d431bd02a55bbd85866f1939f367ddca

This model is a fine-tuned version of google-t5/t5-small on the Helsinki-NLP/opus_books dataset. It achieves the following results on the evaluation set:

• Loss: 2.3415
• Data Size: 1.0
• Epoch Runtime: 6.6882
• Bleu: 3.0177

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	4.6221	0	1.1473	0.1100
No log	1	31	4.6064	0.0078	1.4740	0.1104
No log	2	62	4.5703	0.0156	1.4328	0.1134
No log	3	93	4.5005	0.0312	1.4964	0.1159
No log	4	124	4.4422	0.0625	1.5981	0.0962
No log	5	155	4.2707	0.125	2.1272	0.1219
No log	6	186	4.0733	0.25	2.6881	0.1684
0.7013	7	217	3.8242	0.5	3.7759	0.2088
0.7013	8.0	248	3.5560	1.0	6.4531	0.1758
2.8141	9.0	279	3.3816	1.0	6.2527	0.2138
3.74	10.0	310	3.2521	1.0	6.9367	0.2372
3.74	11.0	341	3.1519	1.0	6.7399	0.9490
3.5558	12.0	372	3.0793	1.0	6.1199	1.0051
3.4043	13.0	403	3.0140	1.0	6.2391	1.2950
3.4043	14.0	434	2.9645	1.0	7.1239	1.2945
3.3021	15.0	465	2.9170	1.0	6.6573	1.4390
3.3021	16.0	496	2.8817	1.0	6.3254	1.5021
3.2109	17.0	527	2.8456	1.0	6.2254	1.5389
3.1266	18.0	558	2.8140	1.0	7.1893	1.6168
3.1266	19.0	589	2.7836	1.0	6.9004	1.7308
3.0775	20.0	620	2.7528	1.0	6.6448	1.7620
3.0026	21.0	651	2.7238	1.0	6.4260	1.8896
3.0026	22.0	682	2.7012	1.0	5.1268	2.0290
2.9563	23.0	713	2.6767	1.0	5.4232	2.1725
2.9563	24.0	744	2.6560	1.0	5.7317	2.2300
2.891	25.0	775	2.6368	1.0	5.3370	2.2402
2.8587	26.0	806	2.6135	1.0	5.3657	2.2324
2.8587	27.0	837	2.5934	1.0	5.2370	2.3208
2.8015	28.0	868	2.5734	1.0	5.3508	2.3066
2.8015	29.0	899	2.5603	1.0	6.2737	2.4315
2.7561	30.0	930	2.5431	1.0	6.4207	2.4338
2.7188	31.0	961	2.5317	1.0	6.1300	2.4899
2.7188	32.0	992	2.5151	1.0	5.5575	2.5777
2.6827	33.0	1023	2.5060	1.0	5.7487	2.5721
2.6416	34.0	1054	2.4929	1.0	5.5721	2.6315
2.6416	35.0	1085	2.4814	1.0	5.7915	2.6630
2.6138	36.0	1116	2.4674	1.0	6.1883	2.7386
2.6138	37.0	1147	2.4581	1.0	5.3096	2.6823
2.5766	38.0	1178	2.4477	1.0	5.3723	2.7470
2.5413	39.0	1209	2.4335	1.0	5.3362	2.7325
2.5413	40.0	1240	2.4254	1.0	6.3199	2.8268
2.5103	41.0	1271	2.4123	1.0	6.5643	2.8315
2.4832	42.0	1302	2.4081	1.0	6.0125	2.9006
2.4832	43.0	1333	2.3951	1.0	6.0619	2.9568
2.4585	44.0	1364	2.3889	1.0	5.9724	3.0480
2.4585	45.0	1395	2.3860	1.0	6.7165	3.0197
2.4241	46.0	1426	2.3754	1.0	6.4182	3.0169
2.3965	47.0	1457	2.3647	1.0	6.5957	3.0111
2.3965	48.0	1488	2.3574	1.0	5.8117	2.9843
2.3654	49.0	1519	2.3501	1.0	5.4900	3.0144
2.3593	50.0	1550	2.3415	1.0	6.6882	3.0177

Framework versions

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