lc 3578🥰

body is overwhelmed, but mind is asking for more. was this the doctrine you were conveying, patanjali?

reading an ai research paper is cool i regret not doing it early

i realised, there are a lot of variables and matrix operations to be understood before i can claim that i have finally understood autoencoders in depth

so, the simple autoencoder fails on something more than 2*2 diagonally similar dataset, time to build more than a simple autoencoder

dimensionality reduction is sexy as fuck!

“Each encoder weight column scans across the flattened input to pick out pattern intensities relevant to that neuron.” “Decoder weights determine how a hidden neuron’s signal spreads back into the 2D image blocks.” 1st yet to understand & analyse clearly

Somewhere along the line in Valerio Velardo's video on autoencoders, I found this comment

binary lifting :/

'Differential programming is not just deep learning' starting out the genius book - 'The Elements of Differential Programming' on the suggestion of Nilesh Trivedi for his seminar in Lossfunk

The x-unit neuron(after hidden_encoder layer) is the Gaussian mean which actually guesses instead of specifying.

Since 𝜖 doesn’t depend on the network’s weights, it behaves like a constant for gradient computation. cool!

i'm still a noob dl guy but till now i realised something the math in ai/ml is a different sort of calculus is there but discussed and understood in the form of matrix operations(linear algebra format), i think i'll update this opinion later on

we need a new tool to play with..something whose user interface consists of images block in canva whiteboard and each block has text and you can pick upon the blocks as the background specific context of the next prompt you are going to give to ai

XOR is not greedy friendly in partitioning because it's non-monotonic.

The audacity to say something like this. How many humans have this confidence? And he had the bite force to match the bark decibels. This quote is emblematic of the 19th century Germanic mind. A fascinating study. The Republic of letters. Each German Man of Letters trying to one

few solved questions in dp today more to go:)

started bitmask dp fucked up my mind in just coming till memoisation only, hope to go swiftly into it

solved a bunch of travelling salesman from bitmask dp Code Travelling salesman problem[GfG] LC 847. Shortest Path Visiting All Nodes LC 3149. Find the Minimum Cost Array Permutation LC 1092. Shortest Common Supersequence

studying some rolling hash