feat: ✨ Learning path ai chatbot demos (#99)

* conf: ⬆️ LangChain version

* fix: 🐛 RAG with simple files

* doc: 📝 Add some documentations

* feat: 🧠 Create Python memory use case

* feat: 🔢 Add steps for learning path demo

* clean: 🗑️ Remove unused import

* clean: 🗑️ Remove code

* fix: 🐛 reorder demo

* feat: ✨ update after learning path

Author: philippart-s

ai/ai-endpoints/python-langchain-chatbot/README.md

@@ -1,10 +1,20 @@
 This project illustrate how to use[LangChain](https://python.langchain.com/v0.2/) Python Framework with [AI Endpoints](https://endpoints.ai.cloud.ovh.net/).
 
+### 🧰 Pre requisites 🧰
+
+  - Python 3.12.x installed
+  - AI Endpoints API token
+  - model to use: any of the LLM instruct models
+  - have the following environment variables created:
+    - OVH_AI_ENDPOINTS_ACCESS_TOKEN: the API token, see [documentation](https://help.ovhcloud.com/csm/en-gb-public-cloud-ai-endpoints-getting-started?id=kb_article_view&sysparm_article=KB0065401#generating-your-first-api-access-key) to know how to generate it
+    - OVH_AI_ENDPOINTS_MODEL_URL: URL of the model, see [AI Endpoints website](https://endpoints.ai.cloud.ovh.net/) to know how to get it.
+    - OVH_AI_ENDPOINTS_MODEL_NAME: model name, see [AI Endpoints website](https://endpoints.ai.cloud.ovh.net/) to know how to get it.
+    - OVH_AI_ENDPOINTS_EMBEDDING_MODEL_NAME: embedding model name, see [AI Endpoints website](https://endpoints.ai.cloud.ovh.net/) to know how to get it.
+
 ## How to use the project
 
-  - set the `OVH_AI_ENDPOINTS_ACCESS_TOKEN` environment variable with your API token (see https://endpoints.ai.cloud.ovh.net/)
-  - install lib magic: `brew install libmagic`
-  - install the required dependencies: `pip3 install -r requirements.txt`
-  - to use the blocking chatbot: `python3 chat-bot.py --question "What is OVHcloud?"`
-  - to use the streaming chatbot: `python3 chat-bot-streaming.py --question "What is OVHcloud?"`
-  - to use the RAG chatbot: `python3 chat-bot-streaming-rag.py --question "What is AI Endpoints?"`
+  - install the required dependencies: `pip install -r requirements.txt`
+  - to use the blocking chatbot: `python chatbot.py`
+  - to use the memory chatbot: `python chatbot-memory.py`
+  - to use the streaming chatbot: `python chatbot-streaming.py`
+  - to use the RAG chatbot: `python chatbot-streaming-rag.py`

ai/ai-endpoints/python-langchain-chatbot/chat-bot-streaming-rag.py

@@ -0,0 +1,46 @@
+import os
+
+from langchain_mistralai import ChatMistralAI
+from langchain.chains import ConversationChain
+from langchain.memory import ConversationBufferWindowMemory
+from langchain_core.prompts import ChatPromptTemplate
+from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
+
+# Load environment variables: see https://endpoints.ai.cloud.ovh.net/ for more information 
+_OVH_AI_ENDPOINTS_ACCESS_TOKEN = os.environ.get('OVH_AI_ENDPOINTS_ACCESS_TOKEN') 
+_OVH_AI_ENDPOINTS_MODEL_NAME = os.environ.get('OVH_AI_ENDPOINTS_MODEL_NAME') 
+_OVH_AI_ENDPOINTS_MODEL_URL = os.environ.get('OVH_AI_ENDPOINTS_MODEL_URL') 
+
+# Callback handler for streaming output
+streaming_handler = StreamingStdOutCallbackHandler()
+
+# Configure the used model: Mistral
+model = ChatMistralAI(model=_OVH_AI_ENDPOINTS_MODEL_NAME, 
+                        api_key=_OVH_AI_ENDPOINTS_ACCESS_TOKEN,
+                        endpoint=_OVH_AI_ENDPOINTS_MODEL_URL, 
+                        max_tokens=1500,
+                        streaming=True,
+                        callbacks=[streaming_handler])
+
+# Define the prompt template for the chatbot
+prompt = ChatPromptTemplate.from_messages([
+    ("system", "You are Nestor, a virtual assistant. Answer to the question. The conversation history is {history}"),
+    ("human", "{input}"),
+  ])
+
+# Set the size of the memory window to 10 exchanges.
+memory = ConversationBufferWindowMemory(k=10)
+
+# Create a chain that "apply" the prompt to the model.
+conversation = ConversationChain(llm=model, memory=memory, prompt=prompt)
+
+# Invoke the chatbot
+question = "Hello, my name is Stéphane"
+print(f"👤: {question}")
+print("🤖:")
+conversation.predict(input=question)
+
+question = "What is my name?"
+print(f"\n👤: {question}")
+print("🤖:")
+conversation.predict(input=question)

ai/ai-endpoints/python-langchain-chatbot/chat-bot-streaming.py

@@ -0,0 +1,56 @@
+import time
+import os
+
+from langchain import hub
+
+from langchain_mistralai import ChatMistralAI
+
+from langchain_chroma import Chroma
+
+from langchain_community.document_loaders import TextLoader
+from langchain_community.embeddings.ovhcloud import OVHCloudEmbeddings
+
+from langchain_core.runnables import RunnablePassthrough
+from langchain_text_splitters import RecursiveCharacterTextSplitter
+
+# Load environment variables: see https://endpoints.ai.cloud.ovh.net/ for more information 
+_OVH_AI_ENDPOINTS_ACCESS_TOKEN = os.environ.get('OVH_AI_ENDPOINTS_ACCESS_TOKEN') 
+_OVH_AI_ENDPOINTS_MODEL_NAME = os.environ.get('OVH_AI_ENDPOINTS_MODEL_NAME') 
+_OVH_AI_ENDPOINTS_MODEL_URL = os.environ.get('OVH_AI_ENDPOINTS_MODEL_URL') 
+_OVH_AI_ENDPOINTS_EMBEDDING_MODEL_NAME = os.environ.get('OVH_AI_ENDPOINTS_EMBEDDING_MODEL_NAME') 
+
+# Configure the used model: Mistral
+model = ChatMistralAI(model=_OVH_AI_ENDPOINTS_MODEL_NAME, 
+                      api_key=_OVH_AI_ENDPOINTS_ACCESS_TOKEN,
+                      endpoint=_OVH_AI_ENDPOINTS_MODEL_URL, 
+                      max_tokens=1500, 
+                      streaming=True)
+
+# Load and split the documents
+loader = TextLoader("./rag-files/content.txt")
+docs = loader.load()
+
+text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
+splits = text_splitter.split_documents(docs)
+
+# Create the vector thanks to OVHcloud embedding model
+vectorstore = Chroma.from_documents(documents=splits, 
+                                    embedding=OVHCloudEmbeddings(model_name=_OVH_AI_ENDPOINTS_EMBEDDING_MODEL_NAME, 
+                                                                  access_token=_OVH_AI_ENDPOINTS_ACCESS_TOKEN))
+
+# Define the prompt template for the chatbot
+prompt = hub.pull("rlm/rag-prompt")
+
+# Create a chain that "apply" the prompt to the model.
+rag_chain = (
+  {"context": vectorstore.as_retriever(), "question": RunnablePassthrough()} |
+  prompt
+  | model
+)
+
+# Invoke the chatbot
+print("👤: Which company created AI Endpoints?")
+print("🤖:")
+for r in rag_chain.stream({"question", "Which company created AI Endpoints?"}):
+  print(r.content, end="", flush=True)
+  time.sleep(0.150)

ai/ai-endpoints/python-langchain-chatbot/chat-bot.py

@@ -0,0 +1,33 @@
+import time
+import os
+
+from langchain_mistralai import ChatMistralAI
+from langchain_core.prompts import ChatPromptTemplate
+
+# Load environment variables: see https://endpoints.ai.cloud.ovh.net/ for more information 
+_OVH_AI_ENDPOINTS_ACCESS_TOKEN = os.environ.get('OVH_AI_ENDPOINTS_ACCESS_TOKEN') 
+_OVH_AI_ENDPOINTS_MODEL_NAME = os.environ.get('OVH_AI_ENDPOINTS_MODEL_NAME') 
+_OVH_AI_ENDPOINTS_MODEL_URL = os.environ.get('OVH_AI_ENDPOINTS_MODEL_URL') 
+
+# Configure the used model: Mistral
+model = ChatMistralAI(model=_OVH_AI_ENDPOINTS_MODEL_NAME, 
+                      api_key=_OVH_AI_ENDPOINTS_ACCESS_TOKEN,
+                      endpoint=_OVH_AI_ENDPOINTS_MODEL_URL, 
+                      max_tokens=1500, 
+                      streaming=True)
+
+# Define the prompt template for the chatbot
+prompt = ChatPromptTemplate.from_messages([
+  ("system", "You are a Nestor, a virtual assistant. Answer to the question."),
+  ("human", "{question}"),
+])
+
+# Create a chain that "apply" the prompt to the model.
+chain = prompt | model
+
+# Invoke the chatbot
+print("👤: What is OVHcloud?")
+print("🤖:")
+for r in chain.stream({"question", "What is OVHcloud?"}):
+  print(r.content, end="", flush=True)
+  time.sleep(0.150)

ai/ai-endpoints/python-langchain-chatbot/chatbot-memory.py

@@ -0,0 +1,30 @@
+import os
+
+from langchain_mistralai import ChatMistralAI
+from langchain_core.prompts import ChatPromptTemplate
+
+# Load environment variables: see https://endpoints.ai.cloud.ovh.net/ for more information 
+_OVH_AI_ENDPOINTS_ACCESS_TOKEN = os.environ.get('OVH_AI_ENDPOINTS_ACCESS_TOKEN') 
+_OVH_AI_ENDPOINTS_MODEL_NAME = os.environ.get('OVH_AI_ENDPOINTS_MODEL_NAME') 
+_OVH_AI_ENDPOINTS_MODEL_URL = os.environ.get('OVH_AI_ENDPOINTS_MODEL_URL') 
+
+# Configure the used model: Mistral
+model = ChatMistralAI(model=_OVH_AI_ENDPOINTS_MODEL_NAME, 
+                      api_key=_OVH_AI_ENDPOINTS_ACCESS_TOKEN,
+                      endpoint=_OVH_AI_ENDPOINTS_MODEL_URL, 
+                      max_tokens=1500)
+
+# Define the prompt template for the chatbot
+prompt = ChatPromptTemplate.from_messages([
+  ("system", "You are Nestor, a virtual assistant. Answer to the question."),
+  ("human", "{question}"),
+])
+
+# Create a chain that "apply" the prompt to the model. 
+chain = prompt | model
+
+# Invoke the chatbot
+response = chain.invoke("What is OVHcloud?")
+
+print("👤: What is OVHcloud?")
+print(f"🤖:{response.content}")