HNHacker News
TopNewBestAskShowJobs

reachtarunhere

578 karma · joined April 13, 2016

reachtarunhere.github.io

GitHub/Twitter - reachtarunhere

submissionscomments
reachtarunhere··on My Apology to Naomi Wu and the Make Community
Also, Dale has some of my respect for realizing what he did was wrong and then reevaluating and trying to make up for it. He could have possibly chosen to let the chaos settle and not apologize. I guess we all make mistakes sometimes. Good that he owned up.
reachtarunhere··on My Apology to Naomi Wu and the Make Community
I second that this was such a great read. I am normally pissed with diversity articles because it looks like they are complaining. I won't say complaining is wrong but it just isn't fun to read. On the other hand, bunny broke the whole issue so well. Amazing read.
reachtarunhere··on Hotshot Tech Founders Face a Dilemma: Lie Low or Live Large
> Also, I find the "years of struggle" trope irritating. Many of people live on beans while working 70-hour weeks and don't end up rich. Plenty of those people work at companies whose executives do get rich.

Your logic is flawed. When I start a company I will be leaving my cushy job that pays damn well and living low in hopes of making it big one day. A founder making a deliberate choice here is very different from unskilled/unlucky/not-qualified (in whatever sense) people living on the beans. They don't have a choice. They are not part of the game.

reachtarunhere··on Understanding LSTM and its diagrams
The paper doesn't speak much. Empirical results.
reachtarunhere··on Understanding LSTM and its diagrams
Interesting paper that compares performance of GRU vs LSTM. http://proceedings.mlr.press/v37/jozefowicz15.pdf

TL;DR - For most tasks other than Language modeling GRU is better.

reachtarunhere··on The jailing of Hong Kong’s young pro-democracy leaders 
I think the parent comment meant the missing context of heavy handedness (HK) in the title.
reachtarunhere··on AllenNLP – An open-source NLP research library, built on PyTorch
I am working with Memory Networks as part of my thesis. If you actually read and implement the FB paper you realise that the system is not half great as the demo shows. It is as bad as the top comment here. Yes we have come far along but frankly the hype is too high.
reachtarunhere··on Autonomous Robots Plant, Tend, and Harvest Entire Crop of Barley
Peter Corke is an amazing guy. Loved his MOOC. Highly recommended if you are just curious about robotics. MATLAB is a sad aspect of it though.
reachtarunhere··on Why I left Japan after 10 years
Is this sarcastic? Just curious.
reachtarunhere··on Jefferies gives IBM Watson a Wall Street reality check
I knew what happens :)
reachtarunhere··on Jefferies gives IBM Watson a Wall Street reality check
Reminds me of https://www.jwz.org/doc/groupware.html
reachtarunhere··on Sony open-sources NNabla – a simple, fast and lightweight NN library
Yes that would be it. Along with Channels and number of images you are processing at a time (batch).
reachtarunhere··on Ask HN: Is it a waste of time to teach yourself data science without a degree?
Can you please elaborate more on the kind of work you do?
reachtarunhere··on Why Isn’t OpenBSD in Google Summer of Code 2017?
Can't really comment on this as I know no such people. The "consulting" companies part is hard to believe given the rules for GSoC. The new payment adjustment taking into account PPP is a welcome move in this regard even though I don't agree on the numbers they have decided.
reachtarunhere··on Why Isn’t OpenBSD in Google Summer of Code 2017?
As a former student I would like to emphasize that it is not about the money. I would have worked on the same project in the summer even if I was not getting paid. There can be various reasons why students won't return or become permanent contributors. For example in my case being a double major in very distant disciplines I do not have the time to contribute when school is on. Then summer I will be working on my thesis at another university which again would leave me little time to make any significant contributions. I am still trying my best to help new people by reviewing requests and making small changes that don't take too much of my time. I plan (hopefully) to get back to contributing on weekends once I am done with school.

For a few friends I have found internships being another reason why they did not go back to their orgs. 5k$ seems like a big amount but it really isn't (even in India!)

Finally, one last reason I can think of is terrible mentors. I have been really lucky to have amazing mentors but I have heard a few horror stories from others.

reachtarunhere··on Open Philanthropy Project awards a grant of $30M to OpenAI
I don't see this as a problem. If I had tons of money and knew someone (by any connection) who was doing amazing work I would definitely chip in.
reachtarunhere··on Distill: a modern machine learning journal
You already are. :) Love your blog.
reachtarunhere··on I've decided to move on to Distill
You are spot on about the impostor syndrome. Colah's confession about being worried bring me some sort of relief given I hold him in really high regard. I am starting out in this area (still an undergrad) and feel really scared at times about what I am doing and at times feel I land up into good internships/projects just by sheer luck. When someone like him who is respected across the community can have these doubts I feel it is completely OK for mere mortals like me have the same issues.

On similar lines, the fact that each of his posts takes 50-200 hours of work is another reality check. I always thought he was writing these on the side in 2-3 hours. This made me believe I could not with years of practice produce such great work. I am still not sure if I would be able to but at least I can now set realistic expectations for myself. Also, knowing it took so much work on his end has made me respect him even more.

reachtarunhere··on Ask HN: Which Berkeley Courses Should I Archive?
Even I don't know what was causing it to be failed. What worked for me was downloading starting from the 3rd video and then downloading the 2nd with the help of a firefox plugin.
reachtarunhere··on Ask HN: Which Berkeley Courses Should I Archive?
I think there is an option in youtube-dl to download the complete channel.
reachtarunhere··on Ask HN: Which Berkeley Courses Should I Archive?
Funny I downloaded exactly the same courses.
reachtarunhere··on The 10,000-hour rule is wrong and perpetuates a cruel myth
I think you and a lot of other people here are missing one very important point. It is not "10K Hours" that does the magic it is "deliberate practice" for 10K hours. Most in the profession of programming can't do 10K hours of deliberate practice because that would mean planning your career and skills like an Olympic Runner. Realizing what you are bad at and leaving everything to work on it. Say you realize that you are great at algorithms but not good at reading huge pieces of code written by others.
reachtarunhere··on Problem Solving with Algorithms and Data Structures using Python (2015)
It is rather short and I would suggest going through all of it. It won't take long. Plus don't forget the exercises.
reachtarunhere··on MIT Media Lab Disobedience Award
How about Alexandra Elbakyan who started sci-hub. Her work is pretty similar to Aaron's. I feel that by defying the rules sci-hub has really helped bring the cause of open-access to everyone's attention and we are all watching it change the system for good.
reachtarunhere··on Want to come to the US? Be prepared to hand over your passwords
Makes me feel RMS is not crazy afterall.
reachtarunhere··on Mark Cuban on Why You Need to Study Artificial Intelligence
Agree on the idea of Classical Machine Learning experts being a little pissed regarding the average joe getting better results with such ease. A lot of these people have put in years of study/work in those techniques and it must be sad to see some form of sudden paradigm shift making it worth less.

On the other hand it is true that DL is not required for many (most?) companies as they don't work with the kind of problems you list (vision, robotics)

reachtarunhere··on 6.S191: Introduction to Deep Learning
Thanks a lot for sharing this. It is great to see that MIT is now catching up with Stanford in terms of Deep Learning classes.
reachtarunhere··on Whiteboard sticker for your laptop
I used to dislike Microsoft till recently. After using OneNote (on Wine) and their keyboard (Natural 4000) I am now a fan.
reachtarunhere··on I cannot continue working on my add-ons
This is how Firefox dies.
reachtarunhere··on Ask HN: What a 2nd tier college student must do to be at par with the best?
I understand the issue of balancing classes and doing projects. If you want to go to grad school GPA will matter a LOT more. Keep it above 8 even at the cost of spending less time on your skills.

If you are sure you do not want to go to grad school (which is very difficult to decide at this point) focus more on projects. Try to get an internship at a good startup and prepare for interviews hard (read DSA). Going to a top employer like Google, MS, Directi will help you get over the second-grade brand. All of these care more about DSA instead of all projects and specialized knowledge in ML and AI. I know preparing for interviews is boring compared to doing projects but take this advice seriously if you want to get hired at one of the places I listed.

← PreviousPage 2 of 4Next →