nopperl

load-reddit-images-directly. Firefox web extension that loads reddit images directly instead of referring to the HTML page containing the image.

marked-lineart-vectorization. Vectorization of clean line-art raster images using an encoder-decoder model.

pokebattle-rl-env. An OpenAI Gym environment for Pokemon battles

torch-image-binarization. `torch.compile`-compatible image binarization algorithm implemented in PyTorch.

corporate_emission_reports. Finetuning and evaluating LLMs to extract GHG emissions from PDF reports using RAG and grammar-based decoding.

clip-synthetic-captions. Tiny-scale experiment showing that CLIP models trained using detailed captions generated by multimodal models (CogVLM and LLaVA 1.5) outperform models trained using the original alt-texts on a range of classification and retrieval tasks.

clip_arxiv_pmc. Training CLIP models on Data from Scientific Papers

pytorch-fused-lamb. LAMB go brrr

ipinyou-ctr. Ad click prediction using the iPinYou dataset

vectorized-image-binarization. Vectorized implementation of the image binarization algorithm of Su et al. (2010)

karaoke-cdg. A Karaoke CDG library for C#

Deep-Vectorization-of-Technical-Drawings. Python

xna-game-console. In-game console for the XNA framework

automatic. Opinionated fork/implementation of Stable Diffusion

data-engineering-exercises. Data engineering exercises including Spark (Scala and PySpark), PostgreSQL query optimization, Hadoop, Hive and Neo4j.

virtual_sketching. Python