Oh btw, I am applying to Ph.D. programs this cycle with a goal to personalise LLMs and help developers write amazing (correct) code. In case you are hiring please look at my apps! I was a pre doctoral researcher at @Microsoft PROSE where I worked closely with @Mukul_Singh_1

@karkarlyly Real

Yeah it’s ‘fair use’ since ALL three used ethically sourced data, which was not scraped. The data owners were asked and informed!!

I have been a bit confused about NLL vs Cross-entropy This video by @HermanKamper proves that both are equivalent.

@peakidiot Congrats king!!!!

RT @josemorgado: UNBELIEVABLE! 19 year old Learner Tien gets an INCREDIBLE win over world #5 and three time runner up Daniil Medvedev 6-3,…

@sidposting The food looks so gooddddd

TPUs are really something, huh?

I recently learned about GLiNER, it's pretty cool. I also love to see a paper repo with these many stars.

@3blue1brown Linear Algebra is the best thing I have ever spent my time on. Hands down the most goated Linear Algebra playlist.

@Aflah02101 Congrats! This is amazing

RT @novasarc01: @teortaxesTex i'll tell what's really happening: 1) there is no strong ecosystem of research here even in the so called top…

RT @whateventeestaa: Apparently I'm a top 1% tipper on Zomato.. GUYS TIP YOUR DELIVERY PARTNERS MORE I DON'T EVEN TIP THAT MUCH

RT @vincentmvdm: great software makes you feel powerful and @modal_labs is a shining example

@prajdabre1 Raj coded

How did they even come up with RoPE

@priyanshu42g But I do see your point as well :)

@priyanshu42g Well it has been trained to start every line of the answer with H,E,L,L,O and seems like it’s just answering the question here perfectly, which I expect for a model like GPT 4o to do.

@akarsh1_u It kinda did, but still pretty clueless, especially on a server less compute

@hornof Umm, I already know about a few of these; the issue is I am running on a serverless system and have already tried reducing the batch size. I am literally using an A-100 for a small encoder model.