usage.md

Mamba adoption

We've been very happy to see Mamba being adopted by many organizations and research labs to speed up their training / inference. This page contains a partial list of places where Mamba is being used. If you'd like to add links to your organization / product / codebase, please open a PR or email us. We'd very much like to hear from you!

Large language models and multi-modal models

• Tencent's Hunyuan-TurboS (560B)

• Nvidia Nemotron-H (8B, 47B, 56B)

• AI21 Jamba (398B)

• TII Falcon-H1 (34B)

• IBM Bamba (9B)

• Mistral's Codestral (7B)

• Nvidia Mamba-2 Hybrid (8B)

• Microsoft Samba (4B)

• TII Falcon-Mamba (7B)

Inference frameworks

• vLLM

• Nvidia's TensorRT-LLM

Hardware

• Nvidia GPUs

• AMD GPUs

• AWS Trainium 2