EAGLE-4 README

项目的 README,来自 github.com/eagle-hq/EAGLE-4。合并为一个 markdown 文件:README.md。

Eagle

EAGLE-4

One Flexible Drafter for All Batch Sizes and Hardware

Paper (EAGLE-4) · EAGLE-4 draft for Qwen3.8-27B · Documentation

eagle-harness on PyPI Maintenance Contributions welcome

Eagle website Eagle on LinkedIn Eagle on X GitHub organization HuggingFace

License


EAGLE-4 single-request speedup over vanilla decoding, from the paper

Figure 1(a) of the EAGLE-4 paper: Qwen3.8-27B on the SPEED-Bench qualitative split, 4× NVIDIA B300, one request, temperature 0. The numbers this release measured with EAGLE-4 are in the support matrix.

EAGLE-4

Leave speculation on. EAGLE-4 adapts to your hardware, batch and workload.

For high load, serve SGLang on four B300 GPUs with the high-load recipe, which keeps pace with plain decoding up to 512 concurrent requests.

EAGLE-4 (Extrapolation Algorithm for Greater Language-model Efficiency) speeds up LLM inference with speculative decoding. EAGLE-4 draft models are built to run with eagle-harness, which decides how far to speculate from measured costs on your setup, so every draft tree fits the batch size, the hardware and the workload. The output stays exactly what the target model would generate on its own: up to 3.4× faster generation on Qwen3.8-27B, with the same answers. Measured speedups

This repository holds eagle-harness, the package that brings EAGLE-4 to SGLang, vLLM and TokenSpeed. Install it, point your engine at an EAGLE-4 draft model, and it serves with EAGLE-4.

Quick start

The quick start serves on one GPU and is tested on one NVIDIA B300 with Linux and a CUDA 13 driver. Qwen3.8-27B and its EAGLE-4 draft download on the first start (about 62 GB). The measured speedups use four B300 GPUs. Add --tp-size 4 to either serve command to serve on four, and see the engine guides for the exact measured commands.

Start an SGLang server with Docker:

docker run --gpus all --ipc host -p 127.0.0.1:30000:30000 \
  -e HF_HOME=/hf -v ~/.cache/huggingface:/hf \
  ghcr.io/eagle-hq/eagle-4-sglang:0.1.3 \
  sglang serve --model-path Qwen/Qwen3.8-27B --host 0.0.0.0 \
  --disable-overlap-schedule --speculative-algorithm EAGLE_HARNESS \
  --speculative-draft-model-path EAGLE-Corp/Qwen3.8-27B-EAGLE4 \
  --speculative-num-steps 12 --speculative-eagle-topk 4 --speculative-num-draft-tokens 16 \
  --speculative-algorithm-config '{}'

Or install with pip, in Python 3.12. Pip installs also need the CUDA toolkit and a C++ compiler:

python3.12 -m venv .venv && source .venv/bin/activate
pip install "eagle-harness[sglang]" "sglang==0.5.23.dev252+eagle.9" \
  --extra-index-url https://eagle-hq.github.io/EAGLE-4/whl/cu130/ \
  --extra-index-url https://download.pytorch.org/whl/cu130 \
  --extra-index-url https://pypi.nvidia.com
sglang serve --model-path Qwen/Qwen3.8-27B \
  --speculative-algorithm EAGLE_HARNESS \
  --speculative-draft-model-path EAGLE-Corp/Qwen3.8-27B-EAGLE4

When the server logs The server is fired up and ready to roll!, send a request from another terminal:

curl http://localhost:30000/v1/chat/completions -H 'Content-Type: application/json' \
  -d '{"model": "Qwen/Qwen3.8-27B", "messages": [{"role": "user", "content": "Explain how a hash table works."}], "max_tokens": 256}'

The first requests after a start compile kernels and measure live costs, so they run slower. Send a few requests before you measure speed, and compare against the same command without the --speculative-* flags (benchmarking).

For vLLM, TokenSpeed and AMD MI355X, follow the engine guides.

Measured speedup

Each speedup divides EAGLE-4 throughput by the same engine's throughput without speculation, on Qwen3.8-27B at temperature 0 on the SPEED-Bench client, with the measured recipe of each engine guide. The support matrix has the tokens/s, the correctness checks and how each number was measured.

Engine GPUs 1 request 8 concurrent 64 concurrent
SGLang 4× B300 3.42× 2.09× 1.39×
vLLM 4× B300 2.83× 1.72× 1.14×
TokenSpeed 4× B300 3.09× 2.02× 1.41×

On one AMD MI355X, SGLang with EAGLE-4 is 2.07× faster at 8 concurrent requests and 1.56× at 32.

More

Citation

If you use EAGLE-4, cite the EAGLE-4 paper. EAGLE-4 builds on EAGLE, EAGLE-2 and EAGLE-3, which CITATION.cff also lists.

@misc{eagle4,
    author = {EAGLE-4 Team},
    title = {{EAGLE-4}: One Flexible Drafter for All Batch Sizes and Hardware},
    year = {2026},
    note = {Preprint}
}

Acknowledgements

License and trademark policy

Eagle 4 is source-available software, not OSI-approved open-source software. The terms of the Eagle 4 Source-Available License govern its use, modification, and distribution, and incorporate the Eagle Trademark Policy. See also the NOTICE file.

Commercial Use of a Competing Offering requires a separate written commercial license regardless of revenue. Other Commercial Use is permitted only within the conditions of the license, including the US $20 million Aggregate Revenue threshold measured over any consecutive 12-month period. If the threshold is exceeded, a separate commercial agreement is required before beginning or continuing any Commercial Use. The license specifies that the requirement continues after a subsequent revenue decrease.

Downloading, cloning, forking, copying, accessing, modifying, using, hosting, distributing, or otherwise exercising any licensed right constitutes acceptance of the Source-Available License and incorporated Eagle Trademark Policy. Mere viewing or inspection does not itself constitute trademark use.

For separate commercial terms, see the Commercial License Agreement and contact [email protected]. Publishing that agreement does not grant commercial-license rights; it becomes binding only as stated in its Section 1. Files derived from sglang, vLLM, FlashInfer and other projects keep those projects' notices, listed in THIRD_PARTY_NOTICES.md.

Contributing

By submitting, uploading, pushing, committing, proposing, or otherwise providing a pull request, patch, issue attachment, model, weight, dataset, documentation, or other contribution to this repository, you agree to the Eagle Contribution License and Submission Terms (PDF). If you do not agree, do not submit a contribution. If you act for an employer, client, partner or other entity, you represent that you have authority to submit the contribution and have obtained all required permissions. Submitting a contribution does not change the license of existing Eagle code; Eagle AI Corp controls outbound licensing for accepted contributions. See CONTRIBUTING.md for the licensing structure and for how to build, test and open a pull request.

EAGLE-4 (eagle-hq/EAGLE-4)

正在搜索数据库…

1.3K ·GitHub
这是什么?EAGLE-4 是一种投机解码方法,可以加速大语言模型推理。本仓库提供的 eagle-harness 是一个软件包,让推理服务引擎能够使用 EAGLE-4 的草稿模型,且不改变模型的输出结果。
这是独立的服务器,还是我可以接到 SGLang 之类引擎里使用的软件包?它是 eagle-harness,一个为现有推理服务引擎(SGLang、vLLM、TokenSpeed)提供 EAGLE-4 支持的软件包,本身并不是新的引擎;它位于引擎与草稿模型之间。
我该怎么让它跑起来?使用提供的 Docker 镜像启动 SGLang 服务器,或在 Python 3.12 下通过 pip 安装 eagle-harness,然后将引擎指向一个 EAGLE-4 草稿模型,待服务器日志显示就绪后再发送请求。
我需要特定的 GPU 配置吗?快速开始的内容是在单块 NVIDIA B300、Linux 系统和 CUDA 13 驱动环境下构建并测试的;实测加速数据使用的是四块 B300 GPU,此外 AMD MI355X 也通过 ROCm 移植版获得支持。
有已知的限制或硬件支持方面的缺口吗?仓库单独维护了一份支持矩阵,按引擎和 GPU 列出 tokens/s、正确性检查及已知限制,因为性能表现和加速效果会随硬件、批处理大小和工作负载而变化。
EAGLE-4 是开源的吗?不是。它属于源代码可获取但未通过 OSI 认证的软件:年营收门槛为每 12 个月 2000 万美元,低于该门槛时商业使用免费;任何“竞争性产品(Competing Offering)”的使用,无论营收多少,都需要单独的商业许可证。
使用投机解码会改变模型的回答吗?不会。EAGLE-4 会用目标模型对草稿进行校验,因此输出结果与不使用投机解码时完全一致,只是速度更快。
使用 EAGLE-4 后推理能快多少?在四块 B300 GPU 上用 SGLang 运行 Qwen3.8-27B 时,单个请求最高可提速 3.42 倍;在更高并发下提速幅度较小,但仍然可观,为 1.1 至 2 倍。
eagle-harness 在运行时实际会决定什么?它会根据你的具体批处理大小、硬件和工作负载所实测的开销,决定要向前投机推测多远,而不是使用固定的投机深度。
它只能配合 SGLang 使用吗?不是,它同样支持 vLLM 和 TokenSpeed,并为每个引擎提供了单独的指南,列出了经过实测的确切命令。
作者
eagle-hq
大小
2.6 MB
星标
1.3K
时长
29天
提交 · 90 天