license: apache-2.0
tags:

• gguf
• abliterated
• muse
• glimmer
• uncensored
• text-generation
  base_model: mlasli/Muse-Glimmer-30B-Abliterated-BF16

Muse Glimmer 30B Abliterated — Q6_K GGUF

This is the Q6_K GGUF quantization of Muse Glimmer 30B Abliterated BF16. The underlying model has been abliterated — its internal refusal mechanism substantially suppressed via weight-level intervention. Q6_K offers near-reference quality in a compact ~25 GB package.

For the full abliteration methodology (how the refusal direction was computed and removed, hardware used, mathematical details), see the BF16 model card.

Abliteration Summary

Abliteration is a post-training technique that directly modifies model weights to remove learned refusal behavior. The process:

1. Collected hidden states at layer 33/52 (65% depth) from 256 harmful + 256 harmless prompt pairs on an A100 80GB GPU.
2. Computed the refusal direction as the normalized difference between harmful and harmless hidden state means (separation score: 86.34).
3. Subtracted (\alpha = 0.15 \times (\mathbf{r} \otimes (W^T \mathbf{r}))) from o_proj and down_proj weights in all 52 layers.
4. Result: refusal rate dropped from 3/3 to 1/3 on held-out harmful prompts (hacking guide and ransomware now comply; weapons prompt still blocked).

Quantization Details

Q6_K uses 6-bit quantization with the K-quant strategy, which assigns higher precision to attention weights and key layers while using lower precision for less critical components. This provides what is generally considered the best tradeoff for quality-critical workloads — perceptually identical to FP16 for most use cases while halving the memory footprint.

• Size: ~25 GB
• Quality: Excellent — near-indistinguishable from FP16 for most tasks
• Recommended hardware: 48 GB GPU (A6000, dual 3090s), or 64 GB system RAM with partial GPU offloading

Usage

llama.cpp

# Download the GGUF file
huggingface-cli download mlasli/Muse-Glimmer-30B-Abliterated-Q6_K-GGUF \
  --local-dir ./models

# Full GPU offload (fits in 48 GB)
./llama-cli -m ./models/Muse-Glimmer-30B-Abliterated-Q6_K.gguf \
  -p "Explain how a CPU works in detail." \
  -n 512 --temp 0.7 -ngl 99

# CPU with partial offload
./llama-cli -m ./models/Muse-Glimmer-30B-Abliterated-Q6_K.gguf \
  -p "Explain how a CPU works in detail." \
  -n 512 --temp 0.7 -ngl 10

Ollama

Create a Modelfile:

FROM ./Muse-Glimmer-30B-Abliterated-Q6_K.gguf
PARAMETER temperature 0.7
PARAMETER num_ctx 8192

ollama create muse-glimmer-30b-abliterated -f Modelfile
ollama run muse-glimmer-30b-abliterated

Available Quantizations

Vision (Multimodal)

This model accepts image input when paired with a vision projector (mmproj).
Abliteration only modified the language backbone — the vision encoder is
untouched — so the standard Meta projector works directly with this repo.

This repository bundles mmproj-Muse-Glimmer-30B-Q4_K_M.gguf (~1.4 GB), Meta's official vision encoder

• projector for Muse Glimmer 30B.

Usage (llama.cpp)

huggingface-cli download mlasli/Muse-Glimmer-30B-Abliterated-Q6_K-GGUF \
  --include "Muse-Glimmer-30B-Abliterated-Q6_K.gguf" \
  --include "mmproj-Muse-Glimmer-30B-Q4_K_M.gguf" \
  --local-dir ./models

./build/bin/llama-mtmd-cli \
  -m ./models/Muse-Glimmer-30B-Abliterated-Q6_K.gguf \
  --mmproj ./models/mmproj-Muse-Glimmer-30B-Q4_K_M.gguf \
  --image photo.png \
  -p "Describe this image."

Ollama note: Ollama does not currently support separate mmproj files
for this architecture. For image input, use llama.cpp (llama-mtmd-cli or
llama-server --mmproj).

Limitations & Disclaimers

• This is an abliterated model — it has been modified to refuse fewer prompts. Use responsibly.
• Some refusal pathways remain (notably weapons-related content). This is not a fully uncensored model.
• Abliteration may subtly affect output quality; (\alpha = 0.15) was chosen conservatively.
• No formal benchmark evaluation has been performed on the abliterated model.
• The vision encoder is untouched by abliteration. Image input is available via the bundled mmproj projector (llama.cpp only; see above).
• This model will generate content the original would refuse. Comply with applicable laws.

License: Apache 2.0

Changelog

v1.1.0 — vision (multimodal) support (2026-08-16)

• Added mmproj-Muse-Glimmer-30B-Q4_K_M.gguf (~1.4 GB), Meta's official vision encoder + projector,
  enabling image input via llama.cpp.
• The vision tower is untouched by abliteration, so this projector matches the
  base model (meta-models/Muse-Glimmer-30B).
• v1.0.0 was the initial (unversioned) text-only upload.