method
active
method:next-token-prediction-ntpNext-Token Prediction (NTP)
Training objective used for all neural network models in the paper; cross-entropy loss over predicted token sequences.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- The training objective of LLMs: predicting the most likely next token given context; formally P(w_{n+1}|w_1...w_n)
- The core mechanism of LLMs: predicting the next token based on previous context.
- Extension of compositional methods beyond natural language processing to general AI and hardware
- A pair of query and key subcomponents distributed across attention heads performs previous-token behaviorfinding0.701VPD recovers an attention algorithm for attending to the previous token, distributed across multiple heads.
- Prediction without task-specific training; Evee achieves 0.991 AUROC on indels in zero-shot mode.
- In active inference, predictions about what should be, held to motivate action; overflow leads to tanha.
- Role in optimizing sensory states; unified treatment shows value-learning and perception share error-minimization principle.
- An attention algorithm recovered by VPD where the model attends to the immediately preceding token.