Switch AI provider to Hugging Face router

README.md

@@ -15,7 +15,7 @@ FastAPI backend for Hadhramout Bank AI customer service system.
 
 ## Features
 - Telegram webhook integration
-- AI-powered responses using Groq
+- AI-powered responses using Hugging Face Inference API
 - Database integration with Supabase
 - Vector search with Pinecone
 

ai_service.py

@@ -1,5 +1,5 @@
 import re
-from config import pc, index, groq_client, EMBED_MODEL, GROQ_MODEL, PROMPT
+from config import pc, index, EMBED_MODEL, HF_MODEL, PROMPT, hf_client
 from database import db_manager
 
 def clean_ai_response(text: str):
@@ -7,8 +7,8 @@ def clean_ai_response(text: str):
     return cleaned_text.strip()
 
 async def get_ai_response(user_query: str, telegram_id: int = None):
-    
-    if not pc or not index or not groq_client:
+
+    if not pc or not index or not hf_client:
         return "Ai service is not available at the moment. Please try again later."
 
     # Save user message if database is available and telegram_id is provided
@@ -16,8 +16,7 @@ async def get_ai_response(user_query: str, telegram_id: int = None):
     if telegram_id and db_manager:
         db_manager.save_message(telegram_id, user_query, "user")
         conversation_history = db_manager.get_formatted_history(telegram_id, limit=6)        
-    
-    
+
     query_embedding = pc.inference.embed(
         model=EMBED_MODEL,
         inputs=[user_query],
@@ -48,19 +47,21 @@ async def get_ai_response(user_query: str, telegram_id: int = None):
         Based on the above information, provide an accurate and helpful response to the customer:
     """
     print("User content:", user_content)
-    completion = groq_client.chat.completions.create(
+
+    completion = hf_client.chat.completions.create(
+        model=HF_MODEL,
         messages=[
             {"role": "system", "content": PROMPT},
-            {"role": "user", "content": user_content}
+            {"role": "user", "content": user_content},
         ],
-        model=GROQ_MODEL,
         temperature=0.1,
-        max_completion_tokens=800,
+        max_tokens=800,
         top_p=0.9,
     )
+
     ai_response = completion.choices[0].message.content
     cleaned_response = clean_ai_response(ai_response)
-    
+
     # Save assistant response if database is available and telegram_id is provided
     if telegram_id and db_manager:
         db_manager.save_message(telegram_id, cleaned_response, "assistant")

config.py

@@ -1,6 +1,6 @@
 import os
 from pinecone import Pinecone
-from groq import Groq
+from openai import OpenAI
 from dotenv import load_dotenv
 
 # Load environment variables from .env file
@@ -8,7 +8,7 @@ load_dotenv()
 
 # Environment Variables
 PINECONE_API_KEY = os.environ.get("PINECONE_API_KEY")
-GROQ_API_KEY = os.environ.get("GROQ_API_KEY")
+HF_TOKEN = os.environ.get("HF_TOKEN") or os.environ.get("HF_API_KEY")
 TELEGRAM_TOKEN = os.environ.get("TELEGRAM_TOKEN")
 SUPABASE_URL = os.environ.get("SUPABASE_URL")
 SUPABASE_KEY = os.environ.get("SUPABASE_KEY")
@@ -18,21 +18,33 @@ SUPABASE_KEY = os.environ.get("SUPABASE_KEY")
 TELEGRAM_URL = f"https://149.154.167.220/bot{TELEGRAM_TOKEN}/sendMessage" if TELEGRAM_TOKEN else None
 
 EMBED_MODEL = os.environ.get("EMBED_MODEL", "multilingual-e5-large")
-GROQ_MODEL = os.environ.get("GROQ_MODEL", "llama-3.1-8b-instant")
-PROMPT = os.environ.get("PROMPT", "You are a helpful customer service assistant for Hadhramout Bank. Answer the user's question based on the provided context. If the context doesn't contain the answer, politely say you don't have enough information to help with that specific query.")
+HF_MODEL = os.environ.get(
+    "HF_MODEL",
+    "dphn/Dolphin-Mistral-24B-Venice-Edition:featherless-ai",
+)
+PROMPT = os.environ.get(
+    "PROMPT",
+    "You are a helpful customer service assistant for Hadhramout Bank. "
+    "Answer the user's question based on the provided context. If the context "
+    "doesn't contain the answer, politely say you don't have enough information "
+    "to help with that specific query."
+)
 
 # Initialize clients only if API keys are available
 pc = None
 if PINECONE_API_KEY:
     pc = Pinecone(api_key=PINECONE_API_KEY)
 
-groq_client = None
-if GROQ_API_KEY:
+hf_client = None
+if HF_TOKEN:
     try:
-        groq_client = Groq(api_key=GROQ_API_KEY)
+        hf_client = OpenAI(
+            base_url="https://router.huggingface.co/v1",
+            api_key=HF_TOKEN,
+        )
     except Exception as e:
-        print(f"Warning: Failed to initialize Groq client: {e}")
-        groq_client = None
+        print(f"Warning: Failed to initialize Hugging Face OpenAI client: {e}")
+        hf_client = None
 
 # Initialize index only if Pinecone client is available
 index = None

requirements.txt

@@ -1,7 +1,7 @@
 fastapi
 uvicorn
 pinecone
-groq
 httpx
 python-dotenv
 supabase
+openai