This is a small project that has a threefold aim:
- learn something new by doing
- summarise interesting companies' news articles and blog posts
- summarise the summaries to extract a global overview of a company's activities and aims
The idea was born when I started going through articles posted in a company's own website, showcasing their successes and focus. At that time I wondered whether I could easily come up with an overall summary of their activities. It was also a good opportunity to test my theory, that it is possible to utilise LLMs on a modest personal computer to increase the volume of information I can assimilate when my time is limited.
This project was implemented in Julia, leveraging the good work of the ollama project. Julia is a joyful language to use, enabling productivity from the get go, with a pretty complete ecosystem that enables Data Scientists to implement a wide gamut of solutions within most domains.
On macOS and Linux, launch a terminal window and type:
$ git clone https://github.com:inferential/NewsSummarised.jl$ cd NewsSummarised.jl$ ./run.sh
On Windows
-
using WSL:
same as above -
using PowerShell:
- Clone
https://github.com:inferential/NewsSummarised.jlwith a tool of your choice - Navigate to
NewsSummarised.jl - Check if Julia is installed on your system. If not, run
winget install julia -s msstore - Run
julia --proj src/NewsSummarised.jl
The first run may take a while, as every article is getting summarised and all summaries are summarised into a final document.
LLM inference on computers with a relatively modern GPU will be faster. Inference on CPU is possible, but expect a significantly lower output.
Subsequent runs will be faster, as the tool loads news/final-summary.txt from storage before printing.
See issues for more information.
*<base url>/robots.txt is respected ✅