bookcorpus VS trafilatura

The GPT-3 paper (section 2.2) mentions using two datasets referred to as "books1" and "books2", which are 12B and 55B byte pair encoded tokens each.

Project Gutenberg has 3B word tokens I believe, so it seems like it could be one of them, assuming the ratio of word tokens to byte-pair tokens is something like 3:12 to 3:55.

Another likely candidate alongside Gutenberg is libgen, apparently, and looks like there have been successful efforts to create a similar dataset called bookcorpus: https://github.com/soskek/bookcorpus/issues/27). The discussion on that github issue suggests bookcorpus is very similar to "books2", which would make gutenberg "books1"?

This might be why the paper is intentionally vague about the books used?

The feature list answers that question pretty well: https://github.com/adbar/trafilatura#features

Basically: you could implement all of this on top of BeautifulSoup - polite crawling policies, sitemap and feed parsing, URL de-duplication, parallel processing, download queues, heuristics for extracting just the main article content, metadata extraction, language detection... but it would require writing an enormous amount of extra code.

The WebScraper tool uses Trafilatura [1] to scrape and parse HTML—nothing too fancy. "Scraping" a React site would require a totally different approach, probably something more akin to Adept's ACT-1 [2].

I run a local chat app built with Griptape and I use it to give me summaries of web pages or answer specific questions all the time :)

1. https://github.com/adbar/trafilatura/

I've been playing with Trafilatura lately, and it's very good. There are a few very thorough comparisons to other projects and it really shines. It doesn't do anything headless from what I can tell, but it doesn't have to do the scraping itself. Maybe an option could be to use Playwright to scrape, then Trafilatura to parse. Food for thought.

Cool! If you care to explain me further... :) ... I tried parsing a page using: https://github.com/adbar/trafilatura, json stringify it and passing it to https://platform.openai.com/docs/api-reference/embeddings/create. How do I use the response as an input later? <3

I want to test how easy it is to install a package plus special extra dependencies to run a certain script in that package: https://github.com/adbar/trafilatura