Upload app.py

app.py

@@ -4,25 +4,30 @@ import numpy as np
 import gradio as gr
 from transformers import AutoTokenizer, AutoModelForSequenceClassification
 
-# Change to your Hugging Face model repo ID
+# 🔹 Change this to your actual Hugging Face model repo ID
 MODEL_ID = "SharvNey/capstone_project"
 
-tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
-model = AutoModelForSequenceClassification.from_pretrained(MODEL_ID)
+# Load model & tokenizer
+tokenizer = AutoTokenizer.from_pretrained(MODEL_ID, use_auth_token=os.getenv("HF_TOKEN"))
+model = AutoModelForSequenceClassification.from_pretrained(MODEL_ID, use_auth_token=os.getenv("HF_TOKEN"))
 device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 model.to(device)
 model.eval()
 
+# Prediction function
 def classify_text(text):
     if not text.strip():
         return {"🧑 Human-Written": 0.0, "🤖 AI-Generated": 0.0}
+
     enc = tokenizer(text, truncation=True, padding=True, max_length=256, return_tensors="pt")
-    enc = {k: v.to(device) for k,v in enc.items()}
+    enc = {k: v.to(device) for k, v in enc.items()}
     with torch.no_grad():
         out = model(**enc)
         probs = torch.nn.functional.softmax(out.logits, dim=-1).cpu().numpy()[0]
+
     return {"🧑 Human-Written": float(probs[0]), "🤖 AI-Generated": float(probs[1])}
 
+# Gradio app
 demo = gr.Interface(
     fn=classify_text,
     inputs=gr.Textbox(lines=8, placeholder="Paste text here..."),
@@ -32,4 +37,12 @@ demo = gr.Interface(
 )
 
 if __name__ == "__main__":
-    demo.launch(share=True)
+    # Detect if running on Hugging Face Spaces
+    in_spaces = os.environ.get("SYSTEM") == "spaces"
+
+    if in_spaces:
+        # Spaces handles public link, no share=True needed
+        demo.launch(server_name="0.0.0.0", server_port=7860)
+    else:
+        # Local/Colab → provide share=True for public link
+        demo.launch(share=True)