Jiaxing Yan

Does it support torch.bmm() or torch.matmul()

1

One part of my model performs the matrix multiplication, i.e torch.bmm() / torch.matmul(), but results show the GMac of this part is 0.0, i want to know if this is...

How can i get pretrained weights?

4

how can i get the pretrained weights for biseNet, not for ResNet?

是否支持fp16的QAT训练

1

是否支持fp16的QAT训练，如何修改配置文件

ScatterND requires foreign node optimizations incompatible with data-dependent shapes, so treating as error.

3

## Description ``` [12/07/2023-19:30:52] [E] Error[9]: [myelinBuilderUtils.cpp::getMyelinSupportType::1105] Error Code 9: Internal Error (ScatterND_213 requires foreign node optimizations incompatible with data-dependent shapes, so treating as error.) [12/07/2023-19:30:52] [E] Error[2]: [builder.cpp::buildSerializedNetwork::751] Error...

如何计算centerpoint模型的参数量？

1

如何计算centerpoint模型的参数量？

RuntimeError: (Unavailable) Memory map failed when create shared memory.

1

错误信息： Process Process-6: Traceback (most recent call last): File "/home/yanjiaxing/anaconda3/envs/paddle_env/lib/python3.8/multiprocessing/process.py", line 315, in _bootstrap self.run() File "/home/yanjiaxing/anaconda3/envs/paddle_env/lib/python3.8/multiprocessing/process.py", line 108, in run self._target(*self._args, **self._kwargs) File "/home/yanjiaxing/anaconda3/envs/paddle_env/lib/python3.8/site-packages/paddle/fluid/dataloader/worker.py", line 371, in _worker_loop six.reraise(*sys.exc_info())...

训练中间evalution的时候老是报CUDA_ERROR_OUT_OF_MEMORY的错误

1

环境信息，cuda版本为11.2 ``` paddle-bfloat 0.1.7 paddle2onnx 1.0.6 paddledet 2.6.0 paddlepaddle-gpu 2.4.2.post112 paddleseg 2.8.0 x2paddle 1.4.1 ``` 训练的时候使用--do_eval，每个若干个epoch就会进行evalution,。训练的时候显存使用正常，还有4G显存空闲。但每次evalution的时候，就会报错CUDA_ERROR_OUT_OF_MEMORY ``` 2023-05-26 15:07:01,583 - INFO - evaluate on validate dataset 2023-05-26 15:07:04,860 - WARNING...

使用container + 多卡训练时，每隔一定epoch就会报错并停止训练

1

系统环境：CentOS + 4 V100 container：paddle3d的容器 训练命令： ``` python -m paddle.distributed.launch --gpus 0,1,2,3 tools/train.py --config configs/centerpoint/centerpoint_pillars_02voxel_apollo.yml --save_dir output/centerpoint/centerpoint_voxels_010voxel_apollo_pillar_test_time_v100/ --save_interval 2 --num_workers 2 --keep_checkpoint_max 50 ``` 训练时，一开始正常没有问题，也可以正常保存weight。但是每次保存了2个权重，也就是训练了2个epoch之后，就会报下面的错误，然后训练停止 ``` 2023-06-18 01:35:04,503 - INFO...

目前是否支持3d iou loss

1

目前是否支持2个3D Oriented Boxes的IOU loss计算？

Thanks for your great work! You have released V2-99 backbones pre-trained on DDAD15M using dense depth estimation as the task, and could you release other vovnet backbone pre-trained on DDAD15M?,...