Update app/server.py

Restore for the original server.py.

app/server.py

@@ -8,7 +8,6 @@ from langchain.prompts import ChatPromptTemplate
 from langchain_core.runnables import RunnablePassthrough, RunnableParallel
 from langchain.schema import StrOutputParser
 from langchain.embeddings.huggingface import HuggingFaceEmbeddings
-from langchain.memory import ConversationBufferMemory
 
 app = FastAPI()
 
@@ -18,8 +17,7 @@ hf_llm = HuggingFaceHub(
     repo_id="mistralai/Mixtral-8x7B-Instruct-v0.1",
     huggingfacehub_api_token=os.environ["HF_TOKEN"],
     task="text-generation",
-    model_kwargs={"temperature":0.01, "max_new_tokens" : 250},
-    memory = ConversationBufferMemory(memory_key="chat_history")
+    model_kwargs={"temperature":0.01, "max_new_tokens" : 250}
 )
 
 embedding_model_id = 'WhereIsAI/UAE-Large-V1'
@@ -30,11 +28,10 @@ retriever = faiss_index.as_retriever()
 # retriever = faiss_index.as_retriever(search_kwargs={"k": 2})
 
 prompt_template = """\
-
-You are a lab assistant who is trying to answer the student's question. The following is the conversation history you are having with the student:
-{chat_history}
-
-Use the provided context below to answer the user's question. If you don't know the answer, say you don't know and ask the student to seek help from the instructor. The answer you provided should be limited to 3 sentences if possible. 
+Use the provided context to answer the user's question. If you don't know the answer, say you don't know. 
+The text output that you are providing should only contain the answer, but not the question text or any other prompt text.
+The answer you provided should be limited to 2 sentences if possible.
+Also, if the answer contains any of the word phrases [pipette tip, vortex mixer, lysis-buffer, covid sample], replace the space in the word phrase with a hyphen (e.g. pipette-tip).
 
 Context:
 {context}