使用kohya_ss训练LoRa模型时遇bitsandbytes模块缺失错误求助

报错信息 #

Traceback (most recent call last):
  File "E:\Homeworklol\Deepfakes\LoRa Modell kram\LoRa Trainer\kohya_ss\library\train_util.py", line 3433, in get_optimizer
    import bitsandbytes as bnb
  File "E:\Homeworklol\Deepfakes\LoRa Modell kram\LoRa Trainer\kohya_ss\venv\lib\site-packages\bitsandbytes\__init__.py", line 6, in <module>
    from . import cuda_setup, utils, research
  File "E:\Homeworklol\Deepfakes\LoRa Modell kram\LoRa Trainer\kohya_ss\venv\lib\site-packages\bitsandbytes\research\__init__.py", line 1, in <module>
    from . import nn
  File "E:\Homeworklol\Deepfakes\LoRa Modell kram\LoRa Trainer\kohya_ss\venv\lib\site-packages\bitsandbytes\research\nn\__init__.py", line 1, in <module>
    from .modules import LinearFP8Mixed, LinearFP8Global
  File "E:\Homeworklol\Deepfakes\LoRa Modell kram\LoRa Trainer\kohya_ss\venv\lib\site-packages\bitsandbytes\research\nn\modules.py", line 8, in <module>
    from bitsandbytes.optim import GlobalOptimManager
  File "E:\Homeworklol\Deepfakes\LoRa Modell kram\LoRa Trainer\kohya_ss\venv\lib\site-packages\bitsandbytes\optim\__init__.py", line 6, in <module>
    from bitsandbytes.cextension import COMPILED_WITH_CUDA
  File "E:\Homeworklol\Deepfakes\LoRa Modell kram\LoRa Trainer\kohya_ss\venv\lib\site-packages\bitsandbytes\cextension.py", line 5, in <module>
    from .cuda_setup.main import evaluate_cuda_setup
  File "E:\Homeworklol\Deepfakes\LoRa Modell kram\LoRa Trainer\kohya_ss\venv\lib\site-packages\bitsandbytes\cuda_setup\main.py", line 21, in <module>
    from .paths import determine_cuda_runtime_lib_path
ModuleNotFoundError: No module named 'bitsandbytes.cuda_setup.paths'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "E:\Homeworklol\Deepfakes\LoRa Modell kram\LoRa Trainer\kohya_ss\train_db.py", line 495, in <module>
    train(args)
  File "E:\Homeworklol\Deepfakes\LoRa Modell kram\LoRa Trainer\kohya_ss\train_db.py", line 181, in train
    _, _, optimizer = train_util.get_optimizer(args, trainable_params)
  File "E:\Homeworklol\Deepfakes\LoRa Modell kram\LoRa Trainer\kohya_ss\library\train_util.py", line 3435, in get_optimizer
    raise ImportError("No bitsandbytes / bitsandbytesがインストールされていないようです")
ImportError: No bitsandbytes / bitsandbytesがインストールされていないようです
Traceback (most recent call last):
  File "C:\Program Files\Python310\lib\runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "C:\Program Files\Python310\lib\runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "E:\Homeworklol\Deepfakes\LoRa Modell kram\LoRa Trainer\kohya_ss\venv\Scripts\accelerate.exe\__main__.py", line 7, in <module>
  File "E:\Homeworklol\Deepfakes\LoRa Modell kram\LoRa Trainer\kohya_ss\venv\lib\site-packages\accelerate\commands\accelerate_cli.py", line 47, in main
    args.func(args)
  File "E:\Homeworklol\Deepfakes\LoRa Modell kram\LoRa Trainer\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 986, in launch_command
    simple_launcher(args)
  File "E:\Homeworklol\Deepfakes\LoRa Modell kram\LoRa Trainer\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 628, in simple_launcher
    raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd)
subprocess.CalledProcessError: Command '["E:\\Homeworklol\\Deepfakes\\LoRa Modell kram\\LoRa Trainer\\kohya_ss\\venv\\Scripts\\python.exe", "./train_db.py", "--pretrained_model_name_or_path=runwayml/stable-diffusion-v1-5", "--train_data_dir=E:/Homeworklol/Deepfakes/LoRa Modell kram/Nele/image", "--resolution=512,512", "--output_dir=E:/Homeworklol/Deepfakes/LoRa Modell kram/Nele/model", "--logging_dir=E:/Homeworklol/Deepfakes/LoRa Modell kram/Nele/log", "--save_model_as=safetensors", "--output_name=Nele_LoRa_1", "--lr_scheduler_num_cycles=1", "--max_data_loader_n_workers=1", "--learning_rate_te=1e-05", "--learning_rate=0.0001", "--lr_scheduler=constant", "--train_batch_size=2", "--max_train_steps=9950", "--save_every_n_epochs=1", "--mixed_precision=bf16", "--save_precision=bf16", "--seed=1234", "--caption_extension=.txt", "--cache_latents", "--optimizer_type=AdamW8bit", "--max_data_loader_n_workers=1", "--clip_skip=2", "--bucket_reso_steps=64", "--xformers", "--bucket_no_upscale", "--noise_offset=0.0"]' returned non-zero exit status 1.

解决方案 #

• 确保在虚拟环境内安装bitsandbytes：打开命令行，进入E:\Homeworklol\Deepfakes\LoRa Modell kram\LoRa Trainer\kohya_ss\venv\Scripts目录，执行activate激活虚拟环境，然后执行pip uninstall -y bitsandbytes彻底卸载，再执行pip install bitsandbytes==0.41.1 --force-reinstall --no-cache-dir安装指定稳定版本。
• 安装Windows适配版bitsandbytes：原版bitsandbytes对Windows支持不佳，卸载现有版本后，安装针对Windows优化的预编译wheel包（需匹配你的Python版本和CUDA版本），激活虚拟环境后执行对应安装命令。
• 检查CUDA版本兼容性：确认你的CUDA版本与bitsandbytes版本匹配，比如CUDA 11.8对应bitsandbytes 0.41.x系列，CUDA 12.x需使用0.43.x及以上版本，不匹配会导致模块加载失败。
• 修复文件缺失问题：手动删除虚拟环境中site-packages下的bitsandbytes文件夹，然后重新安装，避免部分文件损坏或缺失。
• 临时绕过bitsandbytes依赖：如果以上方法都无效，可将训练命令中的--optimizer_type=AdamW8bit替换为--optimizer_type=AdamW，使用普通AdamW优化器，跳过8bit优化器对bitsandbytes的依赖，先完成训练。

内容的提问来源于stack exchange，提问作者Erik Ehlers