Last week was a fairly slow week. Two of the main psychologists were out doing personal things so I was basically on my own.
One of the things that needed to get done was improvement on one of the prompts for generating the second part of the app, but obviously the psychology is not my responsibility or domain so I naturally had to wait for an opportune time to meet with them once they came back.
In the meanwhile, I spent a lot of my time learning more about LLMs and ML. For LLM research, I was following DeepLearning.ai's new course on parsing unstructured data. For ML, I was following freeCodeCamp's course. Both have been teaching me a lot thus far.
I also had two interviews but nothing super interesting to report in that regard.
Friday I was met with a bug from our marketer who said that past message history was carrying over to a new chat, and though I wasn't able to reproduce her issue with the raw code, I recognized that it was probably due to some kind of latency on the actual website that may present itself due to the online nature of things vs. my running it on my local machine which has near-instantaneous connections. My hunch was confirmed when I introduced some artificial latency and I started chatting before the chat history would actually get cleared in the memory.
I got to work implementing some thread locks so that the chat can't be accessed if the chat history is still being cleared. It was a mild pain because the syntax for it is a bit strange in Python, but I eventually got it working flawlessly and even started with adding some logging for important data flow points in the code, which will be useful for the future.
Anyway, that's kinda it for last week. Hopefully I have some more interesting stuff to write for next week. Cheers.
Top comments (0)