Martin Pluskal
pluskalm
Involved Projects and Packages
Media bundle containing image workflow assets for ComfyUI.
Media bundle containing audio, 3D and miscellaneous workflow assets for ComfyUI.
Media bundle containing video workflow assets for ComfyUI.
Kornia is a differentiable computer vision library built on PyTorch.
Python bindings for kornia-rs, a low-level computer vision library written in Rust.
Spandrel loads pre-trained PyTorch image models and auto-detects architecture from .pth / safetensors files.
TorchAudio applies PyTorch to the audio domain with autograd-friendly transforms and C++ extensions.
PyTorch implementation of differentiable SDE solvers with GPU support and efficient backpropagation.
The torchvision package consists of popular datasets, model architectures, and common image transformations for computer vision.
Simple and tiny yield-based trampoline implementation for Python.
Exact trees built before a Leap pool PR is opened; transient.
Staging for RBTools 6.0 port plus python-housekeeping/pydiffx/typelets/kgb (pure-python MIT deps not yet in Factory).
Scratch testing ground for science:GPU:ROCm build fixes: the -fcf-protection / hardening-optflags leak into amdgcn device compilation.
Throwaway/experimental packaging work. Keep home:pluskalm itself clean (it is the MCP deployment project); anything here can be deleted at any time.
Headroom is a context-optimization layer for LLM applications. It compresses
tool output and conversation context so requests use fewer tokens, and ships
a command-line interface plus a compiled pyo3 extension (headroom._core).
This package is the core CLI. The optional [proxy] extra lives in the
headroom-proxy subpackage; without it, "headroom proxy" / "headroom wrap"
exit because ensure_proxy_dependencies() hard-imports that set. Code-aware
slicing shells out to the distro ast-grep binary rather than the
ast-grep-cli PyPI wheel.
Staging area for the pi-coding-agent package before it goes to devel:tools.
Memory-safe zlib implementation, built as a C dynamic library.
Scratch project for packaging SGLang (OBS _multibuild cpu + later cuda) and missing Python deps destined for science:machinelearning and Factory.
Extensive database of location and timezone data for nearly every operational airport and landing strip in the world.
Python library providing a tensorflow.io.gfile-like interface for local, Google Cloud Storage and Azure Blob Storage paths.
HuggingFace community-driven open-source library of datasets.
It provides one-line dataloaders for many public datasets and
efficient data pre-processing for audio, vision and NLP tasks.
Access dict values as attributes (works recursively)
Python library for reading and writing ML models in the GGUF (GGML
Universal File) format used by GGML-based inference engines such as
llama.cpp. It also ships command-line tools to dump metadata, edit
metadata values, copy a GGUF file with new metadata, and convert
endianness.
ModelScope is a Model-as-a-Service SDK for browsing, downloading and
running models from ModelScope Hub. This package ships the hub/library
core only: the cv, nlp, audio and related extras are not required.
The modelscope and ms commands delegate to python-modelscope-hub.
Python SDK and CLI to download, upload and manage models, datasets,
Studio spaces, skills and MCP servers on ModelScope Hub. Provides a
HubApi class and the modelscope-hub and ms-hub commands.