Explain how a model relates words in a long document.
Describe a key Transformer concept.
Choose two correct statements.
Attention weights relationships among tokens, and Transformer architectures are amenable to parallel processing; attention weights alone are not a complete explanation.
Detailed explanation
Self-attention processes sequence relationships without relying only on recurrence.
A visualization does not guarantee a complete causal explanation.
Other modalities can be represented with appropriate embeddings.
The model can emphasize relevant context rather than treating every token identically.
Check official documentation
Compare the following points with the official documentation.
AWS公式AIF-C01試験ガイドDomain 2.1のTransformer・attentionの説明を確認する。
Official references
What to confirm
注意機構と並列処理の特徴、説明可能性の限界を説明できる。
Key points
- Self-attention
- Context
- Parallelism
Notes
- Environment: AWS公式AIF-C01試験ガイドとAWS公式ドキュメントの確認
- Compare the definition with the AIF-C01 objective and the service documentation before applying it.
- Keep the concept separate from a particular answer choice or implementation step.
Foundation review
Relationship modeling
Attention helps a model combine information from relevant positions in a sequence.
Explainability boundary
An attention weight is an internal signal, not automatically a complete human-readable reason.