Update app.py

app.py

@@ -1,27 +1,48 @@
 import gradio as gr
-from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
+import requests
+import json
+import os
 
-# Load the model
-model_name = "microsoft/phi-3-mini-4k-instruct"
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForCausalLM.from_pretrained(model_name)
+# Set your API key as an environment variable
+# Important: In the Hugging Face Space, set this as a secret
+HF_API_KEY = os.environ.get("HF_API_KEY", "")
 
 def analyze_shampoo(prompt):
-    # Set up the model pipeline
-    pipe = pipeline(
-        "text-generation",
-        model=model,
-        tokenizer=tokenizer,
-        max_new_tokens=2048,
-        temperature=0.7,
-        top_p=0.95
-    )
+    """
+    Send the prompt to the Hugging Face Inference API instead of loading the model locally
+    """
+    API_URL = "https://api-inference.huggingface.co/models/microsoft/phi-3-mini-4k-instruct"
+    headers = {
+        "Authorization": f"Bearer {HF_API_KEY}",
+        "Content-Type": "application/json"
+    }
     
-    # Generate the response
-    response = pipe(prompt)[0]["generated_text"]
+    payload = {
+        "inputs": prompt,
+        "parameters": {
+            "max_new_tokens": 2048,
+            "temperature": 0.7,
+            "top_p": 0.95,
+            "return_full_text": False
+        }
+    }
     
-    # Return just the generated part (not the input prompt)
-    return response[len(prompt):]
+    try:
+        if not HF_API_KEY:
+            return "Error: No API key provided. Please add the HF_API_KEY secret to your Space."
+            
+        response = requests.post(API_URL, headers=headers, json=payload)
+        response.raise_for_status()  # Raise an exception for HTTP errors
+        
+        result = response.json()
+        if isinstance(result, list) and len(result) > 0:
+            # Extract the generated text
+            return result[0].get('generated_text', 'No response generated')
+        else:
+            return f"Unexpected response format: {result}"
+            
+    except Exception as e:
+        return f"Error: {str(e)}"
 
 # Create the Gradio interface
 interface = gr.Interface(
@@ -29,7 +50,7 @@ interface = gr.Interface(
     inputs=gr.Textbox(lines=10, placeholder="Enter your analysis prompt here..."),
     outputs=gr.Textbox(lines=20),
     title="Phi-3 Shampoo Analyzer",
-    description="Analyze shampoo ingredients based on user profiles"
+    description="Analyze shampoo ingredients based on user profiles using the Hugging Face Inference API"
 )
 
 interface.launch()