车道线检测GANet网络复现报错解决方法汇总

目录

AttributeError: module 'distutils' has no attribute 'version'

RuntimeError: Distributed package doesn’t have NCCL built in|PyTorch踩坑

windows下“import torch” 报错：“OSError: [WinError 1455] 页面文件太小，无法完成操作” 的问题

---

AttributeError: module 'distutils' has no attribute 'version'

在开始准备训练模型时会报这个错误：

解决： setuptools版本问题”，版本过高导致的问题；setuptools版本

推荐安装：setuptools              57.5.0

pip uninstall setuptools
pip install setuptools==57.5.0 //需要比你之前的低 

注意：以下是windows下会产生的错误

RuntimeError: Distributed package doesn’t have NCCL built in|PyTorch踩坑

在windows系统上复现车道线检测GANet网络时了发生如下错误

raise RuntimeError("Distributed package doesn’t have NCCL "
RuntimeError: Distributed package doesn’t have NCCL built in

原因：windows不支持NCCL，应该修改为gloo

解决方案：在代码distributed_c10d.py里prefix_store = PrefixStore(group_name, store)下添加一段代码：

backend = "gloo"

修改后的片段如下：

prefix_store = PrefixStore(group_name, store)
backend = "gloo"
if backend == Backend.GLOO:
    pg = ProcessGroupGloo(
        prefix_store,
        rank,
        world_size,
        timeout=timeout)
    _pg_map[pg] = (Backend.GLOO, store)
    _pg_names[pg] = group_name
elif backend == Backend.NCCL:
    if not is_nccl_available():
        raise RuntimeError("Distributed package doesn't have NCCL "
                           "built in")

网络上还有其他的一些方法，都是在这个文件里添加backend='gloo'，但是在我用上面的方法解决的。

windows下“import torch” 报错：“OSError: [WinError 1455] 页面文件太小，无法完成操作” 的问题

解决方法：在mmdet\datasets\builder.py里找到num_workers将其赋值为0

    """
    rank, world_size = get_dist_info()
    if dist:
        # DistributedGroupSampler will definitely shuffle the data to satisfy
        # that images on each GPU are in the same group
        if shuffle:
            sampler = DistributedGroupSampler(dataset, samples_per_gpu,
                                              world_size, rank)
        else:
            sampler = DistributedSampler(
                dataset, world_size, rank, shuffle=False)
        batch_size = samples_per_gpu
        num_workers = workers_per_gpu
    else:
        sampler = GroupSampler(dataset, samples_per_gpu) if shuffle else None
        batch_size = num_gpus * samples_per_gpu
        
        # num_workers = num_gpus * workers_per_gpu
    num_workers = 0

    init_fn = partial(
        worker_init_fn, num_workers=num_workers, rank=rank,
        seed=seed) if seed is not None else None

如果还不行，那可能需要调整页面文件大小

参考：解决pycharm中: OSError: [WinError 1455] 页面文件太小，无法完成操作 的问题 - 程序那点事 - 博客园 (cnblogs.com)