initlayers @initlayers

machine learning research • LLMs • UG • pushing my way to the next big thing • trying to understand simple stuff 3228 BCE Joined June 2025

Tweets

708
Followers

248
Following

420
Likes

2K

initlayers @initlayers

9 hours ago

We have to refine our PPT and build a rule engine for the submission on 30th. We are also taking part in two more hackathons in this period. Our second SIH submission should before 17th next month. I am working on a minor group project as well 😐

0 0 4 58 0

initlayers @initlayers

a day ago

Internal hackathon concluded. We cleared this round. Judge liked our presentation very much, he told us that with such an idea, we can directly contact the organisation as well. However he wants us to make our PPT more appealing. So there's some work to do now. I am relieved 😌

2 0 8 176 0

initlayers @initlayers

3 days ago

Guess who didn't sleep all night just for making a PPT. There you go, it's me 🤥 I have my internal hackathon tomorrow (we'll technically today), wish me luck!!

14 0 42 992 2

Download Image

initlayers @initlayers

3 days ago

24 September 2025 I studied 2D filters in CNNs today. How the filters are applied to the depth of the images instead of individual matrices. I tried to find an isolated pixel using custom filters. It was a fun ride.

2 0 18 356 1

Download Image

initlayers @initlayers

4 days ago

23 September 2025 > DL: Filters and Convolutions in 1D image data > MLOps: Introduction to MLOps, need for MLOps > Statistics: Describing datasets, different plotting techniques > MLOps: ML project life cycle, Software development life cycle I will be consistent with my work 🤞🏻

1 0 18 238 0

Download Image

initlayers @initlayers

5 days ago

Some classic papers on the evolution of CNNS (Prereq: Basic Deep Learning): Neocognitron (1980): rctn.org/bruno/public/p… LeNet (1998): yann.lecun.com/exdb/publis/pd… AlexNet (2012): proceedings.neurips.cc/paper_files/pa… Great starting point if you want to see how CNNs evolved over time.

0 1 6 271 4

initlayers @initlayers

5 days ago

22 September 2025 > Woke up at 5:30 AM > Walk + Regular Exercise > Studied basic CNN and history > Studied basic Stats - Unit 1 > Worked on SIH PS, finalized 2 > Couldn't start MLOps, will attack tomorrow It's been a good day, hope this works for 6 more days atleast

0 0 11 175 0

Download Image

initlayers @initlayers

5 days ago

I am starting this book today, I don't know if it's good, I just want to start and learn something ASAP instead of waiting for the right time or resource

0 0 9 152 0

Download Image

initlayers @initlayers

6 days ago

If you're doing a master's in a tier-3 college in India, you're just delaying your unemployment. There, I said it.

1 0 7 167 0

initlayers @initlayers

6 days ago

It feels a bit embarassing to admit I'm making a new plan again, after failing through on so many before. I don't know why I keep procrastinating. Just hoping I don't lose the willpower to start this time.

3 0 11 231 1

initlayers @initlayers

a week ago

20 Sept 2025 I designed a basic neural network for classification using TensorFlow, it felt a lot different to Pytorch. I studied basics of vector algebra and span of two vectors, spent most of the time shortlisting problem statements for SIH

0 0 14 215 0

initlayers @initlayers

a week ago

Okay so this happened as well 😮

initlayers @initlayers

a week ago

Okay so this happened as well 😮 https://t.co/dyyx5IHjFv

2 0 14 730 0

Download Image

1 0 16 366 0

Download Image

initlayers @initlayers

a week ago

Good morning 🌞 The sun's aura is too high that I couldn't see him eye to eye even from 150 million kilometers

0 0 14 173 0

initlayers @initlayers

a week ago

We found 15 interesting problems, but are inclined towards one particular PS from Government of Kerala. We have a similar project from Ministry of Steel as well, we are confused which to choose. I created the outline for these projects but I want to explore the rest of the PS too

0 0 4 145 0

initlayers @initlayers

a week ago

I am home today. It's SIH day. I'll spend the entire day going through all the problem statements. My team will meet me in a few hours, and we will finalize something today.

2 0 11 217 0

Download Image

initlayers @initlayers

a week ago

Two perspectives exist: A neural network can be a complex expressive function, which takes an input to produce output. Or it could be a feature transformation technique which takes the input to very high dimensions and brings it back to a single or a fixed number of dimensions.