profile
viewpoint
Vishal BBloggsbott Chennai https://bbloggsbott.github.io/ Interested in Machine Learning. Likes to talk about it.

BBloggsbott/ghub 23

A Unix like CLI for GitHub

BBloggsbott/Address-Book 5

An Address Book with a simple GUI buit using Java

BBloggsbott/BillingSystem 5

A billing application built using Java best suited for small stores

BBloggsbott/autobound 3

Repository for Google Summer of Code 2019 project

BBloggsbott/k-armed-bandits 2

An implementation of "Deep Bayesian Bandits Showdown: An Empirical Comparison of Bayesian Deep Networks for Thompson Sampling".

BBloggsbott/alexa-mlscrapper 1

The code for the lambda function of the Alexa skill "Machine Learning Paper Fetcher" - https://www.amazon.com/gp/product/B07YSKY7TG

BBloggsbott/autoboundserver 1

Server for AutoBound

BBloggsbott/BBloggsbott.github.io 1

Personal website for Vishal. Based on https://academicpages.github.io, which is a fork from mistakes/minimal-mistakes

BBloggsbott/freshworks-datastore 1

Repository with code for freshworks campus recruitment drive

BBloggsbott/ANN.jl 0

Julia artificial neural networks

push eventBBloggsbott/ready-to-serv

Vishal

commit sha ae6f5d69e44d2ca9e8383414e15d018672edc56e

Remove .idea

view details

push time in 20 hours

create barnchBBloggsbott/ready-to-serv

branch : from-kotlin

created branch time in 20 hours

startedlinkedin/LiFT

started time in 4 days

create barnchBBloggsbott/ready-to-serv

branch : develop

created branch time in 8 days

create barnchBBloggsbott/ready-to-serv

branch : master

created branch time in 8 days

created repositoryBBloggsbott/ready-to-serv

created time in 8 days

push eventBBloggsbott/BBloggsbott.github.io

Vishal R

commit sha 4801e6ef4f702fa3e4008db925e32a55ffe252ec

Direct to calendar

view details

push time in 17 days

push eventBBloggsbott/BBloggsbott.github.io

Vishal R

commit sha 4ad4993cc387fa164195dbcfdabc4764bfefa8ce

Remove schedule in calendar

view details

push time in 17 days

push eventgitter-badger/workout-1

Vigneswaran-Chandrasekaran

commit sha 76ffa029e75938eb084749b1d39d4d802c17b659

Add Readme

view details

Vigneswaran-Chandrasekaran

commit sha 729130efe4f9bbe709bf719f3006fc9668607057

update readme

view details

Vishal

commit sha 4119c2bd2c9152a89db0f84a150e0e84ed0cbafa

More info for README

view details

Vishal

commit sha 16f2959bb8f1d5f22a0788df2784cd3da87e7375

Merge branch 'develop' of https://github.com/BBloggsbott/workout into initial_set

view details

Vishal

commit sha 6817234aa25803c4f3385871d29cacb1eb17d72a

Add basic design of interface for state vector based interfaces

view details

Vishal

commit sha 3e06f7f71f36294a8fd5e8d6a3d0decf7e53ae0f

Merge branch 'develop' into initial_set

view details

Vishal

commit sha 593eecbb06a2d97c3acb1c51c0e9308ffe508770

Minor fix to pass pre-commit checks

view details

Vishal

commit sha 0c56c6cc3bf8fce4bd175c1ebf1c1895319e7232

Merge pull request #1 from BBloggsbott/initial_set Conceptualizing work

view details

Vishal

commit sha 3b417d8a262218a521bd52facef353af1e22d2bd

Merge branch 'develop' into gitter-badge-1

view details

push time in 2 months

push eventBBloggsbott/workout

Vigneswaran-Chandrasekaran

commit sha 76ffa029e75938eb084749b1d39d4d802c17b659

Add Readme

view details

Vigneswaran-Chandrasekaran

commit sha 729130efe4f9bbe709bf719f3006fc9668607057

update readme

view details

Vishal

commit sha 4119c2bd2c9152a89db0f84a150e0e84ed0cbafa

More info for README

view details

Vishal

commit sha 16f2959bb8f1d5f22a0788df2784cd3da87e7375

Merge branch 'develop' of https://github.com/BBloggsbott/workout into initial_set

view details

Vishal

commit sha 6817234aa25803c4f3385871d29cacb1eb17d72a

Add basic design of interface for state vector based interfaces

view details

Vishal

commit sha 3e06f7f71f36294a8fd5e8d6a3d0decf7e53ae0f

Merge branch 'develop' into initial_set

view details

Vishal

commit sha 593eecbb06a2d97c3acb1c51c0e9308ffe508770

Minor fix to pass pre-commit checks

view details

Vishal

commit sha 0c56c6cc3bf8fce4bd175c1ebf1c1895319e7232

Merge pull request #1 from BBloggsbott/initial_set Conceptualizing work

view details

push time in 2 months

delete branch BBloggsbott/workout

delete branch : initial_set

delete time in 2 months

PR merged BBloggsbott/workout

Conceptualizing work brainstorming readme

Good Initial steps would be,

  • [ ] Conceptualizing idea, technical wise (how to implement it?)

  • [ ] Any specific goal that unifies PyTorch's core work?

@BBloggsbott , the technical details need some discussions, so you have any thoughts?

Some, relevant works done are,

https://openai.com/blog/openai-pytorch/ https://gym.openai.com/evaluations/eval_onwKGm96QkO9tJwdX7L0Gw/ https://github.com/roclark/openai-gym-pytorch https://github.com/hominhtri1996/DQN-with-Pytorch-and-OpenAI-Gym

+89 -1

7 comments

9 changed files

Vigneswaran-Chandrasekaran

pr closed time in 2 months

Pull request review commentBBloggsbott/workout

Conceptualizing work

+from gym.spaces import Space, Discrete, Box++import numpy as np+++def get_space_size(space: Space):+    space_type = type(space)+    if space_type == Discrete:+        return space.n+    elif space_type == Box:+        return np.prod(space.shape)

@NikhilMohanKK We know that the Discrete type means categorical data. So that if block is fine for now. Can u take a look at the Box type in the openai gym code and try to alter this if block so it returns what it must (the number of nodes in that layer of the NN)?

Vigneswaran-Chandrasekaran

comment created time in 2 months

MemberEvent

pull request commentBBloggsbott/workout

Add a Gitter chat badge to README.md

We'll marge this PR later.

gitter-badger

comment created time in 2 months

PR closed BBloggsbott/workout

Add a Gitter chat badge to README.md

BBloggsbott/workout now has a Chat Room on Gitter

@BBloggsbott has just created a chat room. You can visit it here: https://gitter.im/BBloggsbott-workout/woukout.

This pull-request adds this badge to your README.md:

Gitter

If my aim is a little off, please let me know.

Happy chatting.

PS: Click here if you would prefer not to receive automatic pull-requests from Gitter in future.

+2 -0

0 comment

1 changed file

gitter-badger

pr closed time in 2 months

pull request commentBBloggsbott/workout

Conceptualizing work

Once that is done, and we are sure it is okay, we'll merge this PR and create another PR for specific environments

Vigneswaran-Chandrasekaran

comment created time in 2 months

pull request commentBBloggsbott/workout

Conceptualizing work

Hey @Vigneswaran-Chandrasekaran any progress?

Vigneswaran-Chandrasekaran

comment created time in 2 months

GollumEvent

push eventBBloggsbott/workout

Vishal

commit sha e753ef3ffbb5eef44a0e2d3ff3bd83e39cde3250

Update action

view details

push time in 3 months

push eventBBloggsbott/workout

Vishal

commit sha 593eecbb06a2d97c3acb1c51c0e9308ffe508770

Minor fix to pass pre-commit checks

view details

push time in 3 months

push eventBBloggsbott/workout

Vishal

commit sha f30ec4c2770a743f0f35917822a7a6603256cf79

Add labeler

view details

Vishal

commit sha 0fdf8831959748145a984a67180370dbbf6bd7fb

Add pre-commit

view details

Vishal

commit sha 3d2fc8c215d34e6c895ad6526a45b5c28f865359

Add precommit as action to check PR

view details

Vishal

commit sha eb136460800d8cad576034aa653a2698278e702c

Merge pull request #2 from BBloggsbott/repo_maintainance Github Actions for Repo maintenance

view details

Vishal

commit sha 3e06f7f71f36294a8fd5e8d6a3d0decf7e53ae0f

Merge branch 'develop' into initial_set

view details

push time in 3 months

pull request commentBBloggsbott/workout

Conceptualizing work

I think to start with, we'll first import some simple gym environment and have hookers to get all the required information out of it. For now, we'll have a basic example (like cart pole) and then later generalize the code to use for all environments. I'll work on importing and extracting gym environment's information and parallelly, you shall create some dummy data structure (assuming it as the extracted information), and try interfacing with torch. And we should definitely have a call (tonight maybe) to discuss about these.

Sounds good. I feel we should also have some interfaces implemented with some good defaults as examples so users can use it as a playground.

Vigneswaran-Chandrasekaran

comment created time in 3 months

push eventBBloggsbott/workout

Vishal

commit sha f30ec4c2770a743f0f35917822a7a6603256cf79

Add labeler

view details

Vishal

commit sha 0fdf8831959748145a984a67180370dbbf6bd7fb

Add pre-commit

view details

Vishal

commit sha 3d2fc8c215d34e6c895ad6526a45b5c28f865359

Add precommit as action to check PR

view details

Vishal

commit sha eb136460800d8cad576034aa653a2698278e702c

Merge pull request #2 from BBloggsbott/repo_maintainance Github Actions for Repo maintenance

view details

push time in 3 months

PR merged BBloggsbott/workout

Reviewers
Github Actions for Repo maintenance readme

Adding labeler and pre-commit for labeling PRs and maintaining code quality. We can add an action for tests once we include them

+75 -2

2 comments

6 changed files

BBloggsbott

pr closed time in 3 months

pull request commentBBloggsbott/workout

Github Actions for Repo maintenance

Merging it

BBloggsbott

comment created time in 3 months

PR opened BBloggsbott/workout

Reviewers
Github Actions for Repo maintainance

Adding labeler and pre-commit

+75 -2

0 comment

6 changed files

pr created time in 3 months

create barnchBBloggsbott/workout

branch : repo_maintainance

created branch time in 3 months

pull request commentBBloggsbott/workout

Conceptualizing work

@Vigneswaran-Chandrasekaran I have

  1. updated the default branch to develop
  2. added a basic architecture of how a state-vector based interface could look like
  3. changed the base branch of this PR to develop
Vigneswaran-Chandrasekaran

comment created time in 3 months

push eventBBloggsbott/workout

Vishal

commit sha 7c773aabe95297ef681dbd50747d4daea9109344

Add vscode to gitignore

view details

Vishal

commit sha 16f2959bb8f1d5f22a0788df2784cd3da87e7375

Merge branch 'develop' of https://github.com/BBloggsbott/workout into initial_set

view details

Vishal

commit sha 6817234aa25803c4f3385871d29cacb1eb17d72a

Add basic design of interface for state vector based interfaces

view details

push time in 3 months

create barnchBBloggsbott/workout

branch : develop

created branch time in 3 months

Pull request review commentBBloggsbott/workout

Conceptualizing work

-# workout\ No newline at end of file+# Workout++### What is Workout?++Workout is an API to import and use `OpenAI-Gym`'s environment with `PyTorch` effortlessly++### Why it is required?++`PyTorch`: Flexible framework to implement deep neural networks and has better GPU integration+`OpenAI-Gym`: Provides extensive and varied Reinforcement Learning environments to use readily++![workout](docs/workout.jpg)++However, the integration between two is not very extensive. Many works have been done to implement+deep network based Reinforcement Learning algorithms using `PyTorch` seperately, then transfer the whole control+to `Gym`'s environment to estimate reward function, state of the system, possible actions for next step, etc.,+and pass it again to `PyTorch`'s model. Therefore, to avoid such complications, `Workout` provides a higher level of abstraction to the `Gym`'s environment, providing an interface to make it more `PyTorch` oriented. By doing so,+the users shall effortlessly use `Gym`'s environment without affecting `PyTorch`'s syntactic sugar. Also, the+translation to `PyTorch` codebase would improve the uniformity of the underlying kernel and helps heavily in+parallelization using GPUs.

Sure. Report work is pretty hectic for me as well :(

Vigneswaran-Chandrasekaran

comment created time in 3 months

Pull request review commentBBloggsbott/workout

Conceptualizing work

-# workout\ No newline at end of file+# Workout++### What is Workout?++Workout is an API to import and use `OpenAI-Gym`'s environment with `PyTorch` effortlessly++### Why it is required?++`PyTorch`: Flexible framework to implement deep neural networks and has better GPU integration+`OpenAI-Gym`: Provides extensive and varied Reinforcement Learning environments to use readily++![workout](docs/workout.jpg)++However, the integration between two is not very extensive. Many works have been done to implement+deep network based Reinforcement Learning algorithms using `PyTorch` seperately, then transfer the whole control+to `Gym`'s environment to estimate reward function, state of the system, possible actions for next step, etc.,+and pass it again to `PyTorch`'s model. Therefore, to avoid such complications, `Workout` provides a higher level of abstraction to the `Gym`'s environment, providing an interface to make it more `PyTorch` oriented. By doing so,+the users shall effortlessly use `Gym`'s environment without affecting `PyTorch`'s syntactic sugar. Also, the+translation to `PyTorch` codebase would improve the uniformity of the underlying kernel and helps heavily in+parallelization using GPUs.

Okay. In that case, I'll create some base classes in the next couple of days. I'm also changing the default branch to develop. We'll use master just for releases.

Vigneswaran-Chandrasekaran

comment created time in 3 months

push eventBBloggsbott/BBloggsbott.github.io

Vishal

commit sha 4868a3973c4d840ae4eaadeac1a61b04656ed112

Minor fix

view details

push time in 3 months

push eventBBloggsbott/BBloggsbott.github.io

Vishal

commit sha c8931142ad9a2b75eb2bb1d4cc35abe6e9c65c51

Minor fix

view details

push time in 3 months

push eventBBloggsbott/BBloggsbott.github.io

Vishal

commit sha 26a6225a80e5eedfb3ec2e217e4b3ad4c9eb734e

Minor fix

view details

push time in 3 months

push eventBBloggsbott/BBloggsbott.github.io

Vishal

commit sha 9cadb9fd0c9b0d40f308e8687d19741c8929152f

Minor fix

view details

push time in 3 months

push eventBBloggsbott/BBloggsbott.github.io

Vishal

commit sha 41dbcb5451a91202990d24040ea3796b2c0ab51d

Minor change

view details

push time in 3 months

push eventBBloggsbott/BBloggsbott.github.io

Vishal

commit sha 1fa85d8a39f884af0ddbdedfd07d25864b25fd96

Update Calendar

view details

push time in 3 months

Pull request review commentBBloggsbott/workout

Conceptualizing work

-# workout\ No newline at end of file+# Workout++### What is Workout?++Workout is an API to import and use `OpenAI-Gym`'s environment with `PyTorch` effortlessly++### Why it is required?++`PyTorch`: Flexible framework to implement deep neural networks and has better GPU integration+`OpenAI-Gym`: Provides extensive and varied Reinforcement Learning environments to use readily++![workout](docs/workout.jpg)++However, the integration between two is not very extensive. Many works have been done to implement+deep network based Reinforcement Learning algorithms using `PyTorch` seperately, then transfer the whole control+to `Gym`'s environment to estimate reward function, state of the system, possible actions for next step, etc.,+and pass it again to `PyTorch`'s model. Therefore, to avoid such complications, `Workout` provides a higher level of abstraction to the `Gym`'s environment, providing an interface to make it more `PyTorch` oriented. By doing so,+the users shall effortlessly use `Gym`'s environment without affecting `PyTorch`'s syntactic sugar. Also, the+translation to `PyTorch` codebase would improve the uniformity of the underlying kernel and helps heavily in+parallelization using GPUs.

@Vigneswaran-Chandrasekaran If we are going to make parallelization possible, we need to identify a way to parallelize gym environments while retaining their states for the different parallel training loops.

I suggest we start designing and building the package with parallelization in mind so we can work on a way to parallelize the environments later and include those changes.

Vigneswaran-Chandrasekaran

comment created time in 3 months

Pull request review commentBBloggsbott/workout

Conceptualizing work

-# workout\ No newline at end of file+# Workout++### What is Workout?++Workout is an API to import and use `OpenAI-Gym`'s environment with `PyTorch` effortlessly++### Why it is required?++`PyTorch`: Flexible framework to implement deep neural networks and has better GPU integration+`OpenAI-Gym`: Provides extensive and varied Reinforcement Learning environments to use readily++![workout](docs/workout.jpg)++However, the integration between two is not very extensive. Many works have been done to implement+deep network based Reinforcement Learning algorithms using `PyTorch` seperately, then transfer the whole control+to `Gym`'s environment to estimate reward function, state of the system, possible actions for next step, etc.,+and pass it again to `PyTorch`'s model. Therefore, to avoid such complications, `Workout` provides a higher level of abstraction to the `Gym`'s environment, providing an interface to make it more `PyTorch` oriented. By doing so,+the users shall effortlessly use `Gym`'s environment without affecting `PyTorch`'s syntactic sugar. Also, the+translation to `PyTorch` codebase would improve the uniformity of the underlying kernel and helps heavily in+parallelization using GPUs.

Yes. But we can't embed pytorch tensors into gym. We'll have to convert the output of gym to tensors and vice versa to work with it.

Vigneswaran-Chandrasekaran

comment created time in 3 months

Pull request review commentBBloggsbott/workout

Conceptualizing work

-# workout\ No newline at end of file+# Workout++### What is Workout?++Workout is an API to import and use `OpenAI-Gym`'s environment with `PyTorch` effortlessly++### Why it is required?++`PyTorch`: Flexible framework to implement deep neural networks and has better GPU integration+`OpenAI-Gym`: Provides extensive and varied Reinforcement Learning environments to use readily++![workout](docs/workout.jpg)++However, the integration between two is not very extensive. Many works have been done to implement+deep network based Reinforcement Learning algorithms using `PyTorch` seperately, then transfer the whole control+to `Gym`'s environment to estimate reward function, state of the system, possible actions for next step, etc.,+and pass it again to `PyTorch`'s model. Therefore, to avoid such complications, `Workout` provides a higher level of abstraction to the `Gym`'s environment, providing an interface to make it more `PyTorch` oriented. By doing so,+the users shall effortlessly use `Gym`'s environment without affecting `PyTorch`'s syntactic sugar. Also, the+translation to `PyTorch` codebase would improve the uniformity of the underlying kernel and helps heavily in+parallelization using GPUs.

I'm not sure if parallelization will be possible for what we have in mind because I'm not sure if the environment can be parallelized (as the environment is not made of pytorch tensors). I'll try to do some research on that.

Vigneswaran-Chandrasekaran

comment created time in 3 months

Pull request review commentBBloggsbott/workout

Conceptualizing work

-# workout\ No newline at end of file+# Workout++### What is Workout?++Workout is an API to import and use `OpenAI-Gym`'s environment with `PyTorch` effortlessly++### Why it is required?++`PyTorch`: Flexible framework to implement deep neural networks and has better GPU integration+`OpenAI-Gym`: Provides extensive and varied Reinforcement Learning environments to use readily++![workout](docs/workout.jpg)++However, the integration between two is not very extensive. Many works have been done to implement+deep network based Reinforcement Learning algorithms using `PyTorch` seperately, then transfer the whole control+to `Gym`'s environment to estimate reward function, state of the system, possible actions for next step, etc.,+and pass it again to `PyTorch`'s model. Therefore, to avoid such complications, `Workout` provides a higher level of abstraction to the `Gym`'s environment, providing an interface to make it more `PyTorch` oriented. By doing so,+the users shall effortlessly use `Gym`'s environment without affecting `PyTorch`'s syntactic sugar. Also, the+translation to `PyTorch` codebase would improve the uniformity of the underlying kernel and helps heavily in+parallelization using GPUs.

@Vigneswaran-Chandrasekaran I don't entirely understand how it affects the kernel and allows parallelization.

Vigneswaran-Chandrasekaran

comment created time in 3 months

push eventBBloggsbott/workout

Vishal

commit sha 4119c2bd2c9152a89db0f84a150e0e84ed0cbafa

More info for README

view details

push time in 3 months

MemberEvent

create barnchBBloggsbott/workout

branch : master

created branch time in 3 months

created repositoryBBloggsbott/workout

created time in 3 months

startedrubenlagus/TelegramBots

started time in 3 months

more