Show loading bars

app.py

@@ -15,24 +15,29 @@ import spaces
 # Dictionary to store loaded models and tokenizers
 loaded_models = {}
 
-def load_model(model_name):
-    """Load the model and tokenizer if not already loaded."""
+def load_model(model_name, progress=gr.Progress()):
+    """Load the model and tokenizer with a progress bar."""
     if model_name not in loaded_models:
-        tokenizer = AutoTokenizer.from_pretrained(model_name)
+        access_token = os.getenv("HF_TOKEN")
+        progress(0, desc="Initializing model loading...")
+        tokenizer = AutoTokenizer.from_pretrained(model_name, use_auth_token=access_token)
+        progress(0.5, desc="Tokenizer loaded. Loading model...")
         model = AutoModelForCausalLM.from_pretrained(
-            model_name, torch_dtype=torch.float16, device_map="auto"
+            model_name, torch_dtype=torch.float16, device_map="auto", use_auth_token=access_token
         )
+        progress(1, desc="Model loaded successfully.")
         loaded_models[model_name] = (tokenizer, model)
     return loaded_models[model_name]
 
 @spaces.GPU
-def generate_text(model_name, prompt):
-    """Generate text using the selected model."""
-    tokenizer, model = load_model(model_name)
+def generate_text(model_name, prompt, progress=gr.Progress()):
+    """Generate text using the selected model with a loading indicator."""
+    tokenizer, model = load_model(model_name, progress)
     inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
     outputs = model.generate(**inputs, max_new_tokens=256)
     return tokenizer.decode(outputs[0], skip_special_tokens=True)
 
+
 # List of models to choose from
 model_choices = [
     "deepseek-ai/DeepSeek-R1-Distill-Llama-8B",
@@ -40,7 +45,6 @@ model_choices = [
     "google/gemma-7b"
 ]
 
-# Gradio interface setup
 with gr.Blocks() as demo:
     gr.Markdown("## Clinical Text Analysis with Multiple Models")
     model_selector = gr.Dropdown(choices=model_choices, label="Select Model")
@@ -51,3 +55,4 @@ with gr.Blocks() as demo:
     analyze_button.click(fn=generate_text, inputs=[model_selector, input_text], outputs=output_text)
 
 demo.launch()
+