Update app.py

app.py

@@ -2,22 +2,22 @@ import os
 import threading
 import discord
 import torch
-import gradio as gr
 from transformers import AutoModelForCausalLM, AutoTokenizer
 from dotenv import load_dotenv
 
-# Load environment variables (from Hugging Face Secrets and .env if available)
+# Load environment variables from Hugging Face Secrets and .env (if available)
 load_dotenv()
 DISCORD_TOKEN = os.getenv("DISCORD_TOKEN")
-HF_TOKEN = os.getenv("HF_TOKEN")  # Optional: only if needed for private models
+HF_TOKEN = os.getenv("HF_TOKEN")  # Optional: only needed if your model repository is private
 
 if not DISCORD_TOKEN:
     raise ValueError("Discord bot token is missing. Set DISCORD_TOKEN in the environment variables.")
 
-# Set the model repository name (public model)
+# Specify the model repository name. For DeepScaleR-1.5B-Preview use the official repo:
 MODEL_NAME = "agentica-org/DeepScaleR-1.5B-Preview"
 
-# Load the tokenizer and model. Use token=HF_TOKEN if provided.
+# Load the tokenizer and model.
+# If HF_TOKEN is provided, use it for authentication.
 if HF_TOKEN:
     tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, token=HF_TOKEN)
     model = AutoModelForCausalLM.from_pretrained(
@@ -29,29 +29,18 @@ else:
         MODEL_NAME, torch_dtype=torch.float16, device_map="auto"
     )
 
-# Function to generate AI responses
+# Define a function to generate responses using the model.
 def generate_response(prompt):
     device = "cuda" if torch.cuda.is_available() else "cpu"
     inputs = tokenizer(prompt, return_tensors="pt", truncation=True, max_length=1024).to(device)
     outputs = model.generate(**inputs, max_new_tokens=200, do_sample=True, top_p=0.9, temperature=0.7)
     response = tokenizer.decode(outputs[0], skip_special_tokens=True)
-    # Ensure the bot always identifies as "Shiv Yantra AI"
+    # Replace any instance of "DeepScaleR" with "Shiv Yantra AI" to enforce the bot's identity.
     response = response.replace("DeepScaleR", "Shiv Yantra AI")
     return response
 
 # ==========================
-# Gradio Interface (optional UI)
-# ==========================
-def gradio_api(input_text):
-    return generate_response(input_text)
-
-iface = gr.Interface(fn=gradio_api, inputs="text", outputs="text", title="Shiv Yantra AI")
-
-def run_gradio():
-    iface.launch(server_name="0.0.0.0", server_port=7860, share=False)
-
-# ==========================
-# Discord Bot Setup (Directly uses local generate_response)
+# Discord Bot Setup
 # ==========================
 intents = discord.Intents.default()
 intents.message_content = True  # Required for reading message content
@@ -63,16 +52,17 @@ async def on_ready():
 
 @client.event
 async def on_message(message):
-    # Avoid replying to itself
+    # Skip messages from the bot itself.
     if message.author == client.user:
         return
 
     user_input = message.content.strip()
     if user_input:
         try:
-            # Directly call the local generate_response function
+            # Directly call the local function instead of making an HTTP request.
             ai_response = generate_response(user_input)
         except Exception as e:
+            print(f"Error generating response: {e}")
             ai_response = "Error processing your request."
         await message.channel.send(ai_response)
 
@@ -80,14 +70,12 @@ def run_discord_bot():
     client.run(DISCORD_TOKEN)
 
 # ==========================
-# Start Both Services Concurrently
+# Start the Discord Bot
 # ==========================
 if __name__ == "__main__":
-    # Start the Gradio interface in a separate thread (optional UI)
-    threading.Thread(target=run_gradio, daemon=True).start()
-    # Start the Discord bot in a separate thread
+    # Run the Discord bot on a separate thread.
     threading.Thread(target=run_discord_bot, daemon=True).start()
 
-    # Keep the main thread alive indefinitely
+    # Keep the main thread alive.
     while True:
         pass