Pinned Loading

1. pegainfer-project/pegainfer pegainfer-project/pegainfer Public

   Pure Rust + CUDA LLM inference engine — no PyTorch, OpenAI-compatible, serves Qwen3 to Kimi-K2

   Rust 737 121

2. InferMatrixLab/vllm-rlt InferMatrixLab/vllm-rlt Public

   A vLLM-inspired inference engine for looped transformers, with loop-level scheduling and execution across composable stages

   Python 221 39

3. InferMatrixLab/system1-omni InferMatrixLab/system1-omni Public

   community maintained vllm/vllm-omni style inference engine for system1 models in Rust for extreme efficient performance

   Python 63 17

4. FeathBow/invar FeathBow/invar Public

   When a new kernel changes your model's numbers, Invar decides whether the evidence supports the replacement for a declared use, with a confidence bound.

   Haskell 3 1

5. NVIDIA-NeMo/labs-molt NVIDIA-NeMo/labs-molt Public

   A scalable, agentic-first, and HuggingFace-native RL framework for research (9k lines).

   Python 1.2k 124

6. NVIDIA-NeMo/Automodel NVIDIA-NeMo/Automodel Public

   🚀 Pytorch Distributed native training library for LLMs/VLMs with OOTB Hugging Face support

   Python 1.1k 370