Upload 2 files

app.py

@@ -56,16 +56,16 @@ with gr.Blocks(title='BotNews') as page:
     with gr.Row():
         output1 = gr.Textbox(label="Notícia gerada por IA", lines=25)
     gr.Markdown("<hr>")
-    gr.Markdown("## Prompt para a Notícia")
+    gr.Markdown("## Ajustar a notícia com IA")
     gr.Markdown(" ")
     gr.Markdown("<b>Instrução:</b> Preencha abaixo com um comando para ser executado sobre a notícia (Por exemplo: 'Resuma em tópicos' ou 'Adicione um tom sarcástico').")
     gr.Markdown(" ")
     with gr.Row():
         input6 = gr.Textbox(label="Ajustar a notícia com IA")
     with gr.Row():
-        button2 = gr.Button("Gerar resposta")
+        button2 = gr.Button("Ajustar notícia")
     with gr.Row():
-        output2 = gr.Textbox(label="Resposta gerada por IA", lines=25)
+        output2 = gr.Textbox(label="Notícia ajustada por IA", lines=25)
 
     button1.click(call_generate_news, inputs=[input1, input2, input3, input4, input5], outputs=[output1])
     button2.click(call_invoke_llm, inputs=[output1, input6, input5], outputs=[output2])

rag.py

@@ -4,6 +4,7 @@
 #from langchain.embeddings import OpenAIEmbeddings
 from langchain_openai import OpenAIEmbeddings
 from langchain_community.embeddings import HuggingFaceEmbeddings
+from langchain_together.embeddings import TogetherEmbeddings
 from langchain_community.vectorstores import Chroma
 from langchain_community.document_loaders import DirectoryLoader
 from langchain.text_splitter import RecursiveCharacterTextSplitter
@@ -14,6 +15,7 @@ from langchain.memory import ConversationBufferMemory
 from langchain.chains import ConversationalRetrievalChain
 import os
 import csv
+import time
 
 def read_csv_to_dict(filename):
     data_dict = {}
@@ -45,9 +47,14 @@ def generate_embeddings_and_vectorstore(path, model):
         #print(docs)
         if model == 'openai':
             fc_embeddings = OpenAIEmbeddings(openai_api_key=os.environ['OPENAI_KEY'])
+            vectorstore = Chroma.from_documents(docs, fc_embeddings)
         else:
-            fc_embeddings = HuggingFaceEmbeddings(model_name = 'intfloat/multilingual-e5-large-instruct')
-        vectorstore = Chroma.from_documents(docs, fc_embeddings)
+            #fc_embeddings = HuggingFaceEmbeddings(model_name = 'intfloat/multilingual-e5-large-instruct')
+            #vectorstore = Chroma.from_documents(docs, fc_embeddings)
+            fc_embeddings = TogetherEmbeddings(model = 'togethercomputer/m2-bert-80M-8k-retrieval', together_api_key = os.environ['TOGETHER_KEY'])
+            for doc in docs:
+                vectorstore = Chroma.from_documents(documents=[doc], embedding=fc_embeddings)
+                time.sleep(0.5)
         print('total de docs no vectorstore=',len(vectorstore.get()['documents']))
 
         return vectorstore