I've been using NotebookLM for a few weeks to organize notes from our system docs and post-mortems. I like it, but sometimes it seems to miss key points from my source files, especially longer logs or messy markdown.
I read that pre-processing the text before uploading can really help. What are the most effective things to do? For example, should I be stripping out repetitive error lines from logs, or breaking a giant file into smaller, thematic chunks? I'm mainly working with .txt and .md files from AWS and Docker setups.
I've noticed the same thing with longer logs. Breaking them into smaller chunks definitely helped in my case - I started grouping error logs by timestamp or session ID.
What would you recommend for cleaning up markdown? I sometimes see weird formatting artifacts from pasted code snippets that seem to throw things off.