valentingol Goto Github PK

followers: 29.0 following: 19.0 repos: 35.0 gists: 7.0

Name: Valentin Goldité

Type: User

Company: ONERA, Palaiseau

Bio: Jax + Haiku fan. Self-attention for the win.

Location: France (Paris)

Blog: www.linkedin.com/in/valentingoldite

Valentin Goldité ~ Valentingol

Welcome to my Github profile! ✨

I hope you will find projects that interest you!

Some words about me

My name is Valentin Goldité and I am a french Deep Learning engineer passionate by deep learning and specialized in the field of computer vision (2D and 3D). I am interested on all the different aspects of deep learning, in every domains and both in the mathematical theory and the practical applications.

My curriculum:

Student at CentraleSupélec 👨‍🎓: top 3 french Engineer School (Deep Learning lessons and many more) - 3 years
Maths Vision Apprentissage (MVA) master at Ecole Normale Supérieure Paris Saclay 👨‍🎓: one of the most renowned masters in europe for machine learning - 1 year
Deep Learning engineer at Sysnav Navigation 🧭: time series prediction in real time in embedded device - 6 months
Deep Learning engineer at Reactive Reality 👨: 3D reconstruction of full head from selfie with PIFu algorithm - 6 months
Deep Learning engineer at IFP Energies Nouvelles 🪨: Underground facies modeling with conditional self-attention GAN - 6 months
Computer Vision research engineer at ONERA The French Aerospace Lab ✈️: Fundamental research on multi-modal few shot object detection. Military applications as part of the KOIOS European project - now

I also did a lot of trainings for CentraleSupelec's AI organization Automatant about both theory and practical application. Some of these trainings are available (in french) in the Automatant website.

My mantra

Python's PEP20 introduces the Zen of Python. I also developed my own mantra 🙏 for ML project:

Linux is far better than Windows for coding purposes.
VSCode is more flexible than Pycharm. It exists other code editor?
Virtualenv and conda have their own advantages...
But Virtualenv plus VirtualenvWrapper are really more convenient than Conda.
Pytorch is better than Tensorflow.
When you can use Jax, you should really consider using it.
Haiku over Flax over all the other nn Jax libs.
Weights and Biases and ClearML are very cool. Tensorboard for sensitive cases only.
Use many linters in real time when coding to write properly.
Local and global tests help you to govern the world.
Github Copilot too.
Self and cross-attention mechanism are revolutionary ideas.
Always look for existing implementations instead of reinventing the wheel.
In return, contribute to open source to thank the community.

Finally, if you like figures

Valentin Goldité's Projects

anybadge

A Python project for generating badges for your projects, with a focus on simplicity and flexibility.

automatants

All scripts of my trainings for the AI association of CentraleSupelec : "Les Automatants"

awesome-gan-inversion

A collection of resources on GAN inversion.

bagofwordscv

Classic computer vision project for images classifications tasks based on keypoints detection.

carl-demo

A demo for the Carl repository

cliconfig

Merging yaml config and set parameters with CLI. Use config processing for more advanced manipulations.

deep-learning-installation

This tutorial provide a step-by-step pipeline to install an effective Python set-up optimized for deep learning for Ubuntu LTS, containing libraries to use efficiently the last versions of Tensorflow and Pytorch with the GPU and a comfortable environment of work with flexible and highly customizable IDE (VSCode) and environment manager (Virtualenv/VirtualenvWrapper).

detectron

FAIR's research platform for object detection research, implementing popular algorithms like Mask R-CNN and RetinaNet.

docstripy

Convert any python docstrings with the docstring format you want (Numpydoc, Google, ReST).

efficientnet-jax

EfficientNet, MobileNetV3, MobileNetV2, MixNet, etc in JAX w/ Flax Linen and Objax

ferrarl

Code our team "FerraRl" for the Reinforcment Learning competition in the AI association of CentraleSupelec (Automatant).

flaxmodels

Pretrained models for Jax/Flax: StyleGAN2, GPT2, VGG, ResNet, etc.

gan-face-editing

Face attribute manipulation with GAN for automatic data augmentation or individual editing.

gan-facies-modeling

Facies modeling using GANs

ganjax

Implementation of different GAN in JAX/Haiku.

jax-models

Unofficial JAX implementations of Deep Learning models

leafnothingbehind

Repository of the 2023 Transfer Learning competition in partnership with World From Space. Team Antoine Aurélien Clement Valentin.

learnrl

A library to use and learn reinforcement learning algorithms

learnrl-pytorch

learntransfo

A personal open source implementation of transformers in Tensorflow 2 with some improvements found on the Internet (sometimes updated). Created in 07/2021 by Valentin Goldité