Create 3d loss surface visualizations, with optimizer path. Issues welcome!

Overview

MLVTK PyPI - Python Version PyPI

A loss surface visualization tool

Png

Simple feed-forward network trained on chess data, using elu activation and Adam optimizer


Gif

Simple feed-forward network trained on chess data, using tanh activation and SGD optimizer


Gif

3 layer feed-forward network trained on hand written letters data, using relu activation, SGD optimizer and learning rate of 2.0. Example of what happens to path when learning rate is too high


Gif

Simple feed-forward network trained on chess data, using hard-sigmoid activation and RMSprop optimizer

Why?

  • :shipit: Simple: A single line addition is all that is needed.
  • âť“ Informative: Gain insight into what your model is seeing.
  • đź““ Educational: See how your hyper parameters and architecture impact your models perception.

Quick Start

Requires version
python >= 3.6.1
tensorflow >= 2.3.1
plotly >=4.9.0

Install locally (Also works in google Colab!):

pip install mlvtk

Optionally for use with jupyter notebook/lab:

Notebook

=5.3" "ipywidgets==7.5"">
pip install "notebook>=5.3" "ipywidgets==7.5"

Lab

pip install jupyterlab "ipywidgets==7.5"

# Basic JupyterLab renderer support
jupyter labextension install [email protected]

# OPTIONAL: Jupyter widgets extension for FigureWidget support
jupyter labextension install @jupyter-widgets/jupyterlab-manager [email protected]

Basic Example

from mlvtk.base import Vmodel
import tensorflow as tf
import numpy as np

# NN with 1 hidden layer
inputs = tf.keras.layers.Input(shape=(None,100))
dense_1 = tf.keras.layers.Dense(50, activation='relu')(inputs)
outputs = tf.keras.layers.Dense(10, activation='softmax')(dense_1)
_model = tf.keras.Model(inputs, outputs)

# Wrap with Vmodel
model = Vmodel(_model)
model.compile(optimizer=tf.keras.optimizers.SGD(),
loss=tf.keras.losses.CategoricalCrossentropy(), metrics=['accuracy'])

# All tf.keras.(Model/Sequential/Functional) methods/properties are accessible
# from Vmodel

model.summary()
model.get_config()
model.get_weights()
model.layers

# Create random example data
x = np.random.rand(3, 10, 100)
y = np.random.randint(9, size=(3, 10, 10))
xval = np.random.rand(1, 10, 100)
yval = np.random.randint(9, size=(1,10,10))

# Only difference, model.fit requires validation_data (tf.data.Dataset, or
# other container
history = model.fit(x, y, validation_data=(xval, yval), epochs=10, verbose=0)

# Calling model.surface_plot() returns a plotly.graph_objs.Figure
# model.surface_plot() will attempt to display the figure inline

fig = model.surface_plot()

# fig can save an interactive plot to an html file,
fig.write_html("surface_plot.html")

# or display the plot in jupyter notebook/lab or other compatible tool.
fig.show()
Owner
Research analyst
Parallel t-SNE implementation with Python and Torch wrappers.

Multicore t-SNE This is a multicore modification of Barnes-Hut t-SNE by L. Van der Maaten with python and Torch CFFI-based wrappers. This code also wo

Dmitry Ulyanov 1.7k Jan 09, 2023
A guide for using Bootstrap 5 classes in Dash Bootstrap Components V1

dash-bootstrap-cheatsheet This handy interactive cheatsheet makes it easy to use the Bootstrap 5 classes with your Dash app made with the latest versi

10 Dec 22, 2022
Python code for solving 3D structural problems using the finite element method

3DFEM Python 3D finite element code This python code allows for solving 3D structural problems using the finite element method. New features will be a

Rémi Capillon 6 Sep 29, 2022
nptsne is a numpy compatible python binary package that offers a number of APIs for fast tSNE calculation.

nptsne nptsne is a numpy compatible python binary package that offers a number of APIs for fast tSNE calculation and HSNE modelling. For more detail s

Biomedical Visual Analytics Unit LUMC - TU Delft 29 Jul 05, 2022
Automatically visualize your pandas dataframe via a single print! 📊 💡

A Python API for Intelligent Visual Discovery Lux is a Python library that facilitate fast and easy data exploration by automating the visualization a

Lux 4.3k Dec 28, 2022
Friday Night Funkin - converts a chart from 4/4 time to 6/8 time, or from regular to swing tempo.

Chart to swing converter As seen in https://twitter.com/i_winxd/status/1462220493558366214 A program written in python that converts a chart from 4/4

5 Dec 23, 2022
Easily configurable, chart dashboards from any arbitrary API endpoint. JSON config only

Flask JSONDash Easily configurable, chart dashboards from any arbitrary API endpoint. JSON config only. Ready to go. This project is a flask blueprint

Chris Tabor 3.3k Dec 31, 2022
Eulera Dashboard is an easy and intuitive way to get a quick feel of what’s happening on the world’s market.

an easy and intuitive way to get a quick feel of what’s happening on the world’s market ! Eulera dashboard is a tool allows you to monitor historical

Salah Eddine LABIAD 4 Nov 25, 2022
Simple python implementation with matplotlib to manually fit MIST isochrones to Gaia DR2 color-magnitude diagrams

Simple python implementation with matplotlib to manually fit MIST isochrones to Gaia DR2 color-magnitude diagrams

Karl Jaehnig 7 Oct 22, 2022
CPG represent!

CoolPandasGroup CPG represent! Arianna Brandon Enne Luan Tracie Project requirements: use Pandas to clean and format datasets use Jupyter Notebook to

Enne 3 Feb 07, 2022
Generate visualizations of GitHub user and repository statistics using GitHub Actions.

GitHub Stats Visualization Generate visualizations of GitHub user and repository statistics using GitHub Actions. This project is currently a work-in-

Aditya Thakekar 1 Jan 11, 2022
Streamlit-template - A streamlit app template based on streamlit-option-menu

streamlit-template A streamlit app template for geospatial applications based on

Qiusheng Wu 41 Dec 10, 2022
Data Analysis: Data Visualization of Airlines

Data Analysis: Data Visualization of Airlines Anderson Cruz | London-UK | Linkedin | Nowa Capital Project: Traffic Airlines Airline Reporting Carrier

Anderson Cruz 1 Feb 10, 2022
Datapane is the easiest way to create data science reports from Python.

Datapane Teams | Documentation | API Docs | Changelog | Twitter | Blog Share interactive plots and data in 3 lines of Python. Datapane is a Python lib

Datapane 744 Jan 06, 2023
Drag’n’drop Pivot Tables and Charts for Jupyter/IPython Notebook, care of PivotTable.js

pivottablejs: the Python module Drag’n’drop Pivot Tables and Charts for Jupyter/IPython Notebook, care of PivotTable.js Installation pip install pivot

Nicolas Kruchten 512 Dec 26, 2022
Interactive plotting for Pandas using Vega-Lite

pdvega: Vega-Lite plotting for Pandas Dataframes pdvega is a library that allows you to quickly create interactive Vega-Lite plots from Pandas datafra

Altair 342 Oct 26, 2022
Bioinformatics tool for exploring RNA-Protein interactions

Explore RNA-Protein interactions. RNPFind is a bioinformatics tool. It takes an RNA transcript as input and gives a list of RNA binding protein (RBP)

Nahin Khan 3 Jan 27, 2022
Profile and test to gain insights into the performance of your beautiful Python code

Profile and test to gain insights into the performance of your beautiful Python code View Demo - Report Bug - Request Feature QuickPotato in a nutshel

Joey Hendricks 138 Dec 06, 2022
A collection of 100 Deep Learning images and visualizations

A collection of Deep Learning images and visualizations. The project has been developed by the AI Summer team and currently contains almost 100 images.

AI Summer 65 Sep 12, 2022
Render tokei's output to interactive sunburst chart.

Render tokei's output to interactive sunburst chart.

134 Dec 15, 2022