Skip to content
Log to TensorBoard with no dependencies
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.


Build Status

TensorBoardLogger.jl is an experimental library for logging arbitrary data to Tensorboard with no dependencies other than ProtoBuf.jl.

Many ideas are taken from UniversalTensorBoard and from TensorBoardX.


To use the library you must create a Logger object and then log data to it.

  • Logger(dir_path) creates a logger saving data to the folder dir_path
  • log_value(logger, name, val) logs to logger the value val under the tag name

Supported values

At the moment, you can log the following values:

  • Real scalar data
  • Complex scalar data, which will show up as two real quantities name/re and name/im
  • Histograms passed as an array which will be automatically binned or passed as a tuple of bins/heights of pre-binned data.


using TensorBoardLogger

lg = Logger("runs/run-12", overwrite=true)

for step=1:100
    ev = log_value(lg, "quan/prova1", step*1.5, step=step)
    ev = log_value(lg, "quan/prova2", step*2.5, step=step)

    x0 = 0.5+step/30; s0 = 0.5/(step/20);
    edges = collect(-5:0.1:5)
    centers = collect(edges[1:end-1] .+0.05)
    histvals = [exp(-((c-x0)/s0)^2) for c=centers]
    data_tuple = (edges, histvals)

    # Log pre-binned data
    log_histogram(lg, "hist/cust", data_tuple, step=step)
    # Automatically bin the data
    log_histogram(lg, "hist/auto", randn(1000).*s0.+x0, step=step)


I would really like to enable logging of more types of data and expand this package. For now I plan on adding log_image and log_scalars very soon.

I would also like to find a way to log whole curves at each timestep.

Contributions are welcome! You can get in touch by opening an issue, sending me an email or by saying hi on slack (@PhilipVinc).

You can’t perform that action at this time.