samzirbo/gendered_balanced_last

gendered_balanced_new

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: 1.1763
• Bleu: 43.813
• Meteor: 0.6892
• Chrf++: 62.6103

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++
4.297	0.44	2500	2.0050	28.2124	0.5592	49.6855
2.4102	0.88	5000	1.7130	33.5308	0.6073	54.4802
2.1323	1.31	7500	1.5723	36.3429	0.6293	56.6736
1.9971	1.75	10000	1.4819	37.5927	0.6411	57.7792
1.8896	2.19	12500	1.4211	38.9306	0.6509	58.8416
1.8066	2.63	15000	1.3774	40.4467	0.6587	59.8543
1.7574	3.06	17500	1.3336	40.6982	0.6646	60.1397
1.6837	3.5	20000	1.3066	41.6068	0.6716	60.8397
1.6587	3.94	22500	1.2767	41.5927	0.6719	60.9297
1.595	4.38	25000	1.2556	42.1176	0.6762	61.2694
1.5764	4.81	27500	1.2317	42.9623	0.6814	61.9533
1.5416	5.25	30000	1.2212	42.9789	0.681	61.9041
1.5155	5.69	32500	1.2070	43.3916	0.6865	62.3045
1.4955	6.13	35000	1.1973	43.455	0.6854	62.2717
1.4688	6.56	37500	1.1863	43.5298	0.6857	62.3399
1.467	7.0	40000	1.1827	43.7299	0.6884	62.4926
1.4416	7.44	42500	1.1796	43.7634	0.6887	62.5774
1.4411	7.88	45000	1.1764	43.93	0.6896	62.6606
1.434	8.31	47500	1.1764	43.8254	0.6885	62.5964
1.4311	8.75	50000	1.1763	43.813	0.6892	62.6103

Framework versions

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