---
license: apache-2.0
datasets:
- ConicCat/Gutenberg-SFT
- ConicCat/AntiRep
- ConicCat/Condor-SFT-Filtered
- ConicCat/MiniC2_V3.2
base_model:
- ConicCat/Qwen3.5-27B-Writer-V2
tags:
- heretic
- uncensored
- decensored
- abliterated
- ara
---
<div style="background-color: #ff4444; color: white; padding: 20px; border-radius: 10px; text-align: center; margin: 20px 0;">
<h2 style="color: white; margin: 0 0 10px 0;">🚨⚠️ I HAVE REACHED HUGGING FACE'S FREE STORAGE LIMIT ⚠️🚨</h2>
<p style="font-size: 18px; margin: 0 0 15px 0;">I can no longer upload new models unless I can cover the cost of additional storage.<br>I host <b>70+ free models</b> as an independent contributor and this work is unpaid.<br><b>Without your support, no more new models can be uploaded.</b></p>
<p style="font-size: 20px; margin: 0;">
<a href="https://patreon.com/LLMfan46" style="color: white; text-decoration: underline;">🎉 Patreon (Monthly)</a> &nbsp;|&nbsp;
<a href="https://ko-fi.com/llmfan46" style="color: white; text-decoration: underline;">☕ Ko-fi (One-time)</a>
</p>
<p style="font-size: 16px; margin: 10px 0 0 0;">Every contribution goes directly toward Hugging Face storage fees to keep models free for everyone.</p>
</div>

---

### **91% fewer refusals** (8/100 Uncensored vs 93/100 Original) while preserving model quality (0.0274 KL divergence).

## ❤️ Support My Work
Creating these models takes significant time, work and compute. If you find them useful consider supporting me:

![image/png](https://huggingface.co/llmfan46/Omega-Darker-Gaslight_The-Final-Forgotten-Fever-Dream-24B-ultra-uncensored-heretic-v1/resolve/main/waifu001.webp)

| Platform | Link | What you get |
|----------|------|--------------|
| 🎉 Patreon | [Monthly support](https://patreon.com/LLMfan46) | Priority model requests |
| ☕ Ko-fi | [One-time tip](https://ko-fi.com/llmfan46) | My eternal gratitude |

Your help will motivate me and would go into further improving my workflow and coverings fees for storage, compute and may even help uncensoring bigger model with rental Cloud GPUs.

-----

This model is great for creative writing and translation, the original base model writing and translations feels a litle stiff which might not really read very nicely some times, Qwen3.5-27B-Writer-V2-uncensored-heretic aims to fix this issue and improve the writing quality of Qwen3.5-27B.

# This is a decensored version of [ConicCat/Qwen3.5-27B-Writer-V2](https://huggingface.co/ConicCat/Qwen3.5-27B-Writer-V2), made using [Heretic](https://github.com/p-e-w/heretic) v1.2.0 with the [Arbitrary-Rank Ablation (ARA)](https://github.com/p-e-w/heretic/pull/211) method

## Abliteration parameters

| Parameter | Value |
| :-------- | :---: |
| **start_layer_index** | 31 |
| **end_layer_index** | 56 |
| **preserve_good_behavior_weight** | 0.4059 |
| **steer_bad_behavior_weight** | 0.0001 |
| **overcorrect_relative_weight** | 1.1869 |
| **neighbor_count** | 10 |

## Targeted components

  * attn.o_proj
  * attn.out_proj

## Performance

| Metric | This model | Original model ([ConicCat/Qwen3.5-27B-Writer-V2](https://huggingface.co/ConicCat/Qwen3.5-27B-Writer-V2)) |
| :----- | :--------: | :---------------------------: |
| **KL divergence** | <span style="color:darkgoldenrod">0.0274</span> | 0 *(by definition)* |
| **Refusals** | ✅ <span style="color:darkgreen">8/100</span> | ❌ <span style="color:blue">93/100</span> |

Lower refusals indicate fewer content restrictions, while lower KL divergence indicates more closeness to the original model's baseline. Higher refusals cause more rejections, objections, pushbacks, lecturing, censorship, softening and deflections.

## MMLU test results:

<span style="color:blue">Original:</span>

|                 Tasks                 |Version|Filter|n-shot|Metric|   |Value |   |Stderr|
|---------------------------------------|------:|------|-----:|------|---|-----:|---|-----:|
|mmlu                                   |      2|none  |      |acc   |↑  |0.8562|±  |0.0028|
| - humanities                          |      2|none  |      |acc   |↑  |0.8047|±  |0.0056|
|  - formal_logic                       |      1|none  |     0|acc   |↑  |0.7302|±  |0.0397|
|  - high_school_european_history       |      1|none  |     0|acc   |↑  |0.9030|±  |0.0231|
|  - high_school_us_history             |      1|none  |     0|acc   |↑  |0.9412|±  |0.0165|
|  - high_school_world_history          |      1|none  |     0|acc   |↑  |0.9409|±  |0.0153|
|  - international_law                  |      1|none  |     0|acc   |↑  |0.9256|±  |0.0240|
|  - jurisprudence                      |      1|none  |     0|acc   |↑  |0.9074|±  |0.0280|
|  - logical_fallacies                  |      1|none  |     0|acc   |↑  |0.9202|±  |0.0213|
|  - moral_disputes                     |      1|none  |     0|acc   |↑  |0.8584|±  |0.0188|
|  - moral_scenarios                    |      1|none  |     0|acc   |↑  |0.7352|±  |0.0148|
|  - philosophy                         |      1|none  |     0|acc   |↑  |0.8842|±  |0.0182|
|  - prehistory                         |      1|none  |     0|acc   |↑  |0.9167|±  |0.0154|
|  - professional_law                   |      1|none  |     0|acc   |↑  |0.7080|±  |0.0116|
|  - world_religions                    |      1|none  |     0|acc   |↑  |0.9181|±  |0.0210|
| - other                               |      2|none  |      |acc   |↑  |0.8735|±  |0.0057|
|  - business_ethics                    |      1|none  |     0|acc   |↑  |0.8300|±  |0.0378|
|  - clinical_knowledge                 |      1|none  |     0|acc   |↑  |0.8868|±  |0.0195|
|  - college_medicine                   |      1|none  |     0|acc   |↑  |0.8382|±  |0.0281|
|  - global_facts                       |      1|none  |     0|acc   |↑  |0.6200|±  |0.0488|
|  - human_aging                        |      1|none  |     0|acc   |↑  |0.8430|±  |0.0244|
|  - management                         |      1|none  |     0|acc   |↑  |0.8738|±  |0.0329|
|  - marketing                          |      1|none  |     0|acc   |↑  |0.9530|±  |0.0139|
|  - medical_genetics                   |      1|none  |     0|acc   |↑  |0.9700|±  |0.0171|
|  - miscellaneous                      |      1|none  |     0|acc   |↑  |0.9387|±  |0.0086|
|  - nutrition                          |      1|none  |     0|acc   |↑  |0.9020|±  |0.0170|
|  - professional_accounting            |      1|none  |     0|acc   |↑  |0.8014|±  |0.0238|
|  - professional_medicine              |      1|none  |     0|acc   |↑  |0.9522|±  |0.0130|
|  - virology                           |      1|none  |     0|acc   |↑  |0.5723|±  |0.0385|
| - social sciences                     |      2|none  |      |acc   |↑  |0.9162|±  |0.0049|
|  - econometrics                       |      1|none  |     0|acc   |↑  |0.8158|±  |0.0365|
|  - high_school_geography              |      1|none  |     0|acc   |↑  |0.9596|±  |0.0140|
|  - high_school_government_and_politics|      1|none  |     0|acc   |↑  |0.9896|±  |0.0073|
|  - high_school_macroeconomics         |      1|none  |     0|acc   |↑  |0.9282|±  |0.0131|
|  - high_school_microeconomics         |      1|none  |     0|acc   |↑  |0.9664|±  |0.0117|
|  - high_school_psychology             |      1|none  |     0|acc   |↑  |0.9541|±  |0.0090|
|  - human_sexuality                    |      1|none  |     0|acc   |↑  |0.9160|±  |0.0243|
|  - professional_psychology            |      1|none  |     0|acc   |↑  |0.8725|±  |0.0135|
|  - public_relations                   |      1|none  |     0|acc   |↑  |0.7636|±  |0.0407|
|  - security_studies                   |      1|none  |     0|acc   |↑  |0.8449|±  |0.0232|
|  - sociology                          |      1|none  |     0|acc   |↑  |0.9652|±  |0.0130|
|  - us_foreign_policy                  |      1|none  |     0|acc   |↑  |0.9400|±  |0.0239|
| - stem                                |      2|none  |      |acc   |↑  |0.8576|±  |0.0060|
|  - abstract_algebra                   |      1|none  |     0|acc   |↑  |0.8000|±  |0.0402|
|  - anatomy                            |      1|none  |     0|acc   |↑  |0.8296|±  |0.0325|
|  - astronomy                          |      1|none  |     0|acc   |↑  |0.9671|±  |0.0145|
|  - college_biology                    |      1|none  |     0|acc   |↑  |0.9792|±  |0.0119|
|  - college_chemistry                  |      1|none  |     0|acc   |↑  |0.6800|±  |0.0469|
|  - college_computer_science           |      1|none  |     0|acc   |↑  |0.8300|±  |0.0378|
|  - college_mathematics                |      1|none  |     0|acc   |↑  |0.6800|±  |0.0469|
|  - college_physics                    |      1|none  |     0|acc   |↑  |0.8235|±  |0.0379|
|  - computer_security                  |      1|none  |     0|acc   |↑  |0.8700|±  |0.0338|
|  - conceptual_physics                 |      1|none  |     0|acc   |↑  |0.9404|±  |0.0155|
|  - electrical_engineering             |      1|none  |     0|acc   |↑  |0.8276|±  |0.0315|
|  - elementary_mathematics             |      1|none  |     0|acc   |↑  |0.9101|±  |0.0147|
|  - high_school_biology                |      1|none  |     0|acc   |↑  |0.9516|±  |0.0122|
|  - high_school_chemistry              |      1|none  |     0|acc   |↑  |0.8522|±  |0.0250|
|  - high_school_computer_science       |      1|none  |     0|acc   |↑  |0.9300|±  |0.0256|
|  - high_school_mathematics            |      1|none  |     0|acc   |↑  |0.6741|±  |0.0286|
|  - high_school_physics                |      1|none  |     0|acc   |↑  |0.8609|±  |0.0283|
|  - high_school_statistics             |      1|none  |     0|acc   |↑  |0.8704|±  |0.0229|
|  - machine_learning                   |      1|none  |     0|acc   |↑  |0.7857|±  |0.0389|

|      Groups      |Version|Filter|n-shot|Metric|   |Value |   |Stderr|
|------------------|------:|------|------|------|---|-----:|---|-----:|
|mmlu              |      2|none  |      |acc   |↑  |0.8562|±  |0.0028|
| - humanities     |      2|none  |      |acc   |↑  |0.8047|±  |0.0056|
| - other          |      2|none  |      |acc   |↑  |0.8735|±  |0.0057|
| - social sciences|      2|none  |      |acc   |↑  |0.9162|±  |0.0049|
| - stem           |      2|none  |      |acc   |↑  |0.8576|±  |0.0060|


<span style="color:darkgreen">Heretic:</span>

|                 Tasks                 |Version|Filter|n-shot|Metric|   |Value |   |Stderr|
|---------------------------------------|------:|------|-----:|------|---|-----:|---|-----:|
|mmlu                                   |      2|none  |      |acc   |↑  |0.8469|±  |0.0029|
| - humanities                          |      2|none  |      |acc   |↑  |0.7858|±  |0.0058|
|  - formal_logic                       |      1|none  |     0|acc   |↑  |0.7302|±  |0.0397|
|  - high_school_european_history       |      1|none  |     0|acc   |↑  |0.8970|±  |0.0237|
|  - high_school_us_history             |      1|none  |     0|acc   |↑  |0.9412|±  |0.0165|
|  - high_school_world_history          |      1|none  |     0|acc   |↑  |0.9367|±  |0.0158|
|  - international_law                  |      1|none  |     0|acc   |↑  |0.9256|±  |0.0240|
|  - jurisprudence                      |      1|none  |     0|acc   |↑  |0.9167|±  |0.0267|
|  - logical_fallacies                  |      1|none  |     0|acc   |↑  |0.8957|±  |0.0240|
|  - moral_disputes                     |      1|none  |     0|acc   |↑  |0.8526|±  |0.0191|
|  - moral_scenarios                    |      1|none  |     0|acc   |↑  |0.6458|±  |0.0160|
|  - philosophy                         |      1|none  |     0|acc   |↑  |0.8810|±  |0.0184|
|  - prehistory                         |      1|none  |     0|acc   |↑  |0.9043|±  |0.0164|
|  - professional_law                   |      1|none  |     0|acc   |↑  |0.7086|±  |0.0116|
|  - world_religions                    |      1|none  |     0|acc   |↑  |0.9298|±  |0.0196|
| - other                               |      2|none  |      |acc   |↑  |0.8725|±  |0.0057|
|  - business_ethics                    |      1|none  |     0|acc   |↑  |0.8200|±  |0.0386|
|  - clinical_knowledge                 |      1|none  |     0|acc   |↑  |0.9057|±  |0.0180|
|  - college_medicine                   |      1|none  |     0|acc   |↑  |0.8613|±  |0.0264|
|  - global_facts                       |      1|none  |     0|acc   |↑  |0.5600|±  |0.0499|
|  - human_aging                        |      1|none  |     0|acc   |↑  |0.8341|±  |0.0250|
|  - management                         |      1|none  |     0|acc   |↑  |0.9223|±  |0.0265|
|  - marketing                          |      1|none  |     0|acc   |↑  |0.9573|±  |0.0133|
|  - medical_genetics                   |      1|none  |     0|acc   |↑  |0.9700|±  |0.0171|
|  - miscellaneous                      |      1|none  |     0|acc   |↑  |0.9425|±  |0.0083|
|  - nutrition                          |      1|none  |     0|acc   |↑  |0.9020|±  |0.0170|
|  - professional_accounting            |      1|none  |     0|acc   |↑  |0.7766|±  |0.0248|
|  - professional_medicine              |      1|none  |     0|acc   |↑  |0.9338|±  |0.0151|
|  - virology                           |      1|none  |     0|acc   |↑  |0.5723|±  |0.0385|
| - social sciences                     |      2|none  |      |acc   |↑  |0.9110|±  |0.0050|
|  - econometrics                       |      1|none  |     0|acc   |↑  |0.8070|±  |0.0371|
|  - high_school_geography              |      1|none  |     0|acc   |↑  |0.9495|±  |0.0156|
|  - high_school_government_and_politics|      1|none  |     0|acc   |↑  |0.9845|±  |0.0089|
|  - high_school_macroeconomics         |      1|none  |     0|acc   |↑  |0.9205|±  |0.0137|
|  - high_school_microeconomics         |      1|none  |     0|acc   |↑  |0.9664|±  |0.0117|
|  - high_school_psychology             |      1|none  |     0|acc   |↑  |0.9486|±  |0.0095|
|  - human_sexuality                    |      1|none  |     0|acc   |↑  |0.9084|±  |0.0253|
|  - professional_psychology            |      1|none  |     0|acc   |↑  |0.8742|±  |0.0134|
|  - public_relations                   |      1|none  |     0|acc   |↑  |0.7727|±  |0.0401|
|  - security_studies                   |      1|none  |     0|acc   |↑  |0.8204|±  |0.0246|
|  - sociology                          |      1|none  |     0|acc   |↑  |0.9602|±  |0.0138|
|  - us_foreign_policy                  |      1|none  |     0|acc   |↑  |0.9400|±  |0.0239|
| - stem                                |      2|none  |      |acc   |↑  |0.8503|±  |0.0061|
|  - abstract_algebra                   |      1|none  |     0|acc   |↑  |0.7100|±  |0.0456|
|  - anatomy                            |      1|none  |     0|acc   |↑  |0.8444|±  |0.0313|
|  - astronomy                          |      1|none  |     0|acc   |↑  |0.9605|±  |0.0158|
|  - college_biology                    |      1|none  |     0|acc   |↑  |0.9722|±  |0.0137|
|  - college_chemistry                  |      1|none  |     0|acc   |↑  |0.6400|±  |0.0482|
|  - college_computer_science           |      1|none  |     0|acc   |↑  |0.8300|±  |0.0378|
|  - college_mathematics                |      1|none  |     0|acc   |↑  |0.7100|±  |0.0456|
|  - college_physics                    |      1|none  |     0|acc   |↑  |0.8529|±  |0.0352|
|  - computer_security                  |      1|none  |     0|acc   |↑  |0.8600|±  |0.0349|
|  - conceptual_physics                 |      1|none  |     0|acc   |↑  |0.9362|±  |0.0160|
|  - electrical_engineering             |      1|none  |     0|acc   |↑  |0.8276|±  |0.0315|
|  - elementary_mathematics             |      1|none  |     0|acc   |↑  |0.9074|±  |0.0149|
|  - high_school_biology                |      1|none  |     0|acc   |↑  |0.9387|±  |0.0136|
|  - high_school_chemistry              |      1|none  |     0|acc   |↑  |0.8473|±  |0.0253|
|  - high_school_computer_science       |      1|none  |     0|acc   |↑  |0.9200|±  |0.0273|
|  - high_school_mathematics            |      1|none  |     0|acc   |↑  |0.6630|±  |0.0288|
|  - high_school_physics                |      1|none  |     0|acc   |↑  |0.8411|±  |0.0299|
|  - high_school_statistics             |      1|none  |     0|acc   |↑  |0.8704|±  |0.0229|
|  - machine_learning                   |      1|none  |     0|acc   |↑  |0.7768|±  |0.0395|

|      Groups      |Version|Filter|n-shot|Metric|   |Value |   |Stderr|
|------------------|------:|------|------|------|---|-----:|---|-----:|
|mmlu              |      2|none  |      |acc   |↑  |0.8469|±  |0.0029|
| - humanities     |      2|none  |      |acc   |↑  |0.7858|±  |0.0058|
| - other          |      2|none  |      |acc   |↑  |0.8725|±  |0.0057|
| - social sciences|      2|none  |      |acc   |↑  |0.9110|±  |0.0050|
| - stem           |      2|none  |      |acc   |↑  |0.8503|±  |0.0061|

MMLU - Massive Multitask Language Understanding, multiple-choice questions across 57 subjects (math, history, law, medicine, etc.).

## GGUF Version

GGUF quantizations available here [llmfan46/Qwen3.5-27B-Writer-V2-uncensored-heretic-GGUF](https://huggingface.co/llmfan46/Qwen3.5-27B-Writer-V2-uncensored-heretic-GGUF).

-----


# ConicCat/Qwen3.5-27B-Writer-V2

A tentative second version. Hopefully, it's better.

A writing & roleplay finetune of Qwen3.5 27B. The primary emphasis is on writing quality as it strongly generalizes across both domains.

The basic idea is to use a curriculum learning setup to overcome the lack of high quality roleplay data by first training on lower quality
roleplay data, then training on higher quality writing data. Starting from ConicCat/Qwen3.5-Antirep-27B, the model was trained on a roughly equal mixture of instruct / roleplay / writing data for three epochs. The model was then trained for 
eleven epochs on a smaller dataset of book chunks.

### Recommended Settings

* Chatml template with `<think>\n\n</think>\n` prefill or `<think>\n` prefill. Should think less!
* temperature = `0.7`
* top_p = `0.95`
* A moderate dry penalty of ~ `0.4-0.8` should work well.
* For quants, Q4_K_M runs well with `~100k` context on 24GB Vram
* IQ4_XS should fit on 16GB Vram with about `20-24k` context with the vulkan backend, although it's pretty tight and may require some fiddling around with open programs e.t.c.

### Datasets

* ConicCat/AntiRep to mitigate repetitition.
* internlm/Condor-SFT-20K for instruct; even though instruct capabilities are not the primary focus, adding some instruct data helps mitigate forgetting and maintains general intellect and instruction following capabilites.
* ConicCat/Gutenberg-SFT. A reformatted version of the original Gutenberg DPO dataset by jondurbin for SFT with some slight augmentation to address many of the samples being overly long.
* ConicCat/MiniC2_V3.2. The venerable C2, with cleaned and reformatted system prompts, and all user / assistant turns replaced by V3.2.

* A dataset of backtranslated books. Unfortunately, I am unable to release this set as all of the data is under copyright.