README.md

Deep Learning Bootcamp

Since 2016, LauzHack has organized hackathons at EPFL in Lausanne, Switzerland. We also organize tech talks during the school year.

This is a repository for our new event: a Deep Learning Summer Bootcamp.

Syllabus

• day01 Introduction to Deep Learning and PyTorch
  • Lecture: Introduction to bootcamp and Deep Learning
  • Seminar: Introduction to pytorch
  • Lecture 2: Python Dev Tools and Git
• day02 Basic Model Architectures
  • Lecture: Fully-connected and Convolutional Neural Networks, ResNet
  • Seminar: Models in pytorch and training pipeline
  • Lecture 2: Recurrent Neural Networks, BatchNorm, LayerNorm
  • Seminar 2: RNN, LSTM, GRU example
• day03 Transformer and R&D Coding
  • Lecture: Transformer
  • Seminar: Implementation of Transformer in pytorch
  • Seminar 2: WandB, experiments configuration and code structure
• day04 Deep Learning in Audio
  • Lecture: Signal Processing basics, ASR, TTS, VC, Speech Denoising,
  • Seminar: Keyword Spotting
  • Lecture 2: Anti-Spoofing and Graph Neural Networks
  • Seminar 2: Graph Neural Networks with PyTorch Geometric and DGL
• day05 Computer Vision
  • Lecture: Object Detection
  • Seminar: YOLO
  • Lecture 2: Image Segmentation
  • Seminar 2: SAM and YOLO for Segmentation
• day06 Efficient Deep Learning and On-Device Learning
  • Lecture: Efficient Single-GPU Training and Distributed Deep Learning
  • Seminar: PyTorch Examples
  • Lecture 2: On-Device Learning, Domain Adaptation, and Continuous Learning
  • Seminar 2: PULP-TrainLib
• day07 Natural Language Processing
  • Lecture: Introduction to NLP: Tokenization, Embeddings, CBOW, BERT
  • Seminar: Embeddings and CBOW
  • Lecture 2: BERT, Knowledge Distillation, GPT and LLMs
  • Seminar 2: Fine-Tuning BERT, Overview of ChatGPT
• day08 DeepRL, XAI, Multimodal Networks and 3D CV
  • Lecture: Deep Reinforcement Learning
  • Lecture 2: Explainable AI (XAI)
  • Seminar 2: Code example
  • Lecture 3: Multimodality (NLP + Computer Vision) and Coordinate Networks
  • Seminar 3: Code examples
• day09 Model Fine-tuning and Hugging Face
  • Lecture: Fine-tuning and Hugging Face
  • Seminar: Fine-Tuning LLM

Projects

Projects on different topics to get some practical experience:

• ASR Automatic Speech Recognition (Speech To Text)
• GAN Generative Adversarial Network (Image Generation and Neural Vocoder)
• AS Anti-Spoofing
• TBD

To do all the heavy computations, use free Google Colab (8h/day) or Kaggle GPUs (30h/week).

Audio Projects are based on HSE DLA Course.

Resources

• Recordings on YouTube

Contributors & bootcamp staff

Bootcamp materials and teaching were delivered by:

• Petr Grinberg
• Seyed Parsa Neshaei
• Eric Bezzam
• Federico Stella
• Atli Kosson
• Cristian Cioflan
• Skander Moalla
• Vinitra Swamy