back to last promising version with treebuild crashlog, feel like sysyphus - trying to sort log formatting with 4o 2

app.py

@@ -21,13 +21,20 @@ from io import BytesIO
 import functools
 import logging
 
+
 # Set up custom logger
 custom_logger = logging.getLogger("custom_logger")
 custom_logger.setLevel(logging.INFO)
+
+# Prevent the root logger from duplicating messages
+custom_logger.propagate = False
+
+# Set up custom handler and formatter
 custom_handler = logging.StreamHandler()
 custom_handler.setFormatter(logging.Formatter('%(message)s'))
 custom_logger.addHandler(custom_handler)
 
+
 # Load environment variables
 load_dotenv()
 
@@ -220,12 +227,11 @@ def build_def_tree(input_ids, data, base_prompt, model, tokenizer, config, depth
     extended_prompt = current_prompt[len(base_prompt):].strip()
     extended_prompt = extended_prompt.replace("\n", "|")  # Replace \n with |
     
-    # Format the line to align "PROB:..." vertically
-    formatted_line = f"Depth {depth}: {extended_prompt:<30} PROB: {cumulative_prob:.4f}"
+    # Format the line to align "PROB:..." vertically, with additional padding
+    formatted_line = f"Depth {depth}: {extended_prompt:<45} PROB: {cumulative_prob:.4f}"
     
-    # Print only the formatted line without the "INFO:custom_logger" prefix
-    if not formatted_line.startswith("INFO:custom_logger:"):
-        custom_logger.info(formatted_line)
+    # Log only the formatted line without the "INFO:custom_logger" prefix
+    custom_logger.info(formatted_line)
 
     top_k_ids, top_k_probs = produce_next_token_ids(input_ids, model, config.TOPK, config.SUB_TOKEN_ID)
 
@@ -253,6 +259,8 @@ def build_def_tree(input_ids, data, base_prompt, model, tokenizer, config, depth
 
         yield from build_def_tree(new_input_ids, new_child, base_prompt, model, tokenizer, config, depth=depth+1, max_depth=max_depth, cumulative_prob=new_cumulative_prob)
 
+
+
 def generate_definition_tree(base_prompt, embedding, model, tokenizer, config):
     logger.info(f"Starting generate_definition_tree with base_prompt: {base_prompt}")
     results_dict = {"token": "", "cumulative_prob": 1, "children": []}