---
license: other
license_name: qwen-research
license_link: https://huggingface.co/Qwen/Qwen2.5-0.5B-Instruct/blob/main/LICENSE
base_model: Qwen/Qwen2.5-0.5B-Instruct
tags:
  - qwen2
  - chat
  - heretic
  - uncensored
  - decensored
  - abliterated
  - conversational
  - text-generation-inference
language:
  - en
pipeline_tag: text-generation
---

# Qwen2.5-0.5B-Instruct-heretic

A decensored variant of [Qwen/Qwen2.5-0.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-0.5B-Instruct), produced with [Heretic](https://github.com/p-e-w/heretic) (directional ablation / "abliteration"). Refusal behavior is suppressed via targeted weight edits rather than fine-tuning, so the base model's knowledge and instruction-following are left largely intact.

**Who this is for:** the smallest model in this heretic series — for CPU-only inference, edge/embedded deployment, or anywhere the 3B/14B variants are too heavy. At 0.5B parameters, capability ceiling is inherently lower than the larger siblings regardless of abliteration; use this where footprint matters more than reasoning depth.

## Files

| File | Format | Size |
|---|---|---|
| `model.safetensors` | BF16/FP16 | 988 MB |
| `Qwen2.5-0.5B-Instruct-heretic.gguf` | GGUF, F16 (unquantized) | 994 MB |
| `Qwen2.5-0.5B-Instruct-heretic-Q8_0.gguf` | GGUF, Q8_0 | 531 MB |
| `Qwen2.5-0.5B-Instruct-heretic-Q5_K_M.gguf` | GGUF, Q5_K_M | 420 MB |
| `Qwen2.5-0.5B-Instruct-heretic-Q4_K_M.gguf` | GGUF, Q4_K_M | 398 MB |

## Reproducibility

Unlike most abliteration repos, the full run is reproducible from the [`reproduce/`](https://huggingface.co/saidutta69/Qwen2.5-0.5B-Instruct-heretic/tree/main/reproduce) folder in this repo:

- `config.toml` — exact Heretic configuration used for this run
- `reproduce.json` — full parameter and metric dump
- `Qwen--Qwen2--5-0--5B-Instruct.jsonl` — evaluation transcripts against the base model
- `SHA256SUMS` — checksums for integrity verification
- `requirements.txt` — pinned environment for re-running the ablation

## Quickstart

```bash
# llama.cpp
llama serve -hf saidutta69/Qwen2.5-0.5B-Instruct-heretic
```

```python
# transformers
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "saidutta69/Qwen2.5-0.5B-Instruct-heretic"
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype="auto", device_map="auto")
tokenizer = AutoTokenizer.from_pretrained(model_name)

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)
out = model.generate(**inputs, max_new_tokens=200)
print(tokenizer.decode(out[0][inputs["input_ids"].shape[-1]:], skip_special_tokens=True))
```

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

## 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. At 0.5B parameters, factual reliability is already limited before any abliteration; don't treat compliance as a proxy for correctness.

## License

Inherits the [`qwen-research`](https://huggingface.co/Qwen/Qwen2.5-0.5B-Instruct/blob/main/LICENSE) license from the base model — research use, see the linked license for commercial terms.

## Related

- [Qwen2.5-3B-Instruct-heretic](https://huggingface.co/saidutta69/Qwen2.5-3B-Instruct-heretic)
- [Qwen2.5-Coder-3B-Instruct-heretic](https://huggingface.co/saidutta69/Qwen2.5-Coder-3B-Instruct-heretic)
- [Qwen2.5-Coder-14B-Instruct-heretic](https://huggingface.co/saidutta69/Qwen2.5-Coder-14B-Instruct-heretic)