Instructions to use WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use WatsonOverHere/Qwen3.8-27B-Uncensored-F16-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 WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16
# Run inference directly in the terminal:
llama cli -hf WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16
# Run inference directly in the terminal:
llama cli -hf WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16

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 WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16
# Run inference directly in the terminal:
./llama-cli -hf WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16

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 WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16
# Run inference directly in the terminal:
./build/bin/llama-cli -hf WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16

Use Docker

docker model run hf.co/WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16

LM Studio
Jan
Ollama
How to use WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF with Ollama:
```
ollama run hf.co/WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16
```
Unsloth Desktop

How to use WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16

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": "WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF with Docker Model Runner:
```
docker model run hf.co/WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16
```

Lemonade

How to use WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16

Run and chat with the model

lemonade run user.Qwen3.8-27B-Uncensored-F16-GGUF-F16

List all available models

lemonade list

Hermes Agent

How to use WatsonOverHere/Qwen3.8-27B-Uncensored-F16-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 WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16

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 WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16

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 "WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF:F16" \
  --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"

Qwen3.8-27B-Uncensored-F16-GGUF

Full F16 GGUF conversion of JonathanColetti/Qwen3.8-27B-Uncensored, derived from Qwen/Qwen3.8-27B.

This repository provides a full 16-bit GGUF build for llama.cpp-compatible runtimes such as LM Studio.

Model Details

Parameters: 27B
GGUF precision: F16
File: Qwen3.8-27B-Uncensored-F16.gguf
File size: 54,657,734,240 bytes
Tensor count: 866
Block count: 65
MTP: retained
Source precision: BF16
Conversion: BF16 Safetensors -> F16 GGUF
llama.cpp commit: a94d563ed

Source Model

The source checkpoint is JonathanColetti/Qwen3.8-27B-Uncensored.

Jonathan Coletti created the uncensored/abliterated checkpoint using Heretic refusal-direction removal against the original Qwen3.8-27B model and restored the MTP tensors from the Qwen base checkpoint.

GGUF Conversion

The BF16 source checkpoint was converted to F16 GGUF using llama.cpp.

python convert_hf_to_gguf.py ..\source --outfile ..\Qwen3.8-27B-Uncensored-F16.gguf --outtype f16

No additional fine-tuning, training, abliteration, or other weight editing was performed during this conversion.

The conversion used llama.cpp commit:

a94d563ed

Verification

The completed conversion reported:

n_tensors = 866
total_size = 54.6G
Model successfully exported

GGUF metadata verification reported:

qwen35.block_count = 65

The resulting file size is:

54,657,734,240 bytes

Intended Runtime

This build is intended for GGUF-compatible runtimes including LM Studio and llama.cpp.

Because this is a full F16 model, substantially more memory is required than Q8, Q6, Q5, or Q4 quantizations.

Attribution

Original Model

Qwen Team — Qwen/Qwen3.8-27B

Abliteration / Uncensored Checkpoint

Jonathan Coletti — JonathanColetti/Qwen3.8-27B-Uncensored

Jonathan Coletti performed the Heretic-based refusal-direction removal and prepared the BF16 checkpoint used as the source for this conversion.

F16 GGUF Conversion

WatsonOverHere

This repository provides the F16 GGUF conversion only. No claim is made to the original model training or abliteration work.

License

Apache License 2.0, inherited from the upstream model.

Users should review the licenses, terms, and documentation of both the original Qwen model and the JonathanColetti source checkpoint.

Downloads last month: 3,822

GGUF

Model size

27B params

Architecture

qwen35

Hardware compatibility

16-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF

Base model

Qwen/Qwen3.8-27B

Finetuned

JonathanColetti/Qwen3.8-27B-Uncensored

Quantized

(29)

this model