第十四篇-问题处理-libcudart.so.11.0: cannot open shared object file: No such file or directory

ta/slim_star --split validation --percentage 1.0
	Traceback (most recent call last):
	  File "/root/TinyLlama/scripts/prepare_slimpajama.py", line 15, in 
		import lit_gpt.packed_dataset as packed_dataset
	  File "/root/TinyLlama/lit_gpt/__init__.py", line 1, in 
		from lit_gpt.model import GPT
	  File "/root/TinyLlama/lit_gpt/model.py", line 13, in 
		from flash_attn import flash_attn_func
	  File "/root/miniconda3/envs/tiny/lib/python3.10/site-packages/flash_attn/__init__.py", line 3, in 
		from flash_attn.flash_attn_interface import (
	  File "/root/miniconda3/envs/tiny/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 8, in 
		import flash_attn_2_cuda as flash_attn_cuda
	ImportError: /root/miniconda3/envs/tiny/lib/python3.10/site-packages/flash_attn_2_cuda.cpython-310-x86_64-linux-gnu.so: undefined symbol: _ZNK3c106SymIntltEl
	 


	python -c "from flash_attn import flash_attn_qkvpacked_func, flash_attn_func"
	Traceback (most recent call last):
	  File "", line 1, in 
	  File "/root/miniconda3/lib/python3.8/site-packages/flash_attn/__init__.py", line 3, in 
		from flash_attn.flash_attn_interface import (
	  File "/root/miniconda3/lib/python3.8/site-packages/flash_attn/flash_attn_interface.py", line 8, in 
		import flash_attn_2_cuda as flash_attn_cuda
	ImportError: libcudart.so.11.0: cannot open shared object file: No such file or directory

解决办法:保持一下版本一致,保持依赖不冲突,可以重新安装

CUDA版本 nvidia-smi
CUDA Toolkit版本 nvcc -V
Torch版本,参考:https://pytorch.org/get-started/previous-versions/
FlashAttn版本,参考 :https://github.com/Dao-AILab/flash-attention/releases

你可能感兴趣的:(AI-LLM-实战,人工智能,chatgpt)