Projects Recently Starred by Experts

Starred 13 hours ago by

Jeffrey Morgan(Cofounder of Ollama).

desktop by openyak

133.3%

278

Local AI desktop assistant for private productivity

Automate office tasks, analyze local data, and draft content securely on your machine.
Integrates 100+ AI models via OpenRouter and supports existing ChatGPT subscriptions.
Features 7 agent modes with multi-step tool calling and sub-agent nesting for complex workflows.
100% local-first architecture ensures complete data privacy with no cloud uploads or telemetry.

Created 1 week ago

Updated 14 hours ago

Starred 14 hours ago by

Jeff Hammerbacher(Cofounder of Cloudera).

bdh by pathwaycom

0.3%

Biologically inspired LLM architecture bridging neuroscience and deep learning

Emergent reasoning and generalization from local neuron dynamics, matching Transformer performance.
Scale-free, graph-based network with Hebbian memory and interpretable, sparse activations.
GPU-friendly state-space formulation enables efficient implementation and analysis.
Offers full interpretability and theoretical grounding, unlike black-box Transformers.

Created 6 months ago

Updated 1 week ago

Starred 17 hours ago by

Tim J. Baek(Founder of Open WebUI). Also starred by

Lysandre Debut(Chief Open-Source Officer at Hugging Face),

Jane Manchun Wong(Security Researcher),

Chaoyu Yang(Founder of Bento), and

8 more.

mlx-lm by ml-explore

2.6%

Python package for LLM text generation and fine-tuning on Apple silicon

Integrates with Hugging Face Hub for easy access to thousands of LLMs.
Supports model quantization and uploading to the Hub.
Enables low-rank and full model fine-tuning, including quantized models.
Offers distributed inference and fine-tuning via mx.distributed.

Created 1 year ago

Updated 13 hours ago

Starred 22 hours ago by

Wing Lian(Founder of Axolotl AI). Also starred by

Daniel Han(Cofounder of Unsloth) and

Dan Guido(Cofounder of Trail of Bits).

ATLAS by itigges22

5257.9%

933

Boosts frozen LLM performance for efficient, self-hosted AI

Achieves 74.6% LiveCodeBench pass@1 using a frozen 14B model on a single consumer GPU.
Employs constraint-driven generation and self-verified iterative refinement for autonomous specialization.
Supports code generation, reasoning, and general knowledge via a fully self-hosted pipeline.
Leverages PlanSearch, Geometric Lens, and PR-CoT for structured generation and self-repair.

Created 1 month ago

Updated 2 days ago

Starred 23 hours ago by

Jeffrey Morgan(Cofounder of Ollama). Also starred by

Tim Suchanek(Founder of expand.ai),

Thomas Dohmke(Former CEO of GitHub),

Kent Dodds(Cofounder of Remix), and

22 more.

pi-mono by badlogic

6.6%

29k

AI agent framework and LLM deployment tools

Unified multi-provider LLM API and agent runtime with tool calling and state management.
Interactive coding agent CLI and terminal UI library with differential rendering.
CLI for managing vLLM deployments on GPU pods.
Web components for AI chat interfaces and a CORS proxy.

Created 7 months ago

Updated 21 hours ago

Starred 1 day ago by

Jeffrey Morgan(Cofounder of Ollama). Also starred by

Matteo Collina(Cofounder of Platformatic).

omlx by jundot

12.7%

LLM inference server for Apple Silicon

Optimized inference with continuous batching and tiered KV caching (RAM + SSD).
Supports LLMs, VLMs, OCR, embeddings, and rerankers via FastAPI.
Managed via macOS menu bar app and admin dashboard; OpenAI API compatible.
Enables practical local LLM use with persistent context caching across restarts.

Created 1 month ago

Updated 10 hours ago

Starred 1 day ago by

Jeff Hammerbacher(Cofounder of Cloudera).

obsidian-gemini by allenhutchison

4.2%

273

Gemini AI agent for Obsidian knowledge management

AI agent actively manages Obsidian vault via tool calling: search, edit, create, research.
Semantic vault search and context-aware AI personalize interactions with your notes.
Features IDE-style text completions, multimodal image analysis, and smart summarization.
Extensible agent skills and MCP server support integrate external tools and workflows.

Created 1 year ago

Updated 22 hours ago

Starred 1 day ago by

Emile Vauge(Founder of Traefik).

HolyClaude by CoderLuii

4268.0%

994

AI development workstation container

Integrated Claude Code, 5 AI CLIs, and web UI for seamless AI-assisted coding.
Includes a headless Chromium browser with Playwright for automated web tasks.
Pre-configured Docker environment resolves common setup issues for immediate use.
Bundles over 50 dev tools and multiple AI agents in a single container.

Created 1 week ago

Updated 14 hours ago

Starred 1 day ago by

Junyang Lin(Core Maintainer at Alibaba Qwen). Also starred by

Shizhe Diao(Author of LMFlow; Research Scientist at NVIDIA),

Meng Zhang(Cofounder of TabbyML),

Ying Sheng(Coauthor of SGLang), and

6 more.

miles by radixark

2.6%

Enterprise RL for large-scale MoE models

Production-grade reinforcement learning framework for massive Mixture-of-Experts models.
Achieves stable inference alignment via true on-policy training and numerical consistency.
Enhances memory robustness and accelerates rollouts using speculative training methods.
Supports new hardware and targets multimodal training for advanced AI applications.

Created 5 months ago

Updated 18 hours ago

Starred 1 day ago by

Junyang Lin(Core Maintainer at Alibaba Qwen). Also starred by

Johannes Hagemann(Cofounder of Prime Intellect),

Travis Fischer(Founder of Agentic),

Jeff Hammerbacher(Cofounder of Cloudera), and

3 more.

prime-rl by PrimeIntellect-ai

1.8%

Decentralized RL training codebase for scaling

Trains models (e.g., DeepSeek-R1-Distill-Qwen-1.5B) on datasets like math-hendrycks-genesys.
Supports multi-node, multi-GPU inference with DP, TP, and PP parallelism.
Uses VLLM for inference; config files define training/inference parameters.

Created 1 year ago

Updated 10 hours ago

...