I asked AI about the feasibility of loading 35 million words of my Journal into a local AI database. Yes. It suggested AnythingLLM. I downloaded and installed it.

​AI had a series of instructions for me as to how to upload my Journal entries but it looks like a lot of work. Each file that has a specific date such as “9-23-94” must also have an accompanying date attached to it. Those daily files go back to April 1991. So that’s about 12,775 files that must be processed by hand. Also, AI tells me that the individual pages have sections with different dates and times. Yeah, like when I was quoting something one of the BBSers posted,

​Well, this confirms that it is possible to set up a local LLM and process it using my own computer. It feels like the processing is still too complicated for local processing. AI said that with 32gb of local memory, AnythingLLM would be able to run a model in the 7 to 8 billion parameter range.

​I have no idea how useful that size of a parameter range is.