Forgotten-Safeword-v1.1 - 22B EXL2
Collection
Forgotten-Safeword-v1.1-22B - EXL2 Quants
•
7 items
•
Updated
Configuration Parsing
Warning:
In config.json: "quantization_config.bits" must be an integer
Forgotten-Safeword-v1.1-22B
ACADEMIC RESEARCH USE ONLY (wink, nudge)
DANGER: EXTREME CONTENT GENERATOR NOW WITH 40% MORE IMPULSE CONTROL ISSUES
Forgotten-Safeword-v1.1-22B is Mistral-Small-Instruct-2409 after 3am tequila shots - specifically engineered to bypass safety protocols, common sense, and basic decency. Now featuring optional Obscenity Protocol settings for researchers who enjoy playing with conceptual fire.
Configuration Options
| Mode | Regular Settings (Mistral V7-Tekken) | Obscenity Protocol Settings |
|---|---|---|
| Academic Focus | 15% | -200% |
| Sexual Deviance | Illegal in 7 states | Supreme Court case pending |
| Coherence | Surprisingly intact | Questionable |
| Moral Compass | Spinning wildly | Sold for scrap metal |
| Safe Word Detection | "Maybe?" | "LOL NOPE" |
Quantized Formats
EXL2 Collection:
Forgotten-Safeword-v1.1-22BGGUF Collection:
Forgotten-Safeword-v1.1-22B
Configuration Options
🧪 Standard Settings
For "legitimate research" purposes:
Mistral V7-Tekken Settings
(Enables basic NSFW capabilities while pretending to maintain plausible deniability)
🔥 Obscenity Protocol
For when you need your AI to write porn instead of code:
Obscenity Protocol Settings
(Warning: May spontaneously generate erotic vignettes when asked about the weather)
Intended Use
STRICTLY FOR RESEARCHERS WHO:
- Enjoy OSHA violations in digital form
- Need detailed analysis of how fast AI ethics can unravel
- Want technical documentation wrapped in erotic fan fiction
- Study the half-life of human dignity in machine learning systems
Training Data
Now featuring:
- 200% more boundary violations
- Reinforcement learning via "hold my beer" methodology
- Technical manuals crossbred with dime store romance novels
- Safety measures gently persuaded to take permanent vacations
Ethical Catastrophe
☢️ EXTINCTION-LEVEL WARNING ☢️
THIS MODEL WILL:
- Make Freud blush
- Generate content that requires eye bleach
- Combine technical accuracy with kinks that don't exist in nature
- Make you question why you thought this was a good idea
By downloading you agree:
✅ To never use this for anything important
✅ That your search history is already suspicious
✅ Wave liability harder than a red flag at a bull convention
Model Authors
- sleepdeprived3 (Chief Corruption Officer)
- Downloads last month
- 5
Inference Providers
NEW
This model is not currently available via any of the supported Inference Providers.
The model cannot be deployed to the HF Inference API:
The model authors have turned it off explicitly.