一些问题归档

comments: true

git

git clone 卡住

解决方法: 1. 别忘了网址最后的.git 2. 先Ctrl+C,尝试再跑一次 3. git config --global http.sslVerify "false"

git推送大型文件

解决方法：缓冲区大小默认是1MB， git config --global http.postBuffer 524288000 修改为500MB

RPC failed; curl 16 Error in the HTTP2 framing layer

解决方法

各种包的安装

import sru 卡住，实际上是出现无限循环

参考https://github.com/asappresearch/sru/issues/196

解决方法：删除~/.cache/torch_extensions文件夹然后重新import,会出现以下warning,但不影响使用

 UserWarning: Just-in-time loading and compiling the CUDA kernels of SRU was unsuccessful. Got the following error:
cannot open shared object file: No such file or directory
  warnings.warn("Just-in-time loading and compiling the CUDA kernels of SRU was unsuccessful. "

pip install torch-cluster包 build很慢

解决方法：因为只是慢，所以耐心等待一般都能跑出来的，不用着急似乎这里有更快的方法参考

biopython版本问题部分函数找不到 importError: cannot import name 'three_to_one' from 'Bio.PDB.Polypeptide'

解决方法: 搜索biopython.org three_to_one, 发现这个函数在biopython 1.75里面有，而本地安装的版本没有. pip修改版本

ESM和ESMFold

ESM和ESMFold都已经封装成了包,直接pip install,不需要clone代码

ESM(只需要语言模型当encoder的情况):

pip install fair-esm  # latest release, OR

ESMFold要容易踩坑一些:

pip install "fair-esm[esmfold]"
# OpenFold and its remaining dependency
pip install 'dllogger @ git+https://github.com/NVIDIA/dllogger.git'
pip install 'openfold @ git+https://github.com/aqlaboratory/openfold.git@4b41059694619831a7db195b7e0988fc4ff3a307'

但是这样pip源匹配的cudatoolkit版本容易出问题。建议用官方给的conda

conda env create -f environment.yml
pip install "fair-esm[esmfold]"
pip install 'openfold @ git+https://github.com/aqlaboratory/openfold.

import torch出现 undefined symbol: iJIT_NotifyEvent

pip install mkl==2024.0.0 不能2024.1及之后的版本,参考

nvcc fatal : Unsupported gpu architecture 'compute_89'

系统CUDA版本,需要>=11.8

RuntimeError: CUDA error: CUBLAS_STATUS_INVALID_VALUE when calling cublasGemmEx( handle, opa, opb, m, n, k, &falpha, a, CUDA_R_16F, lda, b, CUDA_R_16F, ldb, &fbeta, c, CUDA_R_16F, ldc, CUDA_R_32F, CUBLAS_GEMM_DFALT_TENSOR_OP)

动态链接库问题解决方案, unset LD_LIBRARY_PATH即可

def einsum(tensor: Tensor, pattern: str, /) -> Tensor:SyntaxError: invalid syntax

参考issue, 应该安装einops=0.6.1

Linux

RuntimeError: PytorchStreamReader failed reading zip archive: failed finding central directory

一般是模型文件下载不完整重新下载

matplotlib 教程

一些问题归档

git

各种包的安装

ESM和ESMFold

Linux

Comments