NVIDIA Generative AI with LLMs Associate (NCA-GENL) - 400 Questions

1. A data scientist is building a neural network for image classification. The model uses multiple hidden layers where each neuron applies a weighted sum followed by a nonlinear activation function. Which type of neural network architecture is this describing?

2. A machine learning engineer is training a deep neural network and notices that the gradients in the early layers become extremely small, effectively preventing weight updates. Which problem is the engineer encountering?

3. A team is training a transformer-based language model on NVIDIA A100 GPUs. They want to understand the core mechanism that allows the model to weigh the importance of different tokens in a sequence relative to each other. Which mechanism are they referring to?

4. A company is deploying a text classification model and needs an architecture that processes the entire input sequence bidirectionally to produce rich contextual embeddings. Which model family is best suited for this task?

5. A team is fine-tuning a large language model using a technique where a reward model is trained on human preference data, and then the language model is optimized using reinforcement learning with that reward signal. Which training approach is this?