michaldyczko,Michał Dyczko,github

Hi! 👋

Welcome to my profile page! My name is Michał Dyczko, living in Warsaw/Poland.

I work as a Full-Stack Web Developer specialized in Django. By training, I am a Machine Learning Engineer focused on deep learning, doing my research in Computer Vision, in Learned Image Compression specifically.

🔧 Technologies & Tools

Here is my main technological stack:

📈 Github Stats

🔭 My Projects

pytorch-parametrizations - Implementation of Sadam introduced in the paper "Efficient Nonlinear Transforms for Lossy Image Compression" by Johannes Ballé, PCS 2018.
pytorch-smartcrop - SmartCrop PyTorch torchvision-like transform based on libvips algorithm.
gradientnorm-autoclipper - Automatic gradient norm clipping inspired by the paper "AutoClip: Adaptive Gradient Clipping for Source Separation Networks" (https://arxiv.org/abs/2007.14469)

Michał Dyczko's Projects

django-quill-editor

django-quill-editor makes Quill.js easy to use on Django Forms and admin sites

gradientnorm-autoclipper

Allows automatic gradient norm clipping. This feature can help to stabilize training in certain situations by limiting the magnitude of gradient updates. The implementation is inspired by the paper "AutoClip: Adaptive Gradient Clipping for Source Separation Networks"

michaldyczko

GH profile description

ncn-grants

Project for Python for Data Visualisation course (summer semester 2019/2020) at Warsaw University of Technology.

punctuation-restoration

Punctuation Restoration using Transformer Models for High-and Low-Resource Languages

pytorch-parametrizations

PyTorch implementation of the spectral parametrization of the weights of a 2D convolutional layer, as introduced in the paper "Efficient Nonlinear Transforms for Lossy Image Compression" by Johannes Ballé, PCS 2018.