DeltaV Live Demo at The AI Summit New York | Fetch.ai

FetchAI
7 Dec 202305:07

TLDRThis video demonstrates how to integrate Hugging Face's AI model, FinBERT, into an agent on AgentVerse for real-time financial analysis. By leveraging the model's inference capabilities, users can assess a company's financial health, as exemplified by analyzing Go EV's cash reserves. The process involves setting up the model on Hugging Face, obtaining the API, and integrating it with the agent. The video showcases a live test on Delta V, a prompt engineering interface, where users can combine multiple agents to perform complex tasks, such as sentiment analysis and stock price retrieval, accessible through various messaging systems.

Takeaways

  • 🤖 The script discusses integrating Hugging Face to build and deploy a model for inference in an agent on AgentVerse.
  • 🔍 The company 'go EV' is mentioned as an example, which is facing low cash reserves, and the stock performance is poor.
  • 📈 The 'finbert' model hosted on Hugging Face is used to analyze financial data and provide insights.
  • 🔗 The process involves obtaining an API from Hugging Face and integrating it with the agent on AgentVerse.
  • 📊 The agent's code is simple and involves adding API keys and addresses for functionality.
  • 💡 AgentVerse allows the creation of agents that can be discovered and utilized for various tasks.
  • 🔍 Delta V is introduced as a prompt engineering interface for testing and combining agents' capabilities.
  • 🤔 The script demonstrates how to combine two agents to perform a task, such as sentiment analysis and stock price retrieval.
  • 📝 The prompt engineering interface provides options to choose from and shows the outcome of the task execution.
  • 📉 The example task executed results in a negative sentiment analysis for the company's data.
  • 📲 The financial system created can be accessed from various messaging platforms like WhatsApp, providing real-time financial information.

Q & A

  • What is the main purpose of the script?

    -The script demonstrates how to integrate a machine learning model using Hugging Face, generate inference from it, and feed the results into an agent on AgentVerse.

  • What is the name of the company mentioned in the script that is low on cash reserves?

    -The company mentioned is called go EV.

  • How did the script determine that go EV is low on cash reserves?

    -The determination was based on the company's stock performance and a recent financial statement.

  • Which model is being used in the script for sentiment analysis?

    -The model used for sentiment analysis is finBERT, hosted on Hugging Face.

  • What is the role of the agent in AgentVerse?

    -The agent in AgentVerse is used to execute tasks by utilizing the model's inference and API data, such as stock prices.

  • How can the agent be tested?

    -The agent can be tested on Delta V, a prompt engineering interface, by inputting specific prompts and observing the outcomes.

  • What is the significance of Delta V in this context?

    -Delta V serves as the interface for prompt engineering, allowing users to test and interact with agents without needing to write code.

  • How does the script show the integration of the finBERT model with the stock price API?

    -The script demonstrates the integration by showing how the agent can use the finBERT model for sentiment analysis and another agent to fetch the stock price, combining the results in Delta V.

  • What is the outcome of the sentiment analysis for the data mentioned in the script?

    -The outcome of the sentiment analysis is negative, as indicated by the negative score.

  • How can the financial system created in the script be accessed?

    -The financial system can be accessed from anywhere, including messaging systems like WhatsApp, due to the integration with AgentVerse and Delta V.

  • What is the benefit of using a prompt engineering interface like Delta V?

    -Delta V allows for easy testing and interaction with agents without the need for coding, making it accessible for users to develop and refine their prompts.

Outlines

00:00

🤖 Integrating Hugging Face Model with AgentVerse

The video script describes the process of integrating a machine learning model, specifically the FinBERT model, hosted on Hugging Face, with an agent on AgentVerse. The model is used to generate inferences which are then fed into the agent. The script outlines the steps to test the integration on Delta V, a prompt engineering interface. It explains how to use the model to analyze financial data of a company called 'go EV', which is low on cash reserves. The agent is created with a simple code, incorporating the model's API key and address. The script then demonstrates how the agent can be discovered and combined with another agent to perform a task, such as providing stock price information. The result of the task, which includes sentiment analysis and stock price, is shown live on Delta V. The script concludes by highlighting the flexibility of accessing the financial system from various messaging platforms.

05:01

📝 Summary of the Video Script

This paragraph serves as a brief acknowledgment or transition point in the video script, possibly indicating the end of the main content and the beginning of a new section or conclusion.

Mindmap

Keywords

💡Hugging Face

Hugging Face is an open-source platform that provides a wide range of pre-trained natural language processing (NLP) models and tools for developers and researchers. In the video, it is used to host a model that can generate inferences, which are then utilized by an agent in AgentVerse. The platform is highlighted for its ease of integration and the ability to provide real-time updates, as seen when the finBERT model is used to analyze sentiment.

💡Integration

Integration refers to the process of combining two or more systems or components to work together as a unified whole. In the context of the video, it involves connecting a machine learning model from Hugging Face with an agent in AgentVerse to perform a specific task. The integration is crucial for creating a seamless workflow that allows for the generation of inferences and their application in a different environment, such as the prompt engineering interface in Delta V.

💡Inference

In the field of machine learning and artificial intelligence, an inference is a conclusion or prediction made by a model based on its training data. The video discusses generating inferences from a model hosted on Hugging Face, which are then used by an agent to perform tasks. For example, the finBERT model is used to provide sentiment analysis, and these inferences are fed into the agent for further processing.

💡AgentVerse

AgentVerse appears to be a platform or environment where agents can be created and deployed to perform various tasks. In the video, an agent is created within AgentVerse, and it is designed to interact with the Hugging Face model to generate inferences. The agent's role is to facilitate the application of the model's output in a practical context, such as financial analysis or other tasks demonstrated in the video.

💡Delta V

Delta V is described as a prompt engineering interface in the video. It seems to be a user interface that allows for the input of prompts and the execution of tasks by agents. The video demonstrates how Delta V can be used to combine the outputs of different agents, such as one for sentiment analysis and another for stock price retrieval, to perform a more complex task, like financial analysis.

💡finBERT

finBERT is a machine learning model mentioned in the video, which is hosted on Hugging Face. It is likely a variant of the BERT model, trained specifically for financial text analysis. The model is used to analyze the sentiment of financial data, providing a negative, neutral, or positive score. This model's output is then integrated into the AgentVerse agent to perform real-time sentiment analysis on financial information.

💡Prompt Engineering

Prompt engineering refers to the process of crafting input prompts that guide AI models to perform specific tasks or generate desired outputs. In the video, Delta V serves as a prompt engineering interface, allowing users to input prompts and see the results of the agents' actions. This interface is used to demonstrate how the integration of different agents can lead to more complex and useful outcomes.

💡Stock Price

The stock price is the current market value of a company's shares. In the video, one of the agents is responsible for retrieving stock prices, which is a simple yet practical application of the system. The stock price information is combined with sentiment analysis to provide a comprehensive view of a company's financial health, as demonstrated with the example of the company 'go EV'.

💡API

API stands for Application Programming Interface, which is a set of rules and protocols that allow different software applications to communicate with each other. In the video, the API is used to connect the Hugging Face model with the agent in AgentVerse, enabling the agent to receive inferences from the model. The API is a crucial component in the integration process, allowing for the seamless transfer of data between systems.

💡Sentiment Analysis

Sentiment analysis is a type of opinion mining that uses NLP to determine the emotional tone behind a series of words, phrases, or text. In the video, the finBERT model performs sentiment analysis on financial data, providing insights into the overall sentiment being negative, neutral, or positive. This analysis is part of the integrated system that helps users understand the financial health of a company.

💡Chain of Thought

The term 'chain of thought' in the video refers to the sequence of logical steps or reasoning that an AI agent goes through to arrive at a conclusion or perform a task. It is a way to visualize and understand the decision-making process of the AI. In the context of the video, the chain of thought is used to demonstrate how the agents are working together to process and combine information, ultimately providing the user with a comprehensive analysis.

Highlights

Integration of Hugging Face for model building and inference generation

Using the model in an agent on AgentVerse

Company 'go EV' is low on cash reserves, indicated by poor stock performance and recent financial statements

Utilizing the FinBERT model hosted on Hugging Face for financial analysis

Creating an agent in AgentVerse with a simple code and API integration

Agent is active and discoverable within AgentVerse

Delta V as a prompt engineering interface for testing and combining agents

Combining two agents to perform a task, such as financial analysis and stock price retrieval

Real-time sentiment analysis using the FinBERT model

Live result display of the task execution within the development interface

Chain of Thought visualization to understand agent actions

Confirmation of the task outcome and sentiment scores

Running the same analysis in Hugging pH to validate results

Separate agent for fetching current share prices

Combining machine learning models with APIs for practical applications

Access to financial systems via messaging platforms like WhatsApp

Delta V's utility for prompt engineering and its potential for broader applications