LLM Fundamentals

Labels

Approximately how many characters does a single token represent in English text?

2 months ago

Why does tokenisation directly affect the cost of using LLM APIs?

2 months ago

What happens when a prompt exceeds the model's context window limit?

2 months ago

What does prompt caching do in LLM APIs?

2 months ago

Why can't an LLM answer questions about events that occurred after its training data cutoff?

2 months ago

What does setting temperature=0 do when calling an LLM API?

2 months ago

What does the top-p (nucleus sampling) parameter control in text generation?

2 months ago

What does the top-k sampling parameter do during text generation?

2 months ago

Why might an LLM produce slightly different outputs even with temperature=0?

2 months ago

What are stop sequences in LLM API calls?

2 months ago

What is the core mechanism that makes the Transformer architecture so effective for language tasks?

2 months ago

What does self-attention allow a language model to do?

2 months ago

Why did Transformer models largely replace RNNs (recurrent neural networks) for language modelling?

2 months ago

What do the billions of parameters in a large language model represent?

2 months ago

What is the primary training objective during LLM pre-training?

2 months ago

What is fine-tuning in the context of large language models?

2 months ago

What is instruction tuning and why does it make models more useful?

2 months ago

What is RLHF and what problem does it solve in LLM training?

2 months ago

What is Constitutional AI, and what makes it different from standard RLHF?

2 months ago

What are embeddings in the context of language models?

2 months ago