Google previews site for sharing machine learning experiments
TensorBoard.dev allows uploading and sharing of TensorBoards and visualisation of model graphs and metrics
1 November 2019 | 0
Google has unveiled TensorBoard.dev, an online platform where data scientists, researchers, machine learning practitioners, and software developers can share machine learning experiments and collaborate on machine learning projects.
Now in a beta release stage, TensorBoard.dev lets users upload machine learning experiments for sharing with anyone. The platform leverages the TensorBoard visualisation toolkit, which works with Google’s TensorFlow library for machine learning and deep learning.
TensorBoard allows users to track model training metrics such as accuracy and loss, visualise the model graph, and view histograms of weights, biases, and other tensors. Users can upload TensorBoard logs for free and share the URL, although there is a limit on storage. Data uploaded will be visible to anyone with a link, so it is not advisable to upload sensitive data.
The TensorBoard.dev preview starts out with the TensorBoard scalar dashboard, for visualising key machine learning metrics. Over time, more functionality will be added, with the goal of improving the sharing experience.
The TensorBoard.dev preview is now online, and a Google Account is required to use TensorBoard.dev.
IDG News Service