Programming Assignment: Solving Versioning and Dependency conflicts with an LLM

Course Q&A Generative AI for Software Development Team Software Engineering with AI
1

Hello, I am currently doing the team software engineering with ai course module 3 and stuck on the programming part 1. It’s been 11+ hours of chatting with 3 different chatbots about converting code from Python 2.7 to Python 3. I need help. Below is the code that I am stuck on:

[code removed from moderator]

1 Like
2

Same here, but fixed the code. Issue here is compatibility between Python 2 and Python 3 related to division operator, which the LLM is not fully addressing yet by default unless you ask for it.

1 Like
3

Hello there,

You had not posted under the related specialization/course I have moved it to the right place.

Second you are not allowed to post code publicly according to the rules in this forum.

Third, try the above advice and see if it resolves your issue, otherwise send me your solution in a private message and I will have a look at it!

1 Like
4

Thank you, this has helped out and switching chatbots from multiple versions of ChatGPT to the classes gpt.

1 Like
5

Thanks for the hints. I spent a good bit of time with other LLMS as well. Following worked:

  1. Using the provided ChatGPT
  2. Focusing the LLM on division operator difference between python2.7 and 3.
6

I’ll tell you what. That was a pain. No LLM was able to solve the code adaptation. As I work with Python, I had to MANUALLY fix the code in order to pass the tests. With no Python knowledge it cannot be done. Also done the second exercise without LLM. I’m reflecting now that LLMs still have strong limitations and weaknesses when it comes to providing code assistance.

7

This assignment was totally annoying. I used chatgpt 4o for 1hr and assessed both the division handling difference and numpy versioning behaviour. Then i swhitched to chatgpt o1 preview and got it in 20 minutes or so feeding him with the example analyzed by the unittests function. Eventually produced a working version.
Definetly the most annoying test of the unit.

8

I concur.

Or maybe this task isn’t one that chat tools are optimized for.

9

They do have limitations, try to think about natural vs. artificial intelligence! But the point here is that they can use to help and assist you with efficiency, any technology is meant for that not to do the intelligence part!