Upload 3 files

app.py

@@ -1,70 +1,86 @@
+import os
+from dotenv import load_dotenv
 import gradio as gr
-from huggingface_hub import InferenceClient
+from tools import create_agent
+from langchain_core.messages import RemoveMessage
+from langchain_core.messages import trim_messages
 
+# Global params
+AGENT = create_agent()
+theme = gr.themes.Default(primary_hue="red", secondary_hue="red")
+default_msg = "Bonjour ! Je suis là pour répondre à vos questions sur l'actuariat. Comment puis-je vous aider aujourd'hui ?"
 
-def respond(
-    message,
-    history: list[dict[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-    hf_token: gr.OAuthToken,
-):
-    """
-    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(token=hf_token.token, model="openai/gpt-oss-20b")
 
-    messages = [{"role": "system", "content": system_message}]
+def filter_msg(msg_list:list, keep_n:int) -> list:
+    """Keep only last keep_n messages from chat history. Preserves structure user msg -> tool msg -> ai msg"""
+    msg = trim_messages(
+            msg_list,
+            strategy="last",
+            token_counter=len,
+            max_tokens=keep_n,
+            start_on="human",
+            end_on=("tool", "ai"),
+            include_system=True,
+    )
+    return [m.id for m in msg]
 
-    messages.extend(history)
+def agent_response(query, config, keep_n=10):
+    messages = AGENT.get_state(config).values.get("messages", [])
+    
+    if len(messages) > keep_n:
+        keep_msg_ids = filter_msg(messages, keep_n)
+        AGENT.update_state(config, {"messages": [RemoveMessage(id=m.id) for m in messages if m.id not in keep_msg_ids]})
+        print("msg removed")
 
-    messages.append({"role": "user", "content": message})
+    # Generate answer
+    answer = AGENT.invoke({"messages":query}, config=config)
+    return answer["messages"][-1].content
 
-    response = ""
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        choices = message.choices
-        token = ""
-        if len(choices) and choices[0].delta.content:
-            token = choices[0].delta.content
+js_func = """
+function refresh() {
+    const url = new URL(window.location);
 
-        response += token
-        yield response
+    if (url.searchParams.get('__theme') != 'light') {
+        url.searchParams.set('__theme', 'light');
+        window.location.href = url.href;
+    }
+}
+"""
 
 
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-chatbot = gr.ChatInterface(
-    respond,
-    type="messages",
-    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:
-    with gr.Sidebar():
-        gr.LoginButton()
-    chatbot.render()
+def delete_agent():
+    print("del agent")
+    global AGENT
+    AGENT = create_agent()
+    # print(AGENT.get_state(config).values.get("messages"), "\n\n")
+
+with gr.Blocks(theme=theme, js=js_func, title="Dataltist", fill_height=True) as iface:
+    gr.Markdown("# Dataltist Chatbot 🚀")
+    chatbot = gr.Chatbot(show_copy_button=True, show_share_button=False, value=[{"role":"assistant", "content":default_msg}], type="messages", scale=1)
+    msg = gr.Textbox(lines=1, show_label=False, placeholder="Posez vos questions sur l'assurance")  #  submit_btn=True
+    # clear = gr.ClearButton([msg, chatbot], value="Effacer 🗑")
+    config = {"configurable": {"thread_id": "1"}} 
+
+
+    def user(user_message, history: list):
+        return "", history + [{"role": "user", "content": user_message}]
+
+    def bot(history: list):
+        bot_message = agent_response(history[-1]["content"], config)  #AGENT.invoke({"messages":history[-1]["content"]}, config=config)
+        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], [msg, chatbot], queue=False).then(
+        bot, chatbot, chatbot
+    )
+    iface.unload(delete_agent)
 
 if __name__ == "__main__":
-    demo.launch()
+    # load_dotenv()
+    # AUTH_ID = os.environ.get("AUTH_ID")
+    # AUTH_PASS = os.environ.get("AUTH_PASS")
+    iface.launch()  #share=True, auth=(AUTH_ID, AUTH_PASS)

requirements.txt

@@ -0,0 +1,194 @@
+aiofiles==23.2.1
+aiohappyeyeballs==2.4.0
+aiohttp==3.10.5
+aiosignal==1.3.1
+altair==5.4.1
+annotated-types==0.7.0
+anyio==4.4.0
+asgiref==3.8.1
+asttokens==2.4.1
+attrs==24.2.0
+backoff==2.2.1
+bcrypt==4.2.0
+blinker==1.8.2
+build==1.2.1
+cachetools==5.5.0
+certifi==2024.8.30
+charset-normalizer==3.3.2
+chroma-datasets==0.1.5
+chroma-hnswlib==0.7.6
+chromadb==0.5.7
+click==8.1.7
+colorama==0.4.6
+coloredlogs==15.0.1
+comm==0.2.2
+contourpy==1.3.0
+cycler==0.12.1
+dataclasses-json==0.6.7
+datasets==3.0.0
+debugpy==1.8.5
+decorator==5.1.1
+Deprecated==1.2.14
+dill==0.3.8
+distro==1.9.0
+executing==2.1.0
+fastapi==0.112.2
+ffmpy==0.4.0
+filelock==3.15.4
+flatbuffers==24.3.25
+fonttools==4.54.0
+frozenlist==1.4.1
+fsspec==2024.6.1
+gitdb==4.0.11
+GitPython==3.1.43
+google-auth==2.34.0
+googleapis-common-protos==1.65.0
+gradio==4.44.0
+gradio_client==1.3.0
+greenlet==3.0.3
+grpcio==1.66.1
+h11==0.14.0
+httpcore==1.0.5
+httptools==0.6.1
+httpx==0.27.2
+httpx-sse==0.4.0
+huggingface-hub==0.24.6
+humanfriendly==10.0
+idna==3.8
+importlib_metadata==8.4.0
+importlib_resources==6.4.4
+ipykernel==6.29.5
+ipython==8.27.0
+jedi==0.19.1
+Jinja2==3.1.4
+jiter==0.5.0
+joblib==1.4.2
+jsonpatch==1.33
+jsonpointer==3.0.0
+jsonschema==4.23.0
+jsonschema-specifications==2023.12.1
+jupyter_client==8.6.2
+jupyter_core==5.7.2
+kiwisolver==1.4.7
+kubernetes==30.1.0
+langchain==0.3.0
+langchain-chroma==0.1.4
+langchain-community==0.3.0
+langchain-core==0.3.5
+langchain-huggingface==0.1.0
+langchain-openai==0.2.0
+langchain-text-splitters==0.3.0
+langsmith==0.1.126
+langgraph
+markdown-it-py==3.0.0
+MarkupSafe==2.1.5
+marshmallow==3.22.0
+matplotlib==3.9.2
+matplotlib-inline==0.1.7
+mdurl==0.1.2
+mixedbread-ai==2.2.6
+mmh3==4.1.0
+monotonic==1.6
+mpmath==1.3.0
+multidict==6.0.5
+multiprocess==0.70.16
+mypy-extensions==1.0.0
+narwhals==1.6.0
+nest-asyncio==1.6.0
+networkx==3.3
+numpy==1.26.4
+oauthlib==3.2.2
+onnxruntime==1.19.0
+openai==1.43.0
+opentelemetry-api==1.27.0
+opentelemetry-exporter-otlp-proto-common==1.27.0
+opentelemetry-exporter-otlp-proto-grpc==1.27.0
+opentelemetry-instrumentation==0.48b0
+opentelemetry-instrumentation-asgi==0.48b0
+opentelemetry-instrumentation-fastapi==0.48b0
+opentelemetry-proto==1.27.0
+opentelemetry-sdk==1.27.0
+opentelemetry-semantic-conventions==0.48b0
+opentelemetry-util-http==0.48b0
+orjson==3.10.7
+overrides==7.7.0
+packaging==24.1
+pandas==2.2.2
+parso==0.8.4
+pillow==10.4.0
+platformdirs==4.3.2
+posthog==3.6.0
+prompt_toolkit==3.0.47
+protobuf==4.25.4
+psutil==6.0.0
+pure_eval==0.2.3
+pyarrow==17.0.0
+pyasn1==0.6.0
+pyasn1_modules==0.4.0
+pydantic==2.8.2
+pydantic-settings==2.5.2
+pydantic_core==2.20.1
+pydeck==0.9.1
+pydub==0.25.1
+Pygments==2.18.0
+pyparsing==3.1.4
+pypdf==4.3.1
+PyPika==0.48.9
+pyproject_hooks==1.1.0
+pyreadline3==3.4.1
+python-dateutil==2.9.0.post0
+python-dotenv==1.0.1
+python-multipart==0.0.10
+pytz==2024.1
+PyYAML==6.0.2
+pyzmq==26.2.0
+referencing==0.35.1
+regex==2024.7.24
+requests==2.32.3
+requests-oauthlib==2.0.0
+rich==13.8.0
+rpds-py==0.20.0
+rsa==4.9
+ruff==0.6.7
+safetensors==0.4.4
+scikit-learn==1.5.2
+scipy==1.14.1
+semantic-version==2.10.0
+sentence-transformers==3.1.1
+sentencepiece==0.2.0
+setuptools==72.1.0
+shellingham==1.5.4
+six==1.16.0
+smmap==5.0.1
+sniffio==1.3.1
+SQLAlchemy==2.0.32
+stack-data==0.6.3
+starlette==0.38.4
+sympy==1.13.2
+tenacity==8.5.0
+threadpoolctl==3.5.0
+tiktoken==0.7.0
+tokenizers==0.19.1
+toml==0.10.2
+tomlkit==0.12.0
+torch==2.4.0
+tornado==6.4.1
+tqdm==4.66.5
+traitlets==5.14.3
+transformers==4.44.2
+typer==0.12.5
+typing-inspect==0.9.0
+typing_extensions==4.12.2
+tzdata==2024.1
+urllib3==2.2.2
+uvicorn==0.30.6
+watchdog==4.0.2
+watchfiles==0.24.0
+wcwidth==0.2.13
+websocket-client==1.8.0
+websockets==12.0
+wheel==0.43.0
+wrapt==1.16.0
+xxhash==3.5.0
+yarl==1.9.7
+zipp==3.20.1

tools.py

@@ -0,0 +1,134 @@
+from langchain_community.tools import TavilySearchResults
+
+from langchain_core.retrievers import BaseRetriever
+from langchain_core.callbacks import CallbackManagerForRetrieverRun
+from langchain_core.vectorstores import VectorStoreRetriever
+from langgraph.prebuilt import create_react_agent
+from langchain_core.documents import Document
+from langchain_openai import ChatOpenAI
+from langgraph.checkpoint.memory import MemorySaver
+from mixedbread_ai.client import MixedbreadAI
+from langchain.chains import create_retrieval_chain
+from langchain.chains.combine_documents import create_stuff_documents_chain
+from langchain.prompts import ChatPromptTemplate
+from dotenv import load_dotenv
+import os
+from langchain_chroma import Chroma
+import chromadb
+from typing import List
+from datasets import load_dataset
+from langchain_huggingface import HuggingFaceEmbeddings 
+from tqdm import tqdm
+from datetime import datetime
+
+load_dotenv()
+# Global params
+OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY")
+MODEL_EMB = "mxbai-embed-large"
+MODEL_RRK = "mixedbread-ai/mxbai-rerank-large-v1"
+LLM_NAME = "gpt-4o-mini"
+OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY")
+MXBAI_API_KEY = os.environ.get("MXBAI_API_KEY")
+HF_TOKEN = os.environ.get("HF_TOKEN")
+HF_API_KEY = os.environ.get("HF_API_KEY")
+
+# MixedbreadAI Client
+mxbai_client = MixedbreadAI(api_key=MXBAI_API_KEY)
+model_emb = "mixedbread-ai/mxbai-embed-large-v1"  
+
+# # Set up ChromaDB
+memoires_ds = load_dataset("DATALTIST/memoires_vec_800", split="data", token=HF_TOKEN, streaming=True)
+batched_ds = memoires_ds.batch(batch_size=41000)
+client = chromadb.Client()
+collection = client.get_or_create_collection(name="embeddings_mxbai") 
+for batch in tqdm(batched_ds, desc="Processing dataset batches"): 
+    collection.add(
+        ids=batch["id"],
+        metadatas=batch["metadata"],
+        documents=batch["document"],
+        embeddings=batch["embedding"],      
+    )
+print(f"Collection complete: {collection.count()}")
+del memoires_ds, batched_ds
+
+llm_4o = ChatOpenAI(model="gpt-4o-mini", api_key=OPENAI_API_KEY, temperature=0)
+
+    
+
+def init_rag_tool():
+    """Init tools to allow an LLM to query the documents"""
+    # client = chromadb.PersistentClient(path=CHROMA_PATH)
+    db = Chroma(
+        client=client,
+        collection_name=f"embeddings_mxbai",
+        embedding_function = HuggingFaceEmbeddings(model_name=model_emb) 
+    )
+    # Reranker class
+    class Reranker(BaseRetriever):
+        retriever: VectorStoreRetriever
+        # model: CrossEncoder
+        k: int
+
+        def _get_relevant_documents(
+            self, query: str, *, run_manager: CallbackManagerForRetrieverRun
+        ) -> List[Document]:
+            docs = self.retriever.invoke(query)
+            results = mxbai_client.reranking(model=MODEL_RRK, query=query, input=[doc.page_content for doc in docs], return_input=True, top_k=self.k)
+            return [Document(page_content=res.input) for res in results.data]
+
+    # Set up reranker + LLM
+    retriever = db.as_retriever(search_type="similarity", search_kwargs={"k": 25})
+    reranker = Reranker(retriever=retriever, k=4)  #Reranker(retriever=retriever, model=model, k=4)
+    llm = ChatOpenAI(model=LLM_NAME, verbose=True) 
+
+
+    system_prompt = (
+    "Réponds à la question en te basant uniquement sur le contexte suivant: \n\n {context}"
+    "Si tu ne connais pas la réponse, dis que tu ne sais pas."
+    )
+
+    prompt = ChatPromptTemplate.from_messages(
+        [
+            ("system", system_prompt),
+            ("human", "{input}"),
+        ]
+    )
+
+    question_answer_chain = create_stuff_documents_chain(llm, prompt)
+    rag_chain = create_retrieval_chain(reranker, question_answer_chain)
+
+    rag_tool = rag_chain.as_tool(
+        name="RAG_search",
+        description="Recherche d'information dans les mémoires d'actuariat",
+        arg_types={"input": str},
+    )
+    return rag_tool
+
+
+def init_websearch_tool():
+    web_search_tool = TavilySearchResults(
+        name="Web_search",
+        max_results=5,
+        description="Recherche d'informations sur le web",
+        search_depth="advanced",
+        include_answer=True,
+        include_raw_content=True,
+        include_images=False,
+        verbose=False,
+    )
+    return web_search_tool
+
+
+def create_agent():
+    rag_tool = init_rag_tool()
+    web_search_tool = init_websearch_tool()
+    memory = MemorySaver()  
+    llm_4o = ChatOpenAI(model="gpt-4o-mini", api_key=OPENAI_API_KEY, verbose=True, temperature=0, streaming=True)
+    tools = [rag_tool, web_search_tool]
+    system_message = """
+        Tu es un assistant dont la fonction est de répondre à des questions à propos de l'assurance et de l'actuariat.
+        Utilise les outils RAG_search ou Web_search pour répondre aux questions de l'utilisateur.
+    """    # Dans la réponse finale, sépare les informations de l'outil RAG et de l'outil Web. 
+
+    react_agent = create_react_agent(llm_4o, tools, state_modifier=system_message, checkpointer=memory, debug=False)
+    return react_agent