Ollama Example
This document introduces two ways to interact with Ollama: CLI (Command Line Interface) and Python API.
References:
Ollama Commands
Ollama provides a rich set of command-line tools that make it easy to pull, run, delete, and manage models.
You can use the ollama -h command to view the help information for Ollama.
ollama -h
The output will be as follows:
Large language model runner
Usage:
ollama [flags]
ollama [command]
Available Commands:
serve Start ollama
create Create a model from a Modelfile
show Show information for a model
run Run a model
stop Stop a running model
pull Pull a model from a registry
push Push a model to a registry
list List models
ps List running models
cp Copy a model
rm Remove a model
help Help about any command
Flags:
-h, --help help for ollama
-v, --version Show version information
Use "ollama [command] --help" for more information about a command.
CLI Usage
CLI stands for Command-Line Interface, which refers to interacting with a program by entering commands in a terminal or command line.
Pulling a Model
The parameters of Ollama models are related to the device's memory. We need to choose the appropriate model parameters based on the device's memory capacity.
According to Ollama's official documentation: You should have at least 8 GB of RAM available to run the 7B models, 16 GB to run the 13B models, and 32 GB to run the 33B models. This helps us estimate the appropriate model parameters for our device.
For users with limited device memory, we can choose models with smaller parameters, such as deepseek-r1:1.5b.
Use the ollama pull command to download a model. You can find more models and their corresponding commands in the Ollama Library.
ollama pull deepseek-r1:1.5b
The download progress will be displayed in the terminal. Once completed, you'll see output similar to the following:
pulling manifest
pulling aabd4debf0c8: 100% ▕███████████████████████████████████████████████████▏ 1.1 GB
pulling c5ad996bda6e: 100% ▕███████████████████████████████████████████████████▏ 556 B
pulling 6e4c38e1172f: 100% ▕███████████████████████████████████████████████████▏ 1.1 KB
pulling f4d24e9138dd: 100% ▕███████████████████████████████████████████████████▏ 148 B
pulling a85fe2a2e58e: 100% ▕███████████████████████████████████████████████████▏ 487 B
verifying sha256 digest
writing manifest
success
Running a Model
Use the ollama run command to run a model. If the model isn't available locally, Ollama will automatically download it from the remote repository.
ollama run deepseek-r1:1.5b
After successful execution, an interactive interface will appear in the terminal.
You can directly ask questions like Please introduce yourself, and the model will respond.
> > > Please introduce yourself
> > > Hello! I'm DeepSeek-R1, an artificial intelligence assistant created by DeepSeek. I'm at your service and
> > > would be delighted to assist you with any inquiries or tasks you may have.
> > > Send a message (/? for help)
Type /bye or press Ctrl + D to exit the interactive mode.
Python Usage
Ollama provides an easy-to-use Python library for convenient model operations like pulling, running, and deleting models.
Using the Ollama Python library requires Python 3.8 or higher.
We recommend installing and using the ollama Python library in a Conda environment.
You can install the Ollama Python library using the pip command.
pip3 install ollama
Standard Response
Standard responses are synchronous, where the model returns a complete response all at once.
You can use Jupyter Lab cells to run the Python code or copy the code directly into a Python file.
Make sure the model parameter in the Python code matches a model that has been pulled locally.
from ollama import chat
from ollama import ChatResponse
response: ChatResponse = chat(model='deepseek-r1:1.5b', messages=[
{
'role': 'user',
'content': 'Please introduce yourself',
},
])
print(response['message']['content'])
# or access fields directly from the response object
print(response.message.content)
Streaming Response
Streaming responses are asynchronous, where the model returns a stream of responses (output is generated and displayed progressively).
You can use Jupyter Lab cells to run the Python code or copy the code directly into a Python file.
Make sure the model parameter in the Python code matches a model that has been pulled locally.
from ollama import chat
stream = chat(
model='deepseek-r1:1.5b',
messages=[{'role': 'user', 'content': 'Please introduce yourself'}],
stream=True,
)
for chunk in stream:
print(chunk['message']['content'], end='', flush=True)