Add GPT-3 and Llama 2 7b chatbot endpoints

README.md

@@ -4,6 +4,10 @@
 This course provides a comprehensive guide to designing and developing a backend for an AI chatbot using Python. Participants will learn how to create a robust LLM backend, deploy it with modern CI/CD pipelines and GitHub, and monitor its performance using Microsoft Azure. The course is targeted towards software engineers and developers who want to build an LLM-powered chatbot for their applications. Basic understanding of Python programming is required, and previous experience in software development is a plus. Participants will need an Azure subscription, GitHub account, OpenAI API subscription, Visual Studio Code, Python version 3.8 or higher, and Pip installed. The course covers various topics, including the fundamentals of AI and large-language models, utilizing FastAPI for building an LLM backend, setting up development environments, containerization with Docker, implementing CI/CD pipelines with GitHub Actions, and monitoring AI applications using status pages. Upon completion of the course, participants will have gained the necessary skills to develop and deploy a powerful LLM backend for their AI chatbot projects.
 
 ## Getting started
+#### Prerequisites
+- Python 3.8 or higher
+- Docker Installed
+
 #### Installation
 ```
 pip3 install -r requirements.txt

src/ai/__init__.py

@@ -1,13 +1,18 @@
 import os
 from openai import OpenAI
+import requests
+
+API_TOKEN = os.getenv("HUGGINGFACE_API_TOKEN")
+
+API_URL = "https://api-inference.huggingface.co/models/meta-llama/Llama-2-7b-chat-hf"
+headers = {"Authorization": f"Bearer {API_TOKEN}"}
 
 client = OpenAI(
-    organization=os.getenv('OPENAI_ORG_ID'),
-    api_key=os.getenv('OPENAI_API_KEY')
+    organization=os.getenv("OPENAI_ORG_ID"), api_key=os.getenv("OPENAI_API_KEY")
 )
 
 
-def chatbot(user_request:str):
+def gpt_chatbot(user_request: str):
     completion = client.chat.completions.create(
         model="gpt-3.5-turbo",
         messages=[
@@ -17,3 +22,9 @@ def chatbot(user_request:str):
     )
 
     return completion.choices[0].message
+
+
+def llama_chatbot(user_request: str):
+    response = requests.post(API_URL, headers=headers, json={"inputs": user_request})
+
+    return response.json()

src/index/main.py

@@ -1,16 +1,16 @@
 from fastapi import FastAPI, Form
 from fastapi.middleware.cors import CORSMiddleware
-from ai import chatbot
+from ai import gpt_chatbot, llama_chatbot
 
-isProduction = True
+isProduction = False
 
 origins = ["*"]
 
 if isProduction:
     app = FastAPI(
         title="LLM API Endpoints",
-        docs_url=None,  # Disable docs (Swagger UI)
-        redoc_url=None,  # Disable redoc
+        docs_url=None,                  # Disable docs (Swagger UI)
+        redoc_url=None,                 # Disable redoc
     )
 else:
     app = FastAPI(title="LLM API Endpoints")
@@ -30,12 +30,22 @@ def index():
     return {"server ok": True}
 
 
-@app.post("/api/chat")
-async def chat(user_request: str = Form(...)):
+@app.post("/api/chat/gpt3", tags=["OpenAI GPT-3"])
+async def gpt_chat(user_request: str = Form(...)):
     """
-    Chat with LLM Backend
+    Chat with LLM Backend - GPT-3
     """
     # Get the text content in the user request
-    result = chatbot(user_request=user_request)
+    result = gpt_chatbot(user_request=user_request)
+
+    return {"result": result}
+
+@app.post("/api/chat/llama", tags=["Llama 2 7B Chat"])
+async def llama_chat(user_request: str = Form(...)):
+    """
+    Chat with LLM Backend - Llama 2 7b Chat
+    """
+    # Get the text content in the user request
+    result = llama_chatbot(user_request=user_request)
 
     return {"result": result}

src/test_main.py

@@ -8,7 +8,12 @@ def test_read_main():
     assert response.status_code == 200
     assert response.json() == {"server ok": True}
 
-def test_chat():
-    response = client.post("/api/chat", data={"user_request": "What is OpenAI?"})
+def test_gpt_chat():
+    response = client.post("/api/chat/gpt3", data={"user_request": "What is OpenAI?"})
+    assert response.status_code == 200
+    assert response.json()["result"] != ""
+
+def test_llama_chat():
+    response = client.post("/api/chat/llama", data={"user_request": "What is LLM?"})
     assert response.status_code == 200
     assert response.json()["result"] != ""