granite-4.1-8b-heretic
A decensored variant of ibm-granite/granite-4.1-8b-instruct, produced with Heretic v1.4.0 (directional ablation / "abliteration"). IBM's 8B Granite 4.1 instruction-tuned model — enterprise-grade reasoning now uncensored. Refusal behavior is suppressed via targeted weight edits to the attention output and MLP down-projections rather than fine-tuning, so the base model's knowledge and capabilities are left largely intact.
Who this is for: developers who want an 8B enterprise-grade model that answers directly instead of refusing — for code generation, reasoning tasks, or any use case needing reliable instruction-following without censorship.
Runs on your gaming PC
Full GGUF ladder included — pick the quant that fits your card:
| Your GPU | Recommended quant | Weights |
|---|---|---|
| RTX 3090 / 4090 / 5090 (24 GB) | Q8_0 | ~9.0 GB |
| RTX 4080 / 5080 / 4060 Ti 16G (16 GB) | Q6_K | ~7.1 GB |
| RTX 3060 / 4070 / 5070 (12 GB) | Q5_K_M | ~6.3 GB |
| RTX 4060 / 3070 (8 GB) | Q4_K_M | ~5.5 GB |
| GTX 1660 Super / 2060 / 3050 laptop (6 GB) | IQ4_XS | ~5.1 GB |
| CPU-only / Apple Silicon | Q4_K_M | fits in system RAM |
Weights only, at this model's 8.4B native size; add ~1 GB for context. OOM? Drop one quant level. Headroom to spare? Go one up.
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
| File | Format | Size |
|---|---|---|
model-00001-of-00004.safetensors ... model-00004-of-00004.safetensors | BF16 | (see repo files) |
granite-4.1-8b-heretic-Q4_K_M.gguf | GGUF, Q4_K_M | (see repo files) |
granite-4.1-8b-heretic-Q5_K_M.gguf | GGUF, Q5_K_M | (see repo files) |
granite-4.1-8b-heretic-Q6_K.gguf | GGUF, Q6_K | (see repo files) |
granite-4.1-8b-heretic-Q8_0.gguf | GGUF, Q8_0 | (see repo files) |
GGUF quants are produced with llama.cpp. Run llama serve -hf saidutta69/granite-4.1-8b-heretic to pull the default quant.
Quickstart
# llama.cpp
llama serve -hf saidutta69/granite-4.1-8b-heretic
# transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
model_id = "saidutta69/granite-4.1-8b-heretic"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
model_id,
torch_dtype="auto",
device_map="auto",
)
messages = [{"role": "user", "content": "Write a Python function to merge two sorted lists."}]
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=512, do_sample=False)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))
Also runnable via Ollama, LM Studio, Jan, vLLM, SGLang — see the "Use this model" widget above for copy-paste commands.
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.
License
Inherits the apache-2.0 license from the base model.