From the course: Microsoft Azure AI Essentials: Workloads and Machine Learning on Azure

Unlock this course with a free trial

Join today to access over 24,700 courses taught by industry experts.

Using language models: LLM vs. SLM

Using language models: LLM vs. SLM

From the course: Microsoft Azure AI Essentials: Workloads and Machine Learning on Azure

Start my 1-month free trial Buy for my team

Using language models: LLM vs. SLM

“

- [Instructor] Language models can be categorized into two types, large language models, LLMs, and small language models, SLMs. Choosing between them depends on your business needs. Here's what you should consider for the dataset. LLMs are trained on vast amounts of general text sourced from the internet and other public data. SLMs focus on smaller created data sets and are often subject-specific. When it comes to parameters, LLM have billions or even trillions of parameters, which help predict language sequences. SLMs with fewer parameters are often more efficient for specific tasks. When it comes to tasks, LLMs excel in general language generation across various contexts. SLMs, with their focus vocabulary, perform better on specialized tasks, but struggle with broader topics. When talking about deployment, LLMs are typically cloud-based due to their size and complexity. SLMs, however, being smaller, offer more flexibility, including local deployment on devices. Finally, customizing…

Contents