small language models

Travis chat

Filters

Small Language Models (SLMs) are designed to operate with a limited number of parameters, typically ranging from millions to a few billion. This makes them suitable for deployment on devices with lower specifications, such as personal computers and mobile devices. Unlike Large Language Models (LLMs), which require significant memory, processing power, and energy, SLMs offer a more efficient alternative, focusing on agility and cost-effectiveness while maintaining performance ¹³.

The development of SLMs has gained momentum due to the increasing demand for models that can perform specific tasks without the extensive resource requirements of their larger counterparts. Examples of SLMs include Orca 2, DistilBERT, Gemini Nano, and TinyBERT. The latest addition to this category is Phi-2, which has shown impressive performance despite its smaller size compared to LLMs like Llama-2 ³⁴.

In summary, SLMs represent a significant shift in the AI landscape, emphasizing efficiency and adaptability, making them valuable in various applications across different fields ³⁴.

Towards AI

If you are not a Medium member, you can read this article here . Large language models have become very popular recently due to the amazing capabilities shown by these models. Their applicability to a...

Language Models

Towards Data Science

What is a Language Model and how to train one? Machine Learning approaches used to train, outputs and utility of Language Models

It is raining Language Models! All about the new Small Language Models — Phi-2

Towards AI

It is raining Language Models! All about the new Small Language Model— Phi-2 The Dawn of Small Language Models: Introducing Phi-2 that outperformed Llama-2(70B), which is 25 times its size! Image by ...

Small But Mighty — The Rise of Small Language Models

Towards Data Science

Our world has been strongly impacted by the launch of Large Language Models (LLMs). They exploded onto the scene, with GPT-3.5 amassing a million users in a single app in just five days — a testament ...

Language Modeling

Towards Data Science

As you might have guessed by now, language modeling is a use-case employed by us daily, and still, its a complicated concept to grasp. REALM (Retrieval-Augmented Language Model Pre-Training) is the…

Language Models

Dive intro Deep Learning Book

In Section 9.2 , we see how to map text sequences into tokens, where these tokens can be viewed as a sequence of discrete observations, such as words or characters. Assume that the tokens in a text se...

N-Gram Language Models

Towards Data Science

N-Gram Language Models, Laplace Smoothing, MLE, Perplexity, Katz backoff

Not-So-Large Language Models: Good Data Overthrows the Goliath

Towards Data Science

(Image generated by DALL·E) How to make a million-sized language model that tops a billion-size one In this article, we will see how Language Models (LM) can focus on better data and training strategi...

Large Language Models: A Short Introduction

Towards Data Science

And why you should care about LLMs Image by author. There’s an acronym you’ve probably heard non-stop for the past few years: LLM, which stands for Large Language Model. In this article we’re going t...

Some Technical Notes About Phi-3: Microsoft’s Marquee Small Language Model

Towards AI

The model ius able to outperform much larger alternatives and now run locally on mobile devices. Created Using Ideogram I recently started an AI-focused educational newsletter, that already has over ...

The beginners guide to language models

Towards Data Science

The science of extracting information from textual data has changed dramatically over the past decade. As the term Natural Language Processing took over Text Mining as the name of this field, the…

Large Language Models Just Got A Whole Lot Smaller

Towards Data Science

TLDR: Large Language Models (LLMs for short) are currently huge, costly to run, and have a significant carbon footprint . Recent advancements in model compression and system-level optimization methods...