File size: 4,916 Bytes
9ebdf5c
afc6ef4
 
97bb7d1
 
afc6ef4
97bb7d1
afc6ef4
97bb7d1
afc6ef4
97bb7d1
 
afc6ef4
97bb7d1
 
afc6ef4
97bb7d1
afc6ef4
97bb7d1
afc6ef4
97bb7d1
 
afc6ef4
97bb7d1
 
afc6ef4
97bb7d1
 
afc6ef4
97bb7d1
 
9ebdf5c
97bb7d1
 
 
c8d19d4
9ebdf5c
 
 
 
 
97bb7d1
9ebdf5c
 
 
 
 
 
fe57731
 
 
 
 
f043dd7
fe57731
 
 
4d17657
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
![image/png](https://cdn-uploads.huggingface.co/production/uploads/647aa776555b5e199c012d76/ezXwmj1uPJ3XLb-BSdGnB.png)


**Methception 1.4 and LLamaception 1.4: A New Level of Roleplay**  
Welcome to the twin peaks of roleplay presets—the sweet sauce of immersive storytelling now comes in two flavors. Whether you're rocking Mistral-based models or Llama 3.3 70B beasts, you can sprinkle that inception-like secret sauce to craft scenes that feel real, robust, and downright addictive.

---

### A Tale of Two Templates

#### Methception 1.4  
Methception is the classic high-octane preset designed for Mistral-based models using the Metharme prompt format. It’s already proven its mettle with fine-tunes like **Drummers Behemoth 1.2** and **MortalWombat’s Monstral V2**, adding layers of spatial awareness, emotional tension, and dynamic pacing. It keeps your roleplay rolling smoothly while providing a blend of positive, neutral, and negative biases that evoke genuine agency among characters.

#### LLamaception 1.4  
LLamaception is Methception’s brand-new sibling, tailored for models built on **Llama 3.3 70B**. It packs the same delicious sauce but optimized for fine-tunes like **Drummers Anubis**, **Steelskulls Evalibis & Evayale**, and the mind-bending brilliance of **Eva-Unit/Sao10k**. Expect the same focus on sensory immersion, nuanced emotional arcs, and cinematic pacing—just recalibrated for the power of Llama-based architecture.

---

### What’s New in Methception 1.4 (and LLamaception 1.4)?

#### Enhanced Continuity with Instruction 8  
The big star of this update is **Instruction 8**, which doubles down on maintaining a logical flow across multiple scenes. Whether you’re sniffing out faint aromas of spices in Methception or bridging tension from earlier chapters in LLamaception, Instruction 8 teaches the AI how to recall and build on previously mentioned details—making your roleplay smoother and more cohesive.

#### “May I Have Extra Sauce, Please?” – Last Assistant Prefix  
To tackle limited System Prompt support across both Mistral and Llama powerhouses, we’ve introduced the **Last Assistant Prefix**. Before spinning up a new response, the model re-reads the context and instructions, making sure everything stays tight-knit and dripping with that inception-like sauce 🤌. No matter how deep or complex your world-building goes, your AI remains consistent and coherent.

#### Improved Markdown Formatting  
Both **Methception** and **LLamaception** templates now rock a restructured Context Template with crisp markdown headings and end marks. Cleaner formatting means better readability for you and improved structural parsing for the AI—win-win!

#### Strengthened the “Show, Don’t Tell” Philosophy  
We’re not just preaching about continuity; we’re living it. **Instruction 8** demonstrates its own approach by actively referencing earlier context as it explains how to apply continuity. It’s meta in the best way possible, helping the model practice what it preaches.

---

### Why It Matters  
With **Methception 1.4** and **LLamaception 1.4**, you get a synergy of clarity and depth that will catch you by surprise—in a good way. The new instruction set, formatting improvements, and context-handling enhancements all culminate in deeply immersive roleplay. Whether you’re diving into a gritty, character-driven saga or cozying up with your waifu in a dreamy slice-of-life scenario, these templates hit like a euphoric, "quantized" dream. 

![image/png](https://cdn-uploads.huggingface.co/production/uploads/647aa776555b5e199c012d76/AGj6KvmrIr6A0SJcMA0Jg.png)

How to use?

Methception/LLam@ception is formatted to be imported directly into Sillytavern, using "Master Import," located in Advanced Formatting.


![image/png](https://cdn-uploads.huggingface.co/production/uploads/647aa776555b5e199c012d76/pnCJwSljYt6Uh19-0Klat.png)


![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/647aa776555b5e199c012d76/2Gzd4-zkc4raX-OUc9r9v.jpeg)



### **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.

![image/png](https://cdn-uploads.huggingface.co/production/uploads/647aa776555b5e199c012d76/yZKaV6h5wOXG8NwYgjU0m.png)


![image/png](https://cdn-uploads.huggingface.co/production/uploads/647aa776555b5e199c012d76/WX6FwgWdjbKwwBh_N2Bdf.png)