Merge pull request #165 from ks6088ts-labs/feature/issue-164_slm

add SLM based chat app

apps/15_streamlit_chat_slm/README.md

@@ -0,0 +1,25 @@
+# Streamlit Chat with SLM
+
+## Overview
+
+```shell
+# Run Ollama server
+$ ollama serve
+
+# Install dependencies
+$ ollama pull phi3
+
+# Run a simple chat with Ollama
+$ poetry run python apps/15_streamlit_chat_slm/chat.py
+
+# Run summarization with SLM
+$ poetry run python apps/15_streamlit_chat_slm/summarize.py
+
+# Run streamlit app
+$ poetry run python -m streamlit run apps/15_streamlit_chat_slm/main.py
+```
+
+# References
+
+- [ChatOllama](https://python.langchain.com/docs/integrations/chat/ollama/)
+- [Summarize Text](https://python.langchain.com/docs/tutorials/summarization/)

apps/15_streamlit_chat_slm/chat.py

@@ -0,0 +1,43 @@
+import argparse
+import logging
+
+from dotenv import load_dotenv
+from langchain_core.messages import AIMessage
+from langchain_ollama import ChatOllama
+
+
+def init_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(
+        prog="slm_chat",
+        description="Chat with SLM model",
+    )
+    parser.add_argument("-m", "--model", default="phi3")
+    parser.add_argument("-s", "--system", default="You are a helpful assistant.")
+    parser.add_argument("-p", "--prompt", default="What is the capital of France?")
+    parser.add_argument("-v", "--verbose", action="store_true")
+    return parser.parse_args()
+
+
+if __name__ == "__main__":
+    args = init_args()
+
+    # Set verbose mode
+    if args.verbose:
+        logging.basicConfig(level=logging.DEBUG)
+
+    # Parse .env file and set environment variables
+    load_dotenv()
+
+    llm = ChatOllama(
+        model=args.model,
+        temperature=0,
+    )
+
+    ai_msg: AIMessage = llm.invoke(
+        input=[
+            ("system", args.system),
+            ("human", args.prompt),
+        ]
+    )
+    print(ai_msg.model_dump_json(indent=2))
+    # print(ai_msg.content)

apps/15_streamlit_chat_slm/main.py

@@ -0,0 +1,67 @@
+import streamlit as st
+from dotenv import load_dotenv
+from langchain_ollama import ChatOllama
+
+load_dotenv()
+
+SUPPORTED_MODELS = [
+    "phi3",
+]
+with st.sidebar:
+    slm_model = st.selectbox(
+        label="Model",
+        options=SUPPORTED_MODELS,
+        index=0,
+    )
+    "[Azure Portal](https://portal.azure.com/)"
+    "[Azure OpenAI Studio](https://oai.azure.com/resource/overview)"
+    "[View the source code](https://github.com/ks6088ts-labs/workshop-azure-openai/blob/main/apps/15_streamlit_chat_slm/main.py)"
+
+
+def is_configured():
+    return slm_model in SUPPORTED_MODELS
+
+
+st.title("15_streamlit_chat_slm")
+
+if not is_configured():
+    st.warning("Please fill in the required fields at the sidebar.")
+
+if "messages" not in st.session_state:
+    st.session_state["messages"] = [
+        {
+            "role": "assistant",
+            "content": "Hello! I'm a helpful assistant.",
+        }
+    ]
+
+# Show chat messages
+for msg in st.session_state.messages:
+    st.chat_message(msg["role"]).write(msg["content"])
+
+# Receive user input
+if prompt := st.chat_input(disabled=not is_configured()):
+    client = ChatOllama(
+        model=slm_model,
+        temperature=0,
+    )
+
+    st.session_state.messages.append(
+        {
+            "role": "user",
+            "content": prompt,
+        }
+    )
+    st.chat_message("user").write(prompt)
+    with st.spinner("Thinking..."):
+        response = client.invoke(
+            input=st.session_state.messages,
+        )
+    msg = response.content
+    st.session_state.messages.append(
+        {
+            "role": "assistant",
+            "content": msg,
+        }
+    )
+    st.chat_message("assistant").write(msg)

apps/15_streamlit_chat_slm/summarize.py

@@ -0,0 +1,162 @@
+import asyncio
+import operator
+from os import getenv
+from typing import Annotated, Literal, TypedDict
+
+from langchain.chains.combine_documents.reduce import acollapse_docs, split_list_of_docs
+from langchain_community.document_loaders import WebBaseLoader
+from langchain_core.documents import Document
+from langchain_core.output_parsers import StrOutputParser
+from langchain_core.prompts import ChatPromptTemplate
+from langchain_ollama import ChatOllama
+from langchain_openai import AzureChatOpenAI
+from langchain_text_splitters import CharacterTextSplitter
+from langgraph.constants import Send
+from langgraph.graph import END, START, StateGraph
+
+token_max = 1000
+url = "https://lilianweng.github.io/posts/2023-06-23-agent/"
+
+llm_ollama = ChatOllama(
+    model="phi3",
+    temperature=0,
+)
+llm_azure_openai = AzureChatOpenAI(
+    temperature=0,
+    api_key=getenv("AZURE_OPENAI_API_KEY"),
+    api_version=getenv("AZURE_OPENAI_API_VERSION"),
+    azure_endpoint=getenv("AZURE_OPENAI_ENDPOINT"),
+    model=getenv("AZURE_OPENAI_GPT_MODEL"),
+)
+# Use the Ollama model
+llm = llm_ollama
+
+
+def length_function(documents: list[Document]) -> int:
+    """Get number of tokens for input contents."""
+    return sum(llm.get_num_tokens(doc.page_content) for doc in documents)
+
+
+# This will be the overall state of the main graph.
+# It will contain the input document contents, corresponding
+# summaries, and a final summary.
+class OverallState(TypedDict):
+    # Notice here we use the operator.add
+    # This is because we want combine all the summaries we generate
+    # from individual nodes back into one list - this is essentially
+    # the "reduce" part
+    contents: list[str]
+    summaries: Annotated[list, operator.add]
+    collapsed_summaries: list[Document]
+    final_summary: str
+
+
+# This will be the state of the node that we will "map" all
+# documents to in order to generate summaries
+class SummaryState(TypedDict):
+    content: str
+
+
+map_prompt = ChatPromptTemplate.from_messages([("system", "Write a concise summary of the following:\\n\\n{context}")])
+
+map_chain = map_prompt | llm | StrOutputParser()
+
+
+# Here we generate a summary, given a document
+async def generate_summary(state: SummaryState):
+    response = await map_chain.ainvoke(state["content"])
+    return {"summaries": [response]}
+
+
+# Here we define the logic to map out over the documents
+# We will use this an edge in the graph
+def map_summaries(state: OverallState):
+    # We will return a list of `Send` objects
+    # Each `Send` object consists of the name of a node in the graph
+    # as well as the state to send to that node
+    return [Send("generate_summary", {"content": content}) for content in state["contents"]]
+
+
+def collect_summaries(state: OverallState):
+    return {"collapsed_summaries": [Document(summary) for summary in state["summaries"]]}
+
+
+# Also available via the hub: `hub.pull("rlm/reduce-prompt")`
+reduce_template = """
+The following is a set of summaries:
+{docs}
+Take these and distill it into a final, consolidated summary
+of the main themes.
+"""
+
+reduce_prompt = ChatPromptTemplate([("human", reduce_template)])
+
+reduce_chain = reduce_prompt | llm | StrOutputParser()
+
+
+# Add node to collapse summaries
+async def collapse_summaries(state: OverallState):
+    doc_lists = split_list_of_docs(state["collapsed_summaries"], length_function, token_max)
+    results = []
+    for doc_list in doc_lists:
+        results.append(await acollapse_docs(doc_list, reduce_chain.ainvoke))
+
+    return {"collapsed_summaries": results}
+
+
+# This represents a conditional edge in the graph that determines
+# if we should collapse the summaries or not
+def should_collapse(
+    state: OverallState,
+) -> Literal["collapse_summaries", "generate_final_summary"]:
+    num_tokens = length_function(state["collapsed_summaries"])
+    if num_tokens > token_max:
+        return "collapse_summaries"
+    else:
+        return "generate_final_summary"
+
+
+# Here we will generate the final summary
+async def generate_final_summary(state: OverallState):
+    response = await reduce_chain.ainvoke(state["collapsed_summaries"])
+    return {"final_summary": response}
+
+
+async def main():
+    # Construct the graph
+    # Nodes:
+    graph = StateGraph(OverallState)
+    graph.add_node("generate_summary", generate_summary)  # same as before
+    graph.add_node("collect_summaries", collect_summaries)
+    graph.add_node("collapse_summaries", collapse_summaries)
+    graph.add_node("generate_final_summary", generate_final_summary)
+
+    # Edges:
+    graph.add_conditional_edges(START, map_summaries, ["generate_summary"])
+    graph.add_edge("generate_summary", "collect_summaries")
+    graph.add_conditional_edges("collect_summaries", should_collapse)
+    graph.add_conditional_edges("collapse_summaries", should_collapse)
+    graph.add_edge("generate_final_summary", END)
+
+    app = graph.compile()
+
+    # create graph image
+    app.get_graph().draw_mermaid_png(output_file_path="docs/images/15_streamlit_chat_slm.summarize_graph.png")
+
+    text_splitter = CharacterTextSplitter.from_tiktoken_encoder(chunk_size=1000, chunk_overlap=0)
+
+    loader = WebBaseLoader(web_path=url)
+    docs = loader.load()
+
+    split_docs = text_splitter.split_documents(docs)
+    print(f"Generated {len(split_docs)} documents.")
+
+    async for step in app.astream(
+        {"contents": [doc.page_content for doc in split_docs]},
+        {"recursion_limit": 10},
+    ):
+        print(list(step.keys()))
+        print(step)
+
+
+asyncio.run(main())