sp26-cs639 cs639
Practice Questions: Transformers & Architectures Practice Questions: Transformers & Architectures
Content locked
The catalog is public. Reading this content requires approved access.
Sign in to read1. QKV Dot-Product Attention [Numerical Computation]
2. Scaled Dot-Product Attention [Conceptual]
3. Multi-Head Attention [Conceptual]
4. Positional Encoding [Conceptual / Comparative]
5. Encoder vs. Decoder Attention Types [Comparative]
6. Causal Mask Matrix Properties [Constraint Recognition]
7. BERT Pretraining [Conceptual]
8. BERT vs. GPT [Comparative]
9. GPT Autoregressive Generation [Conceptual]
10. GPT-3 Training Data [Factual]
11. Llama 3.1 Innovations [Comparative / Factual]
12. Attention Output Computation [Numerical]