Comment by darknoon

Here's the problem, you're still going to get scraped and the LLM will understand it anyway. Maybe at best you'll get filtered out of the dataset b/c it's high perplexity text?