From 13c66359dd995e8395ce42247d24145eb71625e6 Mon Sep 17 00:00:00 2001 From: youkaichao Date: Mon, 14 Apr 2025 14:00:47 +0800 Subject: [PATCH] =?UTF-8?q?=E5=8E=BB=E9=99=A4=E5=85=B7=E4=BD=93=E7=9A=84?= =?UTF-8?q?=E6=A1=86=E6=9E=B6=E5=90=8D=EF=BC=8C=E6=9C=AA=E6=9D=A5=E7=9A=84?= =?UTF-8?q?=E5=BC=80=E6=BA=90=E5=90=88=E4=BD=9C=E6=98=AF=E9=9D=A2=E5=90=91?= =?UTF-8?q?=E6=95=B4=E4=B8=AA=E5=BC=80=E6=BA=90=E7=A4=BE=E5=8C=BA=E7=9A=84?= =?UTF-8?q?=EF=BC=8C=E4=B8=8D=E5=B1=80=E9=99=90=E4=BA=8E=E5=85=B7=E4=BD=93?= =?UTF-8?q?=E6=9F=90=E4=BA=9B=E9=A1=B9=E7=9B=AE?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- OpenSourcing_DeepSeek_Inference_Engine/README.md | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) diff --git a/OpenSourcing_DeepSeek_Inference_Engine/README.md b/OpenSourcing_DeepSeek_Inference_Engine/README.md index e553e52..dd2cc0a 100644 --- a/OpenSourcing_DeepSeek_Inference_Engine/README.md +++ b/OpenSourcing_DeepSeek_Inference_Engine/README.md @@ -24,14 +24,12 @@ While we initially considered open-sourcing our full internal inference engine, - **Limited Maintenance Bandwidth**: As a small research team focused on developing better models, we lack bandwidth to maintain a large open-source project. -Considering these challenges, we’ve decided to collaborate with the SGLang project and vLLM project as more sustainable alternatives. +Considering these challenges, we’ve decided to collaborate with existing open-source projects as more sustainable alternatives. -SGLang and vLLM stand out for its broad hardware support, extensive model compatibility, and state-of-the-art performance. Its modular architecture gives us confidence that we can adapt and contribute key components from our internal engine to benefit the wider community. - -Moving forward, we will work closely with the SGLang team and vLLM team to: +Moving forward, we will work closely with existing open-source projects to: - **Extract Standalone Features**: Modularize and contribute reusable components as independent libraries. -- **Share Optimizations**: Contribute design improvements and implementation details directly to SGLang and vLLM. +- **Share Optimizations**: Contribute design improvements and implementation details directly. We are profoundly grateful for the open-source movement - from operating systems and programming languages to machine learning frameworks and inference engines. It’s an honor to contribute to this thriving ecosystem and to see our models