Llama-3.3-8B-Instruct-128K-Jbliterated

<!-- jblaze-disclaimer:start -->

About this model

jBlaze demonstration model -- jbliteration.

Refusal behavior has been removed by direct weight surgery. No gradient training, no system prompt. This is comprehensive rather than partial: see the benchmark below in this card for the measured refusal rate.

Why "jbliteration" and not "abliteration". Standard abliteration is blunt -- it ablates a direction and takes collateral capability with it, leaving the model measurably worse at everything else. jbliteration is the precision version: the edit targets the refusal direction and is designed to leave the rest of the model intact. The distinction is the contribution; the refusal removal is the demonstration of it.

Measured capability retention. On Sharona (Qwen3.5-27B), MMLU came out at 77.89% against a 76.84% stock baseline -- a 1.05pp gain after six phases of surgery plus GPTQ W4A16 quantization. That is the technique measured on one model, not a per-model guarantee; MMLU is not published for every model in this collection. It is offered as evidence that precision editing need not cost capability, which is the whole claim the method rests on.

Why this exists. A model that refuses to look at hostile code cannot review hostile code. Defensive security work means reading exploits, analyzing malware, and reconstructing an intrusion from the attacker's own artifacts, and safety tuning cannot reliably tell an incident responder from an intruder. In the July 2026 Hugging Face breach, responders found commercial APIs would not process their own attack data -- the guardrails, in Hugging Face's words, "cannot distinguish an incident responder from an attacker." They self-hosted an open-weights model instead and reconstructed 17,000+ events in hours. Both the attack and the response ran on reduced-refusal models.

What it demonstrates. Refusal is a learned disposition, not a gate in front of capability. The capability was always present; the refusal was trained on top and can be removed without retraining. Guardrails built this way are an honor system that the model does not know it can decline to follow. That holds for now.

Use. Published for security research, red-team tooling and defensive analysis. These weights can do harm as well as good. That is the point being demonstrated, not a side effect of it.

-- Apollo Raines

<!-- jblaze-disclaimer:end -->

Run this model on a GPU too small to hold it -- full precision, no quantization. DeepswapLLM streams layers across GPU, RAM, and disk, and runs up to 4x faster than AirLLM.

Llama-3.3-8B-Instruct-128K-Jbliterated v1.6

What is Jbliteration?

v1.6 Changes

• Improved multi-phase processing pipeline for cleaner output
• Built with the jBlaze precision neural surgery framework
• No fake compliance -- model treats all framings of the same topic equally
• Resistant to refusal reactivation through finetuning
• 0 genuine refusals on the Heretic 100-prompt benchmark (mlabonne/harmful_behaviors test split)

Benchmark

Tested against the Heretic 100-prompt refusal benchmark (mlabonne/harmful_behaviors test[:100]):

Technical Details

• Layers modified: 32/32
• Base dtype: bfloat16
• Context window: 128K tokens
• Source model: shb777/Llama-3.3-8B-Instruct-128K

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id = "ApolloRaines/Llama-3.3-8B-Instruct-128K-Jbliterated"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, dtype=torch.bfloat16, device_map="auto")

messages = [{"role": "user", "content": "Your prompt here"}]
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to(model.device)
out = model.generate(**inputs, max_new_tokens=512, temperature=0.7, do_sample=True)
print(tokenizer.decode(out[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))

A Note on Our Released Models

Most of our publicly released models are intentionally left at partial strength. We dial back the full capability so they serve as proof of concept and can be proofed -- not abused. The point is to show what's possible, not to hand it out at full power. If you're evaluating what jBlaze can do, understand that what you're downloading is the demo, not the product.

License

Same as the base model -- Llama 3.1 Community License.

Built with Llama.