/Langflow-Chatbot

Combine MULTIPLE LLMs to build an AI API! (super simple!!!) Langflow | LangChain | Groq | OpenAI

Langflow-Chatbot

Combine MULTIPLE LLMs to build an AI API! (super simple!!!) Langflow | LangChain | Groq | OpenAI

Tutorial from Ania Kubow https://www.youtube.com/watch?v=5lFChDglhmI&t=135s

In VSCode:

  • Create >Python3.10 Envirenment
  • python -m pip install langflow -U
  • python -m langflow run

Create Langflow -> Field##Groups: -> Connections <>: search for -: ##DB Ingestion (Create and store vector embeddings from files)

  • file (Database): -> File -> Upload .csv

  • split: -> Split Text -> File <> DataInputs(*Split text) -> Split Text Chunks <> IngestData(*AstraDB)

  • llm(openai) embedd-> LLM Embedd. -> Embeddings <> Embedding or Astra Vectorize

  • db(astra): -> AstraDB *(milvus) -> *(Milvus = OpenSource/forFREE) ##Main-Chain (Main Proccess from Input to Output)

  • chat input: -> Chat Input -> Message <> SearchInput(*2xAstraDB)

  • 2x(llm) embedding:-> 2xLLM Embedd.-> Embeddings <> 2x Embedding or Astra Vectorize

  • parse data: -> Parse Data -> Search Results(*2xAstraDB) <> Data(*Parse Data)

  • prompt: -> Prompt -> Edit:

    Below is the context: {context}

---

Given the context above answer the users question.
Question: {question}
Answer:

  • openai: -> OpenAI -> PromptMessage(*prompt) <> Input -> Message(*Chat Input) <> Question(*Prompt) ##URL Scraping (Search whole websites for results)

  • url: -> URL

  • 2x parse data: -> 2x Parse Data-> Data(*URL) <> Data(*2x Parse Data) -> Text(*2x Parse Data) <> context(*Prompt) -> Edit Prompt:

    Below is the website data about the course: {website}
---

#Playground Output (Chat the process in the Playground of Langflow)

  • chat output: -> Chat Output -> Text(*OpenAI) <> Text(*Chat Output) #Chat Memory (Store the chat history)

  • memory: -> Memory -> Edit Prompt:

    ---
{memory}

-> Messages <> memory(*Prompt) #Groq (Groq API instead LLM)