Modelpublic

NotShrirang/t5-base-en-to-svo-converter

sourceHugging Faceapache-2.0updated 3y agoView on Hugging Face

0likes11downloads

Model Card

t5-base-en-to-svo-converter

This model is a fine-tuned version of t5-base on an custom dataset. It achieves the following results on the evaluation set:

—Train Loss: 1.4502
—Validation Loss: 0.9447
—Epoch: 29

Training hyperparameters

The following hyperparameters were used during training:

—optimizer: {'name': 'AdamWeightDecay', 'learningrate': 2e-05, 'decay': 0.0, 'beta1': 0.9, 'beta2': 0.999, 'epsilon': 1e-07, 'amsgrad': False, 'weightdecay_rate': 0.01}
—training_precision: float32

Training results

Train Loss	Validation Loss	Epoch
6.1759	5.0248	0
4.8012	4.0190	1
4.1438	3.3283	2
3.6919	3.0962	3
3.4139	2.7165	4
3.2353	2.6857	5
3.0536	2.5674	6
2.9421	2.3690	7
2.7912	2.2234	8
2.7126	2.0857	9
2.5692	1.9333	10
2.4803	1.8273	11
2.3759	1.7388	12
2.3160	1.6881	13
2.2398	1.6645	14
2.1722	1.6141	15
2.1018	1.6087	16
2.0382	1.4362	17
1.9601	1.3715	18
1.8813	1.3523	19
1.8604	1.2811	20
1.8046	1.3163	21
1.7001	1.2058	22
1.7068	1.2065	23
1.6475	1.1352	24
1.5943	1.1278	25
1.5792	0.9393	26
1.5407	1.0137	27
1.4625	1.0382	28
1.4502	0.9447	29

Framework versions

—Transformers 4.34.1
—TensorFlow 2.14.0
—Datasets 2.14.6
—Tokenizers 0.14.1