Spaces:
Runtime error
Runtime error
File size: 1,879 Bytes
ec0786f bddb702 a7c1ef2 bddb702 a7c1ef2 aa5f7bd a7c1ef2 aa5f7bd bddb702 a7c1ef2 bddb702 a7c1ef2 bddb702 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 |
from dotenv import load_dotenv
# Load environment variables from .env file
load_dotenv()
from langchain.embeddings import OpenAIEmbeddings # for creating embeddings
from langchain.vectorstores import Chroma # for the vectorization part
from langchain.chains import ConversationalRetrievalChain
from langchain.llms import OpenAI # the LLM model we'll use (CHatGPT)
import gradio as gr
max_sources = 4
DB_DIR = "chroma"
embedding = OpenAIEmbeddings()
vectordb = Chroma(persist_directory=DB_DIR, embedding_function=embedding)
pdf_qa = ConversationalRetrievalChain.from_llm(
OpenAI(temperature=0.9, model_name="gpt-3.5-turbo"),
vectordb.as_retriever(),
return_source_documents=True,
)
def chat_pdf(query, chat_history=""):
result = pdf_qa({"question": query, "chat_history": chat_history})
answer = result["answer"]
source_docs = result["source_documents"]
print("source_docs", len(source_docs))
cleaned_docs = []
for doc in source_docs:
cleaned_content = doc.page_content
metadata_info = f"Metadata: {doc.metadata}\n"
cleaned_docs.append(metadata_info + cleaned_content)
# Pad the outputs to match the number of output components in the Gradio interface
padded_outputs = [answer] + cleaned_docs + [""] * (max_sources - len(cleaned_docs))
return padded_outputs
def create_outputs(num_sources):
outputs = [gr.outputs.Textbox(label="Answer")]
for i in range(1, num_sources + 1):
outputs.append(gr.outputs.Textbox(label=f"Source Document {i}"))
return outputs
iface = gr.Interface(
fn=chat_pdf,
inputs=[gr.inputs.Textbox(label="Query")],
outputs=create_outputs(max_sources),
examples=[
["Give 2 species of fulgoroidea"],
["What colors are found among fulgoroidea?"],
["Why are fulgoroidea so cute?"],
],
)
iface.launch(debug=True)
|