Update app.py

app.py

@@ -1,13 +1,31 @@
 from transformers import AutoModelForCausalLM, AutoTokenizer
 import torch
+import pandas as pd
 
-# Cargar el modelo y tokenizer de LLaMA 2
+# Tu token secreto de Hugging Face
+huggingface_token = st.secrets["HUGGINGFACEHUB_API_TOKEN"]
+
+
+# Cargar el modelo y tokenizer de LLaMA 2, usando el token secreto
 model_name = "meta-llama/Llama-2-7b-hf"  # O ajusta según el modelo que prefieras
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForCausalLM.from_pretrained(model_name)
+tokenizer = AutoTokenizer.from_pretrained(model_name, use_auth_token=huggingface_token)
+model = AutoModelForCausalLM.from_pretrained(model_name, use_auth_token=huggingface_token)
+
+# Pedir al usuario que introduzca el path del archivo CSV
+csv_file = input("Por favor introduce la ruta del archivo CSV: ")
+
+# Cargar el CSV y extraer la columna 'job_titles'
+df = pd.read_csv(csv_file)
+job_titles = df['job_titles'].tolist()
 
-# Crear prompt
-prompt = """
+# Crear la lista de job titles en formato de texto para el prompt
+job_titles_text = "\n".join(f"- {title}" for title in job_titles)
+
+# Query del usuario
+user_query = input("Introduce tu query: ")
+
+# Crear el prompt usando los job titles del CSV y la query del usuario
+prompt = f"""
 You are an advanced AI assistant trained to process job titles and user queries. I will provide you with a list of job titles, and a user query. Your task is to:
 
 1. Calculate the cosine similarity score between the query and each job title.
@@ -15,14 +33,9 @@ You are an advanced AI assistant trained to process job titles and user queries.
 3. Return the top 5 job titles with their cosine similarity scores.
 
 Here is the list of job titles from the CSV:
-- Software Engineer
-- Data Scientist
-- Machine Learning Engineer
-- Business Analyst
-- Product Manager
-...
-
-The user's query is: "Machine Learning Expert"
+{job_titles_text}
+
+The user's query is: "{user_query}"
 
 Now, compute the similarity scores, rank the job titles, and return the top 5.
 """