33562f6b9fce67a8c253d38dc51d877c

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

• Loss: 2.5661
• Data Size: 1.0
• Epoch Runtime: 162.8726
• Bleu: 0.7953

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	227.9474	0	12.0214	0.0077
No log	1	367	183.7979	0.0078	13.2936	0.0065
No log	2	734	134.7391	0.0156	15.3314	0.0032
No log	3	1101	69.4337	0.0312	18.8234	0.0011
No log	4	1468	29.1306	0.0625	23.7774	0.0005
3.8408	5	1835	14.9830	0.125	33.6860	0.0155
20.8276	6	2202	11.2351	0.25	51.7824	0.1309
14.5113	7	2569	9.0706	0.5	89.2183	0.0839
10.3356	8.0	2936	6.5654	1.0	162.2343	0.0607
8.1005	9.0	3303	5.6250	1.0	160.6285	0.1078
6.9191	10.0	3670	4.6990	1.0	160.3902	0.1321
6.1299	11.0	4037	4.3165	1.0	159.7420	0.1422
5.4527	12.0	4404	3.9367	1.0	160.9828	0.1397
5.0146	13.0	4771	3.7508	1.0	160.3557	0.1540
4.7112	14.0	5138	3.5266	1.0	160.8869	0.2144
4.4185	15.0	5505	3.4124	1.0	160.9423	0.1896
4.2091	16.0	5872	3.3365	1.0	161.2574	0.2529
4.0543	17.0	6239	3.2665	1.0	161.5701	0.2257
3.908	18.0	6606	3.1801	1.0	160.9361	0.2362
3.7735	19.0	6973	3.2065	1.0	161.9550	0.1961
3.692	20.0	7340	3.0479	1.0	161.8171	0.3107
3.581	21.0	7707	3.0544	1.0	163.2240	0.2429
3.5182	22.0	8074	2.9941	1.0	163.6884	0.3816
3.435	23.0	8441	2.9701	1.0	163.1442	0.3846
3.3766	24.0	8808	2.9647	1.0	162.1650	0.2901
3.3338	25.0	9175	2.9095	1.0	161.4591	0.3830
3.2912	26.0	9542	2.9067	1.0	161.0684	0.3462
3.2124	27.0	9909	2.8573	1.0	162.8073	0.4296
3.1835	28.0	10276	2.8323	1.0	160.5233	0.3739
3.1464	29.0	10643	2.8341	1.0	161.7101	0.3739
3.0991	30.0	11010	2.8026	1.0	161.8965	0.5228
3.059	31.0	11377	2.7899	1.0	161.3309	0.4951
3.0268	32.0	11744	2.7772	1.0	162.3818	0.5396
2.9941	33.0	12111	2.7470	1.0	160.4514	0.6003
2.9486	34.0	12478	2.7180	1.0	160.8334	0.5596
2.9266	35.0	12845	2.7145	1.0	161.7326	0.6708
2.8847	36.0	13212	2.7008	1.0	162.2701	0.5416
2.8645	37.0	13579	2.6849	1.0	162.0892	0.6175
2.8496	38.0	13946	2.6890	1.0	162.7706	0.5850
2.81	39.0	14313	2.6759	1.0	162.3887	0.6315
2.7744	40.0	14680	2.6479	1.0	161.6776	0.6514
2.754	41.0	15047	2.6493	1.0	162.5414	0.6169
2.7357	42.0	15414	2.6309	1.0	163.1594	0.6945
2.686	43.0	15781	2.6149	1.0	165.9331	0.6736
2.6839	44.0	16148	2.6144	1.0	163.4045	0.6818
2.662	45.0	16515	2.6034	1.0	161.7051	0.6951
2.6385	46.0	16882	2.5922	1.0	162.0199	0.7377
2.6123	47.0	17249	2.5817	1.0	160.6540	0.6980
2.5896	48.0	17616	2.5717	1.0	161.6729	0.7286
2.5732	49.0	17983	2.5840	1.0	162.0004	0.7646
2.5499	50.0	18350	2.5661	1.0	162.8726	0.7953

Framework versions

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