RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation

[Tianhao-Qi]

use cascade mask rcnn and add num_convs=2 to rpn_head, however, I get the following errors:

Environment

[ZwwWayne]

Could you provide a config so that we can reproduce the error?

[liming-ai]

Could you provide a config so that we can reproduce the error?

Hi, @ZwwWayne

I have same issue too, it can be fixed by changing this line to:

x = x + self.lateral_convs[i](feat)

[Tianhao-Qi]

Could you provide a config so that we can reproduce the error?

The config file I used is configs/simple_copy_paste/mask_rcnn_r50_fpn_syncbn-all_rpn-2conv_ssj_scp_32x2_270k_coco.py.

[shinya7y]

In my environment, the following fix resolves the error. https://github.com/open-mmlab/mmdetection/blob/ca11860f4f3c3ca2ce8340e2686eeaec05b29111/mmdet/models/dense_heads/rpn_head.py#L65

x = F.relu(x, inplace=False)

• faster_rcnn_r50_fpn_1x_coco.py: no error
• faster_rcnn_r50_fpn_1x_coco.py, --cfg-options "model.rpn_head.num_convs=2": error
• faster_rcnn_r50_fpn_1x_coco.py, --cfg-options "model.rpn_head.num_convs=2", x = F.relu(x, inplace=False): no error

[AT1981]

Solov2 head returns the same problem for me. Has anyone fixed that yet?

[mzillag]

Solov2 head returns the same problem for me. Has anyone fixed that yet?

Yeap, [here](https://github.com/open-mmlab/mmdetection/blob/ca11860f4f3c3ca2ce8340e2686eeaec05b29111/mmdet/models/dense_heads/solov2_head.py#:~:text=%2B%3D%20self.-,convs_all_levels,-%5Bi%5D( ) change to feature_add_all_level = feature_add_all_level + self.convs_all_levels[i](input_p)

[github-actions[bot]]

This issue is marked as stale because it has been marked as invalid or awaiting response for 7 days without any further response. It will be closed in 5 days if the stale label is not removed or if there is no further response.

[github-actions[bot]]

This issue is closed because it has been stale for 5 days. Please open a new issue if you have similar issues or you have any new updates now.