{"id":1558,"date":"2026-04-30T12:39:20","date_gmt":"2026-04-30T07:09:20","guid":{"rendered":"https:\/\/www.techandtrain.com\/blog\/?p=1558"},"modified":"2026-04-30T12:39:24","modified_gmt":"2026-04-30T07:09:24","slug":"learnings-from-assignments-open-book-exams-at-indian-institute-of-technology-gandhinagar-executive-masters-in-data-science-for-decision-making","status":"publish","type":"post","link":"https:\/\/www.techandtrain.com\/blog\/2026\/04\/learnings-from-assignments-open-book-exams-at-indian-institute-of-technology-gandhinagar-executive-masters-in-data-science-for-decision-making\/","title":{"rendered":"Learnings from assignments \/ open book exams at Indian Institute of Technology Gandhinagar – Executive Masters in Data Science for Decision Making"},"content":{"rendered":"\n

One important lesson I learned while working with spatio-temporal graph data on the METR-LA dataset during my Executive Masters open-book assignment:<\/p>\n\n\n\n

Do not keep switching between Claude, ChatGPT, Perplexity, Gemini, and other LLMs or AI tools during the execution stage. This lesson has repeated itself in the two years throughout the Executive Masters whenever we have been allowed to use LLMs.<\/p>\n\n\n\n

My learning:<\/p>\n\n\n\n

\u2022 Different LLMs reason differently<\/p>\n\n\n\n

\u2022 They are trained and fine-tuned differently<\/p>\n\n\n\n

\u2022 They suggest different libraries, assumptions, fixes, and coding styles<\/p>\n\n\n\n

\u2022 Mixing their guidance during debugging can create unnecessary chaos<\/p>\n\n\n\n

\u2022 What looks like \u201cmore intelligence\u201d can become \u201cmore confusion\u201d<\/p>\n\n\n\n

\u2022 Multi-model thinking is useful during brainstorming<\/p>\n\n\n\n

\u2022 It helps in debating, exploring, comparing, and expanding ideas<\/p>\n\n\n\n

\u2022 But once execution begins, consistency matters more than variety<\/p>\n\n\n\n

\u2022 Pick one model and work through the problem step by step<\/p>\n\n\n\n

\u2022 Ask it to explain, debug, simplify, correct, and iterate<\/p>\n\n\n\n

\u2022 Stay with one reasoning path until the solution stabilizes<\/p>\n\n\n\n

My conclusion:<\/p>\n\n\n\n

Use multiple LLMs for exploration.<\/p>\n\n\n\n

Use one LLM for execution.<\/p>\n\n\n\n

Mixing models during ideation can create insight.<\/p>\n\n\n\n

Mixing models during implementation can create chaos.<\/p>\n\n\n\n

This is especially true in technical work involving data science, graph ML, spatio-temporal modeling, package dependencies, tensor shapes, runtime environments, and debugging.<\/p>\n\n\n\n

Progress comes from disciplined iteration, not tool-hopping.<\/p>\n\n\n\n

Note: Enhanced \/ compiled with help of AI \/ LLMs<\/p>\n\n\n\n