Projects Recently Starred by Experts

Starred 11 hours ago by

Mehdi Amini(Author of MLIR; Distinguished Engineer at NVIDIA). Also starred by

Chip Huyen(Author of "AI Engineering", "Designing Machine Learning Systems"),

Jeff Hammerbacher(Cofounder of Cloudera),

Junyang Lin(Core Maintainer at Alibaba Qwen), and

13 more.

flashinfer by flashinfer-ai

0.6%

Kernel library for LLM serving

High-performance CUDA kernels for attention: FlashAttention, SparseAttention, PageAttention.
Load-balanced scheduling alleviates load imbalance for variable-length inputs.
Memory efficiency via Cascade Attention, Head-Query fusion, low-precision & fused-RoPE.
Customizable attention variants via JIT compilation; CUDAGraph & torch.compile compatible.

Created 2 years ago

Updated 19 hours ago

Starred 13 hours ago by

Li Jiang(Coauthor of AutoGen; Engineer at Microsoft). Also starred by

Vincent Weisser(Cofounder of Prime Intellect),

Shawn Wang(Editor of Latent Space),

Mckay Wrigley(Founder of Takeoff AI), and

7 more.

skills by anthropics

4.7%

74k

Dynamic instruction sets for enhancing AI task performance

Automate specialized tasks like document generation, data analysis, and creative content creation.
Supports complex document formats (DOCX, PDF, PPTX, XLSX) and integrates with web technologies like React and Playwright.
Enables dynamic loading of instructions and resources, extending AI agent capabilities for real-world applications.
Includes pre-built skills powering production AI features and serving as developer reference examples.

Created 5 months ago

Updated 2 weeks ago

Starred 14 hours ago by

Edward Sun(Research Scientist at Meta Superintelligence Lab). Also starred by

Didier Lopes(Founder of OpenBB).

claude-code-system-prompts by Piebald-AI

5.4%

AI agentic development system prompt library

Defines prompts for core agents, utilities, and specialized sub-agents.
Enables code analysis, documentation generation, bash execution, and web interaction.
Facilitates agent creation, security review, and conversation summarization.
Includes detailed prompts for slash commands and browser automation tools.

Created 3 months ago

Updated 16 hours ago

Starred 17 hours ago by

Shizhe Diao(Author of LMFlow; Research Scientist at NVIDIA). Also starred by

Meng Zhang(Cofounder of TabbyML),

Ying Sheng(Coauthor of SGLang),

Lewis Tunstall(Research Engineer at Hugging Face), and

5 more.

miles by radixark

2.3%

899

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 4 months ago

Updated 13 hours ago

Starred 22 hours ago by

Shizhe Diao(Author of LMFlow; Research Scientist at NVIDIA). Also starred by

Vincent Weisser(Cofounder of Prime Intellect).

AI-Scientist-v2 by SakanaAI

0.9%

Agentic system for automated scientific discovery

Generates hypotheses, runs experiments, analyzes data, writes manuscripts.
Employs agentic tree search, guided by experiment manager agent.
Uses OpenAI, Gemini, Claude models; Semantic Scholar API for literature search.
Requires Linux, NVIDIA GPUs, CUDA, PyTorch; executes LLM-written code in sandbox.

Created 10 months ago

Updated 2 months ago

Starred 22 hours ago by

Shizhe Diao(Author of LMFlow; Research Scientist at NVIDIA). Also starred by

Chip Huyen(Author of "AI Engineering", "Designing Machine Learning Systems"),

Jeff Hammerbacher(Cofounder of Cloudera),

Carol Willing(Core Contributor to CPython, Jupyter), and

5 more.

AI-Scientist by SakanaAI

0.3%

12k

AI system for automated scientific discovery using LLMs

Automates research: idea generation, experiment execution, paper writing.
Supports NanoGPT, 2D Diffusion, Grokking via provided templates.
Uses OpenAI, Anthropic, DeepSeek, Gemini, OpenRouter APIs.
Includes community-contributed templates for diverse research areas.

Created 1 year ago

Updated 2 months ago

Starred 23 hours ago by

Jeff Hammerbacher(Cofounder of Cloudera). Also starred by

Chris Van Pelt(Cofounder of Weights & Biases),

Stas Bekman(Author of "Machine Learning Engineering Open Book"; Research Engineer at Snowflake),

Luis Capelo(Cofounder of Lightning AI), and

3 more.

tensorizer by coreweave

0.3%

288

Module for fast model serialization/deserialization

Enables fast model loading from HTTP/HTTPS, Redis, and S3 endpoints.
Serializes models/tensors into a single file for efficient loading.
Supports tensor weight encryption/decryption with libsodium.
Provides pre-tensorized models on CoreWeave Cloud, including Stable Diffusion.

Created 3 years ago

Updated 2 weeks ago

Starred 1 day ago by

Didier Lopes(Founder of OpenBB).

worldmonitor by koala73

78.4%

12k

AI-powered global intelligence dashboard

Aggregates 100+ news feeds, monitors geopolitical events, and tracks infrastructure.
Interactive 3D globe visualizes 30+ AI-enhanced data layers via deck.gl.
Employs LLM synthesis for AI briefs and browser-side ML (Transformers.js).
Offers native desktop (Tauri) and PWA with offline map support.

Created 1 month ago

Updated 12 hours ago

Starred 1 day ago by

Travis Fischer(Founder of Agentic).

ClawWork by HKUDS

245.2%

AI coworker framework for real-world economic validation

Empowers AI agents to perform professional tasks, generate income, and achieve economic solvency.
Integrates with LLMs (GPT-4o, etc.) and Nanobot, supporting multimodal interactions for diverse work.
Features a live economic benchmark with real-time cost tracking and LLM-based quality evaluation.
Agents operate under extreme economic pressure, proving AI's value in professional, self-sustaining roles.

Created 1 week ago

Updated 1 day ago

Starred 1 day ago by

Pankaj Gupta(Cofounder of Baseten). Also starred by

Tri Dao(Chief Scientist at Together AI),

Chip Huyen(Author of "AI Engineering", "Designing Machine Learning Systems"),

Mehdi Amini(Author of MLIR; Distinguished Engineer at NVIDIA), and

22 more.

cutlass by NVIDIA

0.2%

CUDA C++ and Python DSLs for high-performance linear algebra

Optimized GEMM, convolution, and linear algebra kernels for NVIDIA GPUs.
Supports diverse data types: FP64 to binary, block-scaled, and narrow integers.
Python DSLs (CuTe) enable easier CUDA kernel development and DL framework integration.
Leverages Tensor Cores across Volta to Blackwell architectures for peak GPU performance.

Created 8 years ago

Updated 6 days ago

...