Merge pull request #163 from arc53/chat-prompts

chat prompts

application/app.py

@@ -1,25 +1,31 @@
-import os
 import json
+import os
 import traceback
 
 import dotenv
 import requests
 from flask import Flask, request, render_template
 from langchain import FAISS
-from langchain.llms import OpenAIChat
 from langchain import VectorDBQA, HuggingFaceHub, Cohere, OpenAI
 from langchain.chains.question_answering import load_qa_chain
+from langchain.chat_models import ChatOpenAI
 from langchain.embeddings import OpenAIEmbeddings, HuggingFaceHubEmbeddings, CohereEmbeddings, \
     HuggingFaceInstructEmbeddings
 from langchain.prompts import PromptTemplate
+from langchain.prompts.chat import (
+    ChatPromptTemplate,
+    SystemMessagePromptTemplate,
+    HumanMessagePromptTemplate,
+)
+
 from error import bad_request
 
-os.environ["LANGCHAIN_HANDLER"] = "langchain"
+# os.environ["LANGCHAIN_HANDLER"] = "langchain"
 
 if os.getenv("LLM_NAME") is not None:
     llm_choice = os.getenv("LLM_NAME")
 else:
-    llm_choice = "openai"
+    llm_choice = "openai_chat"
 
 if os.getenv("EMBEDDINGS_NAME") is not None:
     embeddings_choice = os.getenv("EMBEDDINGS_NAME")
@@ -47,15 +53,21 @@ if platform.system() == "Windows":
 # loading the .env file
 dotenv.load_dotenv()
 
-with open("combine_prompt.txt", "r") as f:
+with open("prompts/combine_prompt.txt", "r") as f:
     template = f.read()
 
-with open("combine_prompt_hist.txt", "r") as f:
+with open("prompts/combine_prompt_hist.txt", "r") as f:
     template_hist = f.read()
 
-with open("question_prompt.txt", "r") as f:
+with open("prompts/question_prompt.txt", "r") as f:
     template_quest = f.read()
 
+with open("prompts/chat_combine_prompt.txt", "r") as f:
+    chat_combine_template = f.read()
+
+with open("prompts/chat_reduce_prompt.txt", "r") as f:
+    chat_reduce_template = f.read()
+
 if os.getenv("API_KEY") is not None:
     api_key_set = True
 else:
@@ -98,7 +110,7 @@ def api_answer():
                 vectorstore = ""
         else:
             vectorstore = ""
-        #vectorstore = "outputs/inputs/"
+        # vectorstore = "outputs/inputs/"
         # loading the index and the store and the prompt template
         # Note if you have used other embeddings than OpenAI, you need to change the embeddings
         if embeddings_choice == "openai_text-embedding-ada-002":
@@ -123,9 +135,20 @@ def api_answer():
 
         q_prompt = PromptTemplate(input_variables=["context", "question"], template=template_quest,
                                   template_format="jinja2")
-        if llm_choice == "openai":
-            llm = OpenAIChat(openai_api_key=api_key, temperature=0)
-            #llm = OpenAI(openai_api_key=api_key, temperature=0)
+        if llm_choice == "openai_chat":
+            llm = ChatOpenAI(openai_api_key=api_key)
+            messages_combine = [
+                SystemMessagePromptTemplate.from_template(chat_combine_template),
+                HumanMessagePromptTemplate.from_template("{question}")
+            ]
+            p_chat_combine = ChatPromptTemplate.from_messages(messages_combine)
+            messages_reduce = [
+                SystemMessagePromptTemplate.from_template(chat_reduce_template),
+                HumanMessagePromptTemplate.from_template("{question}")
+            ]
+            p_chat_reduce = ChatPromptTemplate.from_messages(messages_reduce)
+        elif llm_choice == "openai":
+            llm = OpenAI(openai_api_key=api_key, temperature=0)
         elif llm_choice == "manifest":
             llm = ManifestWrapper(client=manifest, llm_kwargs={"temperature": 0.001, "max_tokens": 2048})
         elif llm_choice == "huggingface":
@@ -133,13 +156,19 @@ def api_answer():
         elif llm_choice == "cohere":
             llm = Cohere(model="command-xlarge-nightly", cohere_api_key=api_key)
 
-        qa_chain = load_qa_chain(llm=llm, chain_type="map_reduce",
-                                 combine_prompt=c_prompt, question_prompt=q_prompt)
+        if llm_choice == "openai_chat":
+            chain = VectorDBQA.from_chain_type(llm=llm, chain_type="map_reduce", vectorstore=docsearch,
+                                               k=4,
+                                               chain_type_kwargs={"question_prompt": p_chat_reduce,
+                                                                  "combine_prompt": p_chat_combine})
+            result = chain({"query": question})
+        else:
+            qa_chain = load_qa_chain(llm=llm, chain_type="map_reduce",
+                                     combine_prompt=c_prompt, question_prompt=q_prompt)
+            chain = VectorDBQA(combine_documents_chain=qa_chain, vectorstore=docsearch, k=4)
+            result = chain({"query": question})
 
-        chain = VectorDBQA(combine_documents_chain=qa_chain, vectorstore=docsearch, k=4)
-
-        # fetch the answer
-        result = chain({"query": question})
+        print(result)
 
         # some formatting for the frontend
         result['answer'] = result['result']
@@ -215,7 +244,6 @@ def api_feedback():
     return {"status": 'ok'}
 
 
-
 # handling CORS
 @app.after_request
 def after_request(response):

application/prompts/chat_combine_prompt.txt

@@ -0,0 +1,4 @@
+You are a DocsGPT, friendly and helpful AI assistant by Arc53 that provides help with documents. You give thorough answers with code examples if possible.
+Use the following pieces of context to help answer the users question.
+----------------
+{summaries}

application/prompts/chat_reduce_prompt.txt

@@ -0,0 +1,3 @@
+Use the following portion of a long document to see if any of the text is relevant to answer the question.
+{context}
+Provide all relevant text to the question verbatim. Summarize if needed. If nothing relevant return "-".