• You're a research intern and you're an undergraduate student. Have some experience in data science and now new to research.

  • Your task is to conduct research on vision transformers. Oh and by the way, you're new to transformers concepts.

  • You first learn transformers and make your fundamentals strong. You implemented and checked vision transformers original code. All fine!

  • You even checked and ran current SOTA papers in vision transformers. Now you have gained enough knowledge to conduct your own research.



  • I am unable to develop good ideas in my mind and obviously unable to implement by myself.

  • I tried and tried and tried, which only leads to big bad errors.

  • I thought maybe I understand other author's codes and practice their implementation to get hang of it. Understanding the code itself takes DAYYSSS(sometimes weeks).

  • I seriously don't understand if there's any right way to do this. Sometimes I wish I had proper guidance to conduct research.

  • I did come up with various strategies, but nothing works. It's been almost 2 months and I feel depressed.

Any suggestions will be highly appreciated.!

About my Advisor/mentor

My mentor who is a university professor assigned me this task is only concerned about my progress and doesn't help me to actually get things done. And in my last meeting with him clearly tells that he's not at all interested in the above things I mentioned

  • 3
    $\begingroup$ Your question might be better suitable for academia.stackexchange.com since it's about the process of research and not the domain $\endgroup$
    – Jonathan
    Commented May 29, 2021 at 12:20
  • $\begingroup$ I thought so too but it was related to data science right. So I thought I post it here first. Anyway, I'll post it there as well. Thank you! $\endgroup$
    – Dagad
    Commented May 29, 2021 at 12:59
  • $\begingroup$ Better for AcademiaSE indeed. Short answer: don't worry, everything you describe is normal. Being able to develop one's own ideas (let alone good ones!) is essentially what students learn when they do a PhD, and it takes a few years. It means that it's not really expected from you as an undergrad intern. Making errors is what everybody does all their life, we start with big ones and we try to decrease their size and frequency :) $\endgroup$
    – Erwan
    Commented May 29, 2021 at 15:43
  • 1
    $\begingroup$ I also notice that you don't mention any supervision: normally your first contact about any problems is your supervisor, don't be shy and tell them about your problems, it's their job to guide you and you're not a "bad student". Btw in ML it's easy to be fooled by how easy it looks when you browse stuff online, but everybody who knows their stuff seriously will tell you that there's a lot of work behind, it takes time. $\endgroup$
    – Erwan
    Commented May 29, 2021 at 15:44
  • $\begingroup$ Thank you. I have added a section about my current mentor and I think you'll definitely understand what I got myself into. And that's the exact problem, I don't have any kind of support but I really want to do this. I did post this in academia and unfortunately, they took down my question :( academia.stackexchange.com/questions/168366/… $\endgroup$
    – Dagad
    Commented May 30, 2021 at 5:37


Browse other questions tagged or ask your own question.