lmsys / SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge

huggingface.co
Total runs: 2.3K
24-hour runs: 52
7-day runs: -38
30-day runs: 637
Model's Last Updated: December 28 2025

Introduction of SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge

Model Details of SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge

specforge_team

EAGLE3 For meta-llama/Llama-3.1-8B-Instruct

About

SpecBundle is an open-source initiative, jointly driven by the community and industry, to democratize speculative decoding by providing high-performance speculative decoding draft weights for mainstream open-source models.

This checkpoint was trained by the SpecForge Team and released as the phase 1 of SpecBundle release . We regenerated the responses in the mlabonne/open-perfectblend and trained the model on 1.4M data samples for 2 epochs. This checkpoint was trained using the SpecForge framework.

Usage

You can use this checkpoint with the command below.

python3 -m sglang.launch_server \
    --model meta-llama/Llama-3.1-8B-Instruct \
    --speculative-algorithm EAGLE3 \
    --speculative-draft-model-path lmsys/SGLang-EAGLE3-Llama-3.1-8B-Instruct-perfect-blend-regenerated \
    --speculative-num-steps 3 \
    --speculative-eagle-topk 1 \
    --speculative-num-draft-tokens 4

Performance

This checkpoint exhibits superior performance on various benchmarks.

Throughput Acceptance Length

You can reproduce the performance with the command below:

# clone specforge
git clone https://github.com/sgl-project/SpecForge.git
cd SpecForge/benchmarks

# run benchmarks
python bench_eagle3.py \
        --model meta-llama/Llama-3.1-8B-Instruct \
        --speculative-algorithm EAGLE3 \
        --speculative-draft-model-path lmsys/SGLang-EAGLE3-Llama-3.3-70B-Instruct-perfect-blend-regenerated \
        --port 30003 \
        --config-list 1,3,1,4 1,5,1,6 1,5,3,6 1,7,1,8 1,7,4,8 \
        --benchmark-list gsm8k math500 mtbench humaneval livecodebench financeqa gpqa  \
        --dtype bfloat16 \
        --name llama3-perfect-blend-regen
Acknowledgement

We sincerely appreciate the collective efforts from both the developers in the open-source community and our industrial partners, especially Ant Group AQ Team, Meituan, Nex-AGI (Qiji Zhifeng), EigenAI for their invaluable contributions to the release of SpecBundle Phase 1.

Runs of lmsys SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge on huggingface.co

2.3K
Total runs
52
24-hour runs
-165
3-day runs
-38
7-day runs
637
30-day runs

More Information About SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge huggingface.co Model

More SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge license Visit here:

https://choosealicense.com/licenses/mit

SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge huggingface.co

SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge huggingface.co is an AI model on huggingface.co that provides SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge's model effect (), which can be used instantly with this lmsys SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge model. huggingface.co supports a free trial of the SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge model, and also provides paid use of the SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge. Support call SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge model through api, including Node.js, Python, http.

SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge huggingface.co Url

https://huggingface.co/lmsys/SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge

lmsys SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge online free

SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge huggingface.co is an online trial and call api platform, which integrates SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge's modeling effects, including api services, and provides a free online trial of SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge, you can try SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge online for free by clicking the link below.

lmsys SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge online free url in huggingface.co:

https://huggingface.co/lmsys/SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge

SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge install

SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge is an open source model from GitHub that offers a free installation service, and any user can find SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge on GitHub to install. At the same time, huggingface.co provides the effect of SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge install, users can directly use SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge installed effect in huggingface.co for debugging and trial. It also supports api for free installation.

SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge install url in huggingface.co:

https://huggingface.co/lmsys/SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge

Url of SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge

SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge huggingface.co Url

Provider of SGLang-EAGLE3-Llama-3.1-8B-Instruct-SpecForge huggingface.co

lmsys
ORGANIZATIONS

Other API from lmsys

huggingface.co

Total runs: 184.5K
Run Growth: -13.6K
Growth Rate: -7.35%
Updated:March 13 2024
huggingface.co

Total runs: 18.3K
Run Growth: -2.9K
Growth Rate: -15.87%
Updated:March 18 2024
huggingface.co

Total runs: 8.5K
Run Growth: -1.7K
Growth Rate: -20.38%
Updated:August 02 2023
huggingface.co

Total runs: 5.2K
Run Growth: -2.1K
Growth Rate: -41.40%
Updated:August 02 2023
huggingface.co

Total runs: 5.1K
Run Growth: 2.4K
Growth Rate: 45.87%
Updated:August 02 2023
huggingface.co

Total runs: 2.3K
Run Growth: -2.5K
Growth Rate: -111.40%
Updated:August 02 2023