Modelpublic

Dogebooch/BioBERT-mnli-snli-scinli-scitail-mednli-stsb-ncbi

sourceHugging Faceupdated 3y agoView on Hugging Face

0likes21downloads

Model Card

BioBERT-mnli-snli-scinli-scitail-mednli-stsb-ncbi

This model is a fine-tuned version of pritamdeka/BioBERT-mnli-snli-scinli-scitail-mednli-stsb on the ncbi_disease dataset. It achieves the following results on the evaluation set:

—Loss: 0.0814
—Precision: 0.8604
—Recall: 0.8990
—F1: 0.8793
—Accuracy: 0.9870

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
—trainbatchsize: 16
—evalbatchsize: 16
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	340	0.0481	0.8308	0.8438	0.8372	0.9840
0.0715	2.0	680	0.0497	0.8337	0.8771	0.8548	0.9857
0.0152	3.0	1020	0.0588	0.8596	0.8802	0.8698	0.9858
0.0152	4.0	1360	0.0589	0.8589	0.8875	0.8730	0.9873
0.0059	5.0	1700	0.0693	0.8412	0.8938	0.8667	0.9852
0.003	6.0	2040	0.0770	0.8701	0.9	0.8848	0.9863
0.003	7.0	2380	0.0787	0.861	0.8969	0.8786	0.9863
0.0014	8.0	2720	0.0760	0.8655	0.8979	0.8814	0.9872
0.0007	9.0	3060	0.0817	0.8589	0.8938	0.8760	0.9865
0.0007	10.0	3400	0.0814	0.8604	0.8990	0.8793	0.9870

Framework versions

—Transformers 4.29.1
—Pytorch 2.0.1+cpu
—Datasets 2.12.0
—Tokenizers 0.13.3