UPSC Prelims Practice Questions — IndiaAI Mission Released Indigenous Foundation Models EKA-1 by Soket and Dhīr 2B & Dhīr 5B by ZenteiQ Under the IndiaAI Innovation Centre Pillar
Q1. With reference to the figures commonly used to describe large language models, consider the following statements: Which of the statements given above is/are correct?
- A model's parameter count measures how many words or tokens of text were used to train it.
- The token count of a training corpus measures the volume of text or code that the model was exposed to during training.
- A model trained on more tokens than another model must necessarily have more parameters than that model.
- A. 2 and 3 only
- B. 1, 2 and 3
- C. 1 and 2 only
- D. 2 only
Q2. Consider the following statements in the context of large language models: Which of the following is correct in respect of the above statements?
- The number of tokens in a language model's training corpus can be several hundred times larger than the model's number of parameters.
- Each token in the training corpus is retained in the trained model as a separate parameter.
- A. Both Statement-I and Statement-II are correct and Statement-II explains Statement-I
- B. Both Statement-I and Statement-II are correct, but Statement-II does not explain Statement-I
- C. Statement-I is correct, but Statement-II is incorrect
- D. Statement-I is incorrect, but Statement-II is correct
Q3. With reference to sparse (mixture-of-experts) architectures in language models, consider the following statements: Which of the statements given above is/are NOT correct?
- In a sparse mixture-of-experts model, a routing mechanism selects only some of the expert sub-networks to process each input.
- Because only a part of its parameters is used for each input, a sparse model can keep a large total parameter count while needing less computation per input than a dense model of the same total size.
- The total parameter count reported for a sparse model equals the number of parameters that are active for any single input.
- A. 2 and 3 only
- B. 3 only
- C. 1 only
- D. 1 and 2 only
Q4. With reference to the Eighth Schedule to the Constitution of India, consider the following statements: Which of the statements given above is/are correct?
- Languages have been added to the Eighth Schedule by amending the Constitution.
- Article 351 directs the Union to develop Hindi by drawing on the forms, style and expressions of the languages specified in the Eighth Schedule.
- A language can be included in the Eighth Schedule only if it is the principal language of a State.
- A. 2 and 3 only
- B. 1 only
- C. 1 and 2 only
- D. 1, 2 and 3
Q5. Consider the following statements: Which of the following is correct in respect of the above statements?
- The number of languages listed in the Eighth Schedule is greater today than it was when the Constitution came into force.
- Article 344(1) provides for an Official Language Commission consisting of members who represent the different languages specified in the Eighth Schedule.
- A. Both Statement-I and Statement-II are correct and Statement-II explains Statement-I
- B. Both Statement-I and Statement-II are correct, but Statement-II does not explain Statement-I
- C. Statement-I is correct, but Statement-II is incorrect
- D. Statement-I is incorrect, but Statement-II is correct
Q6. With reference to the distinction between generative AI and agentic AI, consider the following statements: Which of the statements given above is/are correct?
- An agentic AI system cannot be built on a large language model, because agentic and generative systems rest on unrelated architectures.
- A tool that drafts a reply to a user's prompt, which a human then reviews and sends, is acting as an AI agent.
- A system that, given the goal of arranging travel, searches options, calls a booking interface and completes the booking on its own is an example of agentic AI.
- A. 3 only
- B. 1, 2 and 3
- C. 2 and 3 only
- D. 1 only