'DistributedDataParallel' object has no attribute 'model'

[aixin200005]

🐛 Bug

Hello, I have trained 7 classes of data according to the requirements of Train Custom Data.However, the following errors occurred:
Traceback (most recent call last):
File "train.py", line 400, in
train(hyp)
File "train.py", line 203, in train
check_best_possible_recall(dataset, anchors=model.model[-1].anchor_grid, thr=hyp['anchor_t'])
File "/home/aixin/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/nn/modules/module.py", line 539, in getattr
type(self).name, name))
AttributeError: 'DistributedDataParallel' object has no attribute 'model'

Environment

• OS: [Ubuntu18.04]
• GPU [NVIDIA TITAN Xp * 4]

[github-actions]

Hello @aixin200005, thank you for your interest in our work! Please visit our Custom Training Tutorial to get started, and see our Jupyter Notebook , Docker Image, and Google Cloud Quickstart Guide for example environments.

If this is a bug report, please provide screenshots and minimum viable code to reproduce your issue, otherwise we can not help you.

If this is a custom model or data training question, please note that Ultralytics does not provide free personal support. As a leader in vision ML and AI, we do offer professional consulting, from simple expert advice up to delivery of fully customized, end-to-end production solutions for our clients, such as:

• Cloud-based AI systems operating on hundreds of HD video streams in realtime.
• Edge AI integrated into custom iOS and Android apps for realtime 30 FPS video inference.
• Custom data training, hyperparameter evolution, and model exportation to any destination.

For more information please visit https://www.ultralytics.com.

[ChenYingpeng]

@aixin200005 You can change below code in train.py.

`

if device.type != 'cpu' and torch.cuda.device_count() > 1 and torch.distributed.is_available():
    check_best_possible_recall(dataset, anchors=model.module.model[-1].anchor_grid, thr=hyp['anchor_t'])
else:
    check_best_possible_recall(dataset, anchors=model.model[-1].anchor_grid, thr=hyp['anchor_t'])

`

[aixin200005]

@ChenYingpeng
Thank you for your suggestion, but I still report an error after revising according to your suggestion
Traceback (most recent call last):
File "train_aixin.py", line 400, in
train(hyp)
File "train_aixin.py", line 195, in train
check_best_possible_recall(dataset, anchors=model.module.model[-1].anchor_grid, thr=hyp['anchor_t'])
File "/home/aixin/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/nn/modules/module.py", line 539, in getattr
type(self).name, name))
AttributeError: 'Model' object has no attribute 'module'

[glenn-jocher]

@ChenYingpeng @aixin200005 yes I see, this is caused by the anchor checking trying to pull anchors from a single-gpu model training. I'll see what I can do to fix this, thank you for the suggestion @ChenYingpeng .

[glenn-jocher]

This should be resolved now in the latest commit, please git pull and try again.