AttributeError: 'SyncBatchNorm' object has no attribute '_specify_ddp_gpu_num'

[yangmin666]

I got the error explained above.Is this function "_specify_ddp_gpu_num" exist？Thank you!

[zstarN70]

I got the error explained above.Is this function "_specify_ddp_gpu_num" exist？Thank you!

You need to upgrade the CUDA version or change SyncBN to BN1d

[allezsyh]

Hello, I encountered the same problem as you, I reduced my Pytorch version from 1.9 to 1.8，

[haritha91]

Since am having the same error, I downgraded PyTorch from 1.9 to 1.8. Now it's throwing a CUDA error. Could you please specify which CUDA version to use?

[C-H-Wong]

Since am having the same error, I downgraded PyTorch from 1.9 to 1.8. Now it's throwing a CUDA error. Could you please specify which CUDA version to use?

Requirements of the repo are here

[allezsyh]

Since am having the same error, I downgraded PyTorch from 1.9 to 1.8. Now it's throwing a CUDA error. Could you please specify which CUDA version to use?

Requirements of the repo are here

Thanks, I just checked my CUDA version is 11.2

[haritha91]

I used a Docker container with the said environment configurations which run smoothly. Adding it here for anyone's reference.

FROM nvidia/cuda:10.2-devel-ubuntu16.04

# Install some basic utilities
RUN apt-get update && apt-get install -y \
    curl \
    ca-certificates \
    sudo \
    git \
    bzip2 \
    libx11-6 \
    libsm6 \
    libxext6 \
    libgl1-mesa-glx \
    build-essential \
    libatlas-base-dev \
    libjasper-dev \
    libqtgui4 \
    python3-pyqt5 \
    libqt4-test \
 && rm -rf /var/lib/apt/lists/*

# Create a working directory
RUN mkdir /app
WORKDIR /app

# Create a non-root user and switch to it
RUN adduser --disabled-password --gecos '' --shell /bin/bash user \
 && chown -R user:user /app
RUN echo "user ALL=(ALL) NOPASSWD:ALL" > /etc/sudoers.d/90-user
USER user

# All users can use /home/user as their home directory
ENV HOME=/home/user
RUN chmod 777 /home/user

# Install Miniconda and Python 3.8
ENV CONDA_AUTO_UPDATE_CONDA=false
ENV PATH=/home/user/miniconda/bin:$PATH
RUN curl -sLo ~/miniconda.sh https://repo.continuum.io/miniconda/Miniconda3-py38_4.10.3-Linux-x86_64.sh \
 && chmod +x ~/miniconda.sh \
 && ~/miniconda.sh -b -p ~/miniconda \
 && rm ~/miniconda.sh \
 && conda install -y python==3.8.5 \
 && conda clean -ya

# CUDA 11.1-specific steps
RUN conda install -y -c conda-forge cudatoolkit=10.2 \
 && conda install -y -c pytorch \
    "pytorch=1.8.0=py3.8_cuda10.2_cudnn7.6.5_0" \
    "torchvision=0.9.0=py38_cu102" \
 && conda clean -ya

RUN pip3 install opencv-python

RUN /bin/bash

RUN git clone https://github.com/Media-Smart/vedatad.git
WORKDIR vedatad

RUN pip install -r requirements/build.txt
RUN pip install -v -e .


# # Create directory mounts
RUN sudo mkdir /app/vedatad/data && sudo chown user:user /app/vedatad/data
RUN sudo mkdir /app/vedatad/workdir && sudo chown user:user /app/vedatad/workdir
RUN sudo mkdir /app/vedatad/checkpoints && sudo chown user:user /app/vedatad/checkpoints