Update app.py

app.py

@@ -13,89 +13,77 @@ from langchain.text_splitter import RecursiveCharacterTextSplitter
 import openai
 
 def proper_query(query):
-    prompt = f"El siguiente texto es una pregunta en español: {query}\n\n¿Cómo debería ser la pregunta para que sea correcta en español?\nPregunta corregida:"
-    response = openai.Completion.create(engine="text-davinci-003", prompt=prompt, max_tokens=1000, temperature=0.2)
+    prompt = f"The following text is a user's question: {query}\n\nHow should that question be modified so that it uses correct language?\nReturn the question in the same language.\nCorrected Question:"
+    response = openai.Completion.create(
+        engine="text-davinci-003", prompt=prompt, max_tokens=1000, temperature=0.3)
     return response.choices[0].text
-
-def extract_text_from_pdf(file_path, splitter = "\n\n"):
-    with open(file_path, 'rb') as file:
-        pdf = PyPDF2.PdfReader(file)
-        text = ''
-        for page in pdf.pages:
-            text += page.extract_text()
-        chunks = text.split(splitter)
-        chunks = [splitter + chunk for chunk in chunks[1:]]
-        #create a csv file with the chunks in one column
-        #df = pd.DataFrame(chunks, columns=['text'])
-        #write to csv
-        #df.to_csv(file_path[:-4]+'.csv', index=False)
-        return chunks
     
 embeddings = OpenAIEmbeddings()
-text = extract_text_from_pdf('transito-dgo.pdf','ARTÍCULO')
+#transform a column of a csv into a list
+df = pd.read_csv('reglamento-avianca.csv')
+text = df['text'].tolist()
 
 text_splitter = RecursiveCharacterTextSplitter(
     # Set a really small chunk size, just to show.
-    chunk_size = 500,
+    chunk_size = 1000,
     chunk_overlap  = 0,
     length_function = len,
 )
-
 texts = text_splitter.split_text(text)
 
 docsearch = FAISS.from_texts(texts, embeddings)
 
-def asesor_transito(query):
+def asesor_avianca(query):
     query = proper_query(query)
     docs = docsearch.similarity_search(query)
     refine_prompt_template = (
         "The original question is as follows: {question}\n"
-        "We have provided an existing answer: {existing_answer}\n"
-        "You have the opportunity to refine the existing answer,"
-        "only if needed, exclusively with the context below.\n"
+        "We have provided an answer: {existing_answer}\n"
+        "You have the opportunity to refine that answer,"
+        "only if needed, with the context below.\n"
         "------------\n"
         "{context_str}\n"
         "------------\n"
-        "If that context is not helpful to answer the question, then omit it.\n"
-        "Shorten the answer if possible.\n"
+        "Using no prior knowledge, change the answer only if the given context is related to the question.\n"
+        "Translate the answer to easy-to-understand language.\n"
+        "Shorten the answer as much as possible.\n"
         "Reply in the same language as the question.\n"
-        "If the context is not helpful to answer the question or if it is not a question, then you will refuse to answer.\n"
         "Answer:"
     )
     refine_prompt = PromptTemplate(
         input_variables=["question", "existing_answer", "context_str"],
         template=refine_prompt_template,
     )
-
-
+    
+    
     initial_qa_template = (
         "Context information is below. \n"
         "---------------------\n"
         "{context_str}"
         "\n---------------------\n"
         "Given the context information and not prior knowledge, "
-        "answer the question: {question}\n"
-        "If the context is not helpful to answer the question or if it is not a question, then you will refuse to answer.\n"
+        "answer the question to the user: {question}\n"
+        "If the context is not helpful to answer the question then refuse to answer the question in the same language."
     )
     initial_qa_prompt = PromptTemplate(
         input_variables=["context_str", "question"], template=initial_qa_template
     )
-    chain = load_qa_chain(OpenAI(temperature=0), chain_type="refine", return_refine_steps=False,
-                        question_prompt=initial_qa_prompt, refine_prompt=refine_prompt)
+    chain = load_qa_chain(OpenAI(temperature=0.2), chain_type="refine", return_refine_steps=False,
+                         question_prompt=initial_qa_prompt, refine_prompt=refine_prompt)
     ans = chain({"input_documents": docs, "question": query}, return_only_outputs=True)['output_text']
     return ans
 
 demo = gr.Interface(
     fn=asesor_transito,
     inputs=[
-        gr.Textbox(label="Hola soy tu asesor personal de tránsito de Durango, ¿cuál es tu pregunta? \nHi, I am your Durango transit law personal assistant, ask me anything about Mexico City's transit law in any language.", lines=3,),
+        gr.Textbox(label="Pregunta: / Question: ", lines=3,),
     ],
-    outputs=[gr.Textbox(label="Respuesta: \nAnswer: ")],
-    title="Asesor de Reglamento de Tránsito Durango",
+    outputs=[gr.Textbox(label="Respuesta: \ Answer: ")],
+    title="Asesor de Reglamento de Avianca",
+    description = "Hola soy tu asesor personal de Avianca, Hexagonito. Pregúntame lo que necesites saber sobre las reglas de tu vuelo. \nHi, I am Hexagonito, your Avianca rules personal assistant, ask me anything about Avianca rules in any language.",
     examples=[
-        ["cuál es la multa por no llevar casco?"],
-        ["qué pasa si no tengo licencia de conducir?"],
-        ["What would happen if I drove under the influence of alcohol?"]
+        ["qué documentos necesito para viajar?"],
+        ["no llegó mi equipaje, qué puedo hacer?"]
     ],
 )