By Vilhjalmur A. Vilhjalmsson and Even W. Lauvrak.

(This article was first published in NorwAI Annual Report 2023, 2024-04-01)

This summer, we started our chapter as fresh research assistants at NorwAI. Armed with nothing but our sprawling excitement and hunger for learning, we set out to aid the research of AI in Norway in any way we could.

What started as a daunting task turned into one of the best learning experiences any student could have asked for. We got to build close relationships with professors, post-docs, and the rest of the team at NorwAI, as well as work alongside them in their respective work. All this work has since accumulated in major contributions to the NorGLM project, even co-authoring papers.

My experience (Vilhjalmur):

Throughout the summer, I was tasked with handling the massive dataset used to train the NorGLM model; more precisely, making the NorGLM dataset easily accessible and transparent to the public. Handling a dataset of this scale requires careful management, and I devoted considerable time to optimizing every aspect of its processing.

Highlights

One of the highlights was an optimization I implemented in a processing pipeline, which reduced the processing time from over an hour to less than 30 seconds. I worked on the entire stack of the effort, developing the frontend, setting up the backend, and seamlessly integrating everything.
Taking a step back and reflecting on my time so far at NorwAI, the thing that stands out is the people I have the privilege of working alongside. Not only am I continuously impressed by the academic and intellectual prowess, but also by how kind and welcoming people are. One moment, we’re diving deep into advanced AI concepts; the next, we’re sharing laughs by the coffee machine.

My experience (Even):

Starting from the first week I was given the task of fine tuning and evaluating one of the NorGLM models, as I did not have the greatest expertise on the field of LLMs I were given the freedom to do my own research and learn with guidance from the team. After trial and error, I had learned a lot and my main task from there on was to fine tune and evaluate the different parameter models of NorGLM as well as other baseline models for comparisons.
This led to my contribution to a collaborative research paper that we submitted in December.

No wrong questions

I would say that my favourite part about NorwAI is how open everyone is and that there’s no wrong questions. It may seem intimidating to ask professors or post-docs questions that you think you should know, but they are just happy to have a talk and help, and that what’s make working as a research assistant at NorwAI so great.

Working as research assistant gives you a lot of opportunities that awaits to be taken. How much you want to do and at what level depends on your eagerness to look out for tasks, tasks that mainly comes from talking to the different professors and post-docs, showing interest, and asking to contribute. And the best part is that everyone is open, friendly, and ready for a chat. Some of the best memories we have made from the office is from sitting around the lunch table with all different kinds of people from the office and talking nonsense and just having a good time before we head back to exciting tasks.

Published 2024-08-05