CUDA SETUP: Something unexpected happened. Please compile from source

admin管理员组

文章数量:1590156

在测试 Chinese-Vicuna 时, 运行如下代码报错

bash scripts/finetune.sh 

解决方案:

升级 bitsandbytes 版本到 0.38.0

原因是: 我的显卡 cuda版本是1.21, 而bitsandbytes的0.37.0版本只到1.20

版本修改后,安装一下依赖 pip install -r requirements.txt

然后再运行

bash scripts/finetune.sh 

此报错解决

报错如下:

WARNING:torch.distributed.run:
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************

===================================BUG REPORT===================================
Welcome to bitsandbytes. For bug reports, please submit your error trace to: https://github/TimDettmers/bitsandbytes/issues
================================================================================

===================================BUG REPORT===================================
Welcome to bitsandbytes. For bug reports, please submit your error trace to: https://github/TimDettmers/bitsandbytes/issues
================================================================================
/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/bitsandbytes/cuda_setup/main.py:136: UserWarning: /home/gaosong/anaconda3/envs/vicuna8 did not contain libcudart.so as expected! Searching further paths...
  warn(msg)
/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/bitsandbytes/cuda_setup/main.py:136: UserWarning: WARNING: The following directories listed in your path were found to be non-existent: {PosixPath('/tmp/torchelastic_xzjl0mxr/none__l403kgq/attempt_0/1/error.json')}
  warn(msg)
CUDA_SETUP: WARNING! libcudart.so not found in any environmental path. Searching /usr/local/cuda/lib64...
CUDA SETUP: CUDA runtime path found: /usr/local/cuda/lib64/libcudart.so
CUDA SETUP: Highest compute capability among GPUs detected: 8.0
CUDA SETUP: Detected CUDA version 121
CUDA SETUP: Required library version not found: libbitsandbytes_cuda121.so. Maybe you need to compile it from source?
CUDA SETUP: Defaulting to libbitsandbytes_cpu.so...

================================================ERROR=====================================
CUDA SETUP: CUDA detection failed! Possible reasons:
1. CUDA driver not installed
2. CUDA not installed
3. You have multiple conflicting CUDA libraries
4. Required library not pre-compiled for this bitsandbytes release!
CUDA SETUP: If you compiled from source, try again with `make CUDA_VERSION=DETECTED_CUDA_VERSION` for example, `make CUDA_VERSION=113`.
CUDA SETUP: The CUDA version for the compile might depend on your conda install. Inspect CUDA version via `conda list | grep cuda`.
================================================================================

CUDA SETUP: Something unexpected happened. Please compile from source:
git clone git@github:TimDettmers/bitsandbytes.git
cd bitsandbytes
CUDA_VERSION=121
python setup.py install
/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/bitsandbytes/cuda_setup/main.py:136: UserWarning: /home/gaosong/anaconda3/envs/vicuna8 did not contain libcudart.so as expected! Searching further paths...
  warn(msg)
CUDA_SETUP: WARNING! libcudart.so not found in any environmental path. Searching /usr/local/cuda/lib64...
CUDA SETUP: CUDA runtime path found: /usr/local/cuda/lib64/libcudart.so
CUDA SETUP: Highest compute capability among GPUs detected: 8.0
CUDA SETUP: Detected CUDA version 121
CUDA SETUP: Required library version not found: libbitsandbytes_cuda121.so. Maybe you need to compile it from source?
/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/bitsandbytes/cuda_setup/main.py:136: UserWarning: WARNING: The following directories listed in your path were found to be non-existent: {PosixPath('/tmp/torchelastic_xzjl0mxr/none__l403kgq/attempt_0/0/error.json')}
  warn(msg)
CUDA SETUP: Defaulting to libbitsandbytes_cpu.so...
CUDA_SETUP: WARNING! libcudart.so not found in any environmental path. Searching /usr/local/cuda/lib64...

CUDA SETUP: CUDA runtime path found: /usr/local/cuda/lib64/libcudart.so================================================ERROR=====================================

CUDA SETUP: Highest compute capability among GPUs detected: 8.0CUDA SETUP: CUDA detection failed! Possible reasons:

CUDA SETUP: Detected CUDA version 1211. CUDA driver not installed

CUDA SETUP: Required library version not found: libbitsandbytes_cuda121.so. Maybe you need to compile it from source?2. CUDA not installed

CUDA SETUP: Defaulting to libbitsandbytes_cpu.so...3. You have multiple conflicting CUDA libraries

4. Required library not pre-compiled for this bitsandbytes release!

================================================ERROR=====================================CUDA SETUP: If you compiled from source, try again with `make CUDA_VERSION=DETECTED_CUDA_VERSION` for example, `make CUDA_VERSION=113`.

CUDA SETUP: CUDA detection failed! Possible reasons:CUDA SETUP: The CUDA version for the compile might depend on your conda install. Inspect CUDA version via `conda list | grep cuda`.

1. CUDA driver not installed================================================================================

2. CUDA not installed

3. You have multiple conflicting CUDA librariesCUDA SETUP: Something unexpected happened. Please compile from source:

4. Required library not pre-compiled for this bitsandbytes release!git clone git@github:TimDettmers/bitsandbytes.git

CUDA SETUP: If you compiled from source, try again with `make CUDA_VERSION=DETECTED_CUDA_VERSION` for example, `make CUDA_VERSION=113`.cd bitsandbytes

CUDA SETUP: The CUDA version for the compile might depend on your conda install. Inspect CUDA version via `conda list | grep cuda`.CUDA_VERSION=121

================================================================================python setup.py install

CUDA SETUP: Setup Failed!

CUDA SETUP: Something unexpected happened. Please compile from source:
git clone git@github:TimDettmers/bitsandbytes.git
cd bitsandbytes
CUDA_VERSION=121
CUDA_SETUP: WARNING! libcudart.so not found in any environmental path. Searching /usr/local/cuda/lib64...python setup.py install

CUDA SETUP: CUDA runtime path found: /usr/local/cuda/lib64/libcudart.so
CUDA SETUP: Highest compute capability among GPUs detected: 8.0
CUDA_SETUP: WARNING! libcudart.so not found in any environmental path. Searching /usr/local/cuda/lib64...CUDA SETUP: Detected CUDA version 121

CUDA SETUP: CUDA runtime path found: /usr/local/cuda/lib64/libcudart.soCUDA SETUP: Required library version not found: libbitsandbytes_cuda121.so. Maybe you need to compile it from source?

CUDA SETUP: Defaulting to libbitsandbytes_cpu.so...CUDA SETUP: Highest compute capability among GPUs detected: 8.0

CUDA SETUP: Detected CUDA version 121

================================================ERROR=====================================CUDA SETUP: Required library version not found: libbitsandbytes_cuda121.so. Maybe you need to compile it from source?

CUDA SETUP: CUDA detection failed! Possible reasons:CUDA SETUP: Defaulting to libbitsandbytes_cpu.so...

1. CUDA driver not installed

2. CUDA not installed================================================ERROR=====================================

3. You have multiple conflicting CUDA librariesCUDA SETUP: CUDA detection failed! Possible reasons:

4. Required library not pre-compiled for this bitsandbytes release!1. CUDA driver not installed

CUDA SETUP: If you compiled from source, try again with `make CUDA_VERSION=DETECTED_CUDA_VERSION` for example, `make CUDA_VERSION=113`.2. CUDA not installed

CUDA SETUP: The CUDA version for the compile might depend on your conda install. Inspect CUDA version via `conda list | grep cuda`.3. You have multiple conflicting CUDA libraries

================================================================================4. Required library not pre-compiled for this bitsandbytes release!

CUDA SETUP: If you compiled from source, try again with `make CUDA_VERSION=DETECTED_CUDA_VERSION` for example, `make CUDA_VERSION=113`.CUDA SETUP: Something unexpected happened. Please compile from source:

CUDA SETUP: The CUDA version for the compile might depend on your conda install. Inspect CUDA version via `conda list | grep cuda`.git clone git@github:TimDettmers/bitsandbytes.git

================================================================================cd bitsandbytes

CUDA_VERSION=121

CUDA SETUP: Something unexpected happened. Please compile from source:python setup.py install

git clone git@github:TimDettmers/bitsandbytes.gitCUDA SETUP: Setup Failed!

cd bitsandbytesCUDA SETUP: Something unexpected happened. Please compile from source:

CUDA_VERSION=121git clone git@github:TimDettmers/bitsandbytes.git

python setup.py installcd bitsandbytes

CUDA SETUP: Setup Failed!CUDA_VERSION=121

python setup.py install
CUDA_SETUP: WARNING! libcudart.so not found in any environmental path. Searching /usr/local/cuda/lib64...
CUDA SETUP: CUDA runtime path found: /usr/local/cuda/lib64/libcudart.so
CUDA SETUP: Highest compute capability among GPUs detected: 8.0
CUDA SETUP: Detected CUDA version 121
CUDA SETUP: Required library version not found: libbitsandbytes_cuda121.so. Maybe you need to compile it from source?
CUDA SETUP: Defaulting to libbitsandbytes_cpu.so...

================================================ERROR=====================================
CUDA SETUP: CUDA detection failed! Possible reasons:
1. CUDA driver not installed
2. CUDA not installed
3. You have multiple conflicting CUDA libraries
4. Required library not pre-compiled for this bitsandbytes release!
CUDA SETUP: If you compiled from source, try again with `make CUDA_VERSION=DETECTED_CUDA_VERSION` for example, `make CUDA_VERSION=113`.
CUDA SETUP: The CUDA version for the compile might depend on your conda install. Inspect CUDA version via `conda list | grep cuda`.
================================================================================Traceback (most recent call last):

  File "finetune.py", line 6, in <module>
CUDA SETUP: Something unexpected happened. Please compile from source:
git clone git@github:TimDettmers/bitsandbytes.git
cd bitsandbytes
CUDA_VERSION=121
python setup.py install
CUDA SETUP: Setup Failed!
CUDA SETUP: Something unexpected happened. Please compile from source:
git clone git@github:TimDettmers/bitsandbytes.git
cd bitsandbytes
CUDA_VERSION=121
python setup.py install
    import bitsandbytes as bnb
  File "/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/bitsandbytes/__init__.py", line 7, in <module>
Traceback (most recent call last):
    from .autograd._functions import (  File "finetune.py", line 6, in <module>

  File "/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/bitsandbytes/autograd/__init__.py", line 1, in <module>
    from ._functions import undo_layout, get_inverse_transform_indices
  File "/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/bitsandbytes/autograd/_functions.py", line 9, in <module>
    import bitsandbytes.functional as F
      File "/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/bitsandbytes/functional.py", line 17, in <module>
import bitsandbytes as bnb
  File "/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/bitsandbytes/__init__.py", line 7, in <module>
    from .cextension import COMPILED_WITH_CUDA, lib
  File "/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/bitsandbytes/cextension.py", line 22, in <module>
    from .autograd._functions import (
  File "/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/bitsandbytes/autograd/__init__.py", line 1, in <module>
    raise RuntimeError('''
RuntimeError: 
        CUDA Setup failed despite GPU being available. Inspect the CUDA SETUP outputs above to fix your environment!
        If you cannot find any issues and suspect a bug, please open an issue with detals about your environment:
        https://github/TimDettmers/bitsandbytes/issues    
from ._functions import undo_layout, get_inverse_transform_indices
  File "/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/bitsandbytes/autograd/_functions.py", line 9, in <module>
    import bitsandbytes.functional as F
  File "/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/bitsandbytes/functional.py", line 17, in <module>
    from .cextension import COMPILED_WITH_CUDA, lib
  File "/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/bitsandbytes/cextension.py", line 22, in <module>
    raise RuntimeError('''
RuntimeError: 
        CUDA Setup failed despite GPU being available. Inspect the CUDA SETUP outputs above to fix your environment!
        If you cannot find any issues and suspect a bug, please open an issue with detals about your environment:
        https://github/TimDettmers/bitsandbytes/issues
ERROR:torch.distributed.elastic.multiprocessing.api:failed (exitcode: 1) local_rank: 0 (pid: 372142) of binary: /home/gaosong/anaconda3/envs/vicuna8/bin/python
Traceback (most recent call last):
  File "/home/gaosong/anaconda3/envs/vicuna8/bin/torchrun", line 8, in <module>
    sys.exit(main())
  File "/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
    return f(*args, **kwargs)
  File "/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/torch/distributed/run.py", line 762, in main
    run(args)
  File "/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/torch/distributed/run.py", line 753, in run
    elastic_launch(
  File "/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/torch/distributed/launcher/api.py", line 132, in __call__
    return launch_agent(self._config, self._entrypoint, list(args))
  File "/home/gaosong/anaconda3/envs/vicuna8/lib/python3.8/site-packages/torch/distributed/launcher/api.py", line 246, in launch_agent
    raise ChildFailedError(
torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
============================================================
finetune.py FAILED
------------------------------------------------------------
Failures:
[1]:
  time      : 2023-06-08_15:12:16
  host      : server
  rank      : 1 (local_rank: 1)
  exitcode  : 1 (pid: 372143)
  error_file: <N/A>
  traceback : To enable traceback see: https://pytorch/docs/stable/elastic/errors.html
------------------------------------------------------------
Root Cause (first observed failure):
[0]:
  time      : 2023-06-08_15:12:16
  host      : server
  rank      : 0 (local_rank: 0)
  exitcode  : 1 (pid: 372142)
  error_file: <N/A>
  traceback : To enable traceback see: https://pytorch/docs/stable/elastic/errors.html

本文标签： unexpectedSetupCUDASourcecompile