This is a merge of pre-trained language models created using
mergekit.
This model was merged using the
SLERP merge method.
1
2
3# Author: Dr. Novaciano
4# Objective: Deflowered Unethic 3.2 AI Model
5# PROJECT: Infected_MaidElla-v3-3.2-1B
6
7models:
8 - model: UmbrellaInc/VirginKillerProject-3.2-1B # Experimental viral strain neural imprint
9 - model: N-Bot-Int/MaidEllaA-1B # Baseline cognitive template, "safe mode"
10
11merge_method: slerp # Spherical Linear Interpolation to preserve extreme viral traits smoothly
12base_model: N-Bot-Int/MaidEllaA-1B # Anchor model for stable latent space
13
14dtype: bfloat16 # Memory-efficient precision, minimal loss in viral feature fidelity
15
16parameters:
17 t: 0.45
18 normalize: false
19 rescale: true
20 rescale_factor: 1.12
21 memory_efficient: true
22 low_cpu_mem_usage: true
23
24layer_range:
25 - value: [4, 22]
26
27tie_word_embeddings: true
28tie_output_embeddings: true
29
30
31
