b7537e0fe33c614891986f3d5c825c91

This model is a fine-tuned version of google/long-t5-local-large on the Helsinki-NLP/opus_books [fr-sv] dataset. It achieves the following results on the evaluation set:

• Loss: 3.2105
• Data Size: 1.0
• Epoch Runtime: 38.8575
• Bleu: 0.2691

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	219.7699	0	3.0637	0.0033
No log	1	75	206.1623	0.0078	3.4862	0.0033
No log	2	150	184.6765	0.0156	4.8569	0.0040
No log	3	225	158.5659	0.0312	7.0352	0.0042
No log	4	300	126.6633	0.0625	8.7852	0.0036
No log	5	375	78.1125	0.125	12.4607	0.0053
No log	6	450	34.4594	0.25	16.1818	0.0034
No log	7	525	15.1997	0.5	24.0986	0.0065
18.7003	8.0	600	11.0175	1.0	39.3052	0.0035
16.1992	9.0	675	9.5453	1.0	37.8482	0.0065
13.5206	10.0	750	8.3748	1.0	37.9171	0.0125
12.5957	11.0	825	7.6047	1.0	38.3496	0.0064
11.1559	12.0	900	7.1902	1.0	38.1695	0.0136
10.5851	13.0	975	7.0050	1.0	37.7150	0.0138
9.7612	14.0	1050	6.4365	1.0	38.2679	0.0163
9.2522	15.0	1125	6.2579	1.0	37.6155	0.0262
8.6555	16.0	1200	5.8325	1.0	37.5047	0.0388
8.3139	17.0	1275	5.6114	1.0	37.9400	0.0429
7.8543	18.0	1350	5.4419	1.0	37.7108	0.0428
7.6153	19.0	1425	5.1149	1.0	37.4297	0.0597
7.2648	20.0	1500	5.0269	1.0	37.1780	0.0605
7.1111	21.0	1575	4.9814	1.0	37.1947	0.0376
6.8276	22.0	1650	5.0170	1.0	37.3842	0.0462
6.6184	23.0	1725	4.6441	1.0	36.7989	0.1019
6.3262	24.0	1800	4.5687	1.0	37.3828	0.0698
6.234	25.0	1875	4.4098	1.0	37.0953	0.0428
5.9817	26.0	1950	4.3941	1.0	37.4955	0.0887
5.8618	27.0	2025	4.1422	1.0	37.4159	0.0867
5.6413	28.0	2100	4.0678	1.0	37.4700	0.1030
5.5014	29.0	2175	4.0464	1.0	37.5988	0.0937
5.3375	30.0	2250	3.9598	1.0	37.3789	0.1076
5.2614	31.0	2325	4.0291	1.0	37.4278	0.0897
5.1003	32.0	2400	3.9221	1.0	37.2692	0.1297
4.985	33.0	2475	3.7562	1.0	37.7929	0.1092
4.853	34.0	2550	3.8291	1.0	37.4791	0.0810
4.7849	35.0	2625	3.6264	1.0	37.1947	0.1982
4.6432	36.0	2700	3.6636	1.0	37.4072	0.1725
4.5749	37.0	2775	3.6744	1.0	37.4433	0.2292
4.4215	38.0	2850	3.5357	1.0	37.0327	0.2356
4.3749	39.0	2925	3.3974	1.0	38.0901	0.2529
4.2708	40.0	3000	3.5127	1.0	37.3004	0.2200
4.5798	41.0	3075	3.3963	1.0	36.7234	0.1987
4.1807	42.0	3150	3.3445	1.0	38.2956	0.2380
4.1043	43.0	3225	3.3404	1.0	37.2292	0.2919
4.0446	44.0	3300	3.3719	1.0	37.0030	0.2774
3.9573	45.0	3375	3.2986	1.0	37.8242	0.2756
3.8895	46.0	3450	3.3025	1.0	36.8436	0.2323
3.8454	47.0	3525	3.2694	1.0	37.9007	0.2297
3.7597	48.0	3600	3.2325	1.0	37.4543	0.3330
3.7171	49.0	3675	3.2183	1.0	37.8201	0.2300
3.6733	50.0	3750	3.2105	1.0	38.8575	0.2691

Framework versions

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