Instructions to use mindchain/decider-2b-vision-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use mindchain/decider-2b-vision-GGUF 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 mindchain/decider-2b-vision-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf mindchain/decider-2b-vision-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf mindchain/decider-2b-vision-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf mindchain/decider-2b-vision-GGUF: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 mindchain/decider-2b-vision-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf mindchain/decider-2b-vision-GGUF: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 mindchain/decider-2b-vision-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf mindchain/decider-2b-vision-GGUF:Q4_K_M

Use Docker

docker model run hf.co/mindchain/decider-2b-vision-GGUF:Q4_K_M

LM Studio
Jan
Ollama
How to use mindchain/decider-2b-vision-GGUF with Ollama:
```
ollama run hf.co/mindchain/decider-2b-vision-GGUF:Q4_K_M
```
Unsloth Desktop

How to use mindchain/decider-2b-vision-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf mindchain/decider-2b-vision-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "mindchain/decider-2b-vision-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use mindchain/decider-2b-vision-GGUF with Docker Model Runner:
```
docker model run hf.co/mindchain/decider-2b-vision-GGUF:Q4_K_M
```

Lemonade

How to use mindchain/decider-2b-vision-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull mindchain/decider-2b-vision-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.decider-2b-vision-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use mindchain/decider-2b-vision-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf mindchain/decider-2b-vision-GGUF:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default mindchain/decider-2b-vision-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use mindchain/decider-2b-vision-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf mindchain/decider-2b-vision-GGUF:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "mindchain/decider-2b-vision-GGUF:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

decider-2b-vision Q4_K_M + mmproj — der Bild-Entscheider der decider-Familie

Typed decisions from an image in one forward pass — Kamera-Bild rein, kalibrierte Wahrscheinlichkeiten raus. Der fehlende Baustein für den Kamera-Pfad (Android, alte GPUs, Edge).

File	Größe	Zweck
`decider-2b-vision.Q4_K_M.gguf`	1,27 GB	Sprach-/Entscheidungsteil (Qwen3.5-2B-Basis, v5 Text-Gewichte)
`mmproj-decider-2b-vision-f16.gguf`	668 MB	Vision-Projektor (f16, wie bei Qwen-VL/smolVLM üblich)


Revision	Commit-gepinnt `863e290863655f1d6b69324d77d09ac972d21609` (Repo hat keine Tags)
Toolchain	llama.cpp Commit `9575389`: Text-Pfad `--no-mtp` + bf16 → Q4_K_M; mmproj separat mit `--mmproj --outtype f16` exportiert
⚠️ Toolchain-Warnung	llama.cpp master (`e351231`, Sep 2026) baut kaputte `qwen3_5_text`-GGUFs (falsche Metadaten + verschobene Gewichte, Output `?`). Pin euren Commit; `Qwen3_5ForConditionalGeneration` ist im gepinnten Baum in `qwen.py` (Text) und `qwen3vl.py` (VL) registriert.

Usage (llama-server, zwei Files)

llama-server -m decider-2b-vision.Q4_K_M.gguf \
  --mmproj mmproj-decider-2b-vision-f16.gguf \
  -ngl 99 --alias decider-2b-vision

Rolle im JEV-Stack

Kamera-/Bild-Pfad der Tier-0/1-Kette: Bild rein → Entscheidung + Wahrscheinlichkeiten in einem Forward-Pass. Q4 ~1,3 GB läuft auf einer GTX 1070 neben anderen Services; Text-Schwester decider-2b liest bei T=1.03 (Vision-Temperatur: im decider/-Paket des Upstream-Repos — vor Produktionsnutzung gegen den bf16-Original-Readout verifizieren, wie bei allen unseren Quants).