Quantized builds of gr33r/ux-writing-1 (a UX-writing reviewer fine-tuned from Qwen3.6-27B — 83% blinded human preference over the base model) for llama.cpp, LM Studio, Ollama, and friends. $0 marginal cost: review your product's copy on your own laptop.

file	size	fits
`ux-writing-1-Q4_K_M.gguf`	16.6 GB	24 GB Macs / 24 GB GPUs
`ux-writing-1-Q8_0.gguf`	28.6 GB	32–48 GB machines, near-bf16 quality

Note: converted without the optional multi-token-prediction draft layer (same as the standard Qwen3.6 GGUFs).

Quick start

bash

llama-cli -m ux-writing-1-Q4_K_M.gguf --temp 0 -p "Rewrite this inline error to be
clear and blame-free: 'Invalid' (email field on a signup form). Reply with compact
JSON {\"rewrite\", \"reason\", \"risk\"}."

For the full prompt contract (system prompt, code-context format), the codebase-scanning CLI, and how to fine-tune on your own style guide, see the main model card and repo.

gr33r/ux-writing-1-GGUF

ux-writing-1 — GGUF

Quick start

gr33r/
ux-writing-1-GGUF