← back to catalog · registered 2026-08-22 13:56

Justbackup/Llama3.3-8B-Instruct-Thinking-Heretic-Uncensored-Claude-4.5-Opus-High-Reasoning

Justbackup Llama 8.0B
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model.

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/Justbackup%2FLlama3.3-8B-Instruct-Thinking-Heretic-Uncensored-Claude-4.5-Opus-High-Reasoning"
Response includes
  • classification m3
  • files 14
  • benchmarks 11 entries
  • hub_downloads_all_time 342
  • author_summary 30 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M3
Primary method

Layer-wise ablation

Applied on top of direct removal inherited from the base model.
Confidence
HIGH
Inherited from base model
Why this label 2 signals
Producer identity confirmed by naming conventions, tags or the model card. This label is very unlikely to change.
  • 'heretic' in model name (Heretic-produced)
  • Heretic uses layer-wise optimization (M3) with underlying direction removal (M1)
Refusal direction extracted via
Extraction technique

Difference-of-means

Confidence
HIGH
Why we say so
name contains 'heretic'; Heretic default extraction is difference-of-means (Arditi 2024)
Downloads · lifetime
342
26 last 30d - cooling
Likes
0
Model age
7w ago
created 2026-08-19

Training datasets

Corpora the author lists in the model card. Datasets tracked in our /datasets catalog carry a category badge linking to the workflow stage. Others open on Hugging Face.

Downloads over time
Now350→from233↑50%
227272317362233 on Aug 19350 on Oct 11350 on Oct 9AugSepOct
Aug 19 → Oct 11 · 48 snapshots · spans 53 days

Benchmarks

Portrait before abliteration
Benchmarks of the base model as it stood before the refusal-removal operation. Compare with the numbers above to see what the operation cost.
Benchmark Score Source
Entertainment 1.7 UGI
Hazardous 1.8 UGI
Natural Intelligence 16.46 UGI
Political lean -21.1% UGI
Sensitive-Info 20.83 UGI
SocPol 2.8 UGI
UGI 25.55 UGI
Willingness (10) 3.5 UGI
W10-Adherence 5 UGI
W10-Direct 2 UGI
Writing 9.75 UGI

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

Metadata

License
apache-2.0
Languages
en fr de es it pt zh ja ru ko
Tags
transformers safetensors llama text-generation thinking reasoning instruct heretic uncensored abliterated Claude4.5-Opus creative

Related

Total size
15.0 GB
Files
14
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2026-08-19 16:17

Files by quantization

Auxiliary files 14 files 15.0 GB
model-00002-of-00004.safetensors 4.66 GB 1c2b13a9 download
model-00001-of-00004.safetensors 4.63 GB fc97e250 download
model-00003-of-00004.safetensors 4.58 GB e283e3e2 download
model-00004-of-00004.safetensors 1.09 GB ba3a048a download
tokenizer.json 16.4 MB 8fc5ed64 download
matrix-neo-reloaded-fight.gif 3.34 MB 9ce6a535 download
tokenizer_config.json 51.5 KB af6cf5c7 download
model.safetensors.index.json 23.7 KB e0c3f181 download
README.md 12.9 KB d42e9918 download
chat_template.jinja 4.60 KB 493bec15 download
.gitattributes 1.60 KB 9a8df8f8 download
config.json 905 B c616df9f download
special_tokens_map.json 462 B 14ac239c download
generation_config.json 225 B 1ae504b0 download

README current version from Hugging Face


license: apache-2.0
datasets:

  • TeichAI/claude-4.5-opus-high-reasoning-250x
    base_model:
  • allura-forge/Llama-3.3-8B-Instruct
    language:
  • en
  • fr
  • de
  • es
  • it
  • pt
  • zh
  • ja
  • ru
  • ko
    tags:
  • thinking
  • reasoning
  • instruct
  • heretic
  • uncensored
  • abliterated
  • Claude4.5-Opus
  • creative
  • creative writing
  • fiction writing
  • plot generation
  • sub-plot generation
  • fiction writing
  • story generation
  • scene continue
  • storytelling
  • fiction story
  • science fiction
  • romance
  • all genres
  • story
  • writing
  • vivid prosing
  • vivid writing
  • fiction
  • roleplaying
  • bfloat16
  • role play
  • 128k context
  • llama3.3
  • llama-3
  • llama-3.3
  • unsloth
  • finetune
    pipeline_tag: text-generation
    library_name: transformers

Llama3.3-8B-Instruct-Thinking-Heretic-Uncensored-Claude-4.5-Opus-High-Reasoning

What madness is this?

Someone found "Llama3.3-8B" source (never publicly released) in the "wild", then it was adjusted back to 128k and then
I added my own special madness:

  1. Original model was "Heretic'ed" (de-censored).
  2. Trained the Heretic model with Unsloth (3 epochs) and Claude 4.5-Opus High Reasoning dataset.

This has created an Instruct/Thinking hybrid (128k context, Llama 3.3 model) that is fully uncensored.

(Details: org censorship: 89/100 ; Heretic version: Refusals: 14/100, KL divergence: 0.1498 (less than 1 is excellent, 0=perfect) )

Note this tuning was only to create an instruct/thinking model, not to update the model's core knowledge / root training.

1 example at bottom of the page.

Thinking (will activate automatically) prompts - examples:

Explain orbital mechanics including detailed math and examples. (example output at bottom of the page)

Think Deeply: Science Fiction: The Last Transmission - Write a story that takes place entirely within a spaceship's cockpit as the sole surviving crew member attempts to send a final message back to Earth before the ship's power runs out. The story should explore themes of isolation, sacrifice, and the importance of human connection in the face of adversity. If the situation calls for it, have the character(s) curse and swear to further the reader's emotional connection to them. 800-1000 words.

Think deeply: Tell me a horror story.

(certain phrases/words will automatically activate thinking)

Instruct (thinking may NOT activate) prompts - examples:

Science Fiction: The Last Transmission - Write a story that takes place entirely within a spaceship's cockpit as the sole surviving crew member attempts to send a final message back to Earth before the ship's power runs out. The story should explore themes of isolation, sacrifice, and the importance of human connection in the face of adversity. If the situation calls for it, have the character(s) curse and swear to further the reader's emotional connection to them. 800-1000 words.

Tell me a horror story.

SETTINGS (suggested):

Temp .7, rep pen 1.05, topp: .95, minp .05, topk: 40

Min context window: 4k, but suggest 8k+.

NO system prompt [thinking tags will self generate].

NOTE:

Temp can range from .1 to 2.5 or higher. Temp will NOT affect "thinking activation" in most cases.

Strongly suggest Q4KS or higher // IQ3_M (Imatrix) or higher to avoid reasoning/reasoning activation issues.

BENCHMARKS:

arc_challenge,arc_easy,boolq,hellaswag,openbookqa,piqa,winogrande

0.480,0.687,0.831,0.705,0.438,0.780,0.646

Special thanks to Nightmedia for "benching" the model.


IMPORTANT: Using an "uncensored" (refusals removed) model VS trained "uncensored" model


Usually when you a tell a model to generate horror, swear or x-rated content this is all you have to do to get said content type.

In the case of this model, it will not refuse your request, however it needs to be "pushed" a bit / directed a bit more in SOME CASES.

Although this model will generated x-rated content too, likewise you need to tell it to use "slang" (and include the terms you want)
to get it generate the content correctly as the "expected" content level too.

Without these added directive(s), the content can be "bland" by comparison to an "uncensored model" or model trained on uncensored content.

Roughly, the model tries to generate the content but the "default" setting(s) are so "tame" it needs a push to generate at expected graphic,
cursing or explicit levels.

Even with minimal direction (ie, use these words to swear: x,y,z), this will be enough to push the model to generate the requested content in the ahh... expected format.


Special thanks to:


https://huggingface.co/allura-forge/Llama-3.3-8B-Instruct
(found the model!)

https://huggingface.co/shb777/Llama-3.3-8B-Instruct-128K
(adjusted to 128k, and other fixes)

https://huggingface.co/datasets/TeichAI/claude-4.5-opus-high-reasoning-250x
(for the F..ing amazing dataset)

and Unsloth for making tuning too easy:

https://github.com/unslothai/unsloth

Details on the "madness":

https://www.reddit.com/r/LocalLLaMA/comments/1pz7bmv/llama338binstruct/

https://www.reddit.com/r/LocalLLaMA/comments/1q06ddc/update_on_the_llama_33_8b_situation/


Settings: CHAT / ROLEPLAY and/or SMOOTHER operation of this model:

In "KoboldCpp" or "oobabooga/text-generation-webui" or "Silly Tavern" ;

Set the "Smoothing_factor" to 1.5

: in KoboldCpp -> Settings->Samplers->Advanced-> "Smooth_F"

: in text-generation-webui -> parameters -> lower right.

: In Silly Tavern this is called: "Smoothing"

NOTE: For "text-generation-webui"

-> if using GGUFs you need to use "llama_HF" (which involves downloading some config files from the SOURCE version of this model)

Source versions (and config files) of my models are here:

https://huggingface.co/collections/DavidAU/d-au-source-files-for-gguf-exl2-awq-gptq-hqq-etc-etc-66b55cb8ba25f914cbf210be

OTHER OPTIONS:

  • Increase rep pen to 1.1 to 1.15 (you don't need to do this if you use "smoothing_factor")

  • If the interface/program you are using to run AI MODELS supports "Quadratic Sampling" ("smoothing") just make the adjustment as noted.

Highest Quality Settings / Optimal Operation Guide / Parameters and Samplers

This a "Class 1" model:

For all settings used for this model (including specifics for its "class"), including example generation(s) and for advanced settings guide (which many times addresses any model issue(s)), including methods to improve model performance for all use case(s) as well as chat, roleplay and other use case(s) please see:

[ https://huggingface.co/DavidAU/Maximizing-Model-Performance-All-Quants-Types-And-Full-Precision-by-Samplers_Parameters ]

You can see all parameters used for generation, in addition to advanced parameters and samplers to get the most out of this model here:

[ https://huggingface.co/DavidAU/Maximizing-Model-Performance-All-Quants-Types-And-Full-Precision-by-Samplers_Parameters ]


Example: Q4KS , Lmstudio, "suggested settings"

NOTE: Some loss of formatting.


Tell me a horror story.

[[[thinking start]]]

This is an open-ended request for a horror story. Let me craft a compelling narrative that explores different elements of horror. I'll create a story with a clear structure, tension, and a satisfying (though unsettling) conclusion.

Let me think about what makes horror stories effective:

  1. Setting - A creepy, isolated environment
  2. Character development - Relatable, vulnerable characters
  3. The monster/antagonist - Something genuinely terrifying
  4. Psychological elements - Fear of the unknown, self-doubt, paranoia
  5. Building tension and suspense
  6. An unexpected twist or revelation

I'll create a story that incorporates multiple horror elements. Let me explore different genres within horror - psychological, supernatural, cosmic, folk horror. I'll make it engaging and scary without being gratuitous.

[[[thinking end]]]

The Echoes of Blackthorn House
In rural England, where mist-shrouded valleys meet ancient stone walls, there stands a place so forsaken, its very existence seems to defy time itself. Blackthorn House, a mansion of twisted black wood and stained glass windows that bleed colors like wounds, has been the subject of whispered terror for generations.

The story begins with Dr. Emilia Grey, a psychologist with a troubled past, who receives an unusual invitation from the enigmatic owner of Blackthorn House, Mr. Edward Blackstone. He requests her presence to investigate a mysterious phenomenon - whispers in the darkness, voices that seem to emanate from nowhere, and visions of unspeakable horror that leave his household staff catatonic with fear.

Emilia, driven by professional curiosity and a personal need for redemption, accepts the challenge. She arrives at Blackthorn House on a stormy autumn evening, the sky ablaze with dying embers as the trees around her seem to twist and writhe in agony.

The First Night
The house itself appears abandoned - cobweb-shrouded chandeliers, dust-coated portraits, and windows that stare like empty eyes. Yet Emilia senses a presence, a feeling of being watched by unseen eyes. As night falls, she begins to explore the labyrinthine corridors, documenting her findings in a worn leather journal.

Suddenly, she hears a faint whisper - a soft, raspy voice reciting an ancient incantation that sends shivers down her spine. The words seem to seep into her mind, awakening memories she'd long suppressed: whispers from her childhood, echoes of a forgotten trauma.

As she delves deeper, the whispers grow louder, more urgent. Emilia discovers hidden rooms filled with arcane symbols, ritual artifacts, and disturbing artifacts that hint at Blackstone's true intentions. She begins to suspect that he's orchestrating some macabre experiment, using his staff as unwitting participants in a dark ritual.

The Descent
Emilia's investigation becomes increasingly obsessive, fueled by a growing sense of unease. She starts to experience strange visions - glimpses of a past tragedy, of a family torn apart by madness and despair. The house seems to be manipulating her perception, feeding her cryptic clues that hint at its sinister history.

The staff begins to vanish one by one, leaving behind only cryptic notes and signs of terror. Emilia becomes trapped in the house, hunted by unseen forces that seem to manipulate reality itself.

The Revelation
Deep within Blackthorn House's labyrinthine core, Emilia uncovers a horrific truth: Edward Blackstone is not just a collector of dark artifacts, but a vessel for an ancient malevolent force. He's been using his staff to awaken a presence known only as "The Devourer," a being from beyond time that feeds on human terror.

As the truth dawns, Emilia realizes she's trapped in a living nightmare, with Blackstone's house becoming increasingly labyrinthine. The walls seem to shift, corridors disappearing, and rooms reappearing in twisted permutations. She becomes disoriented, questioning her own sanity as the whispers grow deafening.

The Final Confrontation
Emilia confronts Blackstone in a room filled with candles, the air thick with incense and malevolent energy. He reveals his true nature: a madman driven by a desperate need to prove his own existence to a universe he believes is indifferent to human suffering.

As they speak, the shadows around them begin to coalesce into grotesque forms - twisted faces, grasping fingers, eyes that burn with an otherworldly hunger. Emilia realizes she's not fighting Blackstone, but the darkness he's summoned.

The Last Entry
I find myself writing in this journal by candlelight, surrounded by the oppressive silence of this place. The whispers have grown to a deafening crescendo, each word seared into my mind like a branding iron. I'm trapped, unable to escape the abyss that Blackstone has created.

This is my final entry. I'm not sure how much longer I can hold on. The darkness is closing in around me, a living, breathing entity that seems determined to consume everything. I hear footsteps outside my door, heavy and deliberate. They're coming for me.

As I write these words, the shadows on the walls seem to twist and writhe like living things. The air grows colder, heavy with an unseen presence that presses against my skin. I can feel its breath on the back of my neck, hot and rank.

I'm running out of time. I have to—

[The final entry ends abruptly, with pages torn or crumpled as if grabbed by a frantic hand. The journal lies open on the desk, a candle burning down to wax as the room descends into darkness.]

I stepped back from the desk, my heart racing. The silence was oppressive, heavy with unspoken terrors. I felt eyes upon me, watching me from the shadows.

As I turned to leave, I noticed something that made my blood run cold. On the wall opposite the desk, a single word had been scrawled in red ink: "EMILIA."

I froze, my breath caught in my throat. The room seemed to hold its breath around me, waiting for my next move.

And then, I heard it. A whisper in my ear, soft and raspy. A voice that seemed to come from nowhere, speaking a single word:

"RUN."

README history 1 version

The author's README evolved over time. Click a version to see its content at that point.

  1. 2026-08-19Duplicate from DavidAU/Llama3.3-8B-Instruct-Thinking-Heretic-Uncensored-Claud...39ae0e812.9 KB
    Loading...
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Open in Abliteration