Modelpublic

yacine-djm/fg-bert-sustainability-15-1e-05-0.02-64

sourceHugging Facemitupdated 3y agoView on Hugging Face

0likes5downloads

Model Card

fg-bert-sustainability-15-1e-05-0.02-64

This model is a fine-tuned version of Raccourci/fairguest-bert on the None dataset. It achieves the following results on the evaluation set:

—Loss: 0.0893
—F1: 0.9139
—Roc Auc: 0.9527
—Accuracy: 0.8711

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: 1e-05
—trainbatchsize: 64
—evalbatchsize: 64
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc	Accuracy
No log	1.0	55	0.3449	0.0	0.4999	0.0946
No log	2.0	110	0.3249	0.0	0.4999	0.0946
No log	3.0	165	0.2658	0.0755	0.5195	0.1320
No log	4.0	220	0.2092	0.4475	0.6489	0.3077
No log	5.0	275	0.1706	0.7755	0.8312	0.6663
No log	6.0	330	0.1461	0.8566	0.8998	0.7848
No log	7.0	385	0.1290	0.8929	0.9416	0.8430
No log	8.0	440	0.1161	0.9044	0.9463	0.8649
No log	9.0	495	0.1038	0.9111	0.9505	0.8680
0.2414	10.0	550	0.0993	0.9143	0.9523	0.8711
0.2414	11.0	605	0.0957	0.9106	0.9504	0.8669
0.2414	12.0	660	0.0932	0.9123	0.9516	0.8680
0.2414	13.0	715	0.0910	0.9185	0.9561	0.8784
0.2414	14.0	770	0.0901	0.9151	0.9538	0.8742
0.2414	15.0	825	0.0893	0.9139	0.9527	0.8711

Framework versions

—Transformers 4.30.2
—Pytorch 2.0.1+cu118
—Datasets 2.13.1
—Tokenizers 0.13.3