nopperl's repositories
load-reddit-images-directly
Firefox web extension that loads reddit images directly instead of referring to the HTML page containing the image.
clip_arxiv_pmc
Training CLIP models on Data from Scientific Papers
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.
marked-lineart-vectorization
Vectorization of clean line-art raster images using an encoder-decoder model.
pytorch-fused-lamb
LAMB go brrr
corporate_emission_reports
Finetuning and evaluating LLMs to extract GHG emissions from PDF reports using RAG and grammar-based decoding.
autotrace
bitmap to vector graphics converter
axolotl
Go ahead and axolotl questions
build-scripts
Collection of build scripts, configs and guides for ML systems that don't fit elsewhere
cuda-kernoel
CUDA Kernöl
data-engineering-exercises
Data engineering exercises including Spark (Scala and PySpark), PostgreSQL query optimization, Hadoop, Hive and Neo4j.
gradio
Build and share delightful machine learning apps, all in Python. 🌟 Star to support our work!
lineart-vectorization-demo
Gradio demo of the marked line-art vectorization model developed at https://github.com/nopperl/marked-lineart-vectorization.
llama.cpp
Port of Facebook's LLaMA model in C/C++
llm.c
LLM training in simple, raw C/CUDA
lm-evaluation-harness
A framework for few-shot evaluation of autoregressive language models.
marked-lineart-vectorization-thesis
Master thesis about Semantically Meaningful Vectorization of Line Art in Drawn Animation
nanotron
Minimalistic large language model 3D-parallelism training
neural-rerankers-project
Implementation, training and evaluation of neural rerankers.
Pillow
Python Imaging Library (Fork)
pytorch
Tensors and Dynamic neural networks in Python with strong GPU acceleration
tabby
Self-hosted AI coding assistant
torch-image-binarization
`torch.compile`-compatible image binarization algorithm implemented in PyTorch.
Zicklein-GGML
German alpaca finetune converted to GGML format (compatible with llama.cpp).