THIS SHIT IS SO BEAUTIFUL 😭 > my neural network i made from scratch is connected with the my autograd is LEARNING after each EPOCH in MNIST dataset. > you can see it getting better and better predictions after each epoch, this is not fully done btw, just a first few tests but

Just when i made the post I got bombarded with other work At least, now I am free for the weekends

Took me so long to debug the flow of the program But at last I did it :)

Rectangle done too

Just an interpolated colored triangle generates 129465 fragments Honestly, I am having more fun than I expected

Finally, after struggling more than a hour to fix the texture and fragment shader issues It finally got rendered TT

I am back after a week of break I was a lil sick + didn't feel like doing anything Have midterms from Wednesday Still don't feel like doing anything today Will try to get back on track from tomorrow (hopefully)

I am back (once again)

Super tiring day LET ME REST 🙏 (SOMEONE SAVE ME FROM THIS)

Lesss goooo Congrats twinnn, it's just the beginning Keeping winning ;⁠)

Omgg look who's here with me Harkirat Singh

Writing tests for a Solana SC Faced a silly issue To generate a mint, we need to pass a payer and authority, both are of type Signer and instead of sending the whole keypair object, I was only sending the publickey which was giving undefined At last it worked as intended

Everything workss

Going back home after 4 months Finally time to eat good for the next 10 days :⁠-⁠P

Learned few intresting concepts today Zk (Zero knowledge) Protocol Zk-Proofs, Zk-SNARKs, Zk-STARKs, Circom, and Zk-Circuits Also learned how a on-chain prediction market works (Polymarket, which runs on polygon L2 chain) Interesting stuff will deep dive more into it

Also got to know about Logarithmic Market Scoring Rule (LMSR) AMM Which uses this formula to balance the liquidity pool Cost = b * ln( Σ e^(qi / b) ) b: liquidity parameter qi: number of shares currently in circulation e: base of natural logarithm. ln: natural logarithm Σ: sum

How I Implemented Tensor Operations From Scratch In My Deep Learning Framework: (link in comments)

Read the Solana Whitepaper today It covers 4 main concepts 1) Proof of History 2) Proof of Stake Consensus 3) Proof of Replication 4) System Architecture of Solana It's a good read, only 32 pages give it a go if you are a Solana developer

no longer unemployed.