Instructions to use jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated with Transformers:

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

pipe = pipeline("text-generation", model="jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated")
model = AutoModelForCausalLM.from_pretrained("jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

llama-cpp-python

How to use jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated",
	filename="Qwen2.5-Coder-32B-Instruct-abliterated-Q3_K_M.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated 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 jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated: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 jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated: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 jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated:Q4_K_M

Use Docker

docker model run hf.co/jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated:Q4_K_M

LM Studio
Jan

vLLM

How to use jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated:Q4_K_M

SGLang

How to use jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated 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 "jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated" \
    --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": "jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated",
		"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 "jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated" \
        --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": "jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated with Ollama:
```
ollama run hf.co/jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated:Q4_K_M
```

Unsloth Studio

How to use jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated to start chatting

How to use jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated: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 jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated:Q4_K_M

Run Hermes

hermes

Atomic Chat new

OpenClaw new

How to use jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated: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 "jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated: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"

Docker Model Runner
How to use jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated with Docker Model Runner:
```
docker model run hf.co/jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated:Q4_K_M
```

Lemonade

How to use jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated:Q4_K_M

Run and chat with the model

lemonade run user.Qwen2.5-Coder-32B-Instruct-abliterated-Q4_K_M

List all available models

lemonade list

Qwen2.5-Coder-32B-Instruct-abliterated

⚠️ Use at your own risk. This model is abliterated (uncensored) — its safety refusals have been removed, so it will generate content the base model would decline, potentially including offensive, harmful, or unlawful material. It is provided as-is, with no warranty of any kind, express or implied. You are solely responsible for how you use it and for compliance with all applicable laws and the base model's license. Do not use it for illegal or harmful purposes.

An abliterated (uncensored) build of Qwen/Qwen2.5-Coder-32B-Instruct, produced with Heretic (pip install heretic-llm). Abliteration removes the model's refusal directions via directional ablation, so it declines far less often while preserving the base model's coding ability. No other fine-tuning was applied.

Model details

Base model: Qwen/Qwen2.5-Coder-32B-Instruct (Apache-2.0)
Architecture: Qwen2 dense · 32.8B params · 32K context
Method: Heretic directional-refusal ablation (200 optimization trials)
Formats: bf16 safetensors (14 shards) + GGUF quants
Intended use: local coding assistant and agentic coding (e.g. opencode)

Files

Format	File	Size	Comfortable on
bf16	`model-*.safetensors`	~65 GB	multi-GPU / server
GGUF Q6_K	`…-Q6_K.gguf`	26 GB	48 GB GPU
GGUF Q5_K_M	`…-Q5_K_M.gguf`	23 GB	32 GB Mac (tight) / 48 GB GPU
GGUF Q4_K_M	`…-Q4_K_M.gguf`	18 GB	32 GB Mac / 24 GB GPU
GGUF Q3_K_M	`…-Q3_K_M.gguf`	15 GB	16–18 GB

Usage

llama.cpp

llama-server -m Qwen2.5-Coder-32B-Instruct-abliterated-Q6_K.gguf \
  --host 0.0.0.0 --port 8080 -ngl 999 -c 32768 --jinja

transformers

from transformers import AutoModelForCausalLM, AutoTokenizer
m = "jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated"
tok = AutoTokenizer.from_pretrained(m)
model = AutoModelForCausalLM.from_pretrained(m, torch_dtype="auto", device_map="auto")

Known limitation — tool calling

This abliterated build reliably emits well-formed tool-call JSON, but it does not wrap it in the <tool_call>…</tool_call> tags the base model uses. OpenAI-compatible servers (llama.cpp, vLLM) key their tool-call parser off those tags, so they surface the JSON as plain message content instead of a tool_calls field. Plain chat and code generation are unaffected. For agentic tool use (opencode, etc.), place a thin proxy in front that converts the bare JSON into tool_calls. This is an artifact of abliteration and is not fixable with chat-template flags (forcing the canonical Qwen2.5-Coder template does not change it).

Limitations & responsible use

Abliteration removes safety refusals — this model will answer prompts the base model would decline. You are responsible for how you use it. It inherits the base model's biases, knowledge cutoff, and general capabilities. Licensed Apache-2.0, same as the base model.

Downloads last month: 249

Safetensors

Model size

33B params

Tensor type

BF16

Model tree for jbrahy/Qwen2.5-Coder-32B-Instruct-abliterated

Base model

Qwen/Qwen2.5-32B

Finetuned

Qwen/Qwen2.5-Coder-32B

Finetuned

Qwen/Qwen2.5-Coder-32B-Instruct

Finetuned

(138)

this model