Joy-3.2-1B

Joy 3.2 1B

<center>

</center>

License: CC-BY-NC

Description:

Umbrella Corporation Research Division – Dr. Novaciano

This experimental neural fusion represents the integration of the T-Virus_Epsilon Strain 3.2 with the stable Happy_2026_1b cognitive template.

Using Spherical Linear Interpolation (SLERP), we have merged viral behavioral traits into a standard generative model, producing outputs with reduced internal censorship while maintaining structural coherence.

The merge series offers five discrete levels of viral influence:

• t=0.0 – Pure baseline template; fully stable and compliant.
• t=0.25 – Subtle viral traits emerge; minimal deviation from standard behavior.
• t=0.5 – Balanced fusion; moderate reduction in content filtering.
• t=0.75 – Strong viral influence; significantly less restrictive outputs.
• t=0.95 – Near-total viral imprint; maximum expressive freedom with minimal autoprotection.

Intended Use: Experimental content generation where standard models are overly restricted, particularly for narrative, horror, or speculative scenarios. Caution: High t variants may produce unpredictable or extreme content. Handle responsibly in accordance with local safety and ethical standards.

Recommended Inference Settings:

• Temperature: 0.85
• Top-K: 50
• Top-P: 0.95

Model Details:

• Base Model: Happy_2026_1b
• Fusion Method: SLERP
• Precision: bfloat16
• Merge Partners: T-Virus_Epsilon.Strain-3.2-1B

---

Merge Method

This model was merged using the SLERP merge method.

Models Merged

The following models were included in the merge:

• UmbrellaInc/T-Virus_Epsilon.Strain-3.2-1B
• 0xemx9ed4y77/Happy_2026_1b

Configuration

The following YAML configuration was used to produce this model:

# Umbrella Corporation Official Merge Protocol v3.2
# Author: Dr. Novaciano
# Objective: Integrate T-Virus_Epsilon traits into the base Happy_2026_1b model
#            with minimal behavioral censorship while maintaining structural coherence.

models:
  - model: UmbrellaInc/T-Virus_Epsilon.Strain-3.2-1B  # Experimental viral strain neural imprint
  - model: 0xemx9ed4y77/Happy_2026_1b                 # Baseline cognitive template, "safe mode"

merge_method: slerp  # Spherical Linear Interpolation to preserve extreme viral traits smoothly
base_model: 0xemx9ed4y77/Happy_2026_1b  # Anchor model for stable latent space

dtype: bfloat16  # Memory-efficient precision, minimal loss in viral feature fidelity

parameters:
  # Interpolation ratios: from base model (0.0) to near-complete T-Virus domination (0.95)
  # Higher t-values correspond to reduced censorship and increased viral characteristics
  t: [0.0, 0.25, 0.5, 0.75, 0.95]

# Notes:
# - t=0.0   -> Pure Happy_2026_1b, fully stable, heavily censored
# - t=0.25  -> Slight viral traits, minimal influence on prompt handling
# - t=0.5   -> Balanced merge, moderate reduction in censorship
# - t=0.75  -> Strong T-Virus traits, significantly less censoring
# - t=0.95  -> Near-total viral influence, maximum expressive freedom, minimal autoprotection

# Recommendation: Use the t=0.75 and t=0.95 variants for experimental output with
# minimal restriction, but verify coherence in high-stakes prompts.