EAGLE-LLaMA3-Instruct-8B is an instruction-tuned 8 billion parameter model based on Llama 3 that incorporates EAGLE speculative decoding techniques for faster inference. It supports popular libraries including Transformers and vLLM and is available on Hugging Face. The model is built for developers seeking high-speed text generation while maintaining the quality of Llama 3 instruct capabilities.
EAGLE LLaMA3 Instruct 8B is a Text generation project. High latency in autoregressive text generation for instruction-tuned LLMs. It is built as an open-source project for developers. EAGLE LLaMA3 Instruct 8B is open source under the Open Source license. EAGLE LLaMA3 Instruct 8B is available on the web, the command line, and API.
Behind EAGLE LLaMA3 Instruct 8B is yuhuili, and it first shipped in 2023. The project is developed in the open on GitHub with 2.5k stars. Among its 3 catalogued features are text generation, vLLM support, and transformers integration.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Closest matches by what these projects do