Skip to content

Latest commit

 

History

History
58 lines (44 loc) · 2.32 KB

README.md

File metadata and controls

58 lines (44 loc) · 2.32 KB
Raw

DendroMap Data

This repository contains the data used in the DendroMap Live Site. For use with your own data, keep reading, then navigate to DendroMap Code for example code.

Using your own Data

To give an example, if I downloaded the cifar10 data from this repository, I would put the cifar10 folder in the public folder in my local server in the DendroMap Code.

I would then connect the data to DendroMap Code by adding an object in the options array inside of src/dataOptions.js. It must have the following format:

{
	dataset: "YOUR DATASET NAME",
	model: "YOUR MODEL NAME",
	cluster_filepath: "CLUSTER_FILEPATH",
	class_cluster_filepath: "CLASS_CLUSTER_FILEPATH**OPTIONAL**",
	image_filepath: "IMAGE_FILEPATH",
}

In the case of the cifar10 data in this repository,

📦 cifar10
┣ 📂 clusters
┃ ┣ 📜 cifar10_resnet50.json
┃ ┗ 📜 cifar10_resnet50_classes.json
┣ 📂 images
┃ ┣ 📜 test-0.png
┃ ┣ 📜 test-1.png
...
┗ ┗ 📜 test-9999.png

since it has this file structure, I would connect it by adding this object:

{
	dataset: "CIFAR-10",
	model: "ResNet50",
	cluster_filepath: "cifar10/clusters/cifar10_resnet50.json",
	class_cluster_filepath: "cifar10/clusters/cifar10_resnet50_classes.json",
	image_filepath: "cifar10/images",
}

Then this option will show up in the dataset selection dropdown in the user interface.

You can generate your own data using python functions we created in the DendroMap Code repository with instructions and clustering code in the clustering folder. The notebooks that generated the datafiles in this repository also live there as examples.

Once you've generated your own files, put them in the public folder in the DendroMap Code, then add another option in src/dataOptions.js as shown above with the cifar10 example.

Links