Update app.py

app.py

@@ -58,7 +58,7 @@ from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStream
 # ----------------------------------------------------------------------
 # 1. Setup Model & Tokenizer (Using 8-bit Quantization from Script 2)
 # ----------------------------------------------------------------------
-model_name = 'Tesslate/UIGEN-T3-4B' # Or use Tesslate/UIGEN-T2-7B-3600 if preferred
+model_name = 'Tesslate/UIGEN-T3-4B-Preview' # Or use Tesslate/UIGEN-T2-7B-3600 if preferred
 use_thread = True  # For streaming UI updates
 
 # Determine optimal attention implementation
@@ -135,11 +135,11 @@ logger = logging.getLogger(__name__)
 # 2. Generation Parameter Setup (Values can be adjusted)
 # ----------------------------------------------------------------------
 # Using similar defaults as Script 2, but can be tuned
-MAX_NEW_TOKENS = 12000 # Adjusted back, but can be increased if needed
-TEMPERATURE = 0.7
-TOP_P = 0.9 # Slightly lower than Script 2 default
-TOP_K = 40 # Added based on Script 2 defaults
-REPETITION_PENALTY = 1.1 # Based on Script 2 defaults
+MAX_NEW_TOKENS = 20000 # Adjusted back, but can be increased if needed
+TEMPERATURE = 0.6
+TOP_P = 0.95 # Slightly lower than Script 2 default
+TOP_K = 20 # Added based on Script 2 defaults
+REPETITION_PENALTY = 1.0 # Based on Script 2 defaults
 
 # Note: initialize_gen_kwargs function is removed as params are used directly now