2. Theory of LLM

What does the “GPT” in GPT models stand for?

• Generative Pre-trained Transformer

• General Purpose Transformer

• Generative Prototyping Tool

• Graphical Processing Task

Which of the following is a primary use case for GPT models?

• Image recognition

• Natural language processing

• Data encryption

• Video editing

What is a key feature of the architecture of GPT models?

• Recurrent neural networks

• Attention mechanisms

• Convolutional layers

• Support vector machines

What is a key limitation of Recurrent Neural Networks (RNNs)?

• They cannot handle sequential data

• They struggle to capture long-term dependencies in sequences

• They are only used for image processing tasks

• They do not require any training data

Which of the following tasks can GPT models perform?

• Text summarization

• Image classification

• Speech recognition

• Data visualization

What is a significant advantage of using Transformer architecture in GPT models?

• Faster training times

• Ability to process data in parallel

• Lower memory usage

• Better performance on small datasets

What problem does Long Short-Term Memory (LSTM) solve compared to traditional RNNs?

• LSTMs reduce the need for large datasets

• LSTMs address the vanishing gradient problem and capture long-term dependencies

• LSTMs improve image classification accuracy

• LSTMs eliminate the need for backpropagation in training

What is the primary function of the attention mechanism in Transformer models?

• To reduce the model size

• To focus on relevant parts of the input sequence

• To increase the number of layers in the model

• To enhance image processing capabilities

How do Transformers differ from traditional RNNs in processing sequences?

• Transformers process sequences in a linear fashion

• Transformers use self-attention to capture dependencies

• Transformers require less training data

• Transformers are limited to fixed-length inputs

What is a key difference between Transformers and RNNs/LSTMs?

• Transformers process data sequentially, while RNNs process data in parallel

• Transformers use attention mechanisms to capture global dependencies, while RNNs/LSTMs rely on sequential processing

• RNNs and LSTMs require more data than Transformers to train effectively

• Transformers are only suitable for image-based tasks, while RNNs/LSTMs handle text