TensorSpace Playground is the place we designed for presenting different pre-built models. In the "Playground", we can experience different pre-trained deep learning models, including object classification, object detections, image generations etc.

All models in the "Playground" are interactive. We can move the mouse to see the relation lines among layers; we can click the layer aggregation to check feature maps; we can move the camera to view from any direction... This mirror ships four lightweight models for classroom use: LeNet (0.2MB), ACGAN (3MB), MobileNetv1 (16MB) and YOLOv2-tiny (60MB), all served from this machine. YOLOv2-tiny is the largest one (60MB). It takes a dozen seconds to build on first open, please be patient. To have a better experience in this amusement park, we highly recommend to use medium or large device (device width > 750px).

While we are still developing and expanding our model collections, enjoy the models and have fun~

- left click and drag -

change camera angle

- right click and drag -

change camera position

- scroll up and down -

move camera closer or further