yalun753/moe-l2
MoE expert offload for low-VRAM GPUs — run 100B+ MoE models (DeepSeek, Qwen, Mixtral) on 8 GB cards. Expert cache with LRU hot experts, OpenAI-compatible proxy, GGUF multi-shard.
- Stars
- 146
- Forks
- 5
- Open issues
- 1
- Last push
- 25d ago
About moe-l2
moe-l2 is a free, open-source AI inference tool by yalun753, written mainly in Cuda. MoE expert offload for low-VRAM GPUs — run 100B+ MoE models (DeepSeek, Qwen, Mixtral) on 8 GB cards. Expert cache with LRU hot experts, OpenAI-compatible proxy, GGUF multi-shard. It has 146 GitHub stars and 5 forks, and was last updated 25d ago. It's released under the Apache-2.0 license, a standard open-source license. According to the project, it needs a GPU with at least 9.1 GB of memory.
- License
- Apache-2.0
- Commercial use
- Commercial OK
- Language
- Cuda
- Stars
- 146
- Last update
- 25d ago
- Runs on Mac
- Not stated
- GPU
- 9.1 GB+ (as stated by the project)
- Paid API key
- Not required
- Self-hostable
- Not stated
Similar free tools
llama.cpp132k stars · LLM inference in C/C++
llm-action25k stars · 本项目旨在分享大模型相关技术原理以及实战经验（大模型工程化、大模型应用落地）
TensorRT-LLM15k stars · Optimizes inference for large language models on NVIDIA GPUs.
BitNet41k stars · Optimized inference framework for 1-bit LLMs on CPU and GPU.
ai-infra-book7.0k stars · 《深入理解 AI Infra：量化分析与系统设计》（李博杰 著）开源书稿：从硬件约束和模型架构出发，量化推导 LLM 推理与训练系统设计。含全书正文、PDF、配套计算工具与实验
vllm95k stars · High-throughput inference engine for LLMs.
moe-l2 FAQ
Can I use moe-l2 commercially?
Yes. Apache-2.0 is a permissive license: commercial use is allowed as long as you keep the license notice. This is a plain-English summary, not legal advice.
Is moe-l2 free?
Yes. moe-l2 is open source and free to use. It's released under the Apache-2.0 license, a standard open-source license.
Does moe-l2 run on a Mac?
The project doesn't say. Check its installation guide before you start.
Does moe-l2 need a GPU?
According to the project, moe-l2 needs a GPU with at least 9.1 GB of memory.
Is moe-l2 still maintained?
Yes. It was last updated 25d ago and has 1 open issues on GitHub.
What are alternatives to moe-l2?
Similar free, open-source tools include llama.cpp, llm-action, TensorRT-LLM, BitNet.