-
Notifications
You must be signed in to change notification settings - Fork 0
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
added blog post about academic journey reflections
Signed-off-by: Joel Niklaus <me@joelniklaus.ch>
- Loading branch information
1 parent
f317937
commit e0f3393
Showing
1 changed file
with
69 additions
and
0 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,69 @@ | ||
--- | ||
layout: page | ||
title: Reflections on my Academic Journey | ||
date: 2024-08-17 | ||
--- | ||
|
||
# Reflections on my Last Conference and Academic Journey | ||
|
||
With the conclusion of #ACL2024 some reflections on my last conference of my PhD and the academic journey over the past | ||
four years. | ||
|
||
## 🤝 Conference | ||
|
||
The food at the conference was absolutely amazing! On the academic side, things got spicy beyond just the cuisine. There | ||
was a heated exchange after the second keynote on the reasoning and planning abilities of LLMs which added some | ||
unexpected drama to the mix. | ||
|
||
Attending conferences during my PhD was a highlight, especially since, as the only NLP researcher at the University of | ||
Bern, it was fantastic to connect with so many like-minded people. | ||
|
||
I am honored that the acting ACL president, Emily Bender, attended my oral presentation and actively engaged in the | ||
discussion. Her concerns about synthetic data use in the legal domain, though not directly applicable to our work, added | ||
an interesting dimension to the conversation. The exchange, while intense, was valuable to clarify our methods and | ||
contribute to robust academic discourse. | ||
|
||
Afterward, I had a great chat with Luca Soldaini about their paper OLMo. It’s fantastic to see this huge open science | ||
effort rewarded with a best paper award! | ||
|
||
## 🧑⚖️ Reviews | ||
|
||
Throughout my academic journey, peer review has been challenging and frustrating. Common issues included reviewers | ||
focusing solely on self-disclosed limitations, undervaluing dataset contributions, and requesting impractical | ||
experiments outside the research scope. | ||
|
||
It was particularly disheartening to receive inconsistent review scores and rejection decisions that completely | ||
disregarded our thorough rebuttals and additional experiments. | ||
|
||
I believe many have had similar experiences, which is why the positive feedback from reviewers on my last paper was so | ||
refreshing. One reviewer’s comment, "This is how a dataset paper should be written," made my day. The stellar | ||
meta-review score and the consideration for the best paper award were the cherry on top. This experience was a | ||
much-needed reminder of the rewarding aspects of academia, despite the often arduous process. | ||
|
||
## 🧑🔬 PhD Journey | ||
|
||
When students ask if pursuing a PhD is worth it, I reflect on my own experience. The journey was challenging, filled | ||
with long hours, late nights, and weekends working. Maintaining good mental health was tough, and the review process was | ||
incredibly frustrating. | ||
|
||
However, the rewards were significant. Earning a PhD opened many doors for me, granting access to opportunities that I | ||
might not have had otherwise. It allows me to work on topics that I find genuinely fascinating, travel to conferences, | ||
and enjoy the freedom to choose when and where I work. While the path is demanding, the benefits can be substantial, | ||
particularly if you are passionate about your field. | ||
|
||
## ⚖️ Legal NLP | ||
|
||
It was really nice talking to many researchers at the poster session. I think this may be a reflection of the surge in | ||
interest in applications to the legal domain over the last years. | ||
|
||
It is a very exciting time for legal NLP now, exemplified by so many legal AI startups being founded. When I started, we | ||
were still very limited by the methods available to us. For example, generative capabilities were very limited with | ||
models often still making grammar mistakes. Additionally, the context width was often limited to 512 tokens, making many | ||
applications in the legal domain, such as document drafting and summarization, infeasible. Finally, while models are | ||
still much worse multilingually compared to English, the situation is improving a lot. I am looking forward to seeing | ||
researchers tackle more and more difficult and real world legal tasks. | ||
|
||
Now I am very much looking forward to my next chapter in industry, more about this another time. | ||
|
||
Thanks to all my amazing mentors and collaborators along the way, especially Veton Matoshi, Ilias Chalkidis, Matthias | ||
Stürmer, Daniel E. Ho, and Garrett Honke. |