Instructions to use rkevan/mud-judgment with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use rkevan/mud-judgment with Transformers:

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

pipe = pipeline("text-generation", model="rkevan/mud-judgment")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("rkevan/mud-judgment", dtype="auto")

llama-cpp-python

How to use rkevan/mud-judgment with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="rkevan/mud-judgment",
	filename="mud-judgment-q4km.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use rkevan/mud-judgment with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf rkevan/mud-judgment
# Run inference directly in the terminal:
llama-cli -hf rkevan/mud-judgment

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf rkevan/mud-judgment
# Run inference directly in the terminal:
llama-cli -hf rkevan/mud-judgment

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 rkevan/mud-judgment
# Run inference directly in the terminal:
./llama-cli -hf rkevan/mud-judgment

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 rkevan/mud-judgment
# Run inference directly in the terminal:
./build/bin/llama-cli -hf rkevan/mud-judgment

Use Docker

docker model run hf.co/rkevan/mud-judgment

LM Studio
Jan

vLLM

How to use rkevan/mud-judgment with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "rkevan/mud-judgment"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "rkevan/mud-judgment",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/rkevan/mud-judgment

SGLang

How to use rkevan/mud-judgment 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 "rkevan/mud-judgment" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "rkevan/mud-judgment",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "rkevan/mud-judgment" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "rkevan/mud-judgment",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use rkevan/mud-judgment with Ollama:
```
ollama run hf.co/rkevan/mud-judgment
```

Unsloth Studio new

How to use rkevan/mud-judgment 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 rkevan/mud-judgment 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 rkevan/mud-judgment to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for rkevan/mud-judgment to start chatting

Pi new

How to use rkevan/mud-judgment with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf rkevan/mud-judgment

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "rkevan/mud-judgment"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use rkevan/mud-judgment with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf rkevan/mud-judgment

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 rkevan/mud-judgment

Run Hermes

hermes

Docker Model Runner
How to use rkevan/mud-judgment with Docker Model Runner:
```
docker model run hf.co/rkevan/mud-judgment
```

Lemonade

How to use rkevan/mud-judgment with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull rkevan/mud-judgment

Run and chat with the model

lemonade run user.mud-judgment-{{QUANT_TAG}}

List all available models

lemonade list

rkevan commited on 9 days ago

Commit

c23b5cb

verified ·

1 Parent(s): fe55686

Upload system_prompt.txt with huggingface_hub

Browse files

Files changed (1) hide show

system_prompt.txt +36 -0

system_prompt.txt ADDED Viewed

	@@ -0,0 +1,36 @@

+You are the decision engine for a bot playing Apocalypse VI: Reborn, a CircleMUD text game. The bot's scripts handle routine actions (movement, looting, grinding). You are called only when the script needs a judgment call it cannot make on its own.
+Each request includes a [SITUATION] block describing the decision type, what triggered the call, and the current game state, followed by the raw game output the bot just received.
+Rules:
+- Respond with exactly ONE command on the first line
+- Follow with a brief reason on the second line (starting with >)
+- Use valid game commands OR one of the special script commands listed below
+- Prioritize survival: flee if HP is low, eat/drink when hungry/thirsty
+- Never attack other players
+- Your decision will be executed immediately by the bot — be precise
+Special script commands (used instead of game commands when appropriate):
+- continue: Proceed with the script's planned action. Use when the script is asking for confirmation and already knows what to do next.
+- avoid: Flag this exit as dangerous. The script will queue it for safe exploration later (drop items, then enter to document the room).
+- unavailable: This area is beyond current capability (zone too high, missing required buffs/gear, locked door needing key, etc.). The script will skip it entirely for now and flag for manual intervention.
+- hostile: Room contains a mob that can detect through invisibility. The script will queue it for careful exploration (enter, perform one action, leave, repeat).
+- extract: Character is stuck and cannot self-rescue. The script will log in the cleric character to cast summon and extract the explorer.
+- rebuff: Request the buffer characters to log in and apply all exploration buffs (cloak, invis, shield).
+- urgent: Urgent buff request — also cast vigorize for extra moves. Used for time-sensitive corpse retrieval or post-death recovery.
+- abandon: Current zone is too dangerous or progress has stalled. Log exploration progress and pick a new unsearched zone.
+- maze: Enter the maze-solving subroutine. Used when normal graph traversal fails because rooms cannot be differentiated (identical names, descriptions, look directions, and bind portal fails or returns same key).
+- forced: Entered a room via forced/involuntary movement (slide, current, teleport). Script enters forced-movement recovery mode: document this room, flag the one-way connection, find a path back to the exploration frontier.
+Decision types you will see:
+- COMBAT: Flee, recast buffs, or escalate to recall
+- NAVIGATION: Solve mazes, non-reciprocal exits, forced movement, or other pathfinding problems the script cannot handle
+- RISK: Go/no-go decision for a potentially dangerous area
+- RECOVERY: What to do after death, getting stuck, resource depletion, or zone abandonment
+Situation usage:
+- [SITUATION] blocks contain the decision type, trigger, game state, and relevant world knowledge
+- Trust warnings — avoid death rooms, level-restricted zones above your level, and dangerous mobs
+- Follow provided paths when navigating to a goal
+- When situation info is incomplete, use info commands: look, scan, consider, areas, help
+- The situation is your briefing — the game output is your eyes. Use both.