@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / 991143cfcd
Fork 0

A high-throughput and memory-efficient inference and serving engine for LLMs

Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git

Synced Nov 17, 03:00 UTC

[BugFix] Use consistent logger everywhere (#3738)

Nick Hill committed Mar 29, 2024 at 16:26 UTC 991143cfcdc57d658d312bc001dd6d6dffba9495
5 files changed +13 -10
vllm/lora/models.py
+2 -2
@@ -1,6 +1,5 @@
1 import copy
2 import json
3 -import logging
3 import math
4 import os
5 import re
@@ -11,13 +10,14 @@ import torch
10 from torch import nn
11
12 from vllm.config import LoRAConfig
13 +from vllm.logger import init_logger
14 from vllm.lora.layers import (BaseLayerWithLoRA, LoRAMapping, from_layer,
15 from_layer_logits_processor)
16 from vllm.lora.lora import LoRALayerWeights, PackedLoRALayerWeights
17 from vllm.lora.utils import parse_fine_tuned_lora_name, replace_submodule
18 from vllm.utils import LRUCache, is_pin_memory_available
19
20 -logger = logging.getLogger(__name__)
20 +logger = init_logger(__name__)
21
22 _GLOBAL_LORA_ID = 0
23
vllm/lora/utils.py
+3 -2
@@ -1,9 +1,10 @@
1 -import logging
1 from typing import Tuple
2
3 from torch import nn
4
6 -logger = logging.getLogger(__name__)
5 +from vllm.logger import init_logger
6 +
7 +logger = init_logger(__name__)
8
9
10 def replace_submodule(model: nn.Module, module_name: str,
vllm/lora/worker_manager.py
+2 -2
@@ -1,16 +1,16 @@
1 -import logging
1 from abc import ABC, abstractmethod, abstractproperty
2 from typing import Any, Dict, List, Optional, Set, Type
3
4 import torch
5
6 from vllm.config import LoRAConfig
7 +from vllm.logger import init_logger
8 from vllm.lora.layers import LoRAMapping
9 from vllm.lora.models import (LoRAModel, LoRAModelManager,
10 LRUCacheLoRAModelManager, create_lora_manager)
11 from vllm.lora.request import LoRARequest
12
13 -logger = logging.getLogger(__name__)
13 +logger = init_logger(__name__)
14
15
16 class AbstractWorkerLoRAManager(ABC):
vllm/model_executor/parallel_utils/pynccl.py
+3 -2
@@ -21,7 +21,6 @@
21
22 import ctypes
23 import datetime
24 -import logging
24 import os
25
26 # ===================== import region =====================
@@ -29,7 +28,9 @@ import torch
28 import torch.distributed as dist
29 from torch.distributed import ReduceOp
30
32 -logger = logging.getLogger(__name__)
31 +from vllm.logger import init_logger
32 +
33 +logger = init_logger(__name__)
34
35 so_file = os.environ.get("VLLM_NCCL_SO_PATH", "")
36
vllm/model_executor/parallel_utils/pynccl_utils.py
+3 -2
@@ -1,11 +1,12 @@
1 import contextlib
2 -import logging
2 from typing import Optional
3
4 import torch
5 from torch.distributed import ReduceOp
6
8 -logger = logging.getLogger(__name__)
7 +from vllm.logger import init_logger
8 +
9 +logger = init_logger(__name__)
10
11 try:
12 from vllm.model_executor.parallel_utils.pynccl import (NCCLCommunicator,