ChatGPT Weather Information API

This project is an AI-powered application that provides real-time weather information from various locations around the world. It exposes an HTTP REST endpoint to answer user queries about the current weather conditions in a specific location or from various input file formats (CSV, JSONlines, PDF, Markdown, Txt). The application leverages Pathway’s LLM App features and utilizes the OpenAI API Embeddings and Chat Completion endpoints to generate AI assistant responses.

Features

• Retrieves the latest weather information from various sources.
• Provides an API interface to explore current weather conditions.
• User-friendly UI with Streamlit.
• Filters and presents weather information based on user queries or chosen data sources.
• Supports data and code reusability for offline evaluation.
• Offers the option to use local (cached) or real-time data.
• Extend data sources using Pathway's built-in connectors for JSONLines, CSV, Kafka, Redpanda, Debezium, streaming APIs, and more.

Upcoming Features

• Incorporate additional data from external APIs, various file formats (Jsonlines, PDF, Doc, HTML, or Text), and databases like PostgreSQL or MySQL.
• Merge data from different sources instantly.
• Maintain a data snapshot to observe variations in weather conditions over time.
• Integrate with BI and analytics tools for downstream processing.
• Set up alerts upon detecting weather shifts.

Code Sample

import pathway as pw
from common.embedder import embeddings, index_embeddings
from common.prompt import prompt

def run(host, port):
    query, response_writer = pw.io.http.rest_connector(
        host=host,
        port=port,
        schema=QueryInputSchema,
        autocommit_duration_ms=50,
    )

    sales_data = pw.io.jsonlines.read(
        "./examples/data",
        schema=DataInputSchema,
        mode="streaming"
    )

    embedded_data = embeddings(context=sales_data, data_to_embed=sales_data.doc)
    index = index_embeddings(embedded_data)

    embedded_query = embeddings(context=query, data_to_embed=pw.this.query)

    responses = prompt(index, embedded_query, pw.this.query)
    response_writer(responses)

    pw.run()

class DataInputSchema(pw.Schema):
    doc: str

class QueryInputSchema(pw.Schema):
    query: str

Use Case

OpenAI GPT excels at answering questions but may struggle with unfamiliar topics. LLM App efficiently provides context for searches or answers, especially for real-time weather conditions, specific details about weather in a given location, and current atmospheric conditions.

How the Project Works

1. Prepare Search Data:

   • Collect data from chosen sources or upload a CSV file via the UI file-uploader.
   • Map each row into a JSONline schema for better management of large datasets.
   • Chunk documents into short, self-contained sections for embedding.

2. Embedding and Indexing:

   • Embed each section using the OpenAI API and retrieve the results.
   • Construct an index on the generated embeddings.

3. Search (Once per Query):

   • Generate an embedding for the user's question using the OpenAI API.
   • Retrieve the vector index by relevance to the query.

4. Ask (Once per Query):

   • Insert the question and the most relevant sections into a message to GPT.
   • Return GPT's answer.

How to Run the Project

Run with Docker:

• Set environment variables.
• From the project root folder, run docker compose up in the terminal.
• Navigate to localhost:8501 in your browser.

Prerequisites:

• Ensure Python 3.10 or above is installed.
• Download and install Pip to manage project packages.
• Create an OpenAI account and generate a new API Key.

For more detailed instructions, refer to the documentation.

Feel free to explore, contribute, and enhance this weather information API!

ScreenRecorderProject1.mp4