Amala3/mbart-large-50_customconfig

models

This model is a fine-tuned version of facebook/mbart-large-50 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0060

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: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss
1.4786	0.0699	500	0.0108
0.0542	0.1398	1000	0.0094
0.01	0.2097	1500	0.0086
0.0095	0.2796	2000	0.0080
0.0082	0.3495	2500	0.0076
0.0078	0.4193	3000	0.0075
0.0081	0.4892	3500	0.0074
0.0083	0.5591	4000	0.0071
0.0071	0.6290	4500	0.0070
0.0074	0.6989	5000	0.0069
0.0075	0.7688	5500	0.0067
0.0073	0.8387	6000	0.0066
0.007	0.9086	6500	0.0065
0.0075	0.9785	7000	0.0064
0.006	1.0484	7500	0.0063
0.0181	1.1183	8000	0.0077
0.0066	1.1881	8500	0.0064
0.0067	1.2580	9000	0.0107
0.0087	1.3279	9500	0.0091
0.0065	1.3978	10000	0.0062
0.0054	1.4677	10500	0.0061
0.0055	1.5376	11000	0.0061
0.0053	1.6075	11500	0.0060
0.0052	1.6774	12000	0.0060
0.0051	1.7473	12500	0.0060
0.0051	1.8172	13000	0.0060
0.0063	1.8871	13500	0.0063
0.0054	1.9569	14000	0.0060

Framework versions

• Transformers 4.40.2
• Pytorch 2.1.2
• Datasets 2.18.0
• Tokenizers 0.19.1