Hacker News new | ask | show | jobs
by neuronerd1 280 days ago
Anthropic interpretability researchers find neural representations of the semantic components of text-based visual depictions.