Change set
Pick exam & year, then Go.
Question map
Which of the following statements with regard to Large Language Models (LLMs) used in machine learning is/are correct?
1. LLMs assign probabilities to the next possible words and then pick the one with the highest probability.
2. LLMs process data through mathematical optimization to minimise prediction errors.
3. LLMs produce unbiased outputs.
Select the answer using the code given below:
Explanation
Statement 1 is correct: Large Language Models (LLMs) generate text by predicting the next word (or token) in a sequence. They assign probabilities to all possible next words in their vocabulary based on the preceding context and typically select the one with the highest probability (or sample from the top probabilities) to construct coherent sentences.
Statement 2 is correct: During their training phase, LLMs process massive amounts of text data using mathematical optimization algorithms (such as gradient descent). The goal of this optimization is to adjust the model's internal parameters (weights) to minimize prediction errors (the loss function).
Statement 3 is incorrect: LLMs do not produce unbiased outputs. Because they are trained on vast datasets scraped from the internet and other human-generated text, they inherently absorb and reflect the biases, stereotypes, and prejudices present in that training data.
PROVENANCE & STUDY PATTERN
Guest previewThis is a classic Science & Tech conceptual question driven entirely by the Generative AI boom in current affairs. While standard textbooks won't cover the mathematical nuances of LLMs, basic tech literacy and reading the Science explainers in major dailies makes this highly solvable, especially with the glaring trap in Statement 3.
This question can be broken into the following sub-statements. Tap a statement sentence to jump into its detailed analysis.
- The Union Budget 2025-26 (Economic Survey/Budget documents) explicitly describes the functioning of Generative AI and LLMs.
- It states that the network output assigns probability values to each possible token for the next word.
- It confirms that the tokens with the highest probabilities are selected to generate coherent text.
- Explains the fundamental mechanism of LLMs as 'next-word-prediction machines'.
- Directly mentions that the model generates probabilities for possible next words.
- States that one of the highest probability words is picked to continue the text.
- Academic source from Stanford University detailing the computational principles of LLMs.
- Defines the output of a transformer-based LLM as a distribution over possible next words.
- Explains that generation occurs by sampling from this probability distribution.
This tab shows concrete study steps: what to underline in books, how to map current affairs, and how to prepare for similar questions.
Login with Google to unlock study guidance.
Discover the small, exam-centric ideas hidden in this question and where they appear in your books and notes.
Login with Google to unlock micro-concepts.
Access hidden traps, elimination shortcuts, and Mains connections that give you an edge on every question.
Login with Google to unlock The Vault.