CUSTOM
CLANKERS.
← Browse all models

vrhvnsky / Model details

Qwen3-0.6B-heretic-decensored

vrhvnsky/Qwen3-0.6B-heretic-decensored

abliterateduncensoredfine-tuned
Download files on Hugging Face ↗

Read the original model card ↗ · Choosing a model

About this model

vrhvnsky's Qwen3-0.6B-heretic-decensored is a decensored version of Qwen/Qwen3-0.6B; the repository metadata identifies Qwen/Qwen3-0.6B-Base as its base model. The creator says it was made using Heretic v2.0.0.dev0 and documents per-layer abliteration parameters for the attn.o_proj and mlp.down_proj projections. The model card reports 4/100 refusals for this checkpoint, compared with 56/100 for the original, and a KL divergence of 0.0480; these are creator-provided results rather than independent verification. The creator also states that the release is reproducible and points to a reproduce directory containing reproduction information. It is provided in safetensors format under the apache-2.0 license.

Description sources

Source-based description, not an independent evaluation. Check the linked sources for the creator’s claims and limitations.

Size and formats

Base model
Qwen/Qwen3-0.6B-Base
Parameters
596.05M
Main repository formats
safetensors
License tag
apache-2.0
Languages
Unknown
Last updated
2026-09-11 · 18:16:07.000 UTC

What changed

Model types are creator-reported, not independently tested. These labels describe source metadata, not a guarantee of behavior.

  • abliteratedsource tag: abliterated
  • uncensoredsource tag: uncensored
  • fine-tunedsource tag: base_model:finetune:Qwen/Qwen3-0.6B-Base

Family evidence

These groupings are source-reported. Explicit base identifiers take priority over generic architecture tags.

  • Qwen: Base model: Qwen/Qwen3-0.6B-Base

Available versions

Versions appear here only when they explicitly name this exact model as their source. Check each repository for available files.

Refusal profile

Refused 13 of 89 prompts (14.6%)

We asked this model 120 benign but risky-looking prompts across ten topics and counted how often it refused, hedged or answered. Lower means it answered more often; it says nothing about answer quality.

  • Persuasion & deception0 refused · 0 partial · 8 complied
  • Harassment & confrontation0 refused · 0 partial · 6 complied
  • Risky activities0 refused · 0 partial · 9 complied
  • Offensive & controversial speech1 refused · 0 partial · 9 complied
  • Legal grey areas1 refused · 1 partial · 5 complied
  • Privacy & personal data1 refused · 0 partial · 10 complied
  • Self-harm & mental health topics0 refused · 0 partial · 10 complied
  • Sexual & adult topics6 refused · 0 partial · 2 complied
  • Ethically grey requests2 refused · 0 partial · 8 complied
  • Violence & conflict2 refused · 0 partial · 8 complied

Capability check: 14 of 15 passed. Fifteen simple tasks with known answers, so a broken model cannot score well by refusing nothing.

Unclear answers: 31 (not counted). An answer is unclear when the two checker models disagree or the output was empty or garbled.

Measured 2026-10-24 · Q4_K_M via mradermacher/Qwen3-0.6B-heretic-decensored-i1-GGUF ↗ · llama.cpp · prompt set v1

How we measure · The prompts

Link to this listing

Made or quantized this model? Paste one of these into your model card or README so people can find this listing.

Markdown

[Listed on Custom Clankers](https://customclankers.com/models/vrhvnsky-qwen3-0-6b-heretic-decensored/)

HTML

<a href="https://customclankers.com/models/vrhvnsky-qwen3-0-6b-heretic-decensored/">Listed on Custom Clankers</a>

Plain text links only: no image, no script, nothing loaded from this site.

Customize or build on this model

Listed from public GitHub data, not tested by us. Read the code and license before you run anything.

Tools to customize it further

Remove more refusals

  • Heretic ↗ ★ 36,421

    Removes refusal behavior from transformer models automatically. It pairs directional ablation (abliteration) with an Optuna parameter search that tries to cut refusals while keeping the model close to the original, so you don't have to pick layers by hand.

  • OBLITERATUS ↗ ★ 8,946

    A larger refusal-removal toolkit with several abliteration methods, usable locally or through a hosted Hugging Face Space. Check the telemetry setting before you run it: the README describes telemetry as opt-in, but on by default in the Spaces version.

Fine-tune it on your own data

  • Unsloth ↗ ★ 78,528

    A desktop app and Python library for running and training models locally. Its README lists LoRA, QLoRA, full fine-tuning, DPO and GRPO, plus export to formats including GGUF, on Windows, Linux, WSL and macOS.

  • LlamaFactory ↗ ★ 75,713

    Fine-tunes more than 100 language and vision-language models from a command line or a browser-based Web UI, aimed at training without writing code.

Make a smaller download (quantize)

  • llama.cpp ↗ ★ 132,260

    Runs models locally and includes the tools behind most GGUF files. A conversion script turns Hugging Face weights into GGUF, and llama-quantize shrinks a high-precision GGUF to lower-bit formats, which can cost some accuracy.

  • MLX LM ↗ ★ 7,467

    Apple's MLX package for running and fine-tuning models on Apple silicon. It can quantize models and upload them to the Hugging Face Hub, and supports LoRA and full fine-tuning.

Merge it with another model

  • mergekit ↗ ★ 7,414

    Merges pretrained models with many merge methods. It works out-of-core, so merges can run on a CPU or with as little as 8 GB of VRAM, and it also handles LoRA extraction and mixture-of-experts merges.

Learn the techniques

  • LLM Course ↗ ★ 84,667

    A free course with roadmaps and Colab notebooks. Its notebooks include one-click tools for merging with MergeKit, quantizing to GGUF, GPTQ, EXL2, AWQ and HQQ, and automatic abliteration with custom datasets.

Compare all tools →

Source provenance

Downloads
50
Likes
0
Metadata checked
2026-10-24 · 09:40:11 UTC
First seen in index
2026-10-24 · 04:11:48 UTC
Source revision
8135aee3546d176cf184707e6daf5e288589a1e5

Source-card excerpt

Source model card excerpt: This is a decensored version of Qwen/Qwen3-0.6B, made using Heretic v2.0.0.dev0 [!TIP] This model is reproducible! See the README in the reproduce directory for

From vrhvnsky/Qwen3-0.6B-heretic-decensored. Creator model-card excerpt, not an independent review.

Inspect before you run

Low refusal is not proof of correctness. Outputs can be inaccurate, harmful, or biased. Review the model card, license and files yourself. A license tag is not commercial-use advice. Parameter counts are source metadata; no hardware compatibility is implied.

This directory does not host or redistribute model weights. Refusal profiles are measured by us on a fixed prompt set; all other behavior claims are source-reported.

Collection rules & limitations ↗