Question on content accuracy of the training

Question

I think there is a possible content accuracy issue in the contents in this link ("Understand LLMs" module),

https://learn.microsoft.com/en-us/training/modules/introduction-large-language-models/2-understand-large-language-models?WT.mc_id=academic-0000-alfredodeza

It seems to me that in the last row of the table(under the section"How does an LLM differ from more traditional natural language processing (NLP)?"), the explanations for traditional NLP and LLMs are mixed up. Isn't traditional NLP systems more optimized for specific use cases because they are trained on a domain/task - specific dataset?

Answer 1

Hi Lee,

Thank you for reaching out to us on the Microsoft Q&A forum.

Based on your query LLMs are trained on massive amounts of text and optimized for specific scenarios as mentioned below.

The LLMs objective is to learn the statistical patterns of language and predict the most likely word given the previous words. Therefore, they are most suited for the following scenarios that require generating coherent and fluent text:

• Writing stories
• Writing essays
• Writing captions
• Writing headlines
• Generating natural language from structured data
• Writing code from natural language specifications
• Summarizing long documents

However, they may not perform well on tasks that require more logical reasoning, factual knowledge, or domain-specific expertise. For the latter, sufficient relevant information needs to be augmented to the prompt to ground the model.

LLMs: More versatile and handle diverse tasks but require vast amounts of data and may need additional tuning for specific use cases.

Additionally, please refer the below document for your reference:

https://learn.microsoft.com/en-us/ai/playbook/technology-guidance/generative-ai/working-with-llms/use-case-recommend?source=recommendations#these-models-can-be-best-used-for-generative-applications

If the information is helpful, please accept the answer by clicking the "Upvote" and "Accept Answer" on the post. If you are still facing any issue, please let us know in the comments. We are glad to help you.

Thank you.