hal0-quantize

hal0-quantize is a Local & open models skill for AI agents, published by Hal0ai in the hal0 repository on GitHub. Skillspace rates hal0-quantize C (78/100) and it passed the safety scan; the rating is based on static analysis only.

Creating ROCmFPX-family quants (ROCmFP3/4/6/8, straight or agent) from BF16/F16 GGUF sources on hal0 / Strix Halo, using the Hal0ai/Hal0_ROCmFPX runner. Use when asked to quantize a model to ROCmFP4 (or FP3/FP6/FP8), build the ROCmFPX llama.cpp tree for a GPU arch, or produce an agent-coherent quant for tool-calling / JSON / coding workloads. Encodes the clone→build→quantize contract and the straight-vs-agent tensor-routing choice.

Grade
C
Quality
78 / 100
Safety
Passed
GitHub stars
56

Unproven: rated by static analysis only, not yet compared head-to-head.

This rating describes the version of the repository we crawled; it may have changed since, so go by its current files.

View source on GitHubHal0ai/hal0

Our review

Architecture
96 / 100
Maturity
Structured: ships references or scripts
Security scan
No issues found
License
Apache-2.0
Adoption
Adopted by 0 other repositories, in 0 collections
Scope
Vendor-specific: built around one product. Quantizing GGUF models with the public Hal0_ROCmFPX llama.cpp fork

Not checked yet

  • Not judged on the rubric yet (not a contender in its category)
  • No blind task eval yet
  • Sibling files (scripts, references) not fetched; security scan covers SKILL.md only

More Local & open models skills

Skillspace indexes and rates this skill; it does not host or endorse it. Run third-party skills at your own risk.