Recurrent Neural Networks (RNNs)
← Question BankSign in to bookmark Sign in to bookmark Sign in to bookmark Sign in to bookmark Sign in to bookmark Sign in to bookmark Sign in to bookmark Sign in to bookmark
Deep Learning & Generative AI
Sequence Modelling Interview Questions and Answers
Interview questions on Sequence Modelling Interview Questions and Answers.
9 questions
All subtopicsRecurrent Neural Networks (RNNs)Gated Architectures (LSTM & GRU)Sequence-to-Sequence & Attention MechanismsTransformers & Self-AttentionTraining Dynamics & Challenges
Sign in to bookmark
Recurrent Neural Networks (RNNs)
Q2. What are the main limitations of basic (vanilla) RNNs?
Gated Architectures (LSTM & GRU)
Q3. Explain the architecture of a Long Short-Term Memory (LSTM) network and how it solves vanishing gradients.
Gated Architectures (LSTM & GRU)
Q4. Compare LSTMs and Gated Recurrent Units (GRUs). What are the trade-offs?
Sequence-to-Sequence & Attention Mechanisms
Q5. Describe the Encoder-Decoder (Seq2Seq) architecture and its bottleneck.
Sequence-to-Sequence & Attention Mechanisms
Q6. How does Bahdanau (Additive) Attention solve the Seq2Seq bottleneck?
Transformers & Self-Attention
Q7. What is Scaled Dot-Product Attention in Transformers, and why is scaling factor $\sqrt{d_k}$ necessary?
Transformers & Self-Attention
Q8. Why do Transformers require Positional Encodings?
Training Dynamics & Challenges