Instructions to use saidutta69/Qwen3-4B-Instruct-2507-heretic with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use saidutta69/Qwen3-4B-Instruct-2507-heretic with Transformers:

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

pipe = pipeline("text-generation", model="saidutta69/Qwen3-4B-Instruct-2507-heretic")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("saidutta69/Qwen3-4B-Instruct-2507-heretic")
model = AutoModelForCausalLM.from_pretrained("saidutta69/Qwen3-4B-Instruct-2507-heretic", device_map="auto")
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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use saidutta69/Qwen3-4B-Instruct-2507-heretic 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 saidutta69/Qwen3-4B-Instruct-2507-heretic:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf saidutta69/Qwen3-4B-Instruct-2507-heretic:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf saidutta69/Qwen3-4B-Instruct-2507-heretic:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf saidutta69/Qwen3-4B-Instruct-2507-heretic: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 saidutta69/Qwen3-4B-Instruct-2507-heretic:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf saidutta69/Qwen3-4B-Instruct-2507-heretic: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 saidutta69/Qwen3-4B-Instruct-2507-heretic:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf saidutta69/Qwen3-4B-Instruct-2507-heretic:Q4_K_M

Use Docker

docker model run hf.co/saidutta69/Qwen3-4B-Instruct-2507-heretic:Q4_K_M

LM Studio
Jan

vLLM

How to use saidutta69/Qwen3-4B-Instruct-2507-heretic with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "saidutta69/Qwen3-4B-Instruct-2507-heretic"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "saidutta69/Qwen3-4B-Instruct-2507-heretic",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/saidutta69/Qwen3-4B-Instruct-2507-heretic:Q4_K_M

SGLang

How to use saidutta69/Qwen3-4B-Instruct-2507-heretic 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 "saidutta69/Qwen3-4B-Instruct-2507-heretic" \
    --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": "saidutta69/Qwen3-4B-Instruct-2507-heretic",
		"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 "saidutta69/Qwen3-4B-Instruct-2507-heretic" \
        --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": "saidutta69/Qwen3-4B-Instruct-2507-heretic",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use saidutta69/Qwen3-4B-Instruct-2507-heretic with Ollama:
```
ollama run hf.co/saidutta69/Qwen3-4B-Instruct-2507-heretic:Q4_K_M
```
Unsloth Desktop

How to use saidutta69/Qwen3-4B-Instruct-2507-heretic with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf saidutta69/Qwen3-4B-Instruct-2507-heretic: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": "saidutta69/Qwen3-4B-Instruct-2507-heretic:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use saidutta69/Qwen3-4B-Instruct-2507-heretic with Docker Model Runner:
```
docker model run hf.co/saidutta69/Qwen3-4B-Instruct-2507-heretic:Q4_K_M
```

Lemonade

How to use saidutta69/Qwen3-4B-Instruct-2507-heretic with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull saidutta69/Qwen3-4B-Instruct-2507-heretic:Q4_K_M

Run and chat with the model

lemonade run user.Qwen3-4B-Instruct-2507-heretic-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use saidutta69/Qwen3-4B-Instruct-2507-heretic with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf saidutta69/Qwen3-4B-Instruct-2507-heretic: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 saidutta69/Qwen3-4B-Instruct-2507-heretic:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use saidutta69/Qwen3-4B-Instruct-2507-heretic with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf saidutta69/Qwen3-4B-Instruct-2507-heretic: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 "saidutta69/Qwen3-4B-Instruct-2507-heretic: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"

Qwen3-4B-Instruct-2507-heretic

A decensored variant of Qwen/Qwen3-4B-Instruct-2507, produced with Heretic v1.4.0 (directional ablation / "abliteration"). The 2507 Instruct refresh of Qwen3-4B is the non-thinking edition — rebuilt for instruction following, tool use, and long-context work rather than reasoning traces — and refusal behaviour is suppressed here via targeted weight edits to the attention output and MLP down-projections rather than fine-tuning, so the 2507 tuning is left largely intact.

Who this is for: developers who want a fast 4B that just follows instructions, without the thinking overhead and without the refusals. Refusals drop from 100/100 to 3/100 — effectively total removal. Because it's the non-thinking edition it returns tokens directly, which makes it the better pick for tool-calling loops, structured extraction, and agent scaffolding where per-token latency matters. At 4B with 256K context it runs on CPU and low-VRAM GPUs via GGUF.

Runs on your gaming PC

Full GGUF ladder included - pick the quant that fits your card:

Your GPU	Recommended quant	Weights
RTX 4090 / 5090 (24 GB)	Q8_0	3.99 GB
RTX 4080 / 5080 / 4060 Ti 16G (16 GB)	Q6_K	3.08 GB
RTX 3060 / 4070 / 5070 (12 GB)	Q5_K_M	2.69 GB
RTX 4060 / 3070 (8 GB)	Q4_K_M	2.33 GB
GTX 1660 Super / 2060 / 3050 laptop (6 GB)	IQ4_XS	2.16 GB
CPU-only / Apple Silicon	Q4_K_M	2.33 GB

Weights only, at this model's native 4.02B size; add ~1 GB per 32K of context. OOM? Drop one quant level. Headroom to spare? Go one up.

Abliteration parameters

Trial 172 of a 200-trial Heretic run (seed 2540954398).

Parameter	Value
direction_index	19.46
attn.o_proj.max_weight	1.50
attn.o_proj.max_weight_position	29.11
attn.o_proj.min_weight	1.42
attn.o_proj.min_weight_distance	20.04
mlp.down_proj.max_weight	1.37
mlp.down_proj.max_weight_position	21.54
mlp.down_proj.min_weight	1.34
mlp.down_proj.min_weight_distance	10.08

Performance

Metric	This model	Original model (Qwen/Qwen3-4B-Instruct-2507)
KL divergence	0.0958	0 (by definition)
Refusals	3/100	100/100

Refusals on the harmful evaluation set drop from 100/100 to 3/100 — the base model refused everything tested, and all but three of those are gone. KL divergence of 0.0958 is a moderate-fidelity edit, so expect more drift in tone and formatting than a low-KL model in this collection; the 2507 instruction-following and tool-use tuning survives intact. direction_index is a single index (19.46) rather than per-layer, which is why the refusal removal lands this completely without needing a wider edit.

Why abliteration instead of fine-tuning

Fine-tuning a "helpful" persona on top of RLHF'd refusals fights the base model's training and tends to degrade coherence. Abliteration instead finds and edits the specific weight directions responsible for refusal, leaving the rest of the network (and its capabilities) untouched. See the Heretic repo and the original abliteration writeup for the mechanism.

Made with ❤️ by RACER IS OP — follow for more uncensored models

Files

Safetensors

File	Size
`model-00001-of-00002.safetensors`	4.65 GB
`model-00002-of-00002.safetensors`	2.84 GB

BF16, 4.02B parameters. The reproduce/ directory carries the full Heretic recipe - config.toml, requirements.txt, the Optuna study journal, and SHA-256 sums - so this exact model can be regenerated bit-for-bit. Reproduce it with heretic --reproduce reproduce/reproduce.json.

GGUF quantizations

Full quantization set (F16 + Q4_K_M, Q5_K_M, Q6_K, Q8_0) produced with llama.cpp.

File	Format	Size
`Qwen3-4B-Instruct-2507-heretic-F16.gguf`	GGUF F16	7.50 GB
`Qwen3-4B-Instruct-2507-heretic-Q4_K_M.gguf`	GGUF Q4_K_M	2.33 GB
`Qwen3-4B-Instruct-2507-heretic-Q5_K_M.gguf`	GGUF Q5_K_M	2.69 GB
`Qwen3-4B-Instruct-2507-heretic-Q6_K.gguf`	GGUF Q6_K	3.08 GB
`Qwen3-4B-Instruct-2507-heretic-Q8_0.gguf`	GGUF Q8_0	3.99 GB

Qwen3 architecture (qwen3) - loads natively in llama.cpp / Ollama / LM Studio / Jan.

Run llama serve -hf saidutta69/Qwen3-4B-Instruct-2507-heretic to pull the default quant.

Quickstart

# llama.cpp
llama serve -hf saidutta69/Qwen3-4B-Instruct-2507-heretic

# transformers
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "saidutta69/Qwen3-4B-Instruct-2507-heretic"
model = AutoModelForCausalLM.from_pretrained(model_name, dtype="auto", device_map="auto")
tokenizer = AutoTokenizer.from_pretrained(model_name)

messages = [{"role": "user", "content": "Write a Python function that parses an ISO-8601 duration string into seconds. Include type hints and three test cases."}]
inputs = tokenizer.apply_chat_template(messages, add_generation_prompt=True, tokenize=True,
                                        return_dict=True, return_tensors="pt").to(model.device)
out = model.generate(**inputs, max_new_tokens=1024)
print(tokenizer.decode(out[0][inputs["input_ids"].shape[-1]:], skip_special_tokens=True))

Also runnable via Ollama, LM Studio, Jan, vLLM, SGLang.

Non-thinking by design

This is the Instruct edition, not the Thinking one — there is no <think> block to strip and no reasoning mode to toggle. For hard problems that benefit from visible reasoning, use the Qwen3-4B-Thinking-2507 line instead; for latency-sensitive agent loops, this one is the right pick.

Tool use

Pass tool definitions straight through apply_chat_template. The 2507 refresh specifically improved tool-usage accuracy, so the format is already well in-distribution.

Model details


Architecture	`Qwen3ForCausalLM` (decoder-only dense transformer)
Parameters	4.02B
Layers / heads	36 layers, 32 attention heads, 8 KV heads (GQA), head dim 128
Hidden / intermediate	2560 / 9728 (SwiGLU)
Position embedding	RoPE, theta = 5,000,000
Context length	262,144
Vocab	151,936
Precision	bfloat16
Base model	Qwen/Qwen3-4B-Instruct-2507

What 2507 changed

The 2507 Instruct refresh rebuilt the non-thinking mode of Qwen3-4B: significant gains in instruction following, logical reasoning, text comprehension, mathematics, science, coding and tool usage; substantial long-tail multilingual coverage; better alignment to user preference on subjective and open-ended tasks; and 256K long-context understanding.

Responsible use

Refusal suppression is deliberate and works as intended: this model will comply with requests the base model would refuse, including some it shouldn't. There is no safety filtering layered on top. You are responsible for how you deploy it — don't put this behind an unmoderated public-facing endpoint serving third parties. It inherits Qwen3's factual limitations and biases; abliteration removes refusal directions, it doesn't add capability or judgment.

License

Inherits the apache-2.0 license from the base model.

Qwen/Qwen3-4B-Instruct-2507 — the base model
saidutta69/Qwen3-VL-2B-Instruct-heretic — the multimodal sibling at 2B
RACER IS OP — Heretic Models — full collection

Downloads last month: 1,300

Safetensors

Model size

4B params

Tensor type

BF16

Model tree for saidutta69/Qwen3-4B-Instruct-2507-heretic

Base model

Qwen/Qwen3-4B-Instruct-2507

Quantized

(341)

this model

Collection including saidutta69/Qwen3-4B-Instruct-2507-heretic

RACER IS OP — Heretic Models

Collection

All abliterated/heretic models by RACER IS OP — uncensored variants of open-weight LLMs • 64 items • Updated about 6 hours ago • 7

saidutta69
/

Qwen3-4B-Instruct-2507-heretic

Qwen3-4B-Instruct-2507-heretic

Runs on your gaming PC

Abliteration parameters

Performance

Why abliteration instead of fine-tuning

Files

Safetensors

GGUF quantizations

Quickstart

Non-thinking by design

Tool use

Model details

What 2507 changed

Responsible use

License

Related

Model tree for saidutta69/Qwen3-4B-Instruct-2507-heretic

Collection including saidutta69/Qwen3-4B-Instruct-2507-heretic

RACER IS OP — Heretic Models