We've been working with Docker to make running quantized models easier with Docker. This integration allows users to pull and run Unsloth-optimized models using the native Docker CLI.
Key features:
Uses docker model run syntax.
Supports Dynamic GGUFs (handling quantization overhead).
Cross-platform support.
Example usage: docker model run ai/gpt-oss:20B
Happy to answer questions about the implementation or the quantization methods used.