Generate and debug Python code- with some help from AI
Epsilon Code Banner
Epsilon Code is a tool that can generate Python code from plain-text description (in natural language) for you. It can also get you instant debugging help- much like StackOverflow responses but faster and better than what humans can do!
Everything happens in just 1 click and 2 lines of code! This is THE simplest way that we can imagine coding and debugging - and we are sure you will like it too! The code and debug help comes from the most powerful language model EVER - yes that's GPT-3! We have trained it on thousands of tokens of high quality code and professional responses and it conjures up innovative and resilient solutions to whatever problem you ask it! It has the knowledege of the best of the libraries, APIs, logic loops and Python hacks that it employs to solve your problems in Python in the most efficient way - much like intelligent coders do.
You need to describe what you want your code to do - in plain English. An example use case might be "Print Hello World" or "Build a decision tree classifier on the IRIS dataset". And you will get a working Python code with all of the library imports, indents, intelligent variable names and user interfacing!
You can also get help in debugging by passing in the traceback of the errors you are getting and you can expect a "StackOverflow Like" response!
We clean up your request and pass it in to OpenAI's GPT-3 model which has been fine-tuned on thousands of tokens of sustainable code and pre-processed StackOverflow responses.
It understands what libraries and APIs are needed and what should be the overall structure of the code and logic loops. All this information is used to generate 1 unique code that can fulfill your needs.
Here is an example from the code generator utility. When given the prompt:
Ask user for path to an image. Convert it to grayscale and save it as 'gray.png'.
This is the response:
print("Enter path to an image file: ")
img_path = input()
im = PIL.Image.open(img_path) # open the image file in memory
gray = im.convert('L') # convert it to grayscale mode (L for luminance)
#Generated with Epsilon-Code
As you can see, it imports the necessary libraries, gives intelligent variable names and comments the necessary points.
Here is an example from the debugger utility. When given the traceback snip:
ImportError: No module named Crypto.Cipher
This is the output:
You need to install the cryptography library. You might have installed it in a different location. Try doing "pip3 install cryptography" or "pip install cryptography" depending on your python version and OS. Make sure that you are using the correct path for your installation of cryptography, if you have multiple installations of Python/cryptography in different locations, try adding them all to sys.path so that they can be found by import statements.
As you can see it is REALLY comprehensive and helpful. It not only gives multiple fixes but it also helps you to better understand the root cause and how to prevent it in future.
Whole of Epsilon-Code was developed independently by Shreenabh Agrawal. All information about him on his website: shreenabh.com Github: https://github.com/Shreenabh664/Epsilon-Code Here is the official project contact information: