deep_visu

Visualization and experiments framework for training Deep learning and Machine learning algorithms.

Help to keep track of experiments by saving the results and configurations on single html file.

Features

• Visualization for real time signals
• Written in Python 3
• Based on Bokeh

Requirements

• Python 3
• Bokeh

Use

• Run bokeh in a separated terminal

~$ bokeh serve

• Run the Python script

~$ python3 example_signals.py

How to use it in Python script

# Import the plot
from LogPlotBokeh import LogPlot

# Create an instance
log = LogPlot(name="Title",
        properties=["Reward", "Avr_Reward"],
        properties_telemetry=["speedX", "steer"],
        output_path="."
        )

# Feed data to both plots and the text log
episode_progres = {"Reward":10, "Avr_Reward":20}
log.progress.add(x=i, y=episode_progres)

# telemetry
telemetry = {"speedX":10, "accel":20}
log.telemetry.add(x=i, y=telemetry)

# Add some text to the log
log.terminal("Model  parameters %d" % i)

# Close
log.close()

Todos

• Improve performance
• Remove the use of Bokeh client

License

• GPL v3

Free Software