← Back to Feed
Retweet: Frontier models lack strong visual grounding for document OCR, with LlamaIndex claiming advances in positional
Retweet: Frontier models lack strong visual grounding for document OCR, with LlamaIndex claiming advances in positional accuracy.
Original Post
RT @jerryjliu0: One of the biggest requirements for document OCR is visual grounding, and frontier models (gemini, opus, gpt-5.4) suck at iā¦