Remove calls to logging.basicConfig on import

Fixes #196

instructlab.sdg calls logging.basicConfig on import.

Libraries should not configure logging themselves. It's the job of the
application or main script to configure logging and set up log
handlers according to their needs. logging.basicConfig is a one-shot
API. The second and any following call are a no-op, unless it is
called with force=True.

Remove setup_logging and just use logging.getLogger in the code.

Co-authored-by: Christian Heimes <[email protected]>
Signed-off-by: Mark McLoughlin <[email protected]>

src/instructlab/sdg/block.py

@@ -2,15 +2,13 @@
 # Standard
 from abc import ABC
 from typing import Any, Dict, Union
+import logging
 import os.path
 
 # Third Party
 import yaml
 
-# Local
-from .logger_config import setup_logger
-
-logger = setup_logger(__name__)
+logger = logging.getLogger(__name__)
 
 
 # This is part of the public API.

src/instructlab/sdg/datamixing.py

@@ -1,6 +1,7 @@
 # Standard
 from typing import Optional
 import json
+import logging
 import os.path
 import random
 import uuid
@@ -10,11 +11,10 @@
 import yaml
 
 # First Party
-from instructlab.sdg.logger_config import setup_logger
 from instructlab.sdg.utils import GenerateException, pandas
 
 ALLOWED_COLS = ["id", "messages", "metadata"]
-logger = setup_logger(__name__)
+logger = logging.getLogger(__name__)
 
 
 def _adjust_train_sample_size(ds: Dataset, num_samples: int):

src/instructlab/sdg/eval_data.py

@@ -1,6 +1,7 @@
 # Standard
 from importlib import resources
 from typing import Any
+import logging
 import re
 
 # Third Party
@@ -10,10 +11,7 @@
 # First Party
 from instructlab.sdg.pipeline import EVAL_PIPELINES_PKG, Pipeline
 
-# Local
-from .logger_config import setup_logger
-
-logger = setup_logger(__name__)
+logger = logging.getLogger(__name__)
 
 
 def _extract_options(text: str) -> list[Any]:

src/instructlab/sdg/filterblock.py

@@ -1,15 +1,15 @@
 # SPDX-License-Identifier: Apache-2.0
 # Standard
+import logging
 import operator
 
 # Third Party
 from datasets import Dataset
 
 # Local
 from .block import Block
-from .logger_config import setup_logger
 
-logger = setup_logger(__name__)
+logger = logging.getLogger(__name__)
 
 
 # This is part of the public API.

src/instructlab/sdg/generate_data.py

@@ -76,9 +76,7 @@ def _convert_to_messages(sample):
     return sample
 
 
-def _gen_train_data(
-    machine_instruction_data, output_file_train, output_file_messages
-):
+def _gen_train_data(machine_instruction_data, output_file_train, output_file_messages):
     """
     Generate training data in the legacy system/user/assistant format
     used in train_*.jsonl as well as the legacy messages format used

src/instructlab/sdg/importblock.py

@@ -1,12 +1,14 @@
 # SPDX-License-Identifier: Apache-2.0
+# Standard
+import logging
+
 # Third Party
 from datasets import Dataset
 
 # Local
 from .block import Block
-from .logger_config import setup_logger
 
-logger = setup_logger(__name__)
+logger = logging.getLogger(__name__)
 
 
 # This is part of the public API.

src/instructlab/sdg/llmblock.py

@@ -2,6 +2,7 @@
 # Standard
 from collections import ChainMap
 from typing import Any, Dict
+import logging
 import re
 
 # Third Party
@@ -10,9 +11,8 @@
 
 # Local
 from .block import Block
-from .logger_config import setup_logger
 
-logger = setup_logger(__name__)
+logger = logging.getLogger(__name__)
 
 MODEL_FAMILY_MIXTRAL = "mixtral"
 MODEL_FAMILY_MERLINITE = "merlinite"

src/instructlab/sdg/pipeline.py

@@ -4,6 +4,7 @@
 from dataclasses import dataclass
 from importlib import resources
 from typing import Iterable, Optional
+import logging
 import math
 import os.path
 
@@ -18,9 +19,8 @@
 # Local
 from . import filterblock, importblock, llmblock, utilblocks
 from .block import Block
-from .logger_config import setup_logger
 
-logger = setup_logger(__name__)
+logger = logging.getLogger(__name__)
 
 
 # This is part of the public API.

src/instructlab/sdg/utilblocks.py

@@ -1,4 +1,7 @@
 # SPDX-License-Identifier: Apache-2.0
+# Standard
+import logging
+
 # Third Party
 from datasets import Dataset
 
@@ -7,9 +10,8 @@
 
 # Local
 from .block import Block
-from .logger_config import setup_logger
 
-logger = setup_logger(__name__)
+logger = logging.getLogger(__name__)
 
 
 # This is part of the public API.