Instructions to use h3rb3rn/olmo3-expert-graphrag-7b-v4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use h3rb3rn/olmo3-expert-graphrag-7b-v4 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="h3rb3rn/olmo3-expert-graphrag-7b-v4")# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("h3rb3rn/olmo3-expert-graphrag-7b-v4", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use h3rb3rn/olmo3-expert-graphrag-7b-v4 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 h3rb3rn/olmo3-expert-graphrag-7b-v4:Q4_K_M # Run inference directly in the terminal: llama cli -hf h3rb3rn/olmo3-expert-graphrag-7b-v4:Q4_K_M
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf h3rb3rn/olmo3-expert-graphrag-7b-v4:Q4_K_M # Run inference directly in the terminal: llama cli -hf h3rb3rn/olmo3-expert-graphrag-7b-v4:Q4_K_M
Build from source code
git clone https://github.com/ggml-org/llama.cpp.git cd llama.cpp cmake -B build cmake --build build -j --target llama-app # Start a local OpenAI-compatible server with a web UI: ./build/bin/llama serve -hf h3rb3rn/olmo3-expert-graphrag-7b-v4:Q4_K_M # Run inference directly in the terminal: ./build/bin/llama cli -hf h3rb3rn/olmo3-expert-graphrag-7b-v4:Q4_K_M
Use Docker
docker model run hf.co/h3rb3rn/olmo3-expert-graphrag-7b-v4:Q4_K_M
- Llama new
- LM Studio
- Jan
- vLLM
How to use h3rb3rn/olmo3-expert-graphrag-7b-v4 with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "h3rb3rn/olmo3-expert-graphrag-7b-v4" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "h3rb3rn/olmo3-expert-graphrag-7b-v4", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/h3rb3rn/olmo3-expert-graphrag-7b-v4:Q4_K_M
- SGLang
How to use h3rb3rn/olmo3-expert-graphrag-7b-v4 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 "h3rb3rn/olmo3-expert-graphrag-7b-v4" \ --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": "h3rb3rn/olmo3-expert-graphrag-7b-v4", "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 "h3rb3rn/olmo3-expert-graphrag-7b-v4" \ --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": "h3rb3rn/olmo3-expert-graphrag-7b-v4", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Ollama
How to use h3rb3rn/olmo3-expert-graphrag-7b-v4 with Ollama:
ollama run hf.co/h3rb3rn/olmo3-expert-graphrag-7b-v4:Q4_K_M
- Unsloth Desktop
- Pi
How to use h3rb3rn/olmo3-expert-graphrag-7b-v4 with Pi:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf h3rb3rn/olmo3-expert-graphrag-7b-v4: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:9931/v1", "api": "openai-completions", "apiKey": "none", "models": [ { "id": "h3rb3rn/olmo3-expert-graphrag-7b-v4:Q4_K_M" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Docker Model Runner
How to use h3rb3rn/olmo3-expert-graphrag-7b-v4 with Docker Model Runner:
docker model run hf.co/h3rb3rn/olmo3-expert-graphrag-7b-v4:Q4_K_M
- Lemonade
How to use h3rb3rn/olmo3-expert-graphrag-7b-v4 with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull h3rb3rn/olmo3-expert-graphrag-7b-v4:Q4_K_M
Run and chat with the model
lemonade run user.olmo3-expert-graphrag-7b-v4-Q4_K_M
List all available models
lemonade list
- Hermes Agent
How to use h3rb3rn/olmo3-expert-graphrag-7b-v4 with Hermes Agent:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf h3rb3rn/olmo3-expert-graphrag-7b-v4: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://127.0.0.1:9931/v1 hermes config set model.default h3rb3rn/olmo3-expert-graphrag-7b-v4:Q4_K_M
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use h3rb3rn/olmo3-expert-graphrag-7b-v4 with OpenClaw:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf h3rb3rn/olmo3-expert-graphrag-7b-v4: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://127.0.0.1:9931/v1 \ --custom-model-id "h3rb3rn/olmo3-expert-graphrag-7b-v4: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"
olmo3-expert-graphrag-7b-v4
Kleines Experten-Modell fรผr das MoE-Sovereign-System, Rolle graphrag (Cypher, Neo4j 5). Basis: allenai/Olmo-3-7B-Instruct (OLMo-3-7B-Instruct). Datei: olmo3-expert-graphrag-7b-v4-Q4_K_M.gguf (Q4_K_M, getestet mit KV-Cache q4_0 im Dorf-Standard).
Training
SFT-Rolle graphrag-v3 (LoRA r128), dann 100 Schritte GRPO mit Neo4j-Ausfรผhrungsbelohnung. Training auf LUMI-G (LoRA, TRL-GRPO, Belohnung = ausfรผhrungsgeprรผftes Ergebnis mit Teilkredit 0 / 0,1+0,6ยทZeilenรผberdeckung / 1,0). Das Wissen steckt nicht im Modell: KB-first, Fakten kommen aus der Wissensbasis.
Messung (harter, ausfรผhrungsgeprรผfter Test, bestandene Aufgaben greedy / Sampling Seed 1)
graphrag 13 / 9 von 35 (graphrag-v3: 9 / 7); datainfra 6 / 5; precision 5 / 8.
Der Test besteht aus 109 selbst geschriebenen Aufgaben (datainfra 38 gegen PostgreSQL 17, graphrag 35 gegen Neo4j 5.26, precision 36). Unterschiede unter 3 Aufgaben sind Rauschen.
Einschrรคnkungen
- Gemessen in bf16 auf LUMI-G, nicht in Q4_K_M; die Messung der Q4_K_M-Datei steht aus.
- Die GRPO-Belohnung lief gegen PostgreSQL 16.2 bzw. Neo4j 5.26 mit prozeduralen Aufgaben, der Test gegen PostgreSQL 17.
- Die Spezialisierung kostet andere Rollen Leistung (siehe Zahlen oben). Kleine Stichproben, ein Training je Modell.
- Ohne Wissensbasis erfinden Experten Fakten; nur mit KB-Kontext einsetzen.
SHA-256 der GGUF-Datei: 7223105369e7841088b89a0b633f0ef7b267a71af0e73ce5383e871efbb594f3
- Downloads last month
- 36
4-bit
Model tree for h3rb3rn/olmo3-expert-graphrag-7b-v4
Base model
allenai/Olmo-3-1025-7B