gabrielilharco
(Gabriel Ilharco Magalhaes)
November 14, 2017, 4:39am
1
Neural Information Processing Systems (NIPS) is one of the most prestigious AI research conferences in the world. Since most of the recent papers don’t have open-source implementations, they launched a challenge to incentive people to create and make implementations available for the rest of the world.
Link for the challenge: NIPS Paper Implementation Challenge
It would be really nice if we could gather a couple of interested people to team up and implement some papers! They also give access to GPUs and mentorship =)
13 Likes
KevinB
(Kevin Bird)
November 14, 2017, 4:42am
2
What is the actual challenge? I don’t see what the actual challenge is. Do you just make your own challenge for this and write a paper on it?
gabrielilharco
(Gabriel Ilharco Magalhaes)
November 14, 2017, 4:44am
3
The challenge is just to create open-source, well documented code for an accepted NIPS paper, replicating the results of the paper.
1 Like
jamesrequa
(James Requa)
November 14, 2017, 4:45am
4
I would be interested in participating in this! Looks like a great initiative
1 Like
KevinB
(Kevin Bird)
November 14, 2017, 4:45am
5
@harveyslash just did a cool implementation. He isn’t in v2 of this but he would probably be interested if you aren’t trying to limit it to just v2 students. I am also interested.
I am nearly done implementing this paper :
Its on pytorch, and I am getting these results :
[monaout]
[Unknown-1]
As you can see, this paper produces amazing results(even though mine arent like the paper, its still pretty good)
I really want to make this paper very popular, write a couple blog posts about it. I have some new ideas that may improve the speed of this.
All this requires my results to closely match the paper’s results, which are:
[40 PM]
[01 PM]
If anybody is interested …
1 Like
A_TF57
(Ankit Goila)
November 14, 2017, 4:45am
6
I would love to participate as well!
gabrielilharco
(Gabriel Ilharco Magalhaes)
November 14, 2017, 4:48am
7
You’re right, let me change this from p1v2 to Applications
gabrielilharco
(Gabriel Ilharco Magalhaes)
November 14, 2017, 4:51am
8
PS: deadline is Jan 31, 2018
Thanks @KevinB for showing this thread.
short bio :
I am seriously looking for mentorship , and have decent experience with deep learning/coding.
Have finished part 1 and part 2.
I also blog at medium.com/@harshsayshi
If you want to use my implementation , work on it, improve etc, Ill be more than happy to collaborate.
If there is another paper (I know another paper that uses Patchmatch , very new, very awesome, we can work on that)
if possible, include my name also
3 Likes
dinkar
(Dinkar Juyal)
November 14, 2017, 5:36am
10
Sounds great… I am in for participating!
anandsaha
(Anand Saha)
November 14, 2017, 5:59am
11
This is a fantastic opportunity, though it seems there is stiff competition
Thanks for your interest in implementing a NIPS paper! Could you tell us a little about models you have built before, including Github links if any? (This helps with your application as we are getting an overwhelming response from participants around the world!)
If anybody wants to collaborate with me, say so.
I will apply myself if nobody responds within 2 or 3 days.
2 Likes
anandsaha
(Anand Saha)
November 14, 2017, 6:11am
13
Consider me @harveyslash
I will sync with you offline.
1 Like
sanjeev.b
(Sanjeev Bhalla)
November 14, 2017, 6:12am
14
I am very interested.
Is the list of papers accepted by NIPS available somewhere so we can form groups based on topic of interest?
sanjeev.b
(Sanjeev Bhalla)
November 14, 2017, 6:13am
15
anandsaha
(Anand Saha)
November 14, 2017, 6:14am
16
I tried to search for it, but it seems it will get decided only once you are accepted
anandsaha
(Anand Saha)
November 14, 2017, 6:15am
17
Ok there you go. I was searching in the applications portal.
ravivijay
(Ravi Sekar Vijayakumar)
November 14, 2017, 6:37am
19
I am very interested about this too!
binga
(Phani Srikanth)
November 14, 2017, 7:14am
20
I’m interested in contributing.
I am currently doing v2 part-1 and I can bring good amount of GPU resources and commitment from my end. Let me know how do I join your team.
EDIT: Submitted the application, fingers crossed !