Distrito Telefónica. Innovation & Talent Hub

CodeTalk4Devs: LangChain for building LLM-based applications

Artificial Intelligence Trends

In the new installment of #CodeTalks4Devs, the fascinating universe of Large Language Models (LLM) is explored with the help of Héctor Cordobés, an Artificial Intelligence (AI) expert from Telefónica. A breakdown is provided on how LLMs are revolutionizing AI and the development of applications based on this technology. The role of the LangChain library and its potential use is particularly highlighted.

How have Large Language Models changed the game?

The emergence of LLMs didn't suddenly happen at the end of 2022. There are years of work and investment behind this technology. However, the year 2023 marked a turning point. On the one hand, there was a significant improvement in these technologies, driven by the success of OpenAI models. On the other hand, there was a massive adoption of these technologies by the population in their daily activities. 
The utility of LLMs lies in their ability to understand the way people communicate, that is, natural language. They are also capable of generating high-quality natural language. Another significant advance of LLMs is their ability to solve different tasks, such as sentiment analysis, classification, etc., when before a specific model was required for each task. Their understanding and generation of natural language have made them accessible, facilitating their use without an entry barrier.

What importance do these types of tools have?

The possibilities that LLMs offer are as broad as the imagination allows to develop them. Hence, projects like LangChain work on facilitating developers' integration of LLMs into their applications. LangChain provides a large number of integrations with LLMs from various providers, even some Open Source, as well as with other tools like Gmail, Github, etc. 
As a result, it is possible to create applications where LLMs play a significant role in the interface with the user, facilitating the use of certain tools. Moreover, LangChain opens the door to the so-called "LLM-based Agents" whose "brain" is operated by an LLM. With a specific instruction, these agents can reason and make decisions. 
In summary, Héctor Cordobés introduces in this episode of #CodeTalks4Devs a promising future where Large Language Models and tools like LangChain are protagonists of technological evolution. We invite you not to miss this fascinating journey through the revolution of the LLMs and how LangChain is helping developers to leverage their potential.


Síguenos en el canal de OICampus

Explore our next research

Artificial Intelligence: Applicability of GANs and Autoencoders in Cybersecurity.

Some technologies such as Machine Learning and Deep Learning have become the perfect tool to generate phishing attacks or Fake News, but they can also be used to protect users.

Padlock picture on laptop screen