I know the authors from the blog post quite well. Say what you will about the firm, but one of the authors have been investing in machine learning since 2016, and another has a PhD in CS (including a SIGCOMM test of time award!)
I come from a strong ML background (multiple publications, PhD dropout), I would say that the canon is actually quite good.
I wouldn't go so far. I know the authors quite well, and as someone who has multiple publications in machine learning confeerences (and started a PhD in ML), they know their stuff well.
Yeah, I strongly agree. While Nvidia is working on better hardware (and they're doing a great job at it!), we believe that better training methods should be a big source of efficiency. We've released a new PyTorch library for efficient training at http://github.com/mosaicml/composer.
Our combinations of methods can train CV models ~4x faster to the same accuracy on CV tasks, and ~2x faster to the same perplexity/GLUE score on NLP tasks!
What do you think about the train test discrepancy? ie. will practitioners have to fine-tune Nubia's models on their training dataset in order to evaluate on their test dataset?
Jenks High School class of 2016, MIT class of 2020. Honestly, Jenks was an incredible high school to go to since it was public; having a $20M Math and Science center let me go to places like MIT.
Freshman at MIT here taking this class -- the lectures are actually taught in a flipped classroom format, so I wouldn't imagine they would release the course considering there are no lectures to follow. I could see them releasing problem sets, however.
Anyone have thoughts / ideas on what people need that developers could create for times like these? ie. What are the biggest problems in today's political climate which a developer may be able to solve?
Undergrad at MIT here, thanks for signing! Our community has been fairly troubled by this to say the least, and your outpour of support is incredibly helpful.
MIT 2020 admit here: yes, it is still in the hearts and minds of the newest classes. When discussing some of our idols, one student and I both agreed upon Aaron Swartz. He may be gone, but never forgotten and the new students look up to his legacy.
I'm a high school Senior (17) and I'm currently taking it. It is ridiculously understandable and I often see myself yearning for more. But I've also had Calc 3 + Linear Algebra by now, so its understandable that not everyone would get it. The intuition is simple, however.
Hi. Seventeen-year old high schooler here, I love this idea.
(@sama, anyone else behind YCR):
All my life I had wanted to go into industry, follow that classic Steve Jobs, Elon Musk, Peter Thiel dream that many CS High Schoolers go into.
But recently, I got into research. The CollegeBoard opened up a class called AP Research I am now in, and I spent my last summer researching Machine Learning at my local private university. I was surprised by how much I loved it, the math behind it fascinates me (context: am currently taking Adv. Diff Eq and Linear Algebra my Senior Year.)
I am now trying my own Deep Learning algorithms and working on my paper. I have always been conflicted between industry and academia, since all of my fellow HS CS friends just want to found companies in industry. Things such as YCR make research a bigger dream for youth, whom I would currently argue are too under-exposed to it. It just doesn't seem as glamorous to them.
Thank you for YCR, it reminds me of the cool things research can do. I can't wait to hopefully apply one day.
I disagree. He seems level-headed and focused enough that he could take this as motivation. If he was arrogant, he would have capitalized on it, created a social media account and a gofundme. He seems humble to me.
Source: 17-year old who faced similar media attention, it did inflate my ego a bit despite attempts to remain humble.
I know the authors from the blog post quite well. Say what you will about the firm, but one of the authors have been investing in machine learning since 2016, and another has a PhD in CS (including a SIGCOMM test of time award!)
I come from a strong ML background (multiple publications, PhD dropout), I would say that the canon is actually quite good.