罗崚骁 (Luo, Lingxiao)
罗崚骁 (Luo, Lingxiao)
> Possibly this will be okay now after [be13e22](https://github.com/facebookresearch/xformers/commit/be13e229b52d9d0bdf4422be931c67c492b8092f) if you set XFORMERS_ENABLE_TRITON=1 ? Setting this works for me with xformers v0.0.27. Thanks!
You need to use the [transformer](https://github.com/aenario/tsmirror/blob/e88c846c0cedd523615922dfe90ccad2b60c4f92/packages/reflect/src/transformer.ts) of this project, I'd suggest you doing this with [ttypescript](https://github.com/cevek/ttypescript). I've tested and found out that this works with ts 3.8.3, however, seems to...
倒数第二个代码块上方的公式角标好像没有用大括号括起来。
FYI, I found an related issue in the PyTorch Lightning repository, where it is suggested that one may manually filter the weights to be saved by overriding the `state_dict` method...
@KumoLiu Hello, the format is indeed correct, however, the values of coordinates are mismatched with `flip_boxes`. Let me do it again with this function: ```python import torch from monai.apps.detection.transforms.box_ops import...
> The `spatial_size` parameter dictates the axis along which you intend to flip. I'm sorry if I misunderstand anything, but isn't the `spatial_size` parameter literally indicating the size of the...
@awaelchli I think I find the bug. I don't find `convert_module` defined for `FSDPPrecision`. https://github.com/Lightning-AI/pytorch-lightning/blob/0c8a193d3c46f9ddba46a3ab1818e24ec41698af/src/lightning/pytorch/plugins/precision/fsdp.py#L34 And `FSDPPrecision.convert_module` will finally fallback to `convert_module` of `lightning.fabric.plugins.Precision`, which simply does nothing: https://github.com/Lightning-AI/pytorch-lightning/blob/0c8a193d3c46f9ddba46a3ab1818e24ec41698af/src/lightning/fabric/plugins/precision/precision.py#L48-L54
I'm not sure if this is a limitation or not, currently I actually find no simple way to achieve this. For automatic optimization, gradient unscaling is performed right after the...
@kkoutini No, I give up and use fabric instead.
Hello, sorry for making you wait for so long, since we are working on other stuffs. Did you solve this issue? I guess this may be caused by the code...