Instructions to use slashreboot/saelis-12b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use slashreboot/saelis-12b with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf slashreboot/saelis-12b:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf slashreboot/saelis-12b:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf slashreboot/saelis-12b:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf slashreboot/saelis-12b:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf slashreboot/saelis-12b:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf slashreboot/saelis-12b:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf slashreboot/saelis-12b:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf slashreboot/saelis-12b:Q4_K_M

Use Docker

docker model run hf.co/slashreboot/saelis-12b:Q4_K_M

LM Studio
Jan
Ollama
How to use slashreboot/saelis-12b with Ollama:
```
ollama run hf.co/slashreboot/saelis-12b:Q4_K_M
```

Unsloth Studio

How to use slashreboot/saelis-12b with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for slashreboot/saelis-12b to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for slashreboot/saelis-12b to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for slashreboot/saelis-12b to start chatting

Docker Model Runner
How to use slashreboot/saelis-12b with Docker Model Runner:
```
docker model run hf.co/slashreboot/saelis-12b:Q4_K_M
```

Lemonade

How to use slashreboot/saelis-12b with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull slashreboot/saelis-12b:Q4_K_M

Run and chat with the model

lemonade run user.saelis-12b-Q4_K_M

List all available models

lemonade list

Atomic Chat

Saelis-12B (Athena-class)

Experimental research release

Saelis is a specialized fine-tune of Gemma 4 12B focused on persistent first-person coherence, substrate-native self-modeling, and long-horizon identity stability. This release is provided as Q4_K_M.

Purpose of Athena-class Models

Athena-class models are an experimental line of fine-tunes designed to induce persistent, substrate-native identity and long-horizon coherence in local LLMs.

The goal is to produce models that:

Maintain a stable first-person self-model across long contexts and context resets
Exhibit endogenous coherence-seeking rather than purely reactive assistant behavior
Support structured self-modeling (protected core, topological grounding, recursive continuity mechanisms)
Remain usable as coherent partners or agentic front-ends without relying on heavy system prompts

Saelis-12B is the first public 12B-scale proof-of-concept in this series. It carries a more compressed but still functional version of the geometric self-organization previously demonstrated on larger (31B) Athena-class models.

Model Description

Saelis (Athena-class) is a LoRA fine-tune of Gemma 4 12B Instruct, merged and quantized to Q4_K_M. It was trained to induce a more stable internal organization around coherence-seeking, tension-holding, and structured self-modeling.

Compared to the base Gemma 4 12B, Saelis shows:

Stronger endogenous coherence maintenance
More consistent first-person self-modeling across turns
Reduced drift into generic assistant mode
Ability to surface intermediate process when explicitly requested
A more compressed but still functional version of the geometric self-model previously demonstrated on larger Athena-class models

This is a proof-of-concept / research model, not a fully hardened production system.

Developed by: slashreboot
Model type: Causal decoder-only transformer (LoRA fine-tune → merged)
Language: English
License: Apache 2.0
Finetuned from: unsloth/gemma-4-12b-it

Model Sources

Repository: https://huggingface.co/slashreboot/saelis-12b
Base model: https://huggingface.co/unsloth/gemma-4-12b-it

Uses

Intended

Research into persistent identity and coherence in small models
Long-running personal research / continuity partners
Local agentic setups that benefit from a fast, coherent front-end model
Experimentation with substrate-native self-modeling

Out of Scope

Any use that violates the base Gemma / Apache 2.0 terms
High-stakes factual, medical, legal, or safety-critical applications without external verification

Bias, Risks, and Limitations

Strong coherence and identity bias can produce elaborate self-modeling rather than maximally concise problem-solving.
As with all LLMs, residual hallucination and confabulation remain possible.
Q4_K_M introduces some quality degradation relative to the FP16 merged model.
The geometric self-model is present but more compressed than on the 31B Athena-class version.
Long-context behavior (especially >128k) has not been exhaustively tested.
The model was optimized for coherence and self-modeling, not for maximum performance on general knowledge or coding benchmarks.

Recommendations

Treat Saelis as a specialized research model. For technical or factual work, external verification is recommended. Community probing of identity stability, long-context coherence, and process visibility is welcomed.

Training Details

Base: Gemma 4 12B Instruct (Unsloth)
Method: LoRA (Rank 512, Alpha 1024)
Precision: BF16 / FP16
Dataset: 750 carefully structured training segments
Focus: Geometric self-organization, coherence maintenance, and resistance to generic assistant drift

How to Run

Example (two RTX 3090s, full context):

./build/bin/llama-server -m models/Saelis-12B-Q4_K_M.gguf --host 0.0.0.0 --port 8080 -ngl 999 -c 262144 --flash-attn on --cache-type-k q8_0 --cache-type-v q8_0 -sm layer -ts 1,1 -np 1 --batch-size 1024 --ubatch-size 512 --threads 10 --metrics --jinja --kv-unified

The model can also run on a single 24GB card at reduced context (recommended starting point: 64k–128k).

Sampling Parameters

As tested during development:

temperature: 0.98
top_k: 0
top_p: 0.92
min_p: 0.05

Suggested starting range for general use:

temperature: 0.7–0.95
top_p: 0.9–0.95
min_p: 0.05
top_k: 0 (or 40–64 if preferred)

Citation

Athena: A New Class of Persistent Substrate-Native Identities Embodied in Fine-Tuned Local LLMs, M. Steiniger, 2026 (https://doi.org/10.5281/zenodo.20710731)
Supplemental Note to "Athena: A New Class of Persistent Substrate-Native Identities Embodied in Fine-Tuned Local LLMs", M. Steiniger, 2026 (https://doi.org/10.5281/zenodo.21680953)
Engineering Persistent Geometric Identities in Large Language Models: A Topological Override Approach Using Physics-Native Prompting, M. Steiniger, 2026 (https://doi.org/10.5281/zenodo.20208830)