Ollama Cheatsheet

Here is a comprehensive Ollama cheat sheet containing most often used commands and explanations:

Installation and Setup

• macOS: Download Ollama for macOS using the command:

  curl -fsSL https://ollama.com/install.sh | sh

• Windows (Preview): Download Ollama for Windows.
• Linux: Use the command:

  curl -fsSL https://ollama.com/install.sh | sh

• Docker: Use the official image available at ollama/ollama on Docker Hub.

Running Ollama

• Run Ollama: Start Ollama using the command:

  ollama serve

• Run a Specific Model: Run a specific model using the command:

  ollama run <model_name>

Model Library and Management

• List Models: List all available models using the command:

  ollama list

• Pull a Model: Pull a model using the command:

  ollama pull <model_name>

• Create a Model: Create a new model using the command:

  ollama create <model_name> -f <model_file>

• Remove a Model: Remove a model using the command:

  ollama rm <model_name>

• Copy a Model: Copy a model using the command:

  ollama cp <source_model> <new_model>

Advanced Usage

• Multimodal Input: Use multimodal input by wrapping multiline text in triple quotes (""") and specifying image paths directly in the prompt.
• REST API Examples:
• Generate a Response: Use the command: curl http://localhost:11434/api/generate -d '{"model": "<model_name>", "prompt": "<prompt>"}'
• Chat with a Model: Use the command:
  bash curl http://localhost:11434/api/chat -d '{"model": "<model_name>", "messages": [{"role": "user", "content": "<message>"}]}'

Integration with Visual Studio Code

• Start Ollama: Start a terminal session and execute the command:

  ollama serve

• Run a Model: Start a second terminal session and execute the command:

  ollama run <model_name>

AI Developer Scripts

• ai_review: Scours through your codebase for specific files, provides suggestions and code examples, and saves them in a review-{current_date}.md file.
• ai_commit: Suggests a commit message based on staged changes.
• ai_readme: Creates a README file automatically based on your project.
• ai_pr: Provides a PR review message automatically.

Additional Resources

• GitHub Repository: Find the GitHub repository for AI developer scripts at https://github.com/ikramhasan/AI-Dev-Scripts.

Other Tools and Integrations

• Lobe Chat: An open-source, modern-design LLMs/AI chat framework supporting multiple AI providers and modalities.
• LangChain: A Java version of LangChain.
• AI Vtuber: A virtual YouTuber driven by various AI models, including Ollama, for real-time interaction with viewers.
• AI Code Completion: A locally or API-hosted AI code completion plugin for Visual Studio Code.

Community and Support

• Reddit: Join the Ollama community on Reddit for discussions and support.

Documentation and Updates

• Official Documentation: Refer to the official Ollama documentation for detailed guides and tutorials.
• GitHub Topics: Explore the Ollama topic on GitHub for updates and new projects.

Additional Tips

• GPU Support: Use the command:

  podman run --rm --device nvidia.com/gpu=all --security-opt=label=disable ubuntu nvidia-smi -L

• NVIDIA GPU Support: Generate the CDI spec according to the documentation and check that your GPU is detected.
• Openshift: Use the commands:

  oc new-project darmstadt-workshop
  oc apply -f deployments/ollama.yaml

• Debugging: Use the command:

  oc run mycurl --image=curlimages/curl -it -- sh

Additional References

• Ollama Cheat Sheet: Refer to the Ollama cheat sheet for detailed information on using Ollama.
• LLM AppDev Hands-On: Refer to the LLM AppDev Hands-On repository for additional information on developing applications with local LLMs.

Additional Tools and Resources

• Streamlit: Use Streamlit to run your Ollama application.
• Podman: Use Podman to run your Ollama application in a container.
• NVIDIA GPU Support: Use the command:

  podman run --rm --device nvidia.com/gpu=all --security-opt=label=disable ubuntu nvidia-smi -L

• Openshift: Use the commands:

  oc new-project darmstadt-workshop
  oc apply -f deployments/ollama.yaml

• Debugging: Use the command:

  oc run mycurl --image=curlimages/curl -it -- sh

Additional Tips and Tricks

• Customize a Model: Use the command:

  ollama create <model_name> -f <model_file>

• Customize Prompt: Use the command:

  ollama create <model_name> -f <model_file> -p <prompt>

• Chat with a Model: Use the command:

  curl http://localhost:11434/api/chat -d '{"model": "<model_name>", "messages": [{"role": "user", "content": "<message>"}]}'

• Generate a Response: Use the command:

  curl http://localhost:11434/api/generate -d '{"model": "<model_name>", "prompt": "<prompt>"}'

• Multimodal Input: Use multimodal input by wrapping multiline text in triple quotes (""") and specifying image paths directly in the prompt.
• REST API Examples:
• Generate a Response: Use the command: curl http://localhost:11434/api/generate -d '{"model": "<model_name>", "prompt": "<prompt>"}'
• Chat with a Model: Use the command:
  bash curl http://localhost:11434/api/chat -d '{"model": "<model_name>", "messages": [{"role": "user", "content": "<message>"}]}'

Additional Resources

• GitHub Repository: Find the GitHub repository for AI developer scripts at https://github.com/ikramhasan/AI-Dev-Scripts.
• Reddit: Join the Ollama community on Reddit for discussions and support.
• Official Documentation: Refer to the official Ollama documentation for detailed guides and tutorials.
• GitHub Topics: Explore the Ollama topic on GitHub for updates and new projects.

Additional Tips and Tricks

• GPU Support: Use the command:

  podman run --rm --device nvidia.com/gpu=all --security-opt=label=disable ubuntu nvidia-smi -L

• NVIDIA GPU Support: Generate the CDI spec according to the documentation and check that your GPU is detected.
• Openshift: Use the commands:

  oc new-project darmstadt-workshop
  oc apply -f deployments/ollama.yaml

• Debugging: Use the command:

  oc run mycurl --image=curlimages/curl -it -- sh

Additional References

• Ollama Cheat Sheet: Refer to the Ollama cheat sheet for detailed information on using Ollama.
• LLM AppDev Hands-On: Refer to the LLM AppDev Hands-On repository for additional information on developing applications with local LLMs.

Additional Tools and Resources

• Streamlit: Use Streamlit to run your Ollama application.
• Podman: Use Podman to run your Ollama application in a container.
• NVIDIA GPU Support: Use the command:

  podman run --rm --device nvidia.com/gpu=all --security-opt=label=disable ubuntu nvidia-smi -L

• Openshift: Use the commands:

  oc new-project darmstadt-workshop
  oc apply -f deployments/ollama.yaml

• Debugging: Use the command:

  oc run mycurl --image=curlimages/curl -it -- sh

Additional Tips and Tricks

• Customize a Model: Use the command:

  ollama create <model_name> -f <model_file>

• Customize Prompt: Use the command:

  ollama create <model_name> -f <model_file> -p <prompt>

• Chat with a Model: Use the command:

  curl http://localhost:11434/api/chat -d '{"model": "<model_name>", "messages": [{"role": "user", "content": "<message>"}]}'

• Generate a Response: Use the command:

  curl http://localhost:11434/api/generate -d '{"model": "<model_name>", "prompt": "<prompt>"}'

• Multimodal Input: Use multimodal input by wrapping multiline text in triple quotes (""") and specifying image paths directly in the prompt.
• REST API Examples:
• Generate a Response: Use the command: curl http://localhost:11434/api/generate -d '{"model": "<model_name>", "prompt": "<prompt>"}'
• Chat with a Model: Use the command:
  bash curl http://localhost:11434/api/chat -d '{"model": "<model_name>", "messages": [{"role": "user", "content": "<message>"}]}'

Additional Resources

• GitHub Repository: Find the GitHub repository for AI developer scripts at https://github.com/ikramhasan/AI-Dev-Scripts.
• Reddit: Join the Ollama community on Reddit for discussions and support.
• Official Documentation: Refer to the official Ollama documentation for detailed guides and tutorials.
• GitHub Topics: Explore the Ollama topic on GitHub for updates and new projects.

Additional Tips and Tricks

• GPU Support: Use the command:

  podman run --rm --device nvidia.com/gpu=all --security-opt=label=disable ubuntu nvidia-smi -L

• NVIDIA GPU Support: Generate the CDI spec according to the documentation and check that your GPU is detected.
• Openshift: Use the commands:

  oc new-project darmstadt-workshop
  oc apply -f deployments/ollama.yaml

• Debugging: Use the command:

  oc run mycurl --image=curlimages/curl -it -- sh

Additional References

• Ollama Cheat Sheet: Refer to the Ollama cheat sheet for detailed information on using Ollama.
• LLM AppDev Hands-On: Refer to the LLM AppDev Hands-On repository for additional information on developing applications with local LLMs.

Additional Tools and Resources

• Streamlit: Use Streamlit to run your Ollama application.
• Podman: Use Podman to run your Ollama application in a container.
• NVIDIA GPU Support: Use the command:

  podman run --rm --device nvidia.com/gpu=all --security-opt=label=disable ubuntu nvidia-smi -L

• Openshift: Use the commands:

  oc new-project darmstadt-workshop
  oc apply -f deployments/ollama.yaml

• Debugging: Use the command:

  oc run mycurl --image=curlimages/curl -it -- sh

Additional Tips and Tricks

• Customize a Model: Use the command:

  ollama create <model_name> -f <model_file>

• Customize Prompt: Use the command:

  ollama create <model_name> -f <model_file> -p <prompt>

• Chat with a Model: Use the command:

  curl http://localhost:11434/api/chat -d '{"model": "<model_name>", "messages": [{"role": "user", "content": "<message>"}]}'

• Generate a Response: Use the command:

  curl http://localhost:11434/api/generate -d '{"model": "<model_name>", "prompt": "<prompt>"}'

• Multimodal Input: Use multimodal input by wrapping multiline text in triple quotes (""") and specifying image paths directly in the prompt.
• REST API Examples:
• Generate a Response: Use the command: curl http://localhost:11434/api/generate -d '{"model": "<model_name>", "prompt": "<prompt>"}'
• Chat with a Model: Use the command:
  bash curl http://localhost:11434/api/chat -d '{"model": "<model_name>", "messages": [{"role": "user", "content": "<message>"}]}'

Additional Resources

• GitHub Repository: Find the GitHub repository for AI developer scripts at https://github.com/ikramhasan/AI-Dev-Scripts.
• Reddit: Join the Ollama community on Reddit for discussions and support.
• Official Documentation: Refer to the official Ollama documentation for detailed guides and tutorials.
• GitHub Topics: Explore the Ollama topic on GitHub for updates and new projects.

Additional Tips and Tricks

• GPU Support: Use the command:

  podman run --rm --device nvidia.com/gpu=all --security-opt=label=disable ubuntu nvidia-smi -L

• NVIDIA GPU Support: Generate the CDI spec according to the documentation and check that your GPU is detected.
• Openshift: Use the commands:

  oc new-project darmstadt-workshop
  oc apply -f deployments/ollama.yaml

• Debugging: Use the command:

  oc run mycurl --image=curlimages/curl -it -- sh

Additional References

• Ollama Cheat Sheet: Refer to the Ollama cheat sheet for detailed information on using Ollama.
• LLM AppDev Hands-On: Refer to the LLM AppDev Hands-On repository for additional information on developing applications with local LLMs.

Additional Tools and Resources

• Streamlit: Use Streamlit to run your Ollama application.
• Podman: Use Podman to run your Ollama application in a container.
• NVIDIA GPU Support: Use the command:

  podman run --rm --device nvidia.com/gpu=all --security-opt=label=disable ubuntu nvidia-smi -L

• Openshift: Use the commands:

  oc new-project darmstadt-workshop
  oc apply -f deployments/ollama.yaml

• Debugging: Use the command:

  oc run mycurl --image=curlimages/curl -it -- sh

Additional Tips and Tricks

• Customize a Model: Use the command:

  ollama create <model_name> -f <model_file>

• Customize Prompt: Use the command:

  ollama create <model_name> -f <model_file> -p <prompt>

• Chat with a Model: Use the command:

  curl http://localhost:11434/api/chat -d '{"model": "<model_name>", "messages": [{"role": "user", "content": "<message>"}]}'

• Generate a Response: Use the command:

  curl http://localhost:11434/api/generate -d '{"model": "<model_name>", "prompt": "<prompt>"}'

• Multimodal Input: Use multimodal input by wrapping multiline text in triple quotes (""") and specifying image paths directly in the prompt.
• REST API Examples:
• Generate a Response: Use the command: curl http://localhost:11434/api/generate -d '{"model": "<model_name>", "prompt": "<prompt>"}'
• Chat with a Model: Use the command:
  bash curl http://localhost:11434/api/chat -d '{"model": "<model_name>", "messages": [{"role": "user", "content": "<message>"}]}'

Additional Resources

• GitHub Repository: Find the GitHub repository for AI developer scripts at https://github.com/ikramhasan/AI-Dev-Scripts.
• Reddit: Join the Ollama community on Reddit for discussions and support.
• Official Documentation: Refer to the official Ollama documentation for detailed guides and tutorials.
• GitHub Topics: Explore the Ollama topic on GitHub for updates and new projects.

Additional Tips and Tricks

• GPU Support: Use the command:

  podman run --rm --device nvidia.com/gpu=all --security-opt=label=disable ubuntu nvidia-smi -L

• NVIDIA GPU Support: Generate the CDI spec according to the documentation and check that your GPU is detected.
• Openshift: Use the commands:

  oc new-project darmstadt-workshop
  oc apply -f deployments/ollama.yaml

–