Immersive learning
VR 360 presentation will use this question view when WebXR is available.
Choose a topic
Pick a topic to continue learning.
Choose a format
How would you like to work through it?
Change subject
Choose a subject and topic
Your current selection will stay unchanged until you confirm.
A transformer language model can condition its prediction only on the most recent 2,000 input tokens. A relevant sentence appears 2,500 tokens before the prediction point and is not otherwise represented. What follows?
Explore ideas
Concept library
Browse the ideas behind new questions. Choose a subject and topic, then swipe through concise concept cards.
Assessment complete
Test review
Explanations
Answer review
Your answer
Concept explanation