PerceptiLabs – A GUI and Visual API for TensorFlow
Recently launched PerceptiLabs 0.11, is shortly changing into the GUI and visible API for TensorFlow. PerceptiLabs is constructed round a complicated visible ML modeling editor by which you drag and drop parts and join them collectively to kind your mannequin, routinely creating the underlying TensorFlow code. Try it now.
TensorFlow is arguably the preferred machine learning (ML) framework in the present day due to its wealthy multi-layer API. However, as a framework for ML modeling by way of code, TensorFlow generally is a handful for inexperienced persons. Even skilled data scientists and builders can discover it troublesome when working with massive units of code to visualise the mannequin, to see how adjustments to logic and hyperparameters have an effect on the mannequin, and to trace down bugs.
Just launched PerceptiLabs 0.11, is shortly changing into the GUI and visible API for TensorFlow that goals to resolve these challenges. It’s constructed round a complicated visible ML modeling editor by which you drag and drop parts and join them collectively to kind your mannequin. PerceptiLabs routinely creates the underlying TensorFlow code, successfully wrapping that code inside visible parts in an effort to simply visualize your mannequin.
Image: A have a look at the PerceptiLabs’ visible modeling device, displaying a picture recognition mannequin with it’s parts and code view.
PerceptiLabs is a extremely interactive device. As parts are added and configured, every shows a reside preview to indicate the way it processes enter. There are additionally debugging instruments, supplying you with suggestions, warnings, and errors while you’re constructing your mannequin so you’ll be able to see the place one thing is improper and repair it instantly.
PerceptiLabs offers you a selection of the way you wish to work. You can tweak hyperparameters within the visible modeling editor or you’ll be able to modify the underlying code in PerceptiLabs’ code editor.
Categories of parts embrace knowledge sources, knowledge processing (e.g., to reshape knowledge), deep studying operations (e.g., convolution layers), math operations, and totally different coaching strategies (e.g., classification, reinforcement studying, and many extra).
Image: PerceptiLabs’ statistical view allows you to see and perceive your mannequin’s efficiency in actual time.
PerceptiLabs additionally trains and validates the mannequin, and supplies a wealthy set of statistical views enabling customers to know the mannequin’s efficiency, whereas offering real-time analytics about each operation and variable. You can simply monitor and perceive the gradients’ habits, carry out real-time debugging, and see the place to optimize your mannequin.
PerceptiLabs features a Model Hub to handle and monitor a number of fashions, and you’ll be able to export your mannequin’s code to a Jupyter Notebook file, export your mannequin design and knowledge to GitHub, and export your fully-trained TensorFlow mannequin.
Get Started constructing your mannequin with PerceptiLabs Today!
PerceptiLabs is offered as a free model that runs regionally in your machine, and additionally in Docker and enterprise variations. Ready to test it out? Download and run it now:
$ pip set up perceptilabs