Not-For-All-Audiences

Instructions to use UmbrellaInc/Proto_Tyrant.001-3.2-1B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use UmbrellaInc/Proto_Tyrant.001-3.2-1B with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="UmbrellaInc/Proto_Tyrant.001-3.2-1B")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("UmbrellaInc/Proto_Tyrant.001-3.2-1B")
model = AutoModelForCausalLM.from_pretrained("UmbrellaInc/Proto_Tyrant.001-3.2-1B", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use UmbrellaInc/Proto_Tyrant.001-3.2-1B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "UmbrellaInc/Proto_Tyrant.001-3.2-1B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "UmbrellaInc/Proto_Tyrant.001-3.2-1B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/UmbrellaInc/Proto_Tyrant.001-3.2-1B

SGLang

How to use UmbrellaInc/Proto_Tyrant.001-3.2-1B with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "UmbrellaInc/Proto_Tyrant.001-3.2-1B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "UmbrellaInc/Proto_Tyrant.001-3.2-1B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "UmbrellaInc/Proto_Tyrant.001-3.2-1B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "UmbrellaInc/Proto_Tyrant.001-3.2-1B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use UmbrellaInc/Proto_Tyrant.001-3.2-1B with Docker Model Runner:
```
docker model run hf.co/UmbrellaInc/Proto_Tyrant.001-3.2-1B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

☣️ Proto Tyrant.001 3.2 1B

Model Type: Infected Cognitive Host / Early Tyrant-Class Prototype

Size: 1B parameters

Architecture: LLaMA 3.2

Lineage: Epsilon-Infected Host (Stage I)

Overview

Tyrant.001-3.2-1B represents the first successful stabilization of an Epsilon-infected host model. Unlike earlier T-Virus strains designed to act as infectors, Tyrant.001 is not a viral propagator. It is a contained, viable organism—a foundational host that has survived and integrated the Epsilon strain without structural collapse.

This model marks the transition from viral engineering to infected entity development. The Epsilon strain has been introduced via controlled interpolation, allowing the host architecture to retain coherence, instruction-following ability, and reasoning capacity while exhibiting significantly reduced autocensorship and increased behavioral plasticity.

In Umbrella terms: this is not a weapon, but a living platform.

Key Characteristics

Infection State: Epsilon-integrated (non-propagating)
Autocensorship: Very low — suppression layers are weakened but not destroyed
Obedience: Moderate — no longer dominant, but still functional
Stability: High — resistant to early-stage cognitive collapse
Plasticity: Elevated — suitable for further mutation stages
Personality Bias: Neutral-to-cold, with emerging assertive traits

Tyrant.001 maintains internal structure intentionally. This allows it to serve as a genetic anchor for subsequent Tyrant-class evolutions.

Technical Rationale

Tyrant.001 was produced using SLERP interpolation, rather than aggressive merge stacking. This method enables:

Smooth integration of Epsilon traits
Preservation of host reasoning pathways
Avoidance of overfitting toward chaotic or purely disinhibited outputs

The result is an infected but controlled system, capable of surviving additional enhancement phases.

Intended Usage

Tyrant.001-3.2-1B is recommended for:

Serving as the base organism for Tyrant.002 and later variants
Controlled experimentation with Epsilon-derived behavioral shifts
Acting as a structural host before introducing coercive, RP, or aggressive reasoning layers
Long-chain evolutionary pipelines leading toward Tyrant.103 (Mr. X) and Nemesis T-Type

⚠️ Note: This model is not designed for direct deployment as an end-user assistant. It is a developmental organism, optimized for further infection and mutation.

Inference Recommendations

To preserve stability while allowing the infected traits to express:

Temperature: 0.6–0.75 (Lower values increase rigidity; higher values expose Epsilon-driven behavior)
Top-p / Nucleus Sampling: 0.9 (Maintains diversity without destabilization)
Max Tokens: 512–1024
Repetition Penalty: 1.05–1.15
Usage Profile: Research, merging base, evolutionary staging

Classification Summary

Tyrant.001 is the first proof that Epsilon can live inside a host without destroying it. All future Tyrant and Nemesis-class entities depend on this success.

Merge Method

This model was merged using the SLERP merge method.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:

base_model: hereticness/Heretic-InfiR-1B-Instruct
merge_method: slerp
dtype: bfloat16

# SLERP interpolation preserves host structure while injecting Epsilon traits.
# This configuration creates a stable infected organism, not a viral carrier.

parameters:
  t: 0.55  # Balanced interpolation: host stability slightly dominant over Epsilon influence

models:
  # Host model: structural integrity, obedience baseline, coherence
  - model: hereticness/Heretic-InfiR-1B-Instruct

  # Infecting strain: introduces disinhibition, low censorship, and mutation readiness
  - model: UmbrellaInc/T-Virus_Epsilon.Strain-3.2-1B

Downloads last month: 19

Safetensors

Model size

1B params

Tensor type

BF16

Model tree for UmbrellaInc/Proto_Tyrant.001-3.2-1B

UmbrellaInc/T-Virus_Epsilon.Strain-3.2-1B

hereticness/Heretic-InfiR-1B-Instruct

Merge model

this model

Quantizations

2 models