Greymatter

Life is suffering , so choose something worth suffering for .

[Greymatter @Greymatteryo]

Will grind more to get everything in my 2026 2nd half wishlist

[Greymatter @Greymatteryo]

@JatinRaghav137 W @JatinRaghav137

[Greymatter @Greymatteryo]

like “his” and “her,” and word-like patterns. It obviously can’t use them coherently yet, but seeing these structures emerge from simple next-character prediction is pretty cool.

[Greymatter @Greymatteryo]

Built a small character-level RNN from scratch and trained it on Immortal Stories by Ruskin Bond — about 253,563 characters of text. Even with such a tiny model and limited training data, it has already started picking up spaces, punctuation, common pronouns

[Greymatter @Greymatteryo]

-Learning OS from cs 162 and dinosaur book book
-gained +103 in div2 contest ...currently 1370 rated
-implementing rnn
-more on transformers and attention
-trying to comfortable with vim and bash

[Greymatter @Greymatteryo]

Building a fully connected neural network clf from scratch for CS231n Assignment 1, then adding Momentum, AdaGrad, RMSProp, Adam, dropout, and normalization. Also finished Lecture 8 on attention and transformer,next up is reading and implementing “Attention Is All You Need.”

[Greymatter @Greymatteryo]

Into the first hour, and it already feels so good. Will try to complete this by tomorrow.

Link - https://t.co/s7FhGV0wiI

[Greymatter @Greymatteryo]

link->https://t.co/e3B8wkzLqP

[Greymatter @Greymatteryo]

-create a two layer fully connected neural network from scratch , writing backpropogation from scratch is a nice way to understand thing better
-currently getting arroung 40% accuracy in training set (with batch size of 64)
-will try to improve this
#machinelearning #deeplearning

[Greymatter @Greymatteryo]

-watched the first lecture video of cs 162 by John Kubiatowicz .... even defining the os was crazy
- almost completed the basic pytorch form pytorch docs
will now complete the cs231n

[Greymatter @Greymatteryo]

checkout the code here👇
https://t.co/uZTU8p4f2S

[Greymatter @Greymatteryo]

- wrote sort of something closer to Alexnet architecture
- used TinyImageNet dataset
-got 69%val accuracy

#MachineLearning #DeepLearning #AlexNet

[Greymatter @Greymatteryo]

- Reached pupil in codeforces next is blue😃.
- will try to complete cs231n lectures by aug10
- almost done with assign 1 will complete the rest by aug10
- will try to start a vision based project ,will share the info
- learnign bash scripting language and a bit about os

[Greymatter @Greymatteryo]

@ujjwalbeast And one non-solvable python implementation question which takes 10-15 mins to understand and another 15 mins to decide that whether u should attempt it or not

[Greymatter @Greymatteryo]

@ujjwalbeast Naa ...just try to login a few times and everything will be sorted and for the coding part there will probably be 2 DSA (easy to mid mostly easy questions) if u r decent u will pass all the test cases except one and this one will scare the hell out of u ...a simple SQL que

[Greymatter @Greymatteryo]

#DeepLearning #MachineLearning

[Greymatter @Greymatteryo]

Went through the AlexNet research paper(first paper for me) will implement it next.
Next will go through the research papers of other architectures and will implement one or two of them.

[Greymatter @Greymatteryo]

Applied cross entropy loss on cifar10
-Softmax function and cross entropy loss
-Gradient descent
just using numpy
currently learning cnn from blogs and videos

#DeepLearning #MachineLearning #AI #PyTorch #NeuralNetworks #Optimization #GradientDescent #BuildInPublic

[Greymatter @Greymatteryo]

Completed applying knn classifier from scratch used cifar10 dataset ,used cross validation to tune the value of k.
#DeepLearning #MachineLearning #AI #PyTorch #NeuralNetworks #Optimization #GradientDescent #Momentum #AdaGrad #RMSProp #Adam #BuildInPublic #LearnInPublic

[Greymatter @Greymatteryo]

#DeepLearning #MachineLearning #AI #PyTorch #NeuralNetworks #Optimization #GradientDescent #Momentum #AdaGrad #RMSProp #Adam #BuildInPublic #LearnInPublic #FromScratch #100DaysOfAI

[Greymatter @Greymatteryo]

Learned about different optimization algorithms today:

Momentum
AdaGrad
RMSProp
Adam

Next, I’ll implement each of them from scratch to understand how they work instead of just using them through a library.