app.py sem reader do haystack

app.py

@@ -5,45 +5,26 @@ from haystack.utils import fetch_archive_from_http
 import os
 from haystack.pipelines.standard_pipelines import TextIndexingPipeline
 from haystack.nodes import BM25Retriever
-from haystack.nodes import FARMReader
-from haystack.pipelines import ExtractiveQAPipeline
-from haystack.utils import print_answers
-from pydantic import BaseModel
-from typing import Any
-import pandas as pd
-
-
-
-
-#Criando o objeto documentStore
+from transformers import pipeline
 
+# Criando o objeto documentStore
 document_store = InMemoryDocumentStore(use_bm25=True)
 
-#Exportando os dados necessários sobre o Game Of Thrones
-
+# Exportando os dados necessários sobre o Game Of Thrones
 doc_dir = "data/build_your_first_question_answering_system"
-
 fetch_archive_from_http(
     url="https://s3.eu-central-1.amazonaws.com/deepset.ai-farm-qa/datasets/documents/wiki_gameofthrones_txt1.zip",
     output_dir=doc_dir,
 )
 
-
 files_to_index = [doc_dir + "/" + f for f in os.listdir(doc_dir)]
 indexing_pipeline = TextIndexingPipeline(document_store)
 indexing_pipeline.run_batch(file_paths=files_to_index)
 
 retriever = BM25Retriever(document_store=document_store)
 
-reader = FARMReader(model_name_or_path="deepset/roberta-base-squad2", use_gpu=True)
-
-
-pipe = ExtractiveQAPipeline(reader, retriever)
-
-
-
-
-
+# Utilizando um pipeline da biblioteca Transformers
+pipe = pipeline("question-answering", model="deepset/roberta-base-squad2", tokenizer="deepset/roberta-base-squad2")
 
 image = Image.open('comida.jpg')
 st.image(image)
@@ -52,19 +33,25 @@ e respotas a perguntas de vários domínios sobre esportes para perguntas aberta
 
 st.subheader('QASports',divider='rainbow')
 
-
-
-
-user_input = st.text_input('Digite seu texto aqui:')
-res = pipe(user_input)
-prediction = pipe.run(
-    res, params={"Retriever": {"top_k": 10}, "Reader": {"top_k": 5}}
-)
-
-
+user_input = None
+if not user_input:
+    user_input = st.text_input("Por favor, digite uma pergunta.")
+
+if user_input:
+    res = retriever.retrieve(user_input, top_k=5)  # Recupera os top 5 documentos relevantes
+    if res:
+        st.write(f"Foram encontrados {len(res)} documentos relevantes.")
+        for document in res:
+            prediction = pipe(question=user_input, context=document.content)
+            context = document.content
+            confidence = prediction["score"]
+            answer = prediction["answer"]
+            st.write("Pergunta:", user_input)
+            st.write("Resposta:", answer)
+            st.write("Confiança:", confidence)
+            st.write("Contexto:", context)
+            st.write("-" * 50)
 
 if st.button('Buscar Resposta'):
-    print_answers(prediction, details="minimum")  ## Choose from `minimum`, `medium`, and `all`
-
-
+    st.write(prediction["answer"])
 

search.py

@@ -1,36 +0,0 @@
-from haystack.document_stores import InMemoryDocumentStore
-from haystack.utils import fetch_archive_from_http
-import os
-from haystack.pipelines.standard_pipelines import TextIndexingPipeline
-from haystack.nodes import BM25Retriever
-from haystack.nodes import FARMReader
-from haystack.pipelines import ExtractiveQAPipeline
-from haystack.utils import print_answers
-
-#Criando o objeto documentStore
-
-document_store = InMemoryDocumentStore(use_bm25=True)
-
-#Exportando os dados necessários sobre o Game Of Thrones
-
-doc_dir = "data/build_your_first_question_answering_system"
-
-fetch_archive_from_http(
-    url="https://s3.eu-central-1.amazonaws.com/deepset.ai-farm-qa/datasets/documents/wiki_gameofthrones_txt1.zip",
-    output_dir=doc_dir,
-)
-
-files_to_index = [doc_dir + "/" + f for f in os.listdir(doc_dir)]
-indexing_pipeline = TextIndexingPipeline(document_store)
-indexing_pipeline.run_batch(file_paths=files_to_index)
-
-retriever = BM25Retriever(document_store=document_store)
-
-reader = FARMReader(model_name_or_path="deepset/roberta-base-squad2", use_gpu=True)
-
-
-pipe = ExtractiveQAPipeline(reader, retriever)
-
-
-
-