(AI)nception
Have you ever spent countless hours trying to figure something out, only to discover that there’s a much easier way and that you can solve it in a fraction of the time? This universal experience defined my week.
If you recall, last week’s post, I mentioned running into errors while trying to properly load the data and convert it from numpy arrays into PyTorch tensors. I thought that I had solved the problem, but I really uncovered an underlying error. It turns out that the dataset containing all of the numpy arrays wasn’t uploading properly, meaning that I once again couldn’t access the data to build the model. Just like last week, I was sent on a scavenger hunt across the internet to find just the right strings of code to assemble functional code to upload the dataset properly this time.
As much as I tried to make sense of the different ways in which others online who encountered similar problems to mine managed to solve them, I just could not fix my code. Each time, it always seemed like there was just something small that I was missing that kept sending me back to the same error. The error seemed to stem from the code not being able to locate the file where the dataset is being held. I initially thought it was an issue with the file path and location, but that didn’t seem to be the case after a lot of attempts to fix the code. While I had the correct path to the file, the functions I was using to try and unzip the file to access the data were flawed.
Since I was getting nowhere with searching for solutions online, I asked for help from one of my advisors at the Oasis Lab. One of the important things he mentioned was using ChatGPT to evaluate parts of my code helping to locate errors. While I didn’t see instant success, through the right prompts, I was fortunately able to fix my code and load in the dataset properly this time. As helpful as ChatGPT was, I don’t plan on completely relying on it to build the model, but I am glad to add it as a tool to help me build better code for this machine learning model. I just wish I had used it much sooner to fix some of my errors earlier in coding this model.
Now that that’s over with, I can confidently continue working on the drag model and get to testing it out. Make sure to stay tuned for next week’s post as I continue to make progress on this project!
Comments:
All viewpoints are welcome but profane, threatening, disrespectful, or harassing comments will not be tolerated and are subject to moderation up to, and including, full deletion.