Instructions to use 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M

Use Docker

docker model run hf.co/0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M

SGLang

How to use 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF with Ollama:
```
ollama run hf.co/0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M
```
Unsloth Desktop

How to use 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF with Docker Model Runner:
```
docker model run hf.co/0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M
```

Lemonade

How to use 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Severe Multimodal Degradation in Standard Quants

by vulcan4d - opened 14 days ago

Discussion

vulcan4d

14 days ago

I noticed that the RVN-Q5_K_M model suffers from severe visual hallucinations compared to a smaller Unsloth "UD" models (like IQ4_XS). Even with higher text precision, the Heretic model completely misidentifies major visual anchor objects. I wouldn't use this model for anything vision related. I would not use this model for anything vision related.

The Cause:
Standard uniform quantization bit-crushes the fragile Cross-Attention / Vision-Language bridge layers. This distorts the spatial embeddings being passed from the mmproj to the LLM, causing the model to receive "blurry" data and hallucinate the rest.

How Some Fix it:
Unsloth’s "UD" format uses dynamic quantization to specifically protect the vision alignment layers. It leaves the multimodal bridge and the first/last LLM layers at high precision (FP16 or 8-bit) and only compresses the middle FFN blocks.

0bserverx

Owner 13 days ago

Valid observation — and you're right that uniform quantization isn't ideal for vision-heavy use. A few notes from our side:

The mmproj is already high-precision (mmproj-Qwen3.8-27B-Q8_0.gguf, the official llama.cpp export) — the projector itself isn't the bottleneck. The degradation comes from the quantized LLM input layers that receive the visual embeddings.
For vision-critical work we'd recommend the higher-precision quants (RVN-Q8_0-mtp or RVN-F16/RVN-BF16), or the unquantized model with the mmproj.
We can also ship a "UD-style" vision-protected variant if there's interest: llama-quantize --override-tensor lets us keep the embedding + first layers (the vision-language bridge) at Q8_0/F16 while compressing the middle FFN blocks to the target quant — same approach Unsloth's UD uses.

Thanks for the detailed analysis — that's exactly the kind of feedback that helps.

0bserverx changed discussion status to closed 13 days ago

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment