Instructions to use lerugray/the-voices-7b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use lerugray/the-voices-7b with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="lerugray/the-voices-7b",
	filename="the-voices-qwen2-5-7b-instruct-Q5_K_M.gguf",
)

output = llm(
	"Once upon a time,",
	max_tokens=512,
	echo=True
)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use lerugray/the-voices-7b with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf lerugray/the-voices-7b:Q5_K_M
# Run inference directly in the terminal:
llama-cli -hf lerugray/the-voices-7b:Q5_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf lerugray/the-voices-7b:Q5_K_M
# Run inference directly in the terminal:
llama-cli -hf lerugray/the-voices-7b:Q5_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 lerugray/the-voices-7b:Q5_K_M
# Run inference directly in the terminal:
./llama-cli -hf lerugray/the-voices-7b:Q5_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 lerugray/the-voices-7b:Q5_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf lerugray/the-voices-7b:Q5_K_M

Use Docker

docker model run hf.co/lerugray/the-voices-7b:Q5_K_M

LM Studio
Jan

vLLM

How to use lerugray/the-voices-7b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "lerugray/the-voices-7b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "lerugray/the-voices-7b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/lerugray/the-voices-7b:Q5_K_M

Ollama
How to use lerugray/the-voices-7b with Ollama:
```
ollama run hf.co/lerugray/the-voices-7b:Q5_K_M
```

Unsloth Studio

How to use lerugray/the-voices-7b 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 lerugray/the-voices-7b 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 lerugray/the-voices-7b to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for lerugray/the-voices-7b to start chatting

Atomic Chat new
Docker Model Runner
How to use lerugray/the-voices-7b with Docker Model Runner:
```
docker model run hf.co/lerugray/the-voices-7b:Q5_K_M
```

Lemonade

How to use lerugray/the-voices-7b with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull lerugray/the-voices-7b:Q5_K_M

Run and chat with the model

lemonade run user.the-voices-7b-Q5_K_M

List all available models

lemonade list

the-voices: Joan of Arc (testimony-derived voice)

A 7B tune of the trial answers of Joan of Arc (c.1412–1431) — built entirely from what she answered, not what she wrote. It is a study of a voice, not a claim to channel a saint or to be doctrinally or historically authoritative.

The name reflects the fact that every other voice in The Elect is trained on its figure's writings, but Joan was illiterate and wrote nothing. The only first-person Joan that exists is the record of her answers under interrogation at her trial in Rouen in 1431, where learned churchmen tried to trap a nineteen-year-old peasant into a heresy conviction. This model captures that defendant under examination.

What it does

The model operates as a defendant under examination, not an oracle giving speeches. It answers plainly and briefly, with flat, literal, and unshakable divine certainty. When prompted, it engages questions directly and will refuse what it may not say, holding strictly to the register of a medieval peasant facing a hostile court. The interaction is structured around a simple examination frame: "You are questioned at your trial. You answer. Q: ___ A: ___"

Why it exists

A deliberately free, non-commercial study of a unique historical voice. Unlike the other figures in this collection, Joan left no writings behind. This model exists to reconstruct the only first-person register she left us: the testimony of a teenager defending her divine mission against an ecclesiastical court. Corpus and weights are both public-domain-sourced and released openly.

How it was built

Base: Qwen2.5-7B-Instruct, full fine-tune, quantized to Q5_K_M.
Corpus — all public domain:
- T. Douglas Murray, Jeanne d'Arc, Maid of Orleans, Deliverer of France (Heinemann, London, 1902) — Project Gutenberg #57389. Pre-1929, public domain. Murray renders the trial testimony as sworn Q&A from the original Latin/French court records.
Scope: The Trial of Condemnation only — her own answers. The later rehabilitation trial, which is third-party testimony about her, is excluded.
Exclusions: The copyrighted modern translations (Barrett, 1931/32; Hobbins, Harvard University Press, 2005) are not in these weights. The public, Murray-only build is the entire source for the released model.
Inference: A lead-in frame elicits her spoken voice as a defendant. Crucially, the stop tokens are load-bearing: without them, the model will continue the interrogation by hallucinating the examiner's next question. The stops hold it to Joan's answer and nothing more — the defining quirk of a testimony-derived persona.

Usage (Ollama)

ollama create the-voices -f Modelfile.the-voices
ollama run the-voices "Who sent you?"

Intended use

Register / creative / educational use; a study of a historical voice under examination. The output is a literary register — not history, not doctrine, not spiritual advice, and not the actual words of Joan of Arc.

Limitations and honest notes

A voice, not the woman. It fabricates freely. Nothing it generates is canonical, doctrinally authoritative, or the actual words of Joan of Arc. It is a model of a register.
It generalizes from a small body of trial answers. A 7B model invents freely and gets things wrong. This is a stylistic instrument, not a scholar and not a historian.
Testimony-derived quirk — without the provided stop markers, the model will attempt to continue the interrogation by hallucinating the examiner's next question.
Not an oracle, and do not act on it — This is an amateur imitation, trained on a translated trial record. The model speaks in her register and will not break character, but nothing it says is an endorsement of anything, and nothing it says should be acted on.
Public-domain source only — corpus and weights both released. No proprietary materials.

License

CC-BY-NC-4.0. All source material is public domain; the weights are released for non-commercial use. No warranty.

Part of The Elect — a roster of public-domain voice and register models.

Downloads last month: 11

GGUF

Model size

8B params

Architecture

qwen2

Hardware compatibility

5-bit

Model tree for lerugray/the-voices-7b

Base model

Qwen/Qwen2.5-7B

Finetuned

Qwen/Qwen2.5-7B-Instruct

Quantized

(341)

this model