Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • Jan 21 12:09
    piseabhijeet opened #66
  • Dec 11 2020 06:34
    Atcold closed #65
  • Dec 11 2020 06:34
    Atcold commented #65
  • Dec 09 2020 19:15
    timo-sievernich edited #65
  • Dec 09 2020 18:53
    timo-sievernich opened #65
  • Nov 11 2020 23:04
    Atcold closed #64
  • Nov 11 2020 23:04
    Atcold commented #64
  • Oct 12 2020 14:28
    Gan-Song-CU opened #64
  • Aug 08 2020 19:38
    Atcold commented #61
  • Aug 08 2020 19:20
    Atcold commented #63
  • Aug 08 2020 19:20

    Atcold on master

    Update traffic_gym.py Fix Attr… Merge pull request #63 from jus… (compare)

  • Aug 08 2020 19:20
    Atcold closed #63
  • Aug 03 2020 21:09
    xjh19971 closed #61
  • Aug 03 2020 21:09
    xjh19971 commented #61
  • Aug 01 2020 20:59
    xjh19971 synchronize #61
  • Aug 01 2020 20:54
    xjh19971 synchronize #61
  • Aug 01 2020 20:50
    xjh19971 synchronize #61
  • Aug 01 2020 20:45
    xjh19971 synchronize #61
  • Aug 01 2020 20:17
    xjh19971 synchronize #61
  • Aug 01 2020 02:39
    xjh19971 synchronize #61
Serkan Karakulak
@skarakulak
Hi Alfredo. Hope all is well!
I have created the masks for off-road regions which is shown below in red using the human mode of the simulator
Screenshot from 2019-05-27 02-11-30.png
For the machine mode, my current plan is to use the color (255,0,0) for off road areas, (255x0.20,0,0) for lanes and increase lambda_l from 0.2 to 1, so that cost for lanes stay the same
But changing the color of the lanes will increase the problem that the FM was not trained with this data. Considering the training times and the resource limitations on HPC, should I skip the second task which was training MPUR with the same FM and instead start with training both the FM and the MPUR?
Serkan Karakulak
@skarakulak
@Atcold also, do I need access to the /misc/vlgscratch4/LecunGroup/nvidia-collab/ folder? I can download the current pretrained model from the link provided in the readme, but I don't know if my drive on cassio HPC will be enough for saving checkpoints from all seeds
Alfredo Canziani
@Atcold
I've replied by email to this already, I think. Sorry for being absent least week.
Yair Schiff
@yair-schiff
Can someone provide some details about how to run a jupyter notebook on Cassio but have it open on my local machine's browser? Thanks!
Yair Schiff
@yair-schiff
nevermind, sorry for the spam, I was able to piece it together with some googling
Yair Schiff
@yair-schiff
Hi all, can someone please point me to the location on the Cassio drive of the best policy network trained on I-80? Thanks!
Serkan Karakulak
@skarakulak
Hey Yair! I'm not sure about where it is in nvidia_colab/ folder but you can download it from here: https://drive.google.com/a/nyu.edu/uc?id=1XahspfgFlBVF6ne479LCJgBr0luZGQt7&export=download
Yair Schiff
@yair-schiff
thanks!
Serkan Karakulak
@skarakulak
sure!