-[](https://huggingface.co/models/
/)
+[](https://huggingface.co/lmsys/sglang-EAGLE3-Llama-4-Scout-17B-16E-Instruct-v1)
+[](https://huggingface.co/lmsys/sglang-EAGLE3-Llama-4-Maverick-17B-128E-Instruct-v1)
[](./LICENSE)
-
## Table of Contents
- [📍 Overview](#-overview)
From 3af76ea71fad3c1ef4c4ff1c24def77f62e3a5b5 Mon Sep 17 00:00:00 2001
From: fzyzcjy <5236035+fzyzcjy@users.noreply.github.com>
Date: Wed, 23 Jul 2025 12:32:58 +0800
Subject: [PATCH 02/15] Update README.md (#52)
---
README.md | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/README.md b/README.md
index b5fdeb28..d785b886 100644
--- a/README.md
+++ b/README.md
@@ -53,7 +53,7 @@ We provide two orthogonal paths so everyone can start training in minutes, regar
> **Why does disk matter?**
> During Eagle3 training, the frozen target model will first generate the hidden states for each token given the data sample. The hidden states are fed to the draft model for training.
-> Offline mode stores these hidden states to the local disk, so a small disk fills can be filled up fast.
+> Offline mode stores these hidden states to the local disk, so a small disk can be filled up fast.
> Online mode only generates these hidden states on the fly without storing them to the disk, but needs to keep the draft model alive during training, trading GPU RAM for almost-zero disk footprint.
### ⚡️ SGLang-ready
From bcb6b59544209c77403e3f12d990f611652aef99 Mon Sep 17 00:00:00 2001
From: Yineng Zhang 
+
+We would like to express our sincere gratitude to the official EAGLE team, especially Hongyang Zhang and Yuhui Li, for their invaluable contributions and support. Our thanks also go to the NVIDIA team—particularly Avery H and Izzy Putterman—and to the Google team, especially Ying Wang, for their insightful discussions and generous assistance throughout the project.
+
+We are especially grateful to Meituan for their strong backing and meaningful contributions, which played a vital role in driving this project forward.
This project has also been inspired by many outstanding open-source projects from the LLM community, including [EAGLE](https://github.com/SafeAILab/EAGLE), [BaldEagle](https://github.com/NickL77/BaldEagle), and [TensorRT-Model-Optimizer](https://github.com/NVIDIA/TensorRT-Model-Optimizer) and others. Their contributions and shared knowledge have greatly benefited our work.
diff --git a/assets/acknowledgements.png b/assets/acknowledgements.png
new file mode 100644
index 0000000000000000000000000000000000000000..c903f2ccc433d41569978cf51040c50909442c07
GIT binary patch
literal 382844
zcmeFaRa9Nw(k+U+1qg1zg1as(xCD3C;O-8=9fAfYxVr{-4esvl?r_Ps<=_8lZGVT{
z*6w}cZLT$BGOKD-@73pEtRQJgVR#rU7$6`Zcu^4nSs);oKp-G+G$@FdZwh%c`G9~x
zD2?Tn?UW_N0eY6^G&)}_b@geS%&lJ9fq*!9oUC;8O!e&uboC95Ew~7e8(IhnjK6RZ
zDltpYNm%jg8ySnZ*yzi-NXqNEnCh{AA>`qP;dBDLOkl2Wr$gXmZf0Q%aN;8TYhJ+1
z_3y7~2?_q{VrR-ls4O8(z;9`zPryP$M?*)*4MV_b^ThxlD