ymlab
ymlab
Unfortunately, this is a training instability problem that has plagued us for a long time, and we currently have no good way to avoid this error. Some feasible methods in...
Any suggestions for solving this issue are welcome.
Maybe run the evaluation to see if there is a reasonable accuracy. If not, see if the pretrained model and learning rate need to be adjusted.
> 可能还要稍等等 请问近期会支持吗
> https://github.com/modelscope/ms-swift/blob/main/swift/trainers/loss.py > > Register the loss function using `register_loss_func`, and then specify it using `--loss_name xxx`. 顺便问下怎么自定义在线训练metric呢
I will release a new version of fastbev++ code in the near future, based on the standard setting of bevdet, so stay tuned.
Some new features: - support for fusion of depth information; - no need for custom operators, completely based on onnx native operators, so that onemodel + cross-platform deployment can be...
> I will release a new version of fastbev++ code in the near future, based on the standard setting of bevdet, so stay tuned. https://github.com/ymlab/advanced-fastbev/
> Thanks author for this work, how can I set up the environment for fastbev++. I do not see the getting started documentation there. Please prepare the training environment according...