Skip to content

Instantly share code, notes, and snippets.

View harshraj22's full-sized avatar
✌️
😋 Integrating Continiously

Harsh Raj harshraj22

✌️
😋 Integrating Continiously
View GitHub Profile
- Use concepts of multhreading to implement a thread safe message queue.
@harshraj22
harshraj22 / project_ideas.md
Last active October 30, 2022 05:58
Ideas and requirements for next project

Requirements:

  1. Use Sockets, Microservices, Master Slave DB, ML models, Observability of microservices, load balancer, autoscaling, elastic search

Idea

Build a chatting app, where:

  • users can send, text & images,
  • multiple people can join and leave
  • Authentication (only users with a passcode can join a channel)
  • ML models run synchronously on each message sent
  • Users have an option to see the negative masked message and validate the output of model (becomes data to retrain)
while doing `pip install pycuda==2020.1`
Reference: https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#mandatory-post
mapped python to python3 outside the virtal env
cuda.h was not found:
1. export PATH=/usr/local/cuda-10.2/bin${PATH:+:${PATH}} -> `ls /usr/local` showed us that our cuda version was 10.2
pyconfig.h error:
1. export LD_LIBRARY_PATH=/usr/local/cuda-10.2/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}

--- How to train your Dragon ---

img

  • Given the model architecture (implemented in PyTorch), candidates have to train the given model, and submit the weights.
  • Host competition on kaggle for auto grading of submissions, and manually verify the scores/outputs of winners by their weights.
  • Participants can explore
    • Different data augmentation techniques, including the promising ones like cutmix, mixin etc
  • Various optimizers & activation functions
@harshraj22
harshraj22 / feat_extract.py
Created September 24, 2021 09:17
Bottom up attention: Feature extraction
# import torch
# import detectron2
# from PIL import Image
# import numpy as np
# from detectron2.modeling import build_model
# from detectron2.config import get_cfg
# from detectron2.structures import ImageList
# from torchinfo import summary
@harshraj22
harshraj22 / IdeaList.md
Last active August 13, 2021 11:16
New ideas for BTP: Visual Question Answering
  • 1x1 convolutions have been extensively used to reduce the number of parameters without affecting the results much
  • Deep Mutual Learning: Unlike bagging/ boosting, models learn jointly, and help each other to fit well
  • Skip connections: Help solving degradation problem without adding parameters.
  • Hard Sample Mining
# ~/.bashrc: executed by bash(1) for non-login shells.
# see /usr/share/doc/bash/examples/startup-files (in the package bash-doc)
# for examples
# If not running interactively, don't do anything
case $- in
*i*) ;;
*) return;;
esac

private channels are private to each, doesn't matter who is workspace owner link

replies in slack are theaded keeping the channel clean, while those in discord are sent on the public channel (similar to whatsapp)

emoji react on messages, polls (saves from +1 cancer), personalized reminders,

@harshraj22
harshraj22 / ques.md
Last active August 13, 2020 15:39
MCQ Questions for the first test for internship/ placement.
  1. A process executes the code
  fork ();
  fork ();
  fork ();

The total number of child processes created is
(A) 3
(B) 4
(C) 7

Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.