refactor(logging): rename llama_stack logger categories

This PR renames logging categores of llama_stack logger across the codebase according to https://github.com/llamastack/llama-stack/pull/3065#issuecomment-3187199197.

Signed-off-by: Mustafa Elbehery <melbeher@redhat.com>

llama_stack/models/llama/llama3/multimodal/model.py

@@ -36,7 +36,7 @@ from .utils import get_negative_inf_value, to_2tuple
 
 MP_SCALE = 8
 
-logger = get_logger(name=__name__, category="models")
+logger = get_logger(name=__name__, category="models::llama")
 
 
 def reduce_from_tensor_model_parallel_region(input_):

llama_stack/models/llama/llama3/tool_utils.py

@@ -11,7 +11,7 @@ from llama_stack.log import get_logger
 
 from ..datatypes import BuiltinTool, RecursiveType, ToolCall, ToolPromptFormat
 
-logger = get_logger(name=__name__, category="inference")
+logger = get_logger(name=__name__, category="models::llama")
 
 BUILTIN_TOOL_PATTERN = r'\b(?P<tool_name>\w+)\.call\(query="(?P<query>[^"]*)"\)'
 CUSTOM_TOOL_CALL_PATTERN = re.compile(r"<function=(?P<function_name>[^}]+)>(?P<args>{.*?})")

llama_stack/models/llama/llama4/quantization/loader.py

@@ -18,7 +18,7 @@ from ...datatypes import QuantizationMode
 from ..model import Transformer, TransformerBlock
 from ..moe import MoE
 
-log = get_logger(name=__name__, category="models")
+log = get_logger(name=__name__, category="models::llama")
 
 
 def swiglu_wrapper_no_reduce(

llama_stack/models/llama/quantize_impls.py

@@ -9,7 +9,7 @@ import collections
 
 from llama_stack.log import get_logger
 
-log = get_logger(name=__name__, category="llama")
+log = get_logger(name=__name__, category="models::llama")
 
 try:
     import fbgemm_gpu.experimental.gen_ai  # noqa: F401