RICKand-MORTY comments

Results 13 comments of


                                            RICKand-MORTY

cannot import name '_get_cpp_backtrace' from 'torch._C'请问这种问题该怎样解决呢

pytorch和CUDA版本不对应吧？

An module error when running cm_train.py

suitable flash-attn version: **1.0.2** or **0.2.8** I finally solve the problem.I changed my flash-attn version to **1.0.2** and **0.2.8** to test and both of these version is suitable.Besides, when I...

After running, no .npz file appeared

In cm/logger.py/configure you can find the following code: ` if dir is None: dir = os.getenv("OPENAI_LOGDIR") ` the output data was specified by environment variable “OPENAI_LOGDIR”.If not given, the output...

the libfabric EFA provider is operating in a condition that could result in memory corruption or other system errors.

Maybe the version of pytorch or cuda is incorrect

the libfabric EFA provider is operating in a condition that could result in memory corruption or other system errors.

是多卡训练吗?多卡训练dist_utils.py那个节点gpu数要改成自己的gpu数,另外命令行的mpiexec -n 4的4也要换成自己的gpu数

the libfabric EFA provider is operating in a condition that could result in memory corruption or other system errors.

添加环境变量RDMAV_FORK_SAFE吧看看，可能是为了安全不让直接fork子进程 https://docs.nvidia.com/networking/display/rdmaawareprogrammingv17/ibv_fork_init

the libfabric EFA provider is operating in a condition that could result in memory corruption or other system errors.

> > 添加环境变量RDMAV_FORK_SAFE吧看看，可能是为了安全不让直接fork子进程 https://docs.nvidia.com/networking/display/rdmaawareprogrammingv17/ibv_fork_init > > 在cm.train文件里添加了，但还是不行，报同样的错误在/etc/profile里添加，作为系统环境变量

the libfabric EFA provider is operating in a condition that could result in memory corruption or other system errors.

> 在/etc/profile里添加，作为系统环境变量记得保存后用source刷新一下

from flash_attn.flash_attention import FlashAttention . Where is FlashAttention？Why can't I import it?

is a package.use `pip install flash_attn==0.2.8` to install

from flash_attn.flash_attention import FlashAttention . Where is FlashAttention？Why can't I import it?

Maybe your environment is not suitable. Here is my environment: python:3.11.5 torch:2.0.0 torchvision: 0.15.1 flash-attn:0.2.8 If still has problem in installing flash_attn,try installing manually,see here:https://github.com/Dao-AILab/flash-attention