Skip to content
View AbelHutten's full-sized avatar
0 followers · 4 following
  • Zurich

Achievements

Achievement: Pull Shark

Achievements

Achievement: Pull Shark

Block or report AbelHutten

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Please don't include any personal information such as legal names or email addresses. Maximum 100 characters, markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
AbelHutten/README.md

Hi there 👋

Thank you for taking a look at my profile! Below I describe some of the projects that you can find here.

Highlighted Projects

TEP-Net-V2

  • Experiments on improving TEP-Net, the current SOTA in rail vehicle ego path detection. Added confidence scores, improved out-of-distribution errors when a leading vehicle is close, worked on adding memory to handle otherwise ambiguous cases.
  • Contributed torch.compile integration back to the original TEP-Net, reducing training time by 17.7% on RTX 4090, with potentially greater speedups on datacenter GPUs.

Efficient-Real-Time-Detection

  • Experiments on using ideas from the literature on efficient transformers (e.g., GQA, Sparse Attention) in order to improve the SOTA in real-time object detection.

AskePlaat-DRL-Solutions

  • Solutions to most of the questions and many of the practical exersises in the textbook Deep Reinforcent Learning (2022) by Aske Plaat. Made as a reference for people working through these exercises. At the time of writing, the only one on the internet.

Pinned Loading

  1. TEP-Net-V2 TEP-Net-V2 Public

    Experiments with TEP-Net (Laurent 2024, arxiv.org/abs/2403.13094), with the goal of improving performance through architectural enhancements.

    Python

  2. Efficient-Real-Time-Detection Efficient-Real-Time-Detection Public

    Experiments with the aim of improving the computational efficiency of SOTA object tracking algorithms.

    Python

  3. AskePlaat-DRL-Solutions AskePlaat-DRL-Solutions Public

    My solutions to the exercises in "Deep Reinforcement Learning" by Aske Plaat (arxiv.org/abs/2201.02135v5).

  4. CS-books-I-ve-read CS-books-I-ve-read Public

    A curated list of computer science (mostly RL) books and papers I've read. This repository serves as a personal reference and resource for others interested in similar readings.

  5. x86-64-Assembly-Practice x86-64-Assembly-Practice Public

    Some practice x86-64 Linux Assembly programs I wrote while learning x86-64.

    Assembly

  6. pytorch-dqn-replication pytorch-dqn-replication Public

    Replicating the DQN agent from the 2015 Nature paper by Mnih et. al.