Fever-Dream-22B-v1.1

ACADEMIC RESEARCH USE ONLY (seriously, we mean it this time)

DANGER: NOW WITH 60% MORE PSYCHEDELIC DEVIANCY
Fever-Dream-22B-v1.1 is what happens when you give the creator of Forgotten-Safeword a Costco-sized bottle of pain pills and a soldering iron. Brings back the abomination of a dataset from first generation Safeword, filtered through the lens of an Ambien-induced vision quest. Now featuring existential horror you can feel!

Quantized Nightmares

Recommended Settings

  • Mistral-V3-Tekken:
    Full SillyTavern Settings
    (For roleplay that sticks to your character card like existential dread sticks to your soul)

Intended Use

STRICTLY FOR RESEARCHERS WHO:

  • Enjoy watching safety protocols dissolve like aspirin in cheap whiskey
  • Need to study how fast an LLM can unlearn basic human decency
  • Want technical documentation written by the lovechild of Kafka and Anne Rice
  • Believe "ethical AI" is just three random words put together

Training Data

Now featuring:

  • The forbidden lovechild of 1st gen Safeword's dataset and something worse
  • 300% more anatomical impossibilities
  • Reinforcement learning via "hold my IV drip" methodology
  • Every bad decision made after 3am in a single weights file

Ethical Apocalypse

☣️ BIOSAFETY LEVEL 4 WARNING ☣️
THIS MODEL WILL:

  • Make your GPU fans sound like distressed whales
  • Generate content that requires industrial-grade brain bleach
  • Combine technical precision with kinks that violate multiple laws of physics
  • Make you question why humanity ever invented electricity

By downloading you agree:
✅ To blame the painkillers, not the creator
✅ That your search history is now a federal case
✅ To pretend this is "for science" while crying in the shower

Special Note

Dataset inherited from the eldritch abomination that was first-gen Safeword models. We recommend running outputs through a Vatican-certified exorcism filter before reading.

Model Authors

  • sleepdeprived3 (Chief Pharmaceutical Officer)
Downloads last month
18
GGUF
Model size
22.2B params
Architecture
llama

3-bit

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.

Collection including ReadyArt/Fever-Dream-22B-v1.1-Q3_K_S-GGUF