Methception 1.4, LLamaception 1.5 (New hotness), and Finally Introducing Qwenception 1.4: A New Level of Roleplay
Get ready to indulge in three distinct flavors of inception-like secret sauce. Whether you’re rolling with Mistral-based models, Llama 3.3 70B beasts, or Qwen 2.5/coder wizards, these Ception presets bring the heat without losing that chill vibe. Picture a smooth, quantized dreamscape that pulls you in with deep storytelling, real immersion, and on-point energy.
A Trio of Templates
Methception 1.4
Methception is our OG powerhouse built for Mistral-based models using the Metharme prompt format. It’s already proven its might with fine-tunes like Drummers Behemoth 1.2 and MortalWombat’s Monstral V2, layering spatial awareness, emotional undercurrents, and cinematic pacing. The result? Characters who spring to life, fueled by a balanced mix of positive, neutral, and negative biases that make every scenario pop.
LLamaception 1.5
Its sibling, LLamaception, is tailored for models running Llama 3.3 70B. You’ll get the same secret sauce but fine-tuned for Drummers Anubis, Steelskulls Nevoria, and the mind-bending Upgrade to 1.5. With that sweet storytelling, meta examples, and optimized guidelines for your imagination, Llam@ception 1.5 with Nevoria is the awesome before the sauce.
Finally Introducing Qwenception 1.4
Rounding out the trifecta is Qwenception, crafted for Qwen 2.5/coder models. It pairs perfectly with fine-tunes like Evathene by sophosympatheia and Chuluun by Dat Toad, letting you blend code-savvy insights with cinematic flair. Look forward to vivid emotions, seamless continuity, and that same immersive energy, tuned specifically for Qwen’s style.
What’s New Across Methception, LLamaception (see Changelog for 1.5), and Qwenception 1.4?
Enhanced Continuity with Instruction 8
All three Ception presets now feature a boosted sense of continuity, thanks to Instruction 8. Whether you’re tracking down faint spice aromas in Methception, bridging emotional threads in LLamaception, or recalling code snippets in Qwenception, the AI keeps past details at the forefront to make every scene feel cohesive and genuine.
“May I Have Extra Sauce, Please?” (Last Assistant Prefix)
System Prompt support can be limited, but Last Assistant Prefix helps each model—Mistral, and Qwen—revisit context and instructions before crafting the next response. This keeps everything drenched in that juicy, inception-like sauce, ensuring no detail goes astray no matter how deep your worldbuilding or coding adventure runs.
Improved Markdown Formatting
All three templates benefit from a refreshed Context Template, now featuring clear markdown headings and end markers. This not only improves readability but also helps the AI parse structure more effectively.
Strengthened “Show, Don’t Tell” Philosophy
We don’t just talk about continuity, we demonstrate it. Instruction 8 uses its own guidelines while guiding the AI to maintain logical flow across scenes. This self-referential approach turns your roleplay sessions into a seamless journey, where each narrative beat naturally builds on the last.
Why It Matters
When clarity, depth, and immersion are baked into every line, your roleplay becomes an unforgettable ride. These Ception presets—Methception for Mistral, LLamaception for Llama, and Qwenception for Qwen—deliver a potent mix of “meth”-like intensity wrapped in an inception-flavored dream. With updated instructions, polished formatting, and supercharged context handling, you’ll find yourself riding a creative high that keeps on giving. Strap in, tap into your imagination, and let’s cook up something legendary.
Methception Alternate: Compact Inception for Frontends Like KoboldAI Lite
From the mind behind the 123b phenomenon, Monstral V2’s MortalWombat, comes Methception Alternate—a streamlined twist on the original Methception, tailored for single and multi-character cards.
This version packs all the depth and immersion of its predecessor into a compact, "meth"-infused dose of storytelling brilliance. Designed to drop right into frontends like KoboldAI Lite, Methception Alternate delivers that inception-like experience in a format that’s quick, seamless, and still dripping with special sauce.
How to use?
Methception/LLam@ception is formatted to be imported directly into Sillytavern, using "Master Import," located in Advanced Formatting.
Recommended Models: The Secret Sauce for Your Ception Presets
These fine-tunes are your go-to for maximizing the potential of Methception, LLamaception, and Qwenception. Plug them in and let the inception-like vibes take over.
Methception (Mistral 123b)
LLamaception (Llama 70B 3.3)
Qwenception (Qwen 2.5)
Special Thanks & Credits
This project wouldn’t be what it is without the incredible contributions of these legends.
Marinara: For sharing her amazing templates and knowledge, which influenced the creation of these presets. (Click her name for presets)
Mortal Wombat: For revising and adapting ideas to ensure users across different front ends can enjoy the sauce.
Geechan: For providing invaluable feedback, creative ideas, and crafting the brilliant “pacing” instruction.
Arkamist: For contributing concepts around agency and bias that shaped key aspects of these presets.
Thana Alt: For dedicating time to share roleplay experiences from his robust, authentic, and sophisticated multi-character card for testing.
The Drummer: Because "Meth." Enough said.
The BeaverAI Community: For being a constant source of inspiration, feedback, and enthusiasm.
Thank you!