search

theomarzaki / TrafficOrchestrator

Traffic Orchestrator for Central unit Processing of autonomous vehicle merging through the use of Reinforcment Learning
MIT License
2 stars 0 forks source link

Add Dockerfile #40

Closed turbokadi closed 5 years ago

turbokadi commented 5 years ago

Add cuda dockerfile to run training.

Double_DQN.py run by default.

Modify script to run : sudo docker run -it python Run local script : sudo docker run -it -v $(pwd):/root python

turbokadi commented 5 years ago

Hello @theomarzaki ,

It's just material. Maybe it's not that useful but I use it on our training machine and it can be helpful for further uses (cloud), like the lane detector for example. I don't think it's a big deal to leave it here. We can remove it after if it f* up our stack.

theomarzaki commented 5 years ago

Perfect, no worries ! Just wanted to understand the reasoning behind it :)

But unfortunately, the file changes are a bit behind. Since all the models have been gathered into one file with parser arguments being passed to it instead of calling the whole model script.

I would suggest to pull the latest from master/accurate_waypoints branch so the dockerfile reflects the changes made

theomarzaki commented 5 years ago

@J4BB3R N.B - Completely My bad.

The files in the Master branch hasn't been updated to reflect the changes I made.

Please see accurate_waypoints branch, to see what all comments made above !

I will be merging accurate_waypoints branch very soon, although I need to finish a few things relating the to KPIs before I can merge.

The changes made in accurate_waypoints cover all the main points here. Excluding a docker file however, but you can merge the branch with the docker file only, and that will build on the code I merge, so not to repeat code.

Once again Apologies !

turbokadi commented 5 years ago

Hello @theomarzaki, I misinterpreted the use of this script. I thought it would be more for experiments. I will only merge my dockerfile. Let's talk about it tomorrow.