Chain‑of‑Thought prompting “reduces the chances of oversimplified or incorrect answers”.
I invite the community to try Chain‑of‑Thought prompting for complex queries like deforestation’s impact on climate change. prompting retrieval and tool use
I am experimenting with prompt optimisation for large language models to achieve the most accurate and contextually relevant answers.
I argued that large language models can significantly enhance their abilities by tapping into external APIs. I listed five representative tools—Calculator, Q&A System, Search Engines, Translation System and Calendar—to address weaknesses such as arithmetic, stale knowledge, translation and scheduling.
LLMs can integrate tools such as Calculator, Q&A System, Search Engines, Translation System, and Calendar.
I observed that LLMs are limited by a knowledge freeze at their last training date, so they can miss recent events.
Retrieval Augmentation gives LLMs real‑time relevance, reduces hallucinations, and enables accurate answers about recent global events.