Haystack with OpenAIChatGenerator problem

[hamzachebbigit]

Hello guys I need a little help :d
So I have just started learning Haystack and all its functionality and I really like it but with RAG I seem to have an issue: So I created my vector database with Haystack+Pinenone works like a charm and I dont have an issue but when I apply the retrieval code I get an error revolving arround the " query_pipeline.connect("prompt_builder", "llm") " it says that the llm and prompt_builder does not match in input and ouput variables.... I have tried everything I even tried to switch to the DynamicChatPromptBuilder but the issue still there any help would be appreciated :D here is the full code :

from haystack.utils import Secret
from haystack.components.embedders import SentenceTransformersTextEmbedder
from haystack.components.builders import PromptBuilder
from haystack_integrations.components.retrievers.pinecone import PineconeEmbeddingRetriever
from haystack.components.generators.chat import OpenAIChatGenerator
from haystack import Pipeline
from utility import pinecone_config
from dotenv import load_dotenv

prompt_template = """Answer the following query based on the provided context. If the context does
                     not include an answer, reply with 'I don't know'.\n
                     Query: {{query}}
                     Documents:
                     {% for doc in documents %}
                        {{ doc.content }}
                     {% endfor %}
                     Answer: 
                  """
 
 
 
def get_result(query):                  
    query_pipeline = Pipeline()

    query_pipeline.add_component("text_embedder", SentenceTransformersTextEmbedder())
    query_pipeline.add_component("retriever", PineconeEmbeddingRetriever(document_store=pinecone_config()))
    query_pipeline.add_component("prompt_builder", PromptBuilder(template=prompt_template))
    query_pipeline.add_component("llm", OpenAIChatGenerator(api_key=Secret.from_token("API_HERE"), model="gpt-3.5-turbo"))

    query_pipeline.connect("text_embedder.embedding", "retriever.query_embedding")
    query_pipeline.connect("retriever.documents", "prompt_builder.documents")
    query_pipeline.connect("prompt_builder", "llm")

    query = query

    results = query_pipeline.run(
        {
            "text_embedder": {"text": query},
            "prompt_builder": {"query": query},
        }
    )

    return results['llm']['replies'][0]

if __name__ == '__main__':
    #loading the environment variable
    '''load_dotenv()
    PINECONE_API_KEY = os.getenv("PINECONE_API_KEY")
    os.environ['PINECONE_API_KEY'] = PINECONE_API_KEY
    
    print("Import Successfully")'''
    
    result=get_result("what is rag?")
    print(result)

[rhysdg]

Hey there @hamzachebbigit! I'm sorry that no one has answered your question here. I ran into this same issue and it's just a matter of switching out OpenAIChatGenerator for OpenAIGenerator. Hope that helps!