Yinda (Frédéric) XU

Empirically, I discovered that it is necessary to halve the learning rate in order to training correctly on a 2-GPU machine instead of 4-GPU machine. Hereby I provide the training...

[UPDATE] @fanq15 Halving the learning rate as well as num_gpus reproduced the expected result. Hereby I provide the training log with 2 GPUs & 0.5x SOLVER.BASE_LR for debuggin usage. [fsod_train_log.txt](https://github.com/fanq15/FewX/files/5084515/fsod_train_log.txt)...

Just search `SiamFCppMultiTempTracker` in project and you will get the corresponding config file, for example, `experiments/siamfcpp/test/vot/multi_temp/siamfcpp_googlenet-multi_temp.yaml`

@iTruffle Have you check the link in MODEL_ZOO.md we provide?

@songdejia Training code is done and it is under final test (to make sure it works well after PyTorch version change) before being merged into Master branch. If you are...

@cuikf This is highly probably due to the bottleneck at data providing stage. The ability of data provider is usually due to the CPU and memory of the training machine...

[UPDATE] With the following conditional branch, the import issue was fixed. ``` from pkg_resources import parse_version if parse_version(torch.__version__) >= parse_version("1.0.0"): from torch.utils.cpp_extension import BuildExtension as create_extension else: from torch.utils.ffi import...

@EricKani @neoyang0620 Solved by [this issue](https://github.com/xingyizhou/CenterNet/issues/7)

@zhangwanjie [email protected]

@zhangwanjie Thank you for your reply. BTW, would it be possible to run it under Ubuntu 18.04 since Jeston Xavier only support Ubuntu higher than 18.04? If not, is there...