Gentoo Packages

Search

Install

Install this package:

emerge -a sci-ml/colibri

Package Information

Description:

Colibri runs GLM-5.2 (744B-parameter MoE) on consumer hardware with as little as 25GB RAM. Pure C engine with zero runtime dependencies, streaming experts from disk with an LRU cache. Supports MLA attention, MTP speculative decoding, multiple conversations per server, and optional CUDA/ROCm or Vulkan GPU offloading.

Homepage:

https://github.com/JustVugg/colibri

License:

Apache-2.0

Versions

Metadata

Description

• Colibri runs GLM-5.2 (744B-parameter MoE) on consumer hardware with as little as 25GB RAM. Pure C engine with zero runtime dependencies, streaming experts from disk with an LRU cache. Supports MLA attention, MTP speculative decoding, multiple conversations per server, and optional CUDA/ROCm or Vulkan GPU offloading.

Maintainers

• Benny Powers (web@bennypowers.com) - Type: person

Upstream

• Remote IDs:
  • github: JustVugg/colibri

<pkgmetadata>
	<maintainer type="person">
		<email>web@bennypowers.com</email>
		<name>Benny Powers</name>
	</maintainer>
	<longdescription lang="en">
		Colibri runs GLM-5.2 (744B-parameter MoE) on consumer hardware with
		as little as 25GB RAM. Pure C engine with zero runtime dependencies,
		streaming experts from disk with an LRU cache. Supports MLA attention,
		MTP speculative decoding, multiple conversations per server, and
		optional CUDA/ROCm or Vulkan GPU offloading.
	</longdescription>
	<use>
		<flag name="bench">Install dependencies for quality benchmarks
			(MMLU, HellaSwag, ARC)</flag>
		<flag name="cuda">Build with NVIDIA CUDA GPU offloading for
			resident expert tensors</flag>
		<flag name="python">Install the coli Python CLI for model management,
			chat, serving, and weight conversion</flag>
		<flag name="rocm">Build with AMD ROCm/HIP GPU offloading for
			resident expert tensors</flag>
		<flag name="vulkan">Build the Vulkan compute backend for the
			shared routed-expert tier and dense chains (any Vulkan 1.2
			driver; independent of cuda/rocm)</flag>
	</use>
	<upstream>
		<remote-id type="github">JustVugg/colibri</remote-id>
	</upstream>
</pkgmetadata>

Lint Warnings

• [Warning] Ebuild 2.1.0 has a line exceeding 80 positions on line 12.
• [Warning] Ebuild 2.1.0 has a line exceeding 80 positions on line 72.
• [Warning] Ebuild 2.1.0 has a line exceeding 80 positions on line 122.
• [Warning] Ebuild 2.1.0 has a line exceeding 80 positions on line 125.
• [Warning] Ebuild 2.0.0 has a line exceeding 80 positions on line 12.
• [Warning] Ebuild 2.0.0 has a line exceeding 80 positions on line 72.
• [Warning] Ebuild 2.0.0 has a line exceeding 80 positions on line 122.
• [Warning] Ebuild 2.0.0 has a line exceeding 80 positions on line 125.
• [Notice] Package has only unstable keywords for arch(es): [ amd64 ], all versions are unstable: [ 2.0.0, 2.1.0 ]

USE Flags

Manage flags for this package: euse -i <flag> -p sci-ml/colibri | euse -E <flag> -p sci-ml/colibri | euse -D <flag> -p sci-ml/colibri

Manifest