Analyze table rows with Ollama

[kavehshahedi]

What it does

Tables, such as Events Table, contain various information in their rows, each indicating a specific type of data. However, interpreting this information might be challenging due to not being clear or well structured (i.e., human-readable). Now, this commit aims to provide an LLM-based solution for interpreting the table rows using Ollama, a local LLM infrastructure. Accordingly, the user can select any custom rows of a table, and if Ollama is running on their machine, see the interpretation of the row's data.

How to test

1. Ensure that Ollama is already installed and running on your machine.
2. Load a custom table-based data provider, like Events Table
3. Select any desired row of the table, and right-click on it.
4. Choose "Interpret via Ollama", and then, wait for the process to be completed.
5. See the interpretation!

Follow-ups

Choosing a proper model for interpretation is crucial. By default, the code tends to use llama3.2, which is a decent model regarding size/accuracy. If specifying another model in the code, you should consider its required computation resources.

Review checklist

• As an author, I have thoroughly tested my changes and carefully followed the instructions in this template

[marcdumais-work]

Kaveh, have you considered using this library?
https://www.npmjs.com/package/ollama

Repo: https://github.com/ollama/ollama-js

[kavehshahedi]

Kaveh, have you considered using this library? https://www.npmjs.com/package/ollama

Repo: https://github.com/ollama/ollama-js

Yes, but for our use case (and even many more ones), using direct APIs over HTTP works pretty fine. So, I didn't add an extra dependency to the project.