About the 🚂 Flatland Workshop category

(Replace this first paragraph with a brief description of your new category. This guidance will appear in the category selection area, so try to keep it below 200 characters.)

Use the following paragraphs for a longer description, or to establish category guidelines or rules:

• Why should people use this category? What is it for?

• How exactly is this different than the other categories we already have?

• What should topics in this category generally contain?

• Do we need this category? Can we merge with another category, or subcategory?

View post 0

About the Flatland Workshop-b33f21 category

(Replace this first paragraph with a brief description of your new category. This guidance will appear in the category selection area, so try to keep it below 200 characters.)

Use the following paragraphs for a longer description, or to establish category guidelines or rules:

• Why should people use this category? What is it for?

• How exactly is this different than the other categories we already have?

• What should topics in this category generally contain?

• Do we need this category? Can we merge with another category, or subcategory?

View post 0

About the Flatland Workshop category

(Replace this first paragraph with a brief description of your new category. This guidance will appear in the category selection area, so try to keep it below 200 characters.)

Use the following paragraphs for a longer description, or to establish category guidelines or rules:

• Why should people use this category? What is it for?

• How exactly is this different than the other categories we already have?

• What should topics in this category generally contain?

• Do we need this category? Can we merge with another category, or subcategory?

View post 0

Editor.py missing package jpy_canvas

Hello @pf1, @hagrid67 is the authority on this matter!

View post 0

Metadata.csv is missing from the latest testset

Hello Sahil, which archive are you referring to? I see the metadata.csv in both test-neurips2020-round2-v0.tar.gz and test-neurips2020-round1-v1.tar.gz from the Resources page.

edit: This was an error in the doc that we’ve now corrected, thanks for the report!

View post 0

📑 The Flatland paper is out!

A first version of the Flatland competition paper is out on arXiv!

Flatland-RL : Multi-Agent Reinforcement Learning on Trains

@misc{mohanty2020flatlandrl,
      title={Flatland-RL : Multi-Agent Reinforcement Learning on Trains}, 
      author={Sharada Mohanty and Erik Nygren and Florian Laurent and Manuel Schneider and Christian Scheller and Nilabha Bhattacharya and Jeremy Watson and Adrian Egli and Christian Eichenberger and Christian Baumberger and Gereon Vienken and Irene Sturm and Guillaume Sartoretti and Giacomo Spigler},
      year={2020},
      eprint={2012.05893},
      archivePrefix={arXiv},
      primaryClass={cs.AI}
}

A separate paper will cover competition results, more information about it coming soon.

View post 2

📹 Flatland NeurIPS Townhall

Hey @jakob_stigenberg, yes the event was recorded, we will release the video publicly soon!

View post 4

🏆 NeurIPS 2020 Flatland Winners

View post 0

🏆 NeurIPS 2020 Flatland Winners

View post 0

🏆 NeurIPS 2020 Flatland Winners

View post 0

🏆 NeurIPS 2020 Flatland Winners

Round 2 has finished, concluding the NeurIPS 2020 edition!

We have received over 2’000 submissions from over 700 participants coming from over 51 countries! We have been thrilled by your contributions and enthusiasm.

And the winners are…

RL Track

• JBR_HSE, 214.15 points
• ai-team-flatland, 181.497 points
• SSG-RL, 143.985 points

Non-RL Track

• An_Old_Driver, 297.507 points
• MasterFlatland, 296.347 points
• vetrov_andrew, 273.339 points

Congratulations to all of them! The race was intense

Community Prizes

In parallel to the competition, participants also competed for Community Prizes, contributing to the Flatland ecosystem by building tools and implementing ideas that will help other participants.

And the winners are:

• Shivam Agarwal for his tool Visual Analytics of Flatland Episodes
• Samuel López for his Flatland Symphony
• Nilabha Bhattacharya, ex aequo, for his RLlib Colab Notebooks
• Adrian Egli, ex aequo, for his FastTreeObs observation builder

The submissions were few, but they were very impressive and inspiring. We warmly congratulate all four winners.

What’s next?

This competition may be finished, but Flatland is just getting started!

We are happy to announce that not only will we apply for a new competition at NeurIPS next year, but we are also organizing another challenge before that in the context of the AMLD 2021 conference, which will take place in Lausanne, Switzerland.

Beyond competitions, Flatland will also become a long-term benchmark, accepting submissions on a permanent basis. Our goal is to measure the progress around the real-world problem of train scheduling, and progress doesn’t happen only twice a year!

For this purpose, a new “post-challenge” round will be open soon and will stay open until next competition to allow the community to keep searching.

Thank you all again for your participation, and see you soon on the next leaderboard!

View post 7

The official results will be announced at NeurIPS

Hello everyone,

The review process has been long and complex!

We will ultimately present the official results during the NeurIPS conference, as part of the competition track.

View post 1

🚇 5 days left! Final tips & tricks for the home stretch

Hey @harshadkhadilkar, the review process took more time than expected, as it was challenging to establish if each solution was using RL in a meaningful way.

We will announce the official results at NeurIPS.

View post 1

Will the leaderboard remain open?

Hello @guillaume_alleon,

Yes we plan to keep the submissions open! For now new submissions will appear under “post-challenge submissions” (go to “Leaderboard filters” -> “Include post-challenge submissions” to show them).

We will also announce the launch of the new Flatland event at NeurIPS

This challenge has raised a lot of interest from both participants and partners, the adventure is just getting started!

Cheers,
Florian

View post 2

Submission error: weird error during conda clean

This is probably related to this: Build problem with the current `environment.yml` file

Make sure your requirements.yml only contains what’s strictly necessary!

View post 0

🚇 5 days left! Final tips & tricks for the home stretch

It’s (a): you can submit until the deadline!

View post 1

Different CUDA version

But, as the evaluation environment doesn’t include a GPU, do you really need to setup CUDA in your submissions?

View post 0

Different CUDA version

@shivam may be able to help here?

View post 0

Flatland-Visualization: Visual Analytics of Flatland Episodes

Wow, amazing work

It’s not clear to me how to create new regions: it seems that I need to select multiple cells by keeping my mouse pressed on the Map View? I don’t see the gray overlay on the Map View or the edges between the vertices on the Graph View

View post 0

Evaluation took too long

Hello,

How do I deal with the fact that my 8h of evaluation time ran out and the percentage complete is still > 0.25

This is fine, this means your submission stopped because it reached the overall time limit instead of the “percentage done” limit. Most OR submissions reach the overall timeout (you can see on the leaderboard that their mean “% Done” is very high).

An early submit on the remote client results in an error and doing nothing also results in an error : Timeout : Evaluation took too long.

I don’t understand what you mean?

View post 0