harshraj22

- Use concepts of multhreading to implement a thread safe message queue.

harshraj22

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)

harshraj22

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}}

harshraj22

--- How to train your Dragon ---

• 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

# 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

• 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

harshraj22

# ~/.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

harshraj22

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

1. A process executes the code

  fork ();
  fork ();
  fork ();

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

harshraj22