balanced

This model is a fine-tuned version of samzirbo/mT5.en-es.pretrained on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.1748
• Bleu: 44.0446
• Meteor: 0.6916
• Chrf++: 62.8076

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: 0.0005
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 1000
• training_steps: 50000

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Meteor	Chrf++
4.3249	0.44	2500	2.0076	28.1155	0.5563	49.3185
2.4089	0.88	5000	1.7085	33.4675	0.6049	54.1053
2.1313	1.31	7500	1.5707	36.5102	0.6324	56.7478
1.9952	1.75	10000	1.4745	38.023	0.6421	57.9508
1.8897	2.19	12500	1.4185	39.0718	0.6512	58.9236
1.8052	2.63	15000	1.3728	40.11	0.6593	59.6738
1.7568	3.06	17500	1.3367	40.647	0.6648	60.0802
1.6834	3.5	20000	1.3046	41.6093	0.6713	60.8373
1.6584	3.94	22500	1.2721	41.9856	0.675	61.1498
1.5942	4.38	25000	1.2509	42.4053	0.6807	61.6486
1.5756	4.81	27500	1.2298	42.9706	0.6828	61.9567
1.5408	5.25	30000	1.2189	43.2046	0.6825	62.0617
1.5147	5.69	32500	1.2031	43.404	0.6859	62.2848
1.4953	6.13	35000	1.1941	43.6281	0.6881	62.4345
1.4689	6.56	37500	1.1847	43.8479	0.6889	62.5928
1.4666	7.0	40000	1.1801	43.7555	0.6902	62.5955
1.4424	7.44	42500	1.1770	43.8281	0.6902	62.6384
1.4415	7.88	45000	1.1743	43.9782	0.6912	62.751
1.4347	8.31	47500	1.1751	44.0103	0.6919	62.7896
1.4309	8.75	50000	1.1748	44.0446	0.6916	62.8076

Framework versions

• Transformers 4.38.0
• Pytorch 2.2.1+cu121
• Datasets 2.19.1
• Tokenizers 0.15.2