Update app.py

app.py

@@ -1,32 +1,60 @@
 import gradio as gr
 import torch
 from transformers import T5ForConditionalGeneration, T5Tokenizer
+import os
 
-# 1. Load the model and tokenizer explicitly
-model_id = "amielitos/text-to-markdown-t5" # Replace with your actual username/model
-tokenizer = T5Tokenizer.from_pretrained(model_id)
-model = T5ForConditionalGeneration.from_pretrained(model_id)
+# 1. Configuration - MAKE SURE THIS IS CORRECT
+# Replace 'your-username/text-to-markdown-t5' with your actual repo name
+model_id = "amielitos/text-to-markdown-t5" 
+
+print(f"--- Loading model: {model_id} ---")
+
+try:
+    tokenizer = T5Tokenizer.from_pretrained(model_id, legacy=False)
+    model = T5ForConditionalGeneration.from_pretrained(model_id)
+    print("--- Model and Tokenizer loaded successfully! ---")
+except Exception as e:
+    print(f"--- ERROR LOADING MODEL: {str(e)} ---")
 
 def convert_to_markdown(input_text):
-    # 2. T5 requires the prefix we used in training
-    full_input = f"markdown: {input_text}"
-    
-    # 3. Tokenize input
-    inputs = tokenizer(full_input, return_tensors="pt", max_length=512, truncation=True)
+    print(f"--- Received input: {input_text[:50]}... ---")
     
-    # 4. Generate output
-    with torch.no_grad():
-        outputs = model.generate(inputs.input_ids, max_length=512)
-    
-    # 5. Decode back to text
-    return tokenizer.decode(outputs[0], skip_special_tokens=True)
+    if not input_text.strip():
+        return "Please enter some text."
+
+    try:
+        # T5 prefix logic
+        full_input = f"markdown: {input_text}"
+        
+        # Tokenize
+        inputs = tokenizer(full_input, return_tensors="pt", max_length=512, truncation=True)
+        print("--- Tokenization complete ---")
+        
+        # Generate
+        with torch.no_grad():
+            outputs = model.generate(
+                inputs.input_ids, 
+                max_length=512, 
+                num_beams=2, 
+                early_stopping=True
+            )
+        print("--- Generation complete ---")
+        
+        # Decode
+        result = tokenizer.decode(outputs[0], skip_special_tokens=True)
+        print(f"--- Decoded result: {result[:50]}... ---")
+        return result
+
+    except Exception as e:
+        print(f"--- ERROR DURING CONVERSION: {str(e)} ---")
+        return f"An error occurred: {str(e)}"
 
-# 6. Create the Interface
+# Interface
 demo = gr.Interface(
     fn=convert_to_markdown, 
-    inputs=gr.Textbox(lines=5, label="Plain Text Input", placeholder="Type your text here..."), 
-    outputs=gr.Markdown(label="Markdown Result"),
-    title="AI Text to Markdown Converter"
+    inputs=gr.Textbox(lines=5, label="Plain Text Input"), 
+    outputs=gr.Textbox(label="Result"), # Changed to Textbox temporarily for debugging
+    title="Text to Markdown Debugger"
 )
 
 if __name__ == "__main__":