.0	13.0930	1.3691
3.6817	29.0	2175	3.2329	1.0	13.7874	1.4691
3.6668	30.0	2250	3.2167	1.0	13.7537	1.4748
3.6368	31.0	2325	3.2099	1.0	13.5364	1.4884
3.6085	32.0	2400	3.2005	1.0	13.9050	1.5734
3.5754	33.0	2475	3.1924	1.0	12.7887	1.5612
3.5625	34.0	2550	3.1851	1.0	13.1845	1.6516
3.5247	35.0	2625	3.1741	1.0	13.2080	1.6379
3.5134	36.0	2700	3.1681	1.0	13.9823	1.6851
3.4712	37.0	2775	3.1585	1.0	14.0141	1.7116
3.4621	38.0	2850	3.1557	1.0	14.6896	1.7655
3.448	39.0	2925	3.1470	1.0	14.9696	1.7528
3.4204	40.0	3000	3.1440	1.0	15.3110	1.8229
3.373	41.0	3075	3.1327	1.0	15.0462	1.8151
3.3627	42.0	3150	3.1309	1.0	12.8601	1.8360
3.3506	43.0	3225	3.1262	1.0	12.9121	1.9016
3.3265	44.0	3300	3.1233	1.0	12.9816	1.8646
3.3166	45.0	3375	3.1166	1.0	12.8738	1.8456
3.2904	46.0	3450	3.1127	1.0	12.9575	1.9269
3.269	47.0	3525	3.1092	1.0	12.9906	1.9405
3.2642	48.0	3600	3.1083	1.0	13.8594	1.9314
3.2218	49.0	3675	3.1062	1.0	14.3812	1.9239
3.2166	50.0	3750	3.1004	1.0	15.1206	2.0107

Framework versions

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