liangliang comments

Results 131 comments of


                                            liangliang

为什么最新的拉下来训练的很慢？比半个月拉下来额慢了好几本配置一样

您看一下 GPU 利用率是否一致？或者使用 `git diff commit_id1. commit_id2 ` 看一下主要代码差异。正常情况下不会出现这种问题的。

多机多卡问题

add `--master_port 30001` or other value in start command, for example: ``` python -m torch.distributed.launch --nproc_per_node 8 --master_port 30002 tools/train.py ... ```

Multiple train/val/test dirs

Sorry, currently this project does not support multi directories, we will consider to add this funciton.

state_dict['param_groups'] TypeError: 'NoneType' object is not subscriptable

Are you resuming from a checkpoint which was already trained to the end? Currently, we will tail the checkpoint at the end of the training. However, the checkpoints saved at...

We need to have Confusion matrix, Precision, Recall and F1 score

Ok, we will consider to add those functions as soon as possible.

Model training and evaluation problems

Hi, if you want to get precision and recall at one specific confidence, you can filter out the detect boxes whose confidence are lower than your specific confidence. Second, put...

Model training and evaluation problems

Ok, we will discuss about whether to add this function.

Would be great to add Per-Class "Precision, Recall, mAPs" to the Evaluation Metrics | 请考虑对每一类加上验证指标，如Precision，Recall，mAPs

Hi, thank you for your advice. We will take it into consideration.

Problem about Partial Quantization

感谢您的建议，如果有可能，欢迎提供 PPQ 相关的代码，我们一起来共建此代码库。

TypeError: Object of type Tensor is not JSON serializable

您有对代码进行修改吗？