mbart

This model is a fine-tuned version of facebook/mbart-large-50 on an unknown dataset. It achieves the following results on the evaluation set:

• Train Loss: 0.5342
• Validation Loss: 0.5633
• Epoch: 35

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:

• optimizer: {'name': 'SGD', 'learning_rate': 0.01, 'decay': 0.0, 'momentum': 0.9, 'nesterov': False}
• training_precision: mixed_bfloat16

Training results

Train Loss	Validation Loss	Epoch
4.5626	3.7843	0
2.5836	1.9212	1
1.6546	1.2552	2
1.2499	1.0248	3
1.0088	0.8457	4
0.9100	0.7958	5
0.8290	0.8421	6
0.7999	0.7625	7
0.7633	0.7202	8
0.7439	0.7100	9
0.7182	0.6787	10
0.7092	0.6877	11
0.6823	0.6684	12
0.6738	0.6712	13
0.6603	0.6858	14
0.6462	0.6268	15
0.6373	0.6208	16
0.6424	0.6735	17
0.6259	0.6423	18
0.6249	0.6069	19
0.6148	0.6510	20
0.6063	0.6207	21
0.5987	0.5977	22
0.5917	0.6019	23
0.5800	0.5828	24
0.5779	0.5505	25
0.5765	0.5887	26
0.5667	0.5989	27
0.5623	0.5859	28
0.5564	0.5907	29
0.5523	0.5928	30
0.5478	0.5624	31
0.5472	0.5563	32
0.5462	0.5953	33
0.5324	0.5593	34
0.5342	0.5633	35

Framework versions

• Transformers 4.19.2
• TensorFlow 2.8.0
• Datasets 2.2.1
• Tokenizers 0.12.1