Cheatsheet & Examples: ollama
Run a model
Example Usage:ollama run <model>
What it does:
Starts a model for interactive use, allowing users to input prompts and receive responses directly in the terminal.
Command-line Arguments Explained:
<model>: The name of the model to run (e.g.,llama3,mistral).
Pull a model
Example Usage:ollama pull <model> [--insecure] [--quantize <value>]
What it does:
Downloads a specified model from the Ollama model library to the local machine.
Command-line Arguments Explained:
<model>: The name of the model to pull (e.g.,llama3:latest).--insecure: Disables SSL verification for insecure connections.--quantize <value>: Specifies the quantization level (e.g.,4,8) to reduce model size.
List models
Example Usage:ollama list [--all] [--migrate]
What it does:
Displays all locally installed models, optionally including all available models or migrating outdated ones.
Command-line Arguments Explained:
--all: Shows all models (including those not yet pulled).--migrate: Migrates models to the latest format if needed.
Remove a model
Example Usage:ollama rm <model>
What it does:
Deletes a specified model from the local machine.
Command-line Arguments Explained:
<model>: The name of the model to remove (e.g.,llama3).
Start the Ollama server
Example Usage:ollama serve [--host <address>] [--port <port>] [--modeldir <path>]
What it does:
Launches the Ollama server with custom configuration for hosting models.
Command-line Arguments Explained:
--host <address>: Sets the IP address to bind the server to.--port <port>: Specifies the port number for the server.--modeldir <path>: Defines the directory where models are stored.
Push a model to the Ollama library
Example Usage:ollama push <model> [--insecure]
What it does:
Uploads a locally created model to the Ollama model library for sharing or deployment.
Command-line Arguments Explained:
<model>: The name of the model to push (e.g.,my-custom-model).--insecure: Disables SSL verification for the push operation.
Create a custom model
Example Usage:ollama create <model-name> --file <path-to-config>
What it does:
Generates a new model from a configuration file, allowing customization of parameters or training data.
Command-line Arguments Explained:
<model-name>: The name to assign to the custom model.--file <path-to-config>: Path to the JSON configuration file defining the model.
Manage model configurations
Example Usage:ollama model <model-name> --describe --show
What it does:
Displays detailed information about a model, including its configuration and settings.
Command-line Arguments Explained:
<model-name>: The model to inspect.--describe: Shows a description of the model.--show: Displays the model’s full configuration.
Display help documentation
Example Usage:ollama help [command] [--all]
What it does:
Provides usage instructions and details for a specific command or all commands.
Command-line Arguments Explained:
[command]: The specific command to get help for (e.g.,run,pull).--all: Shows help for all available commands.
