remove __name__ from logging.getLogger() calls to use root logger

Signed-off-by: Khaled Sulayman <[email protected]>

src/instructlab/sdg/datamixing.py

@@ -22,7 +22,7 @@
 #               when |knowledge| << |skills|
 MIN_UPSAMPLE_THRESHOLD = 0.03
 ALLOWED_COLS = ["id", "messages", "metadata"]
-LOGGER = logging.getLogger(__name__)
+LOGGER = logging.getLogger()
 
 
 class DatasetListing(TypedDict):
@@ -739,10 +739,7 @@ def _gen_mixed_data(self, recipe, output_file_recipe, output_file_data):
                 self.num_procs,
             )
 
-    def generate(self, logger=None):
-        if logger is not None:
-            global LOGGER  # pylint: disable=global-statement
-            LOGGER = logger
+    def generate(self):
         self._gen_mixed_data(
             self.knowledge_recipe,
             self.output_file_knowledge_recipe,

src/instructlab/sdg/generate_data.py

@@ -340,7 +340,7 @@ def generate_data(
     document_output_dir = Path(output_dir) / f"documents-{date_suffix}"
 
     leaf_nodes = read_taxonomy_leaf_nodes(
-        taxonomy, taxonomy_base, yaml_rules, document_output_dir, logger=LOGGER
+        taxonomy, taxonomy_base, yaml_rules, document_output_dir
     )
     if not leaf_nodes:
         raise GenerateException("Error: No new leaf nodes found in the taxonomy.")
@@ -406,7 +406,6 @@ def generate_data(
             document_output_dir,
             model_name,
             docling_model_path=docling_model_path,
-            logger=LOGGER,
         )
 
         if not samples:
@@ -458,7 +457,7 @@ def generate_data(
         system_prompt,
     )
 
-    mixer.generate(logger=LOGGER)
+    mixer.generate()
 
     generate_duration = time.time() - generate_start
     LOGGER.info(f"Generation took {generate_duration:.2f}s")

src/instructlab/sdg/pipeline.py

@@ -23,7 +23,7 @@
 from .blocks.block import Block
 from .registry import BlockRegistry
 
-LOGGER = logging.getLogger(__name__)
+LOGGER = logging.getLogger()
 
 
 # This is part of the public API.
@@ -134,16 +134,12 @@ def from_file(cls, ctx, pipeline_yaml):
             pipeline_yaml = os.path.join(resources.files(__package__), pipeline_yaml)
         return cls(ctx, pipeline_yaml, *_parse_pipeline_config_file(pipeline_yaml))
 
-    def generate(self, dataset, checkpoint_name=None, logger=None) -> Dataset:
+    def generate(self, dataset, checkpoint_name=None) -> Dataset:
         """
         Generate the dataset by running the pipeline steps.
         dataset: the input dataset
         checkpoint_name: unique subdir name for the checkpoint within checkpoint_dir
         """
-
-        if logger is not None:
-            global LOGGER  # pylint: disable=global-statement
-            LOGGER = logger
         # The checkpointer allows us to resume from where we left off
         # Saving the output of pipe instances along the way
         checkpoint_dir = None

src/instructlab/sdg/utils/taxonomy.py

@@ -30,7 +30,7 @@
 # Initialize the pdf parser
 PDFParser = pdf_parser_v1()
 
-LOGGER = logging.getLogger(__name__)
+LOGGER = logging.getLogger()
 
 
 def _is_taxonomy_file(fn: str) -> bool:
@@ -372,11 +372,8 @@ def read_taxonomy(
 
 
 def read_taxonomy_leaf_nodes(
-    taxonomy, taxonomy_base, yaml_rules, document_output_dir=None, logger=None
+    taxonomy, taxonomy_base, yaml_rules, document_output_dir=None
 ):
-    if logger is not None:
-        global LOGGER  # pylint: disable=global-statement
-        LOGGER = logger
     seed_instruction_data = read_taxonomy(
         taxonomy, taxonomy_base, yaml_rules, document_output_dir
     )
@@ -466,11 +463,7 @@ def leaf_node_to_samples(
     document_output_dir,
     model_name,
     docling_model_path=None,
-    logger=None,
 ):
-    if logger is not None:
-        global LOGGER  # pylint: disable=global-statement
-        LOGGER = logger
     if not leaf_node:
         return []
     if leaf_node[0].get("documents"):