(say Indian law), and boom: now it answers only legal stuff. Congrats, you’ve made LawGPT.
So why the disrespect?

Is it because: – It’s technically easy now?

OR It’s seen as lazy compared to training from scratch?

What do you think?

I chuckled a bit... but then I paused.

Why does fine-tuning get clowned so hard in the AI community?

After all, it’s not nothing — you're taking an existing LLM (like Mistral or GPT), training it on niche data👇

This worked. Extracted the actual value with data[0] and inserted it correctly. Clean and done.

Better. Now I fetched the data properly—but tried to insert the whole tuple data, which didn’t work.

Final Fix. WIN WIN THICK SKIN

I thought k would store the fetched value, but print() returns None, so I was trying to insert None into the database.

Attempt 2 (Almost there)

This tiny change saved me hours—

…and probably saved my screen from getting smashed too. 🫣💻🥊

Also, give me a follow — it’s free, and honestly, it feels really good to see a notification for once.

I now just do this: 👇

In my prototype of a project “Code of the Day” — where users get daily coding prompts and a leaderboard —

instead of doing this mess: 👇

One day — sobbing, scrolling — I stumbled on a concept.

A concept that would change the way I code forever.

You’re probably asking… what was that concept?

Drumroll... 🥁

Classes.

Let me Explain..

👇