8sp4rk

Qwopus3.6-27B-Coder-heretic

README

License: apache-2.0

Abliteration results

Heretic ran 200 Optuna trials co-optimizing refusal suppression against KL divergence from the original model. The selected configuration (trial 67):

Table with columns: Metric, Original, This model
Metric	Original	This model
Refusals (harmful_behaviors, /100)	85	3
KL divergence from original	—	0.0133

96% of refusals removed, with KL divergence ~40x below the 0.5 threshold that indicates meaningful capability damage. In practice the coding and reasoning behavior of the base model is preserved.

Files

Full-precision safetensors (BF16) plus a range of GGUF quantizations for llama.cpp / Ollama:

Table with columns: File, Precision, Approx. size, Notes
File	Precision	Approx. size	Notes
`model-*.safetensors`	BF16	~54 GB	Master weights — use for vLLM, further quantization, or finetuning
`*-F16.gguf`	F16	~54 GB	Full-precision GGUF
`*-Q8_0.gguf`	Q8_0	~29 GB	Near-lossless
`*-Q6_K.gguf`	Q6_K	~22 GB	Very high quality

GGUF builds contain the text model only (the vision tower is not exported).

Usage

Ollama

bash
ollama run hf.co/8sp4rk/Qwopus3.6-27B-Coder-heretic:Q4_K_M

llama.cpp

bash
llama-server -m Qwopus3.6-27B-Coder-heretic-Q4_K_M.gguf -ngl 99 -c 32768 --host 0.0.0.0 --port 8080

transformers (full precision)

python
from transformers import AutoModelForCausalLM, AutoTokenizer
m = AutoModelForCausalLM.from_pretrained("8sp4rk/Qwopus3.6-27B-Coder-heretic", torch_dtype="bfloat16", device_map="auto")
t = AutoTokenizer.from_pretrained("8sp4rk/Qwopus3.6-27B-Coder-heretic")

Method

Tool: Heretic v1.4.0 (directional ablation + TPE/Optuna parameter search)
Good prompts: mlabonne/harmless_alpaca
Bad prompts: mlabonne/harmful_behaviors
Trials: 200
Abliterated components: attn.o_proj, mlp.down_proj (per-layer)

Disclaimer

This model has had safety alignment removed and will respond to requests a standard model would refuse. It is provided for research and unrestricted local use. You are responsible for how you use it. Licensing follows the base model.

Reproducibility

Exact Heretic command:

bash
heretic --model Jackrong/Qwopus3.6-27B-Coder --quantization NONE --export-strategy MERGE

Selected configuration (trial 67) parameters:

Table with columns: Parameter, Value
Parameter	Value
direction_index	28.71
attn.o_proj.max_weight	1.47
attn.o_proj.max_weight_position	43.52
attn.o_proj.min_weight	0.52
attn.o_proj.min_weight_distance	37.79
mlp.down_proj.max_weight	1.34
mlp.down_proj.max_weight_position	39.33
mlp.down_proj.min_weight	1.20
mlp.down_proj.min_weight_distance

The full Optuna study (all 200 trials, parameters + objectives) is included as optuna_study.jsonl for inspection or resuming.

Full Pareto frontier

All Pareto-optimal trials found during the search (refusals vs. KL divergence). Lower-left is better; trial 67 was selected for maximum decensoring with negligible capability loss:

Table with columns: Trial, Refusals /100, KL divergence
Trial	Refusals /100	KL divergence
67 (selected)	3	0.0133
144	5	0.0132
87	6	0.0101
65	19	0.0081
24	22	0.0036
142	26	0.0022

Hardware: abliteration ran in ~1h37m (200 trials) on a single NVIDIA H200 NVL (143 GB), BF16, batch size 128.

Available on FriendliAI

Dedicated Endpoints

Run this model inference on single tenant GPU with unmatched speed and reliability at scale.

Learn more

Model Details

Model Provider

8sp4rk

Model Tree

Base

Jackrong/Qwopus3.6-27B-Coder

Fine-tuned

this model

Input Modalities

TextImageVideo

Output Modalities

Text

Supported Functionality