Modelpublic

oomu/OOMU-Qwen3-0.6B-Production-MicroRouter-GGUF

sourceHugging Faceapache-2.0updated 1mo agoView on Hugging Face

0likes39downloads

Model Card

OOMU Qwen3 0.6B Production MicroRouter

OOMU-Qwen3-0.6B-Production-MicroRouter-Q8_0.gguf is OOMU's compact, Metal-targeted routing model. It is a LoRA-trained derivative of Qwen/Qwen3-0.6B, merged into the official instruction checkpoint and quantized to Q8_0 with a pinned llama.cpp toolchain.

This is not a general-purpose assistant model. It is intended for OOMU's bounded staged micro-router contract, where native code supplies the allowed capability grammar, validates every model decision, and executes effects only through OOMU's native broker.

Identity

Field	Value
Artifact	`OOMU-Qwen3-0.6B-Production-MicroRouter-Q8_0.gguf`
Size	639,446,784 bytes
SHA-256	`8c9701724873ead66764b16a544f5a15d436654e86e3d96fe64aa6cc7980c5bb`
Quantization	Q8_0
Base checkpoint	`Qwen/Qwen3-0.6B@c1899de289a04d12100db370d81485cdf75e47ca`
Selected LoRA SHA-256	`57866bab45bcca2cfdd71290ceadbc491206288c6b836ebe1e260574538cde64`
Training authority commit	`9c08b9ee7b75dfc8168df30041d9f918984fa58a`

Runtime profile

—Context: 4,608 tokens
—Maximum generated output: 96 tokens
—Thinking mode: disabled
—Required OOMU production backend on Apple silicon: Metal, fail closed if unavailable

License and attribution

This derivative is distributed under Apache License 2.0. Qwen3 attribution and the complete license are preserved in the Hugging Face repository. See the upstream `Qwen/Qwen3-0.6B` model card for the base model's documentation and limitations.