Update app.py

app.py

@@ -16,9 +16,9 @@ import PyPDF2
 import io
 import traceback
 
-from git import Repo
+#from git import Repo
 
-Repo.clone_from("https://huggingface.co/unsloth/Llama-3.2-3B-bnb-4bit", "./local_model_dir")
+#Repo.clone_from("https://huggingface.co/unsloth/Llama-3.2-3B-bnb-4bit", "./local_model_dir")
 
 
 
@@ -41,6 +41,24 @@ model = None
 tokenizer = None
 generation_config = None
 
+def test_llm_generation():
+    try:
+        test_prompt = "Hello, how are you today?"
+        inputs = tokenizer(test_prompt, return_tensors="pt").to(model.device)
+        with torch.no_grad():
+            outputs = model.generate(
+                **inputs,
+                max_new_tokens=10,
+                do_sample=False,
+                pad_token_id=tokenizer.pad_token_id,
+                eos_token_id=tokenizer.eos_token_id
+            )
+        result = tokenizer.decode(outputs[0], skip_special_tokens=True)
+        add_log(f"🧪 Test LLM response: {result[:100]}")
+    except Exception as e:
+        add_log(f"❌ LLM quick test failed: {e}")
+
+
 def initialize_model():
     """Separate model initialization with better error handling"""
     global model, tokenizer, generation_config
@@ -695,6 +713,12 @@ def generate_podcast_gradio(input_text, input_file, language, speaker1, speaker2
         return None, "\n".join(logs)
 
 def create_interface():
+    
+    model_loaded = initialize_model()
+    if model_loaded:
+        test_llm_generation()
+    
+    
     """Create the Gradio interface"""
     language_options = [
         "Auto Detect", "English", "German", "French", "Spanish", "Italian",