Modelpublic

sksayril/masked-lm-tpu

sourceHugging Facemitupdated 3y agoView on Hugging Face

0likes6downloads

Model Card

sksayril/masked-lm-tpu

This model is a fine-tuned version of roberta-base on an unknown dataset. It achieves the following results on the evaluation set:

—Train Loss: 9.9316
—Train Accuracy: 0.0038
—Validation Loss: 9.8372
—Validation Accuracy: 0.0175
—Epoch: 7

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': 'AdamWeightDecay', 'learningrate': {'classname': 'WarmUp', 'config': {'initiallearningrate': 0.0001, 'decayschedulefn': {'classname': 'PolynomialDecay', 'config': {'initiallearningrate': 0.0001, 'decaysteps': 22325, 'endlearningrate': 0.0, 'power': 1.0, 'cycle': False, 'name': None}, '_passiveserialization_': True}, 'warmupsteps': 1175, 'power': 1.0, 'name': None}}, 'decay': 0.0, 'beta1': 0.9, 'beta2': 0.999, 'epsilon': 1e-08, 'amsgrad': False, 'weightdecayrate': 0.001}
—training_precision: float32

Training results

Train Loss	Train Accuracy	Validation Loss	Validation Accuracy	Epoch
10.1929	0.0000	10.1896	0.0	0
10.1870	0.0	10.1773	0.0	1
10.1724	0.0	10.1478	0.0	2
10.1466	0.0000	10.1051	0.0	3
10.1096	0.0000	10.0535	0.0000	4
10.0618	0.0001	9.9953	0.0006	5
9.9983	0.0006	9.9197	0.0057	6
9.9316	0.0038	9.8372	0.0175	7

Framework versions

—Transformers 4.35.2
—TensorFlow 2.12.0
—Tokenizers 0.15.0