Modelpublic

jhaberbe/segformer-b0-finetuned-lipid-droplets-v2

sourceHugging Faceotherupdated 2y agoView on Hugging Face

0likes7downloads

Model Card

segformer-b0-finetuned-lipid-droplets-v2

This model is a fine-tuned version of nvidia/mit-b0 on the jhaberbe/lipid-droplets-v4 dataset. It achieves the following results on the evaluation set:

—Loss: 0.1330
—Mean Iou: 0.004
—Mean Accuracy: 0.008
—Overall Accuracy: 0.008
—Accuracy Unlabeled: nan
—Accuracy Lipid: 0.008
—Iou Unlabeled: 0.0
—Iou Lipid: 0.008

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: 6e-05
—trainbatchsize: 2
—evalbatchsize: 2
—seed: 42
—optimizer: Use OptimizerNames.ADAMWTORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizerargs=No additional optimizer arguments
—lrschedulertype: linear
—num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Unlabeled	Accuracy Lipid	Iou Lipid
0.5568	5.0	20	0.6918	0.0435	0.0870	0.0870	nan	0.0870	0.0870
0.4064	10.0	40	0.6370	0.0311	0.0622	0.0622	nan	0.0622	0.0622
0.298	15.0	60	0.5943	0.0174	0.0348	0.0348	nan	0.0348	0.0348
0.2508	20.0	80	0.4472	0.0069	0.0138	0.0138	nan	0.0138	0.0138
0.2184	25.0	100	0.4683	0.0079	0.0159	0.0159	nan	0.0159	0.0159
0.2213	30.0	120	0.4272	0.004	0.008	0.008	nan	0.008	0.008
0.1669	35.0	140	0.2754	0.004	0.008	0.008	nan	0.008	0.008
0.1871	40.0	160	0.2788	0.004	0.008	0.008	nan	0.008	0.008
0.1202	45.0	180	0.2453	0.004	0.008	0.008	nan	0.008	0.008
0.1081	50.0	200	0.2138	0.004	0.008	0.008	nan	0.008	0.008
0.1062	55.0	220	0.2132	0.004	0.008	0.008	nan	0.008	0.008
0.1026	60.0	240	0.1336	0.004	0.008	0.008	nan	0.008	0.008
0.1009	65.0	260	0.1952	0.004	0.008	0.008	nan	0.008	0.008
0.07	70.0	280	0.1577	0.004	0.008	0.008	nan	0.008	0.008
0.0626	75.0	300	0.1556	0.004	0.008	0.008	nan	0.008	0.008
0.0723	80.0	320	0.1446	0.004	0.008	0.008	nan	0.008	0.008
0.0669	85.0	340	0.1151	0.004	0.008	0.008	nan	0.008	0.008
0.1126	90.0	360	0.1668	0.004	0.008	0.008	nan	0.008	0.008
0.077	95.0	380	0.1536	0.004	0.008	0.008	nan	0.008	0.008
0.0758	100.0	400	0.1330	0.004	0.008	0.008	nan	0.008	0.008

Framework versions

—Transformers 4.49.0
—Pytorch 2.6.0+cu124
—Datasets 3.4.1
—Tokenizers 0.21.1