search

nextml / NEXT

NEXT is a machine learning system that runs in the cloud and makes it easy to develop, evaluate, and apply active learning in the real-world. Ask better questions. Get better results. Faster. Automated.
http://nextml.org
Apache License 2.0
160 stars 54 forks source link

I get this error and do not know how to resolve it. I am working on the Windows 11 WSL2 subsystem. #233

Open zsong96wisc opened 2 years ago

zsong96wisc commented 2 years ago
image
stsievert commented 2 years ago

I'm not sure why that error is happening. I know Linux is supported – NEXT should run there.

The more pertinent question: why do you want to use NEXT? Who are you collaborating with?

zsong96wisc commented 2 years ago

I am also from UW Madison. I am working on a marketing project about a donation platform. We want to use the NEXT platform to simulate active learning on recommending donation projects to new users. I am working with Prof. Qing at Bussiness School. For this issue, I tried with different versions of Ubuntu as WSL2 systems and it occurred. The command I run is "./docker_up.sh", but I got an error message 'env: bash\r: No such file or directory'. I typed command 'sed $'s/\r$//' ./docker_up.sh > ./docker_up.Unix.sh' and run './docker_up.Unix.sh'.

stsievert commented 2 years ago

What type of query do you want to ask to the user? Do you have features for each user?

By "NEXT should run on Linux," I meant that it should work on a base installation of Ubuntu/etc. It might work on WSL2 as well, but I would verify with Ubuntu/etc first. Let's figure out your use case first.

zsong96wisc commented 2 years ago

I was trying to learn how NEXT works. I just want to set it up and run the example cases in your example folder. Then I faced this issue when I was installing it.

zsong96wisc commented 2 years ago

I tried to add command: tail -F anything to the minionworker in yaml file and the server works. It looks fine but when I was running the strange_fruit_triplet experiment I got this error

image
stsievert commented 2 years ago

I was trying to learn how NEXT works.

For that task, I'd most recommend watching the PyData or SciPy talk mentioned at https://github.com/nextml/next//wiki. It briefly mentions some of the interface – and if you're launching Docker machines by yourself, I think you're ready.

Launching with the AMI should work (but may not): https://github.com/nextml/next//wiki/AMI-launch

I am working on a marketing project about a donation platform. We want to use the NEXT platform to simulate active learning on recommending donation projects to new users. I am working with Prof. Qing at Bussiness School.

What information do you have about the users and donation tasks? What does "success" look like – are you trying to increase revenue with A/B testing, or are you trying to increase click-through rate for online ads? I ask these questions because I suspect there will be more interest, especially given Prof. Qing's research interests.

stsievert commented 2 years ago

Any update @zsong96wisc? Have you launched successfully?

Prof. Robert Nowak is a lot more involved in this project than me (I've since graduated from UW), and has been involved in paper we list at http://nextml.org/papers. I think he'd be interested in this research project, and will be able to provide some good problem-specific insights and/or help. I recommend reaching out to him.

zsong96wisc commented 2 years ago

Hi there, so I am trying to launch it via Amazon EC2 AMI and I can not find the AMI in the Oregon region server. Do you have any idea where I can find it?

image
stsievert commented 2 years ago

We've had some trouble with Rob's Amazon AWS account. I've emailed UW's public cloud support team, and cc'd Rob. I'll keep you updated. In the meantime, I'd suggest using a macOS or Linux machine (possibly through Amazon EC2) to get up and running... I haven't seen the error in https://github.com/nextml/NEXT/issues/233#issuecomment-1188432154 before, and these commands worked for me on macOS 10.14.6:

$ cd ~/Developer/nextml/NEXT/
$ cd local
$ ./docker_up.sh
$ # visited http://localhost:8000/home, successfully launched strange_fruit_triplet

What question do you want to ask humans? (e.g., "how good is item A?", "is item A or B better?", "is item A or B more similar to item H?") Do you have context for each user? There are some active learning software platforms that could be useful. If you're interested in asking about three items ("triplet queries" above), I would look at Salmon: https://docs.stsievert.com/salmon/ (and also talk to Prof. Nowak; I'm not sure the active ML algorithms in there are suited for marketing/etc).

stsievert commented 2 years ago

Any update here?