@Pivot2Centre I have never used apple pay, but use Google wallet. Does it require to load credit card or debit card or any kind of card, if yes then I don't see a fuss around apple pay. Upi allows account to account transfers without any card or anything.
@Neetivaan These are just cheap handouts that the govt is giving out for vote bank politics, In realty, most SC/ST folks will be stuck in unending paper work. Local cash injections could have been better.
@kejimao Democracy is not a feature, it's a necessity for India to avoid civil wars and balkanization. Democracy can't stop corruption, nor can it fix feudalism. People sleep better at night knowing they can remove their leader, no matter how much bribe they have to pay in daily life.
@original_ngv Being eligible doesn't mean getting the job in placements, companies have every applicants full profile and can choose not to hire you, and the reason could be anything.
@adityarjn_ True, my friend who is currently building his startup in the US convinced his parents that BITS is equally good and only prepped for bitsat. He left the whole JEE grind and worked on ML projects with college kids on the side. This helped him a lot in Phd applications.
@coolcoder56 Compulsion to include everything in one page. Its better to give a github website link. It's easy and cheap. No one send hard copy CV these days.
@jsensarma In kgp, dep. highest ranker was called as "kholu" (opener, someone who opened the dep.) and the last ranker was called "dhakkan" (lid, someone who closed the dep.
My latest pastime is to watch theists using 'logic' to disprove atheists. Someone was arguing that you will never find Rowling in Harry Potter book, because creator can't be seen in the creation. Well, I don't think harry was chanting Rowling hymns to win quidditch.
@stuartbuck1 That's so true. Coming from a stats background, the jargon kept me away from ML for a long time, only to find out later that it's just the application of Least squares and MLE. Now I directly try to read the equations of any new fancy ML algo, it makes it easier to understand.
I believe that LSTM based architectures (constant hidden state, constant flops / step) will crush transformers as we know them in 2-3 years.
we will laugh about KV Caches and how dumb we were..
to achieve this, the LSTM will need to have a large external memory bank with tool-use like Mem0 / RLMs / RAG to read/write memories into a db, but this will be more like us writing things in a notebook vs. being forced to memorize everything in a KV-cache.
this will require the model NOT to memorize facts inside the weights, but ONLY learn functions to manipulate, write, and retrieve facts, which of course is the better system.
if I am vibe coding, why do I need to know who Kevin Costner is, and how many people are in paris, and have all of those facts in my weight file.. seems like a waste.
and the whole model + fixed hidden state can sit in SRAM and the "memories" will sit in cpu ram (or disk?), as it will be addressed in small amounts and perhaps even regex-able vs. big heavy knn mat mul.
what is distinct about this structure is the model itself will need to learn a saliency function of what to store in short term, long term, vs. external memory such that it can be easily retrieved, and what to forget entirely... without ANY human example traces. just via RL.
the biggest implication as it comes to chips and DRAM is that it will prefer chips w large SRAM and a huge amount of SMs and likely little DRAM if any at all.
so chips that support this will be good investments.
does anyone else see this yet?
@late_bloomer_7 That's not completely true. In a moving average fashion, the difficulty of questions have only increased. I remember solving the same optics question in multiple choice which had come as a 15 mark subjective some years back. The aptitude to clear the exam has increased.