AdamW8 bit and Lion8bit not working - CUDA_SETUP: WARNING! libcudart.so not found in any environmental path.

I am on windows

It is looking for libcudart.so however i think this only exists in linux

this is what we have in windows

i have cuda installed and there is also no such file

**the problem is your code only looking for linux not for window cuda. it is cudart.lib in windows**


this is cuda folder screenshot


```
===================================BUG REPORT===================================
Welcome to bitsandbytes. For bug reports, please submit your error trace to: https://github.com/TimDettmers/bitsandbytes/issues
For effortless bug reporting copy-paste your error into this form: https://docs.google.com/forms/d/e/1FAIpQLScPB8emS3Thkp66nvqwmjTEgxp8Y9ufuWTzFyr9kJ5AoI47dQ/viewform?usp=sf_link
================================================================================
CUDA_SETUP: WARNING! libcudart.so not found in any environmental path. Searching /usr/local/cuda/lib64...
F:\kohya_ss\venv\lib\site-packages\bitsandbytes\cuda_setup\paths.py:27: UserWarning: WARNING: The following directories listed in your path were found to be non-existent: {WindowsPath('/usr/local/cuda/lib64')}
warn(
WARNING: No libcudart.so found! Install CUDA or the cudatoolkit package (anaconda)!
CUDA SETUP: Loading binary F:\kohya_ss\venv\lib\site-packages\bitsandbytes\libbitsandbytes_cpu.so...
╭─────────────────────────────── Traceback (most recent call last) ────────────────────────────────╮
│ F:\kohya_ss\train_network.py:873 in │
│ │
│ 870 │ args = parser.parse_args() │
│ 871 │ args = train_util.read_config_from_file(args, parser) │
│ 872 │ │
│ ❱ 873 │ train(args) │
│ 874 │
│ │
│ F:\kohya_ss\train_network.py:262 in train │
│ │
│ 259 │ │ ) │
│ 260 │ │ trainable_params = network.prepare_optimizer_params(args.text_encoder_lr, args.u │
│ 261 │ │
│ ❱ 262 │ optimizer_name, optimizer_args, optimizer = train_util.get_optimizer(args, trainable │
│ 263 │ │
│ 264 │ # dataloaderを準備する │
│ 265 │ # DataLoaderのプロセス数：0はメインプロセスになる │
│ │
│ F:\kohya_ss\library\train_util.py:2699 in get_optimizer │
│ │
│ 2696 │ │
│ 2697 │ if optimizer_type == "AdamW8bit".lower(): │
│ 2698 │ │ try: │
│ ❱ 2699 │ │ │ import bitsandbytes as bnb │
│ 2700 │ │ except ImportError: │
│ 2701 │ │ │ raise ImportError("No bitsand bytes / bitsandbytesがインストールされていない │
│ 2702 │ │ print(f"use 8-bit AdamW optimizer | {optimizer_kwargs}") │
│ │
│ F:\kohya_ss\venv\lib\site-packages\bitsandbytes\__init__.py:6 in │
│ │
│ 3 # This source code is licensed under the MIT license found in the │
│ 4 # LICENSE file in the root directory of this source tree. │
│ 5 │
│ ❱ 6 from .autograd._functions import ( │
│ 7 │ MatmulLtState, │
│ 8 │ bmm_cublas, │
│ 9 │ matmul, │
│ │
│ F:\kohya_ss\venv\lib\site-packages\bitsandbytes\autograd\_functions.py:5 in │
│ │
│ 2 import warnings │
│ 3 │
│ 4 import torch │
│ ❱ 5 import bitsandbytes.functional as F │
│ 6 │
│ 7 from dataclasses import dataclass │
│ 8 from functools import reduce # Required in Python 3 │
│ │
│ F:\kohya_ss\venv\lib\site-packages\bitsandbytes\functional.py:13 in │
│ │
│ 10 from typing import Tuple │
│ 11 from torch import Tensor │
│ 12 │
│ ❱ 13 from .cextension import COMPILED_WITH_CUDA, lib │
│ 14 from functools import reduce # Required in Python 3 │
│ 15 │
│ 16 # math.prod not compatible with python < 3.8 │
│ │
│ F:\kohya_ss\venv\lib\site-packages\bitsandbytes\cextension.py:41 in │
│ │
│ 38 │ │ return cls._instance │
│ 39 │
│ 40 │
│ ❱ 41 lib = CUDALibrary_Singleton.get_instance().lib │
│ 42 try: │
│ 43 │ lib.cadam32bit_g32 │
│ 44 │ lib.get_context.restype = ct.c_void_p │
│ │
│ F:\kohya_ss\venv\lib\site-packages\bitsandbytes\cextension.py:37 in get_instance │
│ │
│ 34 │ def get_instance(cls): │
│ 35 │ │ if cls._instance is None: │
│ 36 │ │ │ cls._instance = cls.__new__(cls) │
│ ❱ 37 │ │ │ cls._instance.initialize() │
│ 38 │ │ return cls._instance │
│ 39 │
│ 40 │
│ │
│ F:\kohya_ss\venv\lib\site-packages\bitsandbytes\cextension.py:31 in initialize │
│ │
│ 28 │ │ │ self.lib = ct.cdll.LoadLibrary(binary_path) │
│ 29 │ │ else: │
│ 30 │ │ │ print(f"CUDA SETUP: Loading binary {binary_path}...") │
│ ❱ 31 │ │ │ self.lib = ct.cdll.LoadLibrary(binary_path) │
│ 32 │ │
│ 33 │ @classmethod │
│ 34 │ def get_instance(cls): │
│ │
│ C:\Python3108\lib\ctypes\__init__.py:452 in LoadLibrary │
│ │
│ 449 │ │ return getattr(self, name) │
│ 450 │ │
│ 451 │ def LoadLibrary(self, name): │
│ ❱ 452 │ │ return self._dlltype(name) │
│ 453 │ │
│ 454 │ __class_getitem__ = classmethod(_types.GenericAlias) │
│ 455 │
│ │
│ C:\Python3108\lib\ctypes\__init__.py:364 in __init__ │
│ │
│ 361 │ │ │ else: │
│ 362 │ │ │ │ import nt │
│ 363 │ │ │ │ mode = nt._LOAD_LIBRARY_SEARCH_DEFAULT_DIRS │
│ ❱ 364 │ │ │ │ if '/' in name or '\\' in name: │
│ 365 │ │ │ │ │ self._name = nt._getfullpathname(self._name) │
│ 366 │ │ │ │ │ mode |= nt._LOAD_LIBRARY_SEARCH_DLL_LOAD_DIR │
│ 367 │
╰──────────────────────────────────────────────────────────────────────────────────────────────────╯
TypeError: argument of type 'WindowsPath' is not iterable
╭─────────────────────────────── Traceback (most recent call last) ────────────────────────────────╮
│ C:\Python3108\lib\runpy.py:196 in _run_module_as_main │
│ │
│ 193 │ main_globals = sys.modules["__main__"].__dict__ │
│ 194 │ if alter_argv: │
│ 195 │ │ sys.argv[0] = mod_spec.origin │
│ ❱ 196 │ return _run_code(code, main_globals, None, │
│ 197 │ │ │ │ │ "__main__", mod_spec) │
│ 198 │
│ 199 def run_module(mod_name, init_globals=None, │
│ │
│ C:\Python3108\lib\runpy.py:86 in _run_code │
│ │
│ 83 │ │ │ │ │ __loader__ = loader, │
│ 84 │ │ │ │ │ __package__ = pkg_name, │
│ 85 │ │ │ │ │ __spec__ = mod_spec) │
│ ❱ 86 │ exec(code, run_globals) │
│ 87 │ return run_globals │
│ 88 │
│ 89 def _run_module_code(code, init_globals=None, │
│ │
│ in :7 │
│ │
│ 4 from accelerate.commands.accelerate_cli import main │
│ 5 if __name__ == '__main__': │
│ 6 │ sys.argv[0] = re.sub(r'(-script\.pyw|\.exe)?$', '', sys.argv[0]) │
│ ❱ 7 │ sys.exit(main()) │
│ 8 │
│ │
│ F:\kohya_ss\venv\lib\site-packages\accelerate\commands\accelerate_cli.py:45 in main │
│ │
│ 42 │ │ exit(1) │
│ 43 │ │
│ 44 │ # Run │
│ ❱ 45 │ args.func(args) │
│ 46 │
│ 47 │
│ 48 if __name__ == "__main__": │
│ │
│ F:\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py:918 in launch_command │
│ │
│ 915 │ elif defaults is not None and defaults.compute_environment == ComputeEnvironment.AMA │
│ 916 │ │ sagemaker_launcher(defaults, args) │
│ 917 │ else: │
│ ❱ 918 │ │ simple_launcher(args) │
│ 919 │
│ 920 │
│ 921 def main(): │
│ │
│ F:\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py:580 in simple_launcher │
│ │
│ 577 │ process.wait() │
│ 578 │ if process.returncode != 0: │
│ 579 │ │ if not args.quiet: │
│ ❱ 580 │ │ │ raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd) │
│ 581 │ │ else: │
│ 582 │ │ │ sys.exit(1) │
│ 583 │
╰──────────────────────────────────────────────────────────────────────────────────────────────────╯
CalledProcessError: Command '['F:\\kohya_ss\\venv\\Scripts\\python.exe', 'train_network.py', '--enable_bucket', '--pretrained_model_name_or_path=F:/0 models/v1-5-pruned.ckpt',
'--train_data_dir=F:\\test1\\img', '--resolution=512,512', '--output_dir=F:\\test1\\model', '--logging_dir=F:\\test1\\log', '--network_alpha=1', '--save_model_as=safetensors',
'--network_module=networks.lora', '--text_encoder_lr=5e-05', '--unet_lr=0.0001', '--network_dim=128', '--output_name=test1', '--lr_scheduler_num_cycles=8', '--learning_rate=0.0001',
'--lr_scheduler=cosine', '--lr_warmup_steps=260', '--train_batch_size=1', '--max_train_steps=2600', '--save_every_n_epochs=1', '--mixed_precision=bf16', '--save_precision=bf16',
'--cache_latents', '--optimizer_type=AdamW8bit', '--max_data_loader_n_workers=0', '--bucket_reso_steps=64', '--mem_eff_attn', '--gradient_checkpointing', '--xformers', '--bucket_no_upscale']'
returned non-zero exit status 1.

```

Start at library/train_util.py:get_optimizer, then trace the reported bitsandbytes cuda_setup/paths.py and cextension.py imports on Windows. Reproduce the CUDA_SETUP warning with the shown AdamW8bit path and determine the expected Windows CUDA library handling; done means the reported Windows setup loads the CUDA-backed optimizer without the shown fallback or traceback.

Written by the indexing model from the issue text.