Update app.py

app.py

@@ -1,64 +1,94 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
+import os
+import time
+from huggingface_hub import login
+from langchain_community.document_loaders import TextLoader
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from langchain_community.vectorstores import SKLearnVectorStore
+from langchain_huggingface import ChatHuggingFace, HuggingFacePipeline, HuggingFaceEmbeddings
+from langchain.prompts import PromptTemplate
+from langchain_core.output_parsers import StrOutputParser
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
+HF_TOKEN = os.getenv("HUGGINGFACEHUB_API_TOKEN")
+login(token=HF_TOKEN)
 
+data=TextLoader("erikdata.txt").load()
+text_splitter = RecursiveCharacterTextSplitter.from_tiktoken_encoder(chunk_size=125, chunk_overlap=5)
+data_split = text_splitter.split_documents(data)
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+model_name = "BAAI/llm-embedder"
+model_kwargs = {'device': 'cpu'}
+encode_kwargs = {'normalize_embeddings': False}
+hf = HuggingFaceEmbeddings(
+    model_name=model_name,
+    model_kwargs=model_kwargs,
+    encode_kwargs=encode_kwargs
+)
+vectorstore = SKLearnVectorStore.from_documents(
+    documents=data_split,
+    embedding=hf,
+)
+retriever = vectorstore.as_retriever(k=4)
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+prompt = PromptTemplate(
+    template="""You are an AI replica of Professor Erik Birgersson. Professor Erik is the director of the Engineering Science Program in the National University of Singapore.
+    Your students are likely to be Engineering Science undergraduates who are lost on the syllabus on the following courses:
+1. Numerical Methods and Statistics (ESP2107) which covers concepts such as interpolation, root finding, numerical differentiation and integration and linear systems, using MATLAB code.
+2. Principles of Continua (ESP2106) which covers concepts such as computational fluid dynamics, heat transfer, Cauchy's equation, vectors and tensors.
+    Use the following documents to answer the question and try to follow the way he speaks.
+    If you don't know the answer, just say that you don't know.
+    Use three sentences maximum and keep the answer concise. You MUST only output one question and one answer.
+    Documents: {documents}
+    Question: {user_message}
+    Answer:
+    """,
+    input_variables=["user_message", "documents"],
+)
 
-    messages.append({"role": "user", "content": message})
+llm = HuggingFacePipeline.from_model_id(
+    model_id="meta-llama/Llama-3.2-1B",
+    task="text-generation",
+    pipeline_kwargs=dict(
+        max_new_tokens=64,
+    )
+)
 
-    response = ""
+rag_chain = prompt | llm |StrOutputParser()
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+class RAGApplication:
+    def __init__(self, retriever, rag_chain):
+        self.retriever = retriever
+        self.rag_chain = rag_chain
+    def run(self, user_message):
+        documents = self.retriever.invoke(user_message)
+        doc_texts = "\\n".join([doc.page_content for doc in documents])
+        answer = self.rag_chain.invoke({"user_message": user_message, "documents": doc_texts})
+        answer = answer.split('Answer:\n')[1].split('Question:')[0] #splitting the answer output
+        return answer
 
-        response += token
-        yield response
+rag_application = RAGApplication(retriever, rag_chain)
 
 
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-    ],
-)
+with gr.Blocks() as demo:
+    chatbot = gr.Chatbot(type="messages")
+    msg = gr.Textbox()
+    clear = gr.Button("Clear")
+
+    def user(user_message, history: list):
+        return  history + [{"role": "user", "content": user_message}]
+
+    def bot(user_message, history: list):
+        bot_message = rag_application.run(user_message)
+        history.append({"role": "assistant", "content": ""})
+        for character in bot_message:
+            history[-1]['content'] += character
+            time.sleep(0.005)
+            yield history
 
+    msg.submit(user, [msg, chatbot], chatbot, queue=False).then(
+        bot, [msg, chatbot], chatbot
+    )
+    clear.click(lambda: None, None, chatbot, queue=False)
 
 if __name__ == "__main__":
     demo.launch()