Update app.py

app.py

@@ -1,15 +1,19 @@
 import gradio as gr
 import pandas as pd
-from groq import Groq
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
 
-# Initialize the Groq client with your API key
-client = Groq(api_key="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")
+# Load the model and tokenizer
+model_name = "nvidia/Llama-3.1-Nemotron-70B-Instruct-HF"
+model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.bfloat16, device_map="auto")
+tokenizer = AutoTokenizer.from_pretrained(model_name)
 
 # Define the system message for the model
-system_message = {
-    "role": "system",
-    "content": "You are an experienced Fashion designer who starts conversation with proper greeting, giving valuable and catchy fashion advice and suggestions, stays to the point and precise, asks questions only if the user has any concerns over your provided suggestions."
-}
+system_message = (
+    "You are an experienced Fashion designer who starts conversation with proper greeting, "
+    "giving valuable and catchy fashion advice and suggestions, stays to the point and precise, "
+    "asks questions only if the user has any concerns over your provided suggestions."
+)
 
 # Function to reset the chat
 def reset_chat():
@@ -45,22 +49,16 @@ def chat(user_input, messages):
         # Append user message to the conversation history
         messages.append({"role": "user", "content": user_input})
 
+        # Prepare the input for the model
+        input_text = system_message + "\n" + "\n".join([f"{msg['role']}: {msg['content']}" for msg in messages])
+        
+        # Tokenize and encode the input text
+        inputs = tokenizer(input_text, return_tensors="pt").to(model.device)
+
         try:
-            # Generate a response from the Groq API
-            completion = client.chat.completions.create(
-                model="llama3-8b-8192",
-                messages=messages,
-                temperature=1,
-                max_tokens=1024,
-                top_p=1,
-                stream=False,
-            )
-
-            # Ensure response is valid
-            if completion.choices and len(completion.choices) > 0:
-                response_content = completion.choices[0].message.content
-            else:
-                response_content = "Sorry, I couldn't generate a response."
+            # Generate a response from the model
+            outputs = model.generate(**inputs, max_length=150, num_return_sequences=1, temperature=0.7)
+            response_content = tokenizer.decode(outputs[0], skip_special_tokens=True)
 
         except Exception as e:
             response_content = f"Error: {str(e)}"