Lecture Notes: Testing DeepSeq R1 Model for a Mapping Task

Introduction to DeepSeq R1

• DeepSeq R1 is a large language model similar to OpenAI's ChatGPT.
• Known for its advanced capabilities in language processing.
• The task: Plot the 20 most visited cities in the world on an interactive map.

Setting Up DeepSeq R1

• Access DeepSeq at chat.deepseq.com.
• Task specifications:
  • Plot interactive map.
  • Display city name, rank, and number of visitors when a point is clicked.
  • Use Python in Google Colab environment.

Features of DeepSeq R1

• Web Browsing Option: Allows the model to refer to the web for up-to-date information.
  • The feature is free in the current version.
• Reasoning Model: Helps in making decisions by resolving conflicting information found on the web.

Task Execution

• Extracts the 20 most visited cities with visitor data and ranks.
• Automates gathering of latitude and longitude data.
• Data organized in a pandas DataFrame using Python.
• Uses the Folium library to plot data on an interactive map.

Using Google Colab

• Google Colab is a simple online coding environment accessible through Google Drive.
• Steps to set up:
  • Log into Google Drive.
  • Navigate to working folder, right-click, select Google Colab.

Implementation Workflow

1. Install Libraries: Necessary libraries are installed automatically.
2. Code Execution:
   • Import libraries and create DataFrame.
   • Plot the map with city markers.
   • Add functionality to display city details on click.
3. Color Coding:
   • Red for top 5 cities, Blue for next 5, Green for following 5, Yellow for last 5.

Results and Observations

• Interactive map displays cities with visitor counts and ranks.
• Automatically includes the corresponding year of visitor data.
• Provides a quick visual representation of most visited cities using color-coded markers.
• Example: Bangkok is the top city in 2024 with 32.4 million visitors.

Conclusion

• Demonstrates the ease of using large language models for data processing tasks.
• Encourages experimentation with both DeepSeq and ChatGPT for comparison.
• Highlights future potential for complex task handling by comparing model outputs.

Final Thoughts

• Large language models simplify data tasks significantly.
• The importance of user-friendliness and performance evaluation in choosing models.
• Recommendation to explore and test different models for specific needs.