Skip to content

Latest commit

 

History

History
97 lines (70 loc) · 4.17 KB

README.md

File metadata and controls

97 lines (70 loc) · 4.17 KB

Overview

Practice and tutorial-style notebooks covering my Machine Learning and Deep Learning experiments/projects.

Repository Structure

  • data is an empty folder which is used as a destination for the datasets.
  • Notebooks are kept in the root of the project for now.
  • models folder contains the various trained models and their custom objects, such as pickle files.
  • images folder contains media which is being referenced in Notebooks to add visuals.
  • mathematics folder contains python files and jupyter notebooks explaining all the mathematics required for machine learning and running statistical computations.
  • data_science folder contains the python files and jupyter notebooks for explaining and running the code required to analyze and process data.
  • visualization folder contains snippets to generate various graphs and plots.

Notebooks:

Text Classification:

Classification of Newsgroup documents using four different approaches/algorithms.

Notebook: classification-newsgroup-dataset.ipynb

Blog post on the same: https://shenoy-anurag.github.io/text-classification-on-newsgroup-data.html

Hindi Digit Recognizer:

Classification of Handwritten Hindi (Devanagari script) digits using Convolutional Neural Networks.

Notebook: hindi-digit-recognition.ipynb

Achieved 99.59% accuracy on Test Dataset!

Blog post can be found here: https://shenoy-anurag.github.io/hindi-mnist-recognizer.html

Intent Classification:

Classification of Intents using LSTMs (RNN).

Notebook: intent-classification.ipynb

This model can be used for a chatbot along with an NER model to pick up entities.

System

MacBook Air (M1, 2020)

ARM64 architecture (arm64)

Hardware:

Apple M1 chip 8-core CPU with 4 performance cores and 4 efficiency cores 7-core GPU, 8-core GPU 16-core Neural Engine 16 GB Ram

Operating System:

MacOS Monterey 12.2.1 (21D62)

Environment

Conda (miniforge3):

  • conda version : 4.11.0
  • python version : 3.9.7.final.0

Libraries:

  • tensorflow-macos==2.8.0
  • tensorflow-metal==0.4.0

Setting up an M1 for tensorflow:

  1. First install xcode-select command-line utilities.

    xcode-select --install

  2. Installing Miniforge3

    1. Either using Homebrew:

      brew install miniforge

    2. Or, go to the releases section of miniforge's github page, and find the Miniforge3 file which corresponds to your system.

      Like: Miniforge3-4.11.0-0-MacOSX-arm64.sh

      1. Download the file to a folder.
      2. Open a terminal and change to the folder where you downloaded the install script.
      3. Run the command chmod +x Miniforge3-4.11.0-0-MacOSX-arm64.sh (don't forget to replace the file name with the name of the file you downloaded).
      4. Then install from the file by running sh Miniforge3-4.11.0-0-MacOSX-arm64.sh in your terminal.
      5. source ~/miniforge3/bin/activate

  3. Initialize Miniforge using the command:

    conda init

  4. Use this Conda Cheatsheet to create a conda environment.

  5. Activate the newly created conda environment.

  6. To use your environment in Jupyter notebooks

    1. conda install -y jupyter (this command installs jupyter)

    2. conda install nb_conda (this command installs nb_conda, which adds conda env support to jupyter notebooks)

    3. And finally, add your environment to jupyter using

      python -m ipykernel install --user --name <env_name> --display-name <display_name>

      Don't forget to replace <env_name> and <display_name> with the name you want.

If you face any issues in setting up your environment for M1 Macbooks, take a look at these resources: