大佬，单卡训练的时候有一个问题

[End1ezs]

AttributeError: 'MMT4Caption' object has no attribute 'module' 不知道哪里的问题

[Kamino666]

把更详细一点的报错放出来吧

[End1ezs]

Traceback (most recent call last): File "train.py", line 315, in mmt4captiontrain(cfg.data, args_) File "train.py", line 245, in mmt4caption_train val_loss = val_epoch(model, val_dataloader, mode=cfg['train']['task'], local_args=local_args) File "/home/endless/anaconda3/envs/pytorchtest/lib/python3.8/site-packages/torch/autograd/grad_mode.py", line 27, in decorate_context return func(*args, **kwargs) File "train.py", line 161, in val_epoch loss = model.module(v_feats, v_masks, captions) File "/home/endless/anaconda3/envs/pytorchtest/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1185, in getattr raise AttributeError("'{}' object has no attribute '{}'".format( AttributeError: 'MMT4Caption' object has no attribute 'module'

问题是在一轮结束后出现的。

[Kamino666]

哦哦，单卡训练参数有没有设置成--gpu呢，--gpu是单卡训练，--multi_gpu是多卡的

[Kamino666]

尝试一下使用--multi_gpu，然后把显卡数设置为1

[End1ezs]

哦哦，单卡训练参数有没有设置成--gpu呢，--gpu是单卡训练，--multi_gpu是多卡的 感谢大佬的回复， python train.py -c ./configs/caption-task_baseline_modal_clip4clip_config.json --gpu 大佬，我之前用的参数是这个 然后刚才用多卡试了一下， python train.py -c ./configs/caption-task_baseline_modal_clip4clip_config.json -ws 1 --multigpu 现在这样报错 Traceback (most recent call last): File "train.py", line 305, in args = configurehardware(args) File "/home/endless/pythonProject/Video-Captioning-Transformer-master/utils.py", line 138, in configure_hardware local_rank = int(os.environ["LOCAL_RANK"]) File "/home/endless/anaconda3/envs/pytorchtest/lib/python3.8/os.py", line 675, in getitem raise KeyError(key) from None KeyError: 'LOCAL_RANK'

[Kamino666]

用这个命令试一下 python -m torch.distributed.run --nproc_per_node 1 train.py -c ./configs/caption-task_baseline_modal_clip4clip_config.json --multi_gpu -ws 1

[End1ezs]

感谢大佬。