i have postgresql database with over 100 000 verbatims (with tonality link to each verbatim) in it.
I want to do a resume of all negative verbatim on this database.
Right now my main problem is it looks like llamaindex send only 2 data (over my 100 000 data) to openai. So openai to a resume of my 2 verbatims only … I think i miss understand something but i can’t figure what.

Here my code :

from llama_index.llms.azure_openai import AzureOpenAI
from llama_index.embeddings.azure_openai import AzureOpenAIEmbedding
from llama_index.core import Settings
from llama_index.core import VectorStoreIndex, SimpleDirectoryReader

llm = AzureOpenAI(
    model="gpt-4",
    deployment_name="gpt-4",
    api_key=api_key,
    azure_endpoint=azure_endpoint,
    api_version=api_version,
)



embed_model = AzureOpenAIEmbedding(
    model="text-embedding-ada-002",
    deployment_name="text-embedding-ada-002",
    api_key=api_key,
    azure_endpoint=azure_endpoint,
    api_version=api_version,
)
Settings.llm = llm
Settings.embed_model = embed_model
vector_store = PGVectorStore.from_params(
    database="DATABASE",
    host="HOST",
    password="PASSWORD",
    port=5432,
    user="USERNAME",
    table_name="cars",
    embed_dim=1536,
    debug=True
)
index = VectorStoreIndex.from_vector_store(vector_store=vector_store)
query_engine = index.as_query_engine()

response = query_engine.query("Resume negative verbatim of my client")

I think llamaindex is “sorting” my data before sending it, thank to vector. It’s ok, but i would like to send to openAi more “context” like 100 verbatims or 1000 verbatims to feed openAi with more details. Maybe i’m not using the good component to do what i want to achieve.
Any idea ?

Thank for your help !