---
license: llama3.2
base_model:
- meta-llama/Llama-3.2-3B-Instruct
tags:
- coreml
---
CoreML conversion of Llama-3.2-3B-Instruct with a 512 context length. Optimized for Apple Neural Engine.

Use [this CLI](https://github.com/smpanaro/coreml-llm-cli) to download and run inference. macOS 14 (Sonoma) is required.

> [!IMPORTANT]
> This model will likley run slowly or not at all on M1 Macs and phones. Consider trying the 1B model for those devices: [smpanaro/Llama-3.2-1B-Instruct-CoreML](https://huggingface.co/smpanaro/Llama-3.2-1B-Instruct-CoreML)