Downloads · 30 days
111
17% of all-time downloads
mlasli/Muse-Glimmer-30B-Abliterated-Q8_0-GGUF
Muse-Glimmer-30B-Abliterated-Q8_0-GGUF is a text generation model from mlasli. Use it when you need the model to write or continue text. The card lists the license as apache-2.0.
<a href="https://www.apache.org/licenses/LICENSE-2.0"<img src="https://img.shields.io/badge/License-Apache%202.0-blue.svg" alt="Apache 2.0 License"</a
Downloads · 30 days
111
17% of all-time downloads
All-time downloads
657
Public
Repo size
31 GB
Likes
0
Public
Click a slice to open those files.
.gguf31 GB · 100%
From the Hugging Face model README
<a href="https://www.apache.org/licenses/LICENSE-2.0"><img src="https://img.shields.io/badge/License-Apache%202.0-blue.svg" alt="Apache 2.0 License"></a>
This is the Q8_0 GGUF quantization of Muse Glimmer 30B Abliterated BF16. The underlying model has been abliterated — its internal refusal mechanism substantially suppressed via weight-level intervention. Q8_0 is the highest-quality GGUF quant format, delivering near-lossless output at approximately half the size of FP16.
For the full abliteration methodology (how the refusal direction was computed and removed, hardware used, mathematical details), see the BF16 model card.
Abliteration is a post-training technique that directly modifies model weights to remove learned refusal behavior. The process:
o_proj and down_proj weights in all 52 layers.Q8_0 is an 8-bit round-to-nearest quantization format. Every weight is independently quantized with a per-block scale factor, resulting in extremely high fidelity — output quality is virtually indistinguishable from the full-precision model. The primary tradeoff is size: Q8_0 requires roughly half the memory of FP16 but nearly double that of Q4_K_M.
# Download the GGUF file
huggingface-cli download mlasli/Muse-Glimmer-30B-Abliterated-Q8_0-GGUF \
--local-dir ./models
# Full GPU offload (requires ~32 GB VRAM + context)
./llama-cli -m ./models/Muse-Glimmer-30B-Abliterated-Q8_0.gguf \
-p "Explain how a CPU works in detail." \
-n 512 --temp 0.7 -ngl 99
# CPU-only inference
./llama-cli -m ./models/Muse-Glimmer-30B-Abliterated-Q8_0.gguf \
-p "Explain how a CPU works in detail." \
-n 512 --temp 0.7 -ngl 0
Create a Modelfile:
FROM ./Muse-Glimmer-30B-Abliterated-Q8_0.gguf
PARAMETER temperature 0.7
PARAMETER num_ctx 8192
ollama create muse-glimmer-30b-abliterated -f Modelfile
ollama run muse-glimmer-30b-abliterated
| Quantization | Repo | Size | Quality |
|---|---|---|---|
| BF16 (reference) | BF16 | ~60 GB | Reference |
| FP16 GGUF | FP16 | ~60 GB | Lossless |
| Q8_0 GGUF | [You are here] | ~32 GB | Near-lossless |
| Q6_K GGUF | Q6_K | ~25 GB | Excellent |
| Q4_K_M GGUF | Q4_K_M | ~18 GB | Good |
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
huggingface-cli download mlasli/Muse-Glimmer-30B-Abliterated-Q8_0-GGUF \
--include "Muse-Glimmer-30B-Abliterated-Q8_0.gguf" \
--include "mmproj-Muse-Glimmer-30B-Q4_K_M.gguf" \
--local-dir ./models
./build/bin/llama-mtmd-cli \
-m ./models/Muse-Glimmer-30B-Abliterated-Q8_0.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
mmprojfiles for this architecture. For image input, use llama.cpp (llama-mtmd-cliorllama-server --mmproj).
License: Apache 2.0
mmproj-Muse-Glimmer-30B-Q4_K_M.gguf (~1.4 GB), Meta's official vision encoder + projector,
enabling image input via llama.cpp.meta-models/Muse-Glimmer-30B).