← back to catalog · registered 2026-08-22 13:56

mlasli/Muse-Glimmer-30B-Abliterated-Q6_K-GGUF

mlasli 30B GGUF second-order 131K ctx
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model.

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/mlasli%2FMuse-Glimmer-30B-Abliterated-Q6_K-GGUF"
Response includes
  • classification m8
  • files 5
  • hub_downloads_all_time 1,808
  • author_summary 23 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M8
Primary method

Repackaging (quantization)

Applied on top of direct removal inherited from the base model.
Confidence
MEDIUM
Inherited from base model
Why this label 3 signals
Method inferred from partial signals - repository name, related files, or tag patterns. Producer identity not confirmed; label may sharpen or shift as we gather more evidence.
  • 'abliterated' in name/tags
  • is_gguf=1
  • assume M1 (base ablation) + M8 (GGUF quant) - default when producer unknown
Refusal direction extraction

No specific extraction method could be identified for this model. The producer either did not document it or used a proprietary pipeline.

What is a refusal direction? →
Downloads · lifetime
2K
164 last 30d - cooling
Likes
1
Model age
2mo ago
created 2026-08-11
Downloads over time
Now1.9K→from0↑0%
06861.4K2.1K0 on Aug 121.9K on Oct 111.9K on Oct 10AugSepOct
Aug 12 → Oct 11 · 49 snapshots · spans 60 days

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

Variants by this author 4 formats · 1K downloads combined

The same weights this author released in different packaging. Pick the format that matches your runtime.

Metadata

License
apache-2.0
Quantizations
Q6_K
Tags
gguf abliterated muse glimmer uncensored text-generation base_model:mlasli/Muse-Glimmer-30B-Abliterated-BF16 base_model:quantized:mlasli/Muse-Glimmer-30B-Abliterated-BF16 license:apache-2.0 endpoints_compatible region:us conversational

Related

Total size
21.3 GB
Files
5
Quantizations
3
Registered
2026-08-22 13:56
Last updated on HF
2026-08-16 10:25

Files by quantization

Q6_K 1 file 21.3 GB
Muse-Glimmer-30B-Abliterated-Q6_K.gguf 21.3 GB e7f3f3ac download
Q4_K 1 file 1.30 GB
mmproj-Muse-Glimmer-30B-Q4_K_M.gguf 1.30 GB f48b4523 download
Auxiliary files 3 files 8.12 KB
README.md 5.62 KB 6a5a6b43 download
.gitattributes 1.63 KB b1292aa4 download
LICENSE 900 B 0bd66457 download

README current version from Hugging Face


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

Apache 2.0 License

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

Quantization Repo Size Quality
BF16 (reference) BF16 ~60 GB Reference
FP16 GGUF FP16 ~60 GB Lossless
Q8_0 GGUF Q8_0 ~32 GB Near-lossless
Q6_K GGUF [You are here] ~25 GB Excellent
Q4_K_M GGUF Q4_K_M ~18 GB Good

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.

README history 3 versions

The author's README evolved over time. Click a version to see its content at that point.

  1. 2026-08-16docs: document vision (mmproj) usage + changelog6c9536b5.6 KB
    Loading...
  2. 2026-08-11docs: comprehensive README with abliteration methodology4e98eea4.2 KB
    Loading...
  3. 2026-08-11Add model card for Q6_K GGUF116bc9b1022 B
    Loading...
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Open in Abliteration