Fetching the paper…

Scalable Influence and Fact Tracing for Large Language Model Pretraining · Around