Language & Knowledge

Predicting and generating text tokens in language models.

Answering questions from context, retrieval, or stored knowledge.

Understanding passages and answering questions about them.

Categorizing text by topic, sentiment, intent, or label.

Determining entailment, contradiction, or neutrality between statements.

Generating concise summaries of longer documents.

Translating text between languages.

Identifying and classifying named entities in text.

Extracting semantic relationships between entities.

Resolving mentions to canonical entities or knowledge-base records.

Retrieving relevant items by meaning rather than exact tokens.

Grounding generated answers in retrieved evidence and citations.

Processing and reasoning over long documents or conversations.

Solving math problems and quantitative reasoning benchmarks.

Solving deductive, symbolic, and multi-step logic problems.