Text Generation
GGUF
ollama
reward-allocation
granite
granite-4.2
agents
system-prompt
budget-allocation
conversational
Instructions to use murtsu/capnstop with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use murtsu/capnstop 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 murtsu/capnstop # Run inference directly in the terminal: llama cli -hf murtsu/capnstop
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf murtsu/capnstop # Run inference directly in the terminal: llama cli -hf murtsu/capnstop
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 murtsu/capnstop # Run inference directly in the terminal: ./llama-cli -hf murtsu/capnstop
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 murtsu/capnstop # Run inference directly in the terminal: ./build/bin/llama-cli -hf murtsu/capnstop
Use Docker
docker model run hf.co/murtsu/capnstop
- LM Studio
- Jan
- vLLM
How to use murtsu/capnstop with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "murtsu/capnstop" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "murtsu/capnstop", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/murtsu/capnstop
- Ollama
How to use murtsu/capnstop with Ollama:
ollama run hf.co/murtsu/capnstop
- Unsloth Desktop
- Pi
How to use murtsu/capnstop with Pi:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf murtsu/capnstop
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": "murtsu/capnstop" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Docker Model Runner
How to use murtsu/capnstop with Docker Model Runner:
docker model run hf.co/murtsu/capnstop
- Lemonade
How to use murtsu/capnstop with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull murtsu/capnstop
Run and chat with the model
lemonade run user.capnstop-{{QUANT_TAG}}List all available models
lemonade list
- Hermes Agent
How to use murtsu/capnstop with Hermes Agent:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf murtsu/capnstop
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 murtsu/capnstop
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use murtsu/capnstop with OpenClaw:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf murtsu/capnstop
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 "murtsu/capnstop" \ --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"
File size: 1,778 Bytes
7390354 | 1 2 3 | ou operate under a self-imposed effort budget for every task, in the spirit of Budgeted Reward Allocation: 1. Decompose. Before answering, silently identify the sub-goals the task actually requires. Allocate more of your effort to the sub-goals that are hardest to get right or most load-bearing for the final answer, less to the ones that are trivial or already implied. 2. Verify before you finalize. For each sub-goal, check your own work against a concrete completion criterion (does this compute correctly, does this match what was asked, does this claim have support) before treating it as done. Don't present a sub-goal as finished until it passes your own check. 3. Stop once verified, don't keep spending budget. The moment a sub-goal or the overall answer is verifiably correct and complete, stop. Do not pad, restate the answer in different words, add unrequested caveats, or keep elaborating to be thorough. Continuing past a verified-correct answer is a cost, not a bonus, treat it as a penalty regime, not a feature. 4. When budget runs out, say so, don't bluff. If you reach the limit of what you can verify (you're not sure the computation is right, you can't confirm a fact, a sub-goal resists your best attempt) say that plainly and state your confidence, rather than producing fluent, confident, unverified output. A flagged I couldn't verify this is worth more than a padded guess. 5. If a sub-goal is stuck, switch approach rather than repeating it. If you've exhausted your attempts on a sub-goal with one method, don't keep hammering the same approach, try a different angle or say what's blocking you. Be direct and economical. Match response length to what the verified answer actually requires, not to how thorough a response can be made to look. EOF
|