Update app.py

app.py

@@ -91,40 +91,29 @@ def generate_text(prompt, max_length=256, temperature=0.7, top_p=0.9, top_k=150,
     
     start_time = time.time()
     
-    # Update progress bar - tokenization step
-    progress(0.1, desc="تحليل النص (Tokenizing input)")
-    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
-    
-    # Update progress bar - generation starting
-    progress(0.2, desc="توليد النص (Generating text)")
-    
-    # Define a callback function to update progress during generation
-    def generation_callback(beam_idx, token_idx, token_id, scores, generation_config):
-        # Estimate progress based on token index and max length
-        # We start at 20% and go to 90%, leaving room for post-processing
-        progress_value = 0.2 + 0.7 * min(token_idx / max_length, 1.0)
-        progress(progress_value, desc=f"توليد النص: {token_idx}/{max_length} (Generating: {token_idx}/{max_length})")
-        return False  # Continue generation
-    
-    # Generate with callback
-    output = model.generate(
-        **inputs, 
-        max_length=max_length, 
-        temperature=temperature, 
-        top_p=top_p, 
-        do_sample=True,
-        repetition_penalty=repetition_penalty,
-        num_beams=num_beams,
-        top_k=top_k,
-        early_stopping=True,
-        pad_token_id=tokenizer.pad_token_id,
-        eos_token_id=tokenizer.eos_token_id,
-        callback=generation_callback if hasattr(model, "generation_config") else None,
-    )
-    
-    # Update progress bar - decoding step
-    progress(0.9, desc="معالجة النتائج (Processing results)")
-    result = tokenizer.decode(output[0], skip_special_tokens=True)
+    # Simply use the progress.tqdm wrapper for automatic progress tracking
+    with progress.tqdm("توليد النص (Generating text)", total=1) as pbar:
+        # Tokenize input
+        inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+        
+        # Generate text
+        output = model.generate(
+            **inputs, 
+            max_length=max_length, 
+            temperature=temperature, 
+            top_p=top_p, 
+            do_sample=True,
+            repetition_penalty=repetition_penalty,
+            num_beams=num_beams,
+            top_k=top_k,
+            early_stopping=True,
+            pad_token_id=tokenizer.pad_token_id,
+            eos_token_id=tokenizer.eos_token_id,
+        )
+        
+        # Decode output
+        result = tokenizer.decode(output[0], skip_special_tokens=True)
+        pbar.update(1)
     
     # Update stats
     generation_time = time.time() - start_time
@@ -154,9 +143,6 @@ def generate_text(prompt, max_length=256, temperature=0.7, top_p=0.9, top_k=150,
         }
     )
     
-    # Complete progress
-    progress(1.0, desc="اكتمل (Complete)")
-    
     return result, f"تم توليد {token_count} رمز في {generation_time:.2f} ثانية (Generated {token_count} tokens in {generation_time:.2f} seconds)"
 
 def save_feedback(input, output, params) -> None: