User Tools

Site Tools


ml:educational_framework_edf

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision
Previous revision
ml:educational_framework_edf [2020/08/19 11:41] jmflanigml:educational_framework_edf [2023/06/15 07:36] (current) – external edit 127.0.0.1
Line 1: Line 1:
 ====== Educational Framework ====== ====== Educational Framework ======
-The Educational Framework (EDF) is a deep learning framework for educational purposes written in 150 lines of python and numpy code.  It was written by David McCallester for the course [[https://mcallester.github.io/ttic-31230/|TTIC 31230: Fundamentals of Deep Learning]]+The Educational Framework (EDF) is a deep learning framework for educational purposes written in 150 lines of Python and NumPy code.  It was written by David McCallester for the course [[https://mcallester.github.io/ttic-31230/|TTIC 31230: Fundamentals of Deep Learning]], ([[https://mcallester.github.io/ttic-31230/Fall2020/|Fall 2020]] with [[https://www.youtube.com/channel/UCciVrtrRR3bQdaGbti9-hVQ/featured|videos]].)
   * [[https://mcallester.github.io/ttic-31230/02MLP/EDFslides.pdf|Slides]]   * [[https://mcallester.github.io/ttic-31230/02MLP/EDFslides.pdf|Slides]]
   * [[https://mcallester.github.io/ttic-31230/02MLP/edf.py|Code]]   * [[https://mcallester.github.io/ttic-31230/02MLP/edf.py|Code]]
 +
 +Here is the complete source code for the framework:
  
 <code python> <code python>
Line 10: Line 12:
 learning_rate = .001 learning_rate = .001
  
-############################################ Clearing the Computation graph #################################        +######################## Clearing the Computation graph ######################        
 # This should be done immediately before creating a new computation graph. # This should be done immediately before creating a new computation graph.
  
Line 18: Line 20:
     Parameters = []     Parameters = []
  
-############################################ Forward-Backward and SGD #################################+########################### Forward-Backward and SGD ######################
  
 def Forward(): def Forward():
Line 33: Line 35:
     for p in Parameters: p.SGD()     for p in Parameters: p.SGD()
  
-############################################ Computation Graphs #################################+########################## Computation Graphs #############################
  
 """ """
-There are three kinds of nodes in a computation graph, inputs, parameters, and computed nodes (CompNodes). +There are three kinds of nodes in a computation graph, inputs, parameters, 
-Computed nodes are defined on a case by case bases.  The Sigmoid class is defined as an example.  Other classes are +and computed nodes (CompNodes). 
-defined below.+Computed nodes are defined on a case by case bases.  The Sigmoid class 
 +is defined as an example.  Other classes are defined below.
 """ """
  
Line 66: Line 69:
         self.grad += delta         self.grad += delta
  
-############### Parameter Packages and some Compnodes ###########################+############### Parameter Packages and some Compnodes #######################
  
 class ParameterPackage: class ParameterPackage:
ml/educational_framework_edf.1597837263.txt.gz · Last modified: 2023/06/15 07:36 (external edit)

Donate Powered by PHP Valid HTML5 Valid CSS Driven by DokuWiki