Modelpublic

LiukG/mus_promoter-finetuned-lora-500m-1000g

sourceHugging Facecc-by-nc-sa-4.0updated 3y agoView on Hugging Face

0likes6downloads

Model Card

mus_promoter-finetuned-lora-500m-1000g

This model is a fine-tuned version of InstaDeepAI/nucleotide-transformer-500m-1000g on the None dataset. It achieves the following results on the evaluation set:

—Loss: 0.2791
—F1: 0.9211
—Matthews Correlation: 0.8076
—Accuracy: 0.9062
—F1 Score: 0.9211

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: 0.0005
—trainbatchsize: 8
—evalbatchsize: 64
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Matthews Correlation	Accuracy	F1 Score
0.6279	0.43	100	0.4652	0.8986	0.7910	0.8906	0.8986
0.3719	0.85	200	0.3562	0.9167	0.8113	0.9062	0.9167
0.3615	1.28	300	0.6468	0.8718	0.6790	0.8438	0.8718
0.3425	1.71	400	0.4302	0.8889	0.7210	0.8594	0.8889
0.3106	2.14	500	0.3645	0.9041	0.7773	0.8906	0.9041
0.3218	2.56	600	0.2542	0.9333	0.8395	0.9219	0.9333
0.2135	2.99	700	0.4137	0.9211	0.8076	0.9062	0.9211
0.2512	3.42	800	0.3547	0.9351	0.8414	0.9219	0.9351
0.1963	3.85	900	0.2171	0.9333	0.8395	0.9219	0.9333
0.1304	4.27	1000	0.2791	0.9211	0.8076	0.9062	0.9211

Framework versions

—Transformers 4.38.1
—Pytorch 2.1.0+cu121
—Datasets 2.17.1
—Tokenizers 0.15.2