srivatsankrishnan / oss-arch-gym Goto Github PK
View Code? Open in Web Editor NEWOpen source version of ArchGym project.
License: Apache License 2.0
Open source version of ArchGym project.
License: Apache License 2.0
It would be nice if we could add the arch gym figure to this page as well since it is the landing read the docs page:
https://oss-archgym.readthedocs.io/en/latest/index.html
After installing by Docker, there are sorts of conflicts
I assume that some packages are installed by default with higher version. However there exists many packages, it's hard to detect the error.
So if it's possible to update all the requirement.txt with detailed version of dependencies. Thank You!
Hi,
Congratulations for this well-needed contribution. I would like to know if you have an estimated timeline about when the CFU-Playground is supposed to be integrated? Also, do you welcome contributions in this area?
Let's add AstraSim to the install_sim.sh script so that we can simply have folks run install_sim.sh astrasim
and everything is installed for them out of the box.
Then we can clean up the documentation in Astrasim for setup/install to simply be:
Run the following command from the oss-arch-gym repository root
install_sim.sh astrasim
We have multiple example dummy environments under arch_gym/envs/
. Let's just keep the example we like the best to clean it up.
Let's have all train scripts follow the same naming convention to clean things up:
train_[algo]_[sim].py
It appears many agents are not working out of the box. We should verify all of them and add the necessary documentation,
We should update this section's documentation to have a standalone script for them to run in arch gym rather than asking them to do these steps:
-Make a copy of this script vizier_dse.py
-Remove line 10
-Replace line 40 to return random int ; for example: cycles, cells = 1, 1
-Run this file, sample output should look something like this: EXITING DSE…
which looks messy for us to ask of them.
All agents work out of the box except for the following:
-train_brute_force.py
-train_ppo.py
-train_single_agent.py
-train_simpleEnv.py
@ShravaniKode I was testing the custom env agents again from scratch and they are not working out of the box. Can you please resolve this?
@aadyapipersenia04 @shreyasgrampurohit I am not sure where the CustomEnv
went?
We need to port over arch_gym_configs.py
under configs
dir to run our previous agents:
(arch-gym) ~/oss-arch-gym/sims/DRAM$ python train_bo_DRAMSys.py
Traceback (most recent call last):
File "/home/oss-arch-gym/sims/DRAM/train_bo_DRAMSys.py", line 8, in <module>
from bo.DRAMSysEstimator import DRAMSysEstimator
File "/home/oss-arch-gym/bo/DRAMSysEstimator.py", line 4, in <module>
from configs import arch_gym_configs
ImportError: cannot import name 'arch_gym_configs' from 'configs' (unknown location)
Was this intentional or we just forgot?
@manan-garg1504 @ShravaniKode Can you guys add and close the CFU documentation thread?
Please work with @ShvetankPrakash on the structure. I think we can reuse some of the stuff from the CFU materials itself.
I am trying to run our step by step here on a fresh VM instance in preparation for tutorial to make sure everything works properly based on our documentation but am getting the following issue:
(base) ~/oss-arch-gym$ conda env create -f environment.yml
Collecting package metadata (repodata.json): done
Solving environment: failed
ResolvePackageNotFound:
- torch
- qhoptim
This may just be an issue with this VM setup since it has worked previously but I am adding this issue to see if it is reproducible by others so we can resolve it if it is.
After this section in the documentation we should have them run like a working example (i.e. train*.py) to verify install is working before going on to the vizier installation. Also the vizier installation should be on a separate page I think cause the setup page feels kind of cluttered at the moment.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.