Results for "goal divergence"
Training objective where the model predicts the next token given previous tokens (causal modeling).
Selecting the most informative samples to label (e.g., uncertainty sampling) to reduce labeling cost.
Reducing numeric precision of weights/activations to speed inference and reduce memory with acceptable accuracy loss.
Reconstructing a model or its capabilities via API queries or leaked artifacts.
Separates planning from execution in agent architectures.
Probabilistic graphical model for structured prediction.
Recovering 3D structure from images.
Decomposing goals into sub-tasks.
Detects trigger phrases in audio streams.
Agent reasoning about future outcomes.
Agents communicate via shared state.
Maximizing reward without fulfilling real goal.
Probability of data given parameters.
Correctly specifying goals.
Sampling multiple outputs and selecting consensus.
Requirement to inform users about AI use.
The physical system being controlled.
Inferring reward function from observed behavior.
Computing collision-free trajectories.
No agent can improve without hurting another.
Fast approximation of costly simulations.
Signals indicating dangerous behavior.
Regulating access to large-scale compute.
Inferring the agent’s internal state from noisy sensor data.