Been Kim and colleagues, 2018
TCAV uses directional derivatives to compute models’ conceptual sensitivity across entire classes of inputs.
Don't Just Read it – Explore It
TCAV uses directional derivatives to compute models’ conceptual sensitivity across entire classes of inputs.
PPO strikes a balance between ease of implementation, sample complexity, and ease of tuning.
How do we create agents that behave in accordance with the user’s intentions?
We explore a new direction for scaling up reinforcement learning from human feedback.
Computational properties can emerge as collective properties of systems having a large number of simple equivalent components.
A neural network model for a mechanism of visual pattern recognition is proposed in this paper.
Performance has a power-law relationship with each of the three scale factors: model size, dataset size, and computation.
Language models begin to learn these tasks without any explicit supervision when trained on a new dataset of millions of webpages.
Scaling up language models greatly improves task-agnostic, few-shot performance, sometimes reaching competitiveness with prior fine-tuning approaches.
We present a general end-to-end approach to sequence learning that makes minimal assumptions on the sequence structure.
AlphaGo achieved a 99% win rate against the strongest other Go programs, and defeated European champion Fan Hui 5–0.
The agent is not given any information about the rules or purpose of the game.