In the process of learning PDF files, AI does not handle many symbols well.
Some files are recognized by OCR and may contain some symbols and pictures.
AI will produce various misunderstandings during the recognition learning process.
For example, a Chinese music theory book.
Any progress to make it not a toy ?
Last time I checked, Gemini does not look through images in a PDF file and really just extracts the raw text content (in the case of AI Studio)
I think so, and the problem is some OCR symbols couldn’t be extracted correctly, it may contain spaces, wrong line breaks. etc…
I guess working files processing is very needed if AI being man’s helper