Updates Oct 14
PGrid
- Added new cpu and cpu generation tables to db
- Got popular cpu names and benchmarks of those cpus from a single article into json
- Added a new scraper to get specs for those cpus and populate the db.
- Added new method of classifying “generic” items that I have been scraping starting with cpus. Its a new table that maps a generic item to a cpu row.
- Updated amazon scraper to be more generic and handle different kinds of products
- Playing around with prompts for cpu classification. Had 3 iterations including using cheaper ai models to classify then verify each other, or using claude to classify multiple cpus in one go. Still not fully reliable but using claude might be the way.
- Got worker state timeline working again but running into same issues as before where it would send incorrect state as soon as a task was started. Oddly one of the servers is working and sending state correctly.

- Initial cpu table on frontend

- Initial cpu details page but not complete

- Refactoring pages and price history chart to work to not be gpu specific
- Added little navigation bar

Fujifilm PGrid
- Procrastinated on getting amazon scraper for different countries.