Downloads three posts of Vitaliks blog and creates embeddings with langchain4j which can then be queried with OpenAI GPT (see https://hackmd.io/@alculexum/embedding4j)

TalkToEmbeddedButerin.java

//DEPS dev.langchain4j:langchain4j:0.23.0
//DEPS dev.langchain4j:langchain4j-open-ai:0.23.0
//DEPS dev.langchain4j:langchain4j-embeddings-all-minilm-l6-v2:0.23.0
//DEPS commons-io:commons-io:2.14.0
//DEPS org.apache.commons:commons-text:1.10.0
//DEPS org.jsoup:jsoup:1.16.1
//DEPS org.slf4j:slf4j-simple:2.0.9

package dev.indus340;

import dev.langchain4j.chain.ConversationalRetrievalChain;
import dev.langchain4j.data.document.Document;
import dev.langchain4j.data.document.DocumentType;
import dev.langchain4j.data.document.parser.TextDocumentParser;
import dev.langchain4j.data.document.splitter.DocumentSplitters;
import dev.langchain4j.data.segment.TextSegment;
import dev.langchain4j.memory.chat.MessageWindowChatMemory;
import dev.langchain4j.model.embedding.AllMiniLmL6V2EmbeddingModel;
import dev.langchain4j.model.embedding.EmbeddingModel;
import dev.langchain4j.model.openai.OpenAiChatModel;
import dev.langchain4j.retriever.EmbeddingStoreRetriever;
import dev.langchain4j.store.embedding.EmbeddingStore;
import dev.langchain4j.store.embedding.EmbeddingStoreIngestor;
import dev.langchain4j.store.embedding.inmemory.InMemoryEmbeddingStore;
import org.apache.commons.io.IOUtils;
import org.apache.commons.text.WordUtils;
import org.jsoup.Jsoup;
import org.jsoup.nodes.Element;

import java.io.IOException;
import java.time.Duration;
import java.util.Scanner;

public class TalkToEmbeddedButerin {
  
    private static final String[] blogPosts = new String[] {
      
        // Should Ethereum be okay with enshrining more things in the protocol?
        "https://vitalik.ca/general/2023/09/30/enshrinement.html",
        
        // What do I think about Community Notes?
        "https://vitalik.ca/general/2023/08/16/communitynotes.html",
        
        // What do I think about biometric proof of personhood?
        "https://vitalik.ca/general/2023/07/24/biometric.html"
    };

    // args[0] should be OPENAI_API_KEY
    public static void main(String[] args) throws IOException {
        if (args.length != 1) {
            System.out.println("OPENAI_API_KEY has to be provided as first argument.");
            return;
        }
        String filesContent = downloadPostsContent();
        interactWithEmbeddedDocuments(args[0], filesContent);
    }

    private static void interactWithEmbeddedDocuments(String openaiKey, String content) {
        EmbeddingModel embeddingModel = new AllMiniLmL6V2EmbeddingModel();
        EmbeddingStore<TextSegment> embeddingStore = new InMemoryEmbeddingStore<>();
        EmbeddingStoreIngestor ingestor = EmbeddingStoreIngestor.builder()
                // the Devoxx sample used 500, for my texts 1000 seems better.
                .documentSplitter(DocumentSplitters.recursive(1000, 0))
                .embeddingModel(embeddingModel)
                .embeddingStore(embeddingStore)
                .build();

        System.out.println("creating embeddings");
        Document document = new TextDocumentParser(DocumentType.TXT).parse(IOUtils.toInputStream(content));
        ingestor.ingest(document);
        OpenAiChatModel assistant =
                OpenAiChatModel.builder()
                        .modelName("gpt-3.5-turbo")
                        .apiKey(openaiKey)
                        .timeout(Duration.ofMinutes(5)).build();

        ConversationalRetrievalChain chain = ConversationalRetrievalChain.builder()
                .chatLanguageModel(assistant)
                .retriever(EmbeddingStoreRetriever.from(embeddingStore, embeddingModel))
                // be careful when uncommenting the chat memory window.
                // to verify the difference between embeddings and plain GPTs
                // the chat memory has to be deactivated as plain GPT does not 
                // use the chain but the model directly which has no chat memory.
                //.chatMemory(MessageWindowChatMemory.withMaxMessages(20))
                .build();

        String question = "";
        Scanner scanner = new Scanner(System.in);
        while (!question.equalsIgnoreCase("q")) {
            System.out.print("\nWhat's your question: ");
            question = scanner.nextLine();
            System.out.println("\n### With embeddings\n" + WordUtils.wrap(chain.execute(question), 80));
            System.out.println("\n### Plain GPT\n" + WordUtils.wrap(assistant.generate(question), 80));
        }
        scanner.close();
    }

    private static String downloadPostsContent() throws IOException {
        StringBuilder allPostsCombined = new StringBuilder();
        for (String url : blogPosts) {
            System.out.println("downloading " + url);
            org.jsoup.nodes.Document document = Jsoup.connect(url).get();
            Element postBody = document.body();
            allPostsCombined.append(postBody.text());
        }
        return allPostsCombined.toString();
    }

}

Can be run with JBang:

jbang https://gist.github.com/ice09/ae5409de706fbd5e08183f7632b44434 <OPENAI_API_KEY>