Update app.py

app.py

@@ -1,14 +1,22 @@
 import torch
+import os
 import requests
 from fastapi import FastAPI
 from transformers import pipeline
 
+# 🔱 CPU Core Management: Stop 99% CPU Usage
+# HF Free Space එකක සාමාන්‍යයෙන් CPU Cores 2ක් තියෙන නිසා අපි 2කට සීමා කරමු
+os.environ["OMP_NUM_THREADS"] = "2"
+os.environ["MKL_NUM_THREADS"] = "2"
+torch.set_num_threads(2)
+
 main = FastAPI()
 
-# 🔱 Inachi Core Configuration
+# 🔱 Inachi Identity Settings
 MODEL_ID = "google/gemma-3-1b-it"
 
-# Load the model with CPU optimizations
+# 🔱 Optimized Pipeline
+# bfloat16 පාවිච්චි කිරීමෙන් RAM සහ CPU මතකය ඉතිරි වේ
 pipe = pipeline(
     "text-generation",
     model=MODEL_ID,
@@ -17,35 +25,35 @@ pipe = pipeline(
     trust_remote_code=True
 )
 
-# 🔱 Simple Web Search Tool (DuckDuckGo API - No Key Required)
 def web_search(query):
     try:
+        # Simple DuckDuckGo API for search context
         url = f"https://api.duckduckgo.com/?q={query}&format=json"
-        response = requests.get(url).json()
-        return response.get("AbstractText", "No specific web data found.")
+        response = requests.get(url, timeout=5).json()
+        return response.get("AbstractText", "No specific data found.")
     except:
-        return "Search failed."
+        return "Search unavailable."
 
 @main.post("/v1/chat")
 async def chat(data: dict):
     user_query = data.get("message", "")
     
-    # 🔱 Inachi Identity & System Instruction
-    # මෙතනින් තමයි එයාට තමන් කවුද කියලා කියලා දෙන්නේ
+    # 🔱 System Identity: Developed by Inachi Team
     system_prompt = (
         "You are Inachi AI, a highly advanced assistant developed by the Inachi Team. "
-        "Your goal is to provide technical, precise, and helpful information. "
-        "Always identify yourself as Inachi AI when asked."
+        "You are an expert in system architecture and web development. "
+        "Always identify as Inachi AI."
     )
     
-    # Web search check (සරලව search කරන්න ඕනෙද කියලා බලනවා)
+    # Search logic
     search_context = ""
-    if "search" in user_query.lower() or "latest" in user_query.lower():
-        search_context = f"\nWeb Search Result: {web_search(user_query)}"
+    if "search" in user_query.lower():
+        search_context = f"\nWeb Context: {web_search(user_query)}"
 
-    # Prompt එක සැකසීම
-    full_prompt = f"{system_prompt}\nContext: {search_context}\nUser: {user_query}\nInachi AI:"
+    # Prompt construction
+    full_prompt = f"{system_prompt}\n{search_context}\nUser: {user_query}\nInachi AI:"
     
+    # 🔱 Inference with limited tokens for speed
     results = pipe(
         full_prompt, 
         max_new_tokens=512, 
@@ -54,10 +62,10 @@ async def chat(data: dict):
         top_p=0.9
     )
     
-    # පිරිසිදු පිළිතුර ලබා ගැනීම
     reply = results[0]['generated_text'].split("Inachi AI:")[-1].strip()
     return {"reply": reply}
 
 if __name__ == "__main__":
     import uvicorn
+    # HF Spaces uses port 7860 by default
     uvicorn.run(main, host="0.0.0.0", port=7860)