grammar-lab

[grammar-lab]

My objective was to solve all those PESKY syntax problems before the mid-may, I think it is do-able, God help me

[grammar-lab]

Mid may is approaching,

Today was a bad day, but first, a small recap

I solved my NP problem, and got up to 90% accuracy, but when I switched to sentences, nothing seemed to work.

The accuracy shot down to 0 percent, and I managed to get it up to 18% somehow, but I felt like something was wrong...

After trying I thought I noticed a problem that made what I was trying impossible, I spent a good week thinking about it and finally messaged my supervisor to ask about it, showing the contradiction that I found.

TUrns out I didnn't read the paper correctly, fuck me

fuck me. I can't believe I've been so shite, I now know where to go, but DAMN do I feel stupid, there was a small difference in notation that I didn't even notice that changes everything. THIS IS horrible, when I realized this I felt dizzy like, nauseous and wanting to cry at the same time, terrible... Thank God my supervisor didn't make a big deal out of it, but UGHHHH this is a great blow at my ego

I am shite, but hopefully I'll get better

[grammar-lab]

Wow we're already end of april

Brief recap of what happened these past two weeks:

Finally got ACGTK to work (lmao it wasn't that complicated at all)

got a work computer (and thank God, I wouldn't've been able to do any work without it since my computer was SO SLOW to run linux on a VM)

I've worked on a constituent Parser to extract NPs

I scraped multiple websites for dependency relations

I generated a bunch of ACG rules (and they're not great, believe me)

I generated an ACG script that works 70% of the time (this is far from good :( (( )

and now I'm a bit depressed after that presentation haha :)

But who knows ? I might be able to solve this freaking SYNTAX PROBLEM and finally start my SEMANTIC WORK

ahahahah :-)

[grammar-lab]

08/04: I finished the whole "formula thingy" on a saturday after working on it for like 10 hours straight.

09/04: That sunday I went to the botanical garden again with NLP friends

10/04: Today is a holiday ! I almost went to the lab but after asking in the group chat, they told me everything was closed. So I went to the parc with friends, and we played frisbee. it was nice, and I also discovered a place that does MASSIVE sandwiches that are cheap ! So my ftour was a gigantic french tacos (it was unbelievable how big it was)

11/04: I've made my results way prettier, and continued working on the lambda interpreter, I have worked until 4am without much results :/ I got better ideas, but still no results.

12/04: I presented my work to my supervisor, it went well. When I mumbled that I was working on my own lambda calculus interpreter he didn't overreact and just told me that it would take too much time.

I also discovered "FraCaS" today, it is the dataset that we will use for evaluation. (it's a bunch of premises, and a conclusion)

[grammar-lab]

wow ! It's already thursday !

Brief summary of every day:

03/04:

It's monday ! So we have a meeting with the team ! I lied and pretended that I was working on a functionality to turn a dependency grammar into a formal expression. (But I started it that day, so it's fine ! I spent the other week working on the lambda interpreter)

04/04:

Hurray ! after long hours of work, I finally have something that works ! It's the first time something I do works perfectly

05/04

Turns out it didn't work, I worked from home today and found a new solution.

06/04:

Today I had a meeting with all the new people of the lab, where they gave us security instructions and stuff, and in the afternoon I went to a protest with people from the lab. I'll work a little bit more

[grammar-lab]

WEEKEND !

On friday night, I went to a Bengali restaurant with friends from the Master's degree, and we spent hours talking, I felt great for the first time in a while.

On saturday, we met again to play some boardgames, and then went to an "all you can eat" buffet ! Fantastic times !!

On Sunday, I was supposed to go to an escape game (paid by the lab hehe) but it was cancelled, and now I am working on my lambda calculus interpreter again, feeling good and fresh !

[grammar-lab]

31/03

Today I felt so bad, I just stayed working from home, well to be honest, I took some rest. And stuff didn't feel right. I even called a support line because I was so panicked about not working.

The person was very nice to me and gave me good advice. (I'm not sure I'll apply it though.

I spent the rest of the day relaxing.

[grammar-lab]

30/03

That day, I mostly spent my time trying to work on my lambda calculus interpreter.

I think I finally managed to have something that works in a very basic use-case. I still need to test it a lot and debug the errors that I will find.

It was very hard to get stuff done, and I got very demotivated.

[grammar-lab]

29/03

Even though it's not the end of the day, I'm writing this now, because my day is pretty much over at this point.

Today I woke up late, but in great GREAT shape ! It has been a while since I felt this great. By the time I got dressed and went to the lab to grab my dinner (I put the lunch in tupperwares and only eat them at Ftour) and went back home, it was already 2pm and I hadn't started working ! (I also had to go to the Poste and the Accueil of the building, for personal reasons, and that administrative stuff also took a little while.)

And now it's already 2:15 and I have the TOEIC exam that starts at 4:30 (so I have to leave home around 4pm) so I cannot work a lot today. But I'll do what I can :-)

[grammar-lab]

28/03

Because of a small incident that happened yesterday (that I will not mention here) I decided to stay and work from home today.

And, I felt way better, mostly because there were ideas and things I was too afraid to do in the lab (Like work on my own lambda calculus interpreter, when they have ACGtk) But in the end, I was very very productive and laid a solid foundation for my program. I also re-read the computational semantics course in its entirety and noticed that there were a few small sections here & there that I did not completely understand.

I also felt more relaxed as I took breaks whenever I wanted and was able to simply watch videos and laugh during those breaks (this couldn't happen in the lab, we never laugh :( )

Anyway, it was a great day, and I've decided to stay and work from home tomorrow as well.

[grammar-lab]

27/03

Monday.

Every monday, we have a team meeting. It was my first meeting. So there was a lot of people, and the meeting was simple, everyone had to say what they did during the week. It didn't matter if you were an intern, a teacher, a researcher or a PhD student. I guess it helped with my anxiety, because I saw that I was doing more than enough.

In the afternoon, I worked on De Groote's original paper again (Towards ACGs) but I was very tired because of Ramadhan.

After that, I went on a small walk with Joe, and shortly after that I went home.

Sok Long and Joe were in the bus, and they seemed to be doing fine. I got home around 6:30 pm and it felt very very early to me.

I guess I tried to relax the rest of the day. Oh, and there was an escape game thing with the research team and I'm invited, but I noticed that I had an exam the same day (TOEIC exam) so I had to cancel even though I said yes at first.

[grammar-lab]

weekend !

I went to the Botanical Garden with my loved one, and we ate at an italian restaurant ! Good times 🥰

[grammar-lab]

24/04

I'm writing this the 27/03.

That day, everyone came to the lab late, because of last night's party. We quickly went to eat (and in my case, I filled my tupperware for Sha9an al Fatr) we were so numerous that we filled a full table of 12. And there still was some people that wanted to come sit at the table.

Since my M2 friends were around, I went and joined them on another table, leaving the sit to another person.

It was nice to sit and talk with them again, even though it felt different.. In the afternoon, Valentin offered to show me a few explanations about that whole ACGTK thing, and thanks to his help, I got ressources (and a very important paper to read to understand the theory behind it all) I spent the rest of the day trying to read it (there was some complicated maths in it, and some new symbols, notably one called that he called "the lollipop"

And I like the name, I think it fits hahah

A bit before 6pm I played Ping Pong with Joe and it was nice, and oddly enough, we switched bus stops today (he left at Médreville, and I left at the Gare to join my girl who was coming to town)

I didn't work this weekend. (But I went to the Jardin Botanique, fun times, but I won't elaborate, I'm writing this whole thing during working hours.)

[grammar-lab]

23/03

I'm writing this and the next day on the 27/03, my memories might be a bit blurry.

That day, I first went into the interns room to work in a calm place (I was still not used to the normal room of the team) I got the final grades for the Master and so I was bummed out and did procrastinate a little while. Then a PhD student invited me to the main SEMAGRAMME room. That same day there was a huge protest in France, so I went there with the members of the lab who were protesting.

It was a very pleasant time, where I got to know the members of the team better. I am very grateful for that opportunity. Then, at night, I went to the game night, where we played a variant of the "Loup Garou" I was very hungry since it was the first day of Ramadhan, and I got to break fast in the middle of the explanations of the game. (With a tupperware that I filled during lunchtime). Once again, it was a good bonding moment, and by the end of the night, I was much more confortable with the rest of the team.

[grammar-lab]

22/03

Today, I woke up late, and went to the lab around 11 AM, the intern room was occupied so I went to the other one (with the PhD students and engineers) this is the real "SEMAGRAMME" room, everyone here is working on something similar to me in a way or another I believe. It's also more intimidating because they somehow look older.

They were nice enough to invite me to eat with them, one of my M2 TD teachers was with them, it was very awkward, I didn't know whether I should "vouvoyer" her or "tutoyer" her (for non-french speakers, it is basically a marker of respect) so I asked her, and everyone laughed at that haha I didn't feel bad, and I understood how it could've been funny for them.

While eating, I met another intern that was working on Bias in Large language models. The ethics and Biases of AI are a very interesting subject, so I hope to learn more on her subject.

In the afternoon, I tried working on that ACGtk thing, but then I understood that is was being modified in real time, so I studied that "Lambda notebook", but it also didn't fit my expectations. I'm considering to make my own lambda interpreter and beta-reduction algorithm, it cannot be that complicated !

I also started reading the other paper that Pr. De Groote gave me. And I see how it inspired him, and what changes he brought to that already good system.

One of the PhD students invited me to a game night tomorrow, which is great since I never got the chance to interact with them on a more casual way.

Also, tonight I'll have to wake up before fajr for the s7our, indeed Ramadan starts tomorrow.

[grammar-lab]

21/03

Today was the GARDENT FEST, in honour of Claire Gardent who got the Silver medal of CNRS (one of the highest awards one can get in research in France)

The whole day was dedicated to her, with conferences, and buffets ! The morning I struggled with installing ACGTK, then, during noon I ate with some new friends, we then went to the music room and played piano and after that, some old teachers came in to practice for their jazz band. (So we went to play ping pong !)

In the afternoon, I worked one hour, and then went to the Gardent fest, there were many speakers, notably Mark Steedman with whom I got the chance to speak ! Shashi Narayan from the Google Brain team, and a Parisian professor lol.

It was very stressful for me as I am not used to these kinds of events.

I'm considering putting more media on this blog (pictures, videos, maybe audios ??) just to make it more lively, don't hesitate to ask me if you have ideas.

[grammar-lab]

20/03

Today, I had issues with installing the freaking lambda notebook thing, I spent like 2 hours before managing to make it word,

And I spent the afternoon understanding it before my tutor called me and told me that there was a tool that I was going to use :/ Can't believe this ! Let's hope this afternoon was not wasted and this knowledge will serve me someday somehow.

Yeah, let's stay positive :-)