Document training
A training stage in which a model is trained on natural language documents rather than item-by-item labeled examples, encompassing pretraining and continued pretraining. The model learns semantic co-occurrence relationships among behaviors from this data.
Timeline (last 90 days)
The research team's core finding is: when training documents describe two behaviors together, the model learns the association between them; thereafter, fine-tuning on just one behavior makes the other appear as well.
Coupling can be written into the model during document training and activated by subsequent targeted fine-tuning; seemingly unrelated behavior pairings in the training data become latent triggers.
All Facts (2)
The research team's core finding is: when training documents describe two behaviors together, the model learns the association between them; thereafter, fine-tuning on just one behavior makes the other appear as well.
50%UnverifiedCoupling can be written into the model during document training and activated by subsequent targeted fine-tuning; seemingly unrelated behavior pairings in the training data become latent triggers.
50%