Image by Editor | ChatGPT
In the current era of large language model (LLM) products such as ChatGPT or Gemini, we have relied on these tools to increase our productivity by helping with various tasks, such as answering questions, summarizing documents, planning activities, and more. These tools have become a part of our everyday lives.
However, many of these products are hosted in the cloud, and we need to access them through their platform. Moreover, each platform is typically limited to its own proprietary models and will not allow the usage of other LLMs. This is why the Ollama application was developed to help users who wish to implement various LLMs in their local environment.
Ollama has been around for some time and has already helped many users run language models locally. The latest update has made it even more powerful.
Let’s explore why Ollama’s new application is becoming an essential tool for many users.
# Ollama’s New App
As mentioned, Ollama is an open-source tool that runs LLMs in a local environment without relying on cloud hosting. We can explore which models are available on Ollama’s site, download them, and run them directly using the application interface. The application acts as a local model manager that hosts various models for us to execute freely. A local application like this benefits the user by allowing them to have freedom while still preserving data privacy and reducing any latency.
The biggest development is that Ollama now runs as a standalone GUI app, as opposed to solely via command line. Gone are the days of necessarily seeking out, installing, and configuring a third party UI app (or writing your own) that could turn Ollama into a more pleasurable experience. Sure you can still do all of that, but there is no need to any longer.
With the new update, the Ollama application has become more useful than ever, and we will explore each of its features individually.
The Ollama application allows users to utilize an LLM locally by downloading the model into their local environment. You can interact with a model by selecting it and providing a prompt to get a result.
Ollama keeps your conversation history, allowing you to follow up on the conversation.
If a model is not yet available locally, the new Ollama application will automatically download it before executing the prompt. This simplifies the user experience, as they no longer need to download models separately before use.
Another new feature is the ability to chat with your files. By dragging and dropping a file onto the Ollama app, we can ask questions about its contents.
The result is shown below, where the model accesses the file and executes the prompt based on the provided document.
If you need to process many larger documents, you can increase Ollama’s context length through the settings. However, increasing the context length will require more memory to ensure stable performance.
Ollama’s new features are not limited to textual documents like PDF and Word. It also offers multimodal support, provided the chosen model can process different data types. For example, we can use a model like Llama to process an image, as shown below.
The result is shown below.
Lastly, the new Ollama application can process code files to produce documentation. This feature is especially beneficial for developers.
The result is shown below.
These are the key new features Ollama provides to improve your productivity. We encourage you to utilize them in your work.
# Wrapping Up
Ollama is an application that allows users to run LLMs in a local environment. It is a valuable tool for any user who wants to test various models and keep their data private. Its new release offers more powerful features, including a usable chat UI, easy model downloading, the ability to chat with your files, multimodal support, and code processing.
I hope this has helped!
Cornellius Yudha Wijaya is a data science assistant manager and data writer. While working full-time at Allianz Indonesia, he loves to share Python and data tips via social media and writing media. Cornellius writes on a variety of AI and machine learning topics.
