Fabio Milentiansen Sim

Hi @GhYang0519, thank you for your interest. I'd be glad to help. What problems are you facing when exporting GeoTransformer to ONNX?

I've seen this error before (see this [related stackoverflow issue](https://stackoverflow.com/questions/70166946/pytorch-model-tracing-not-working-we-dont-have-an-op-for-atenfill)). I believe this occurs in the following lines: https://github.com/qinzheng93/GeoTransformer/blob/e7a135af4c318ff3b8d7f6c963df094d7e4ea540/geotransformer/modules/ops/pointcloud_partition.py#L87-L88 ```python node_masks = torch.zeros(nodes.shape[0], dtype=torch.bool).cuda() # (M,) node_masks.index_fill_(0, point_to_node, True) ```...

I see.. Then perhaps GeoTransformer is too complicated to convert to ONNX

Yes, if you're using VSCode to debug the export process, you can place breakpoints in the forward passes of modules, starting from the beginning and gradually advancing further until VSCode...

Hi @antithing , thank you for your interest in DeDoDe-ONNX-TensorRT. I'm not an expert at C++, so I'd refer you to the documentation at https://onnxruntime.ai/docs/execution-providers/TensorRT-ExecutionProvider.html#cc.

Hi @xzh929, thank you for your interest in Depth-Anything-ONNX. The PyTorch ONNX exporter doesn't register the xformers ops, so if you would like to export the models, please uninstall xformers...

Hi @YoniSmolin, thank you for your interest. Yes, you are correct. I set it to `False` when I was measuring the latency.

Hi @heorhiikalaichev , thank you for your interest in DocShadow-ONNX-TensorRT. I believe you were referred here from https://github.com/CXH-Research/DocShadow-SD7K/issues/6, but CoreML conversion does not need ONNX anymore. It is now recommended...

Hi @Pizzawookiee, thank you for your interest in Depth Anything ONNX. Yeah, dynamo exporter is still in early stages, so it could probably be due to things like torch version...

Hi @Siwakonrome, thank you for your interest in Depth-Anything-ONNX. If I understand the script correctly, inference is the same up to this line: https://github.com/DepthAnything/Depth-Anything-V2/blob/31dc97708961675ce6b3a8d8ffa729170a4aa273/metric_depth/depth_to_pointcloud.py#L93 ```python pred = depth_anything.infer_image(image, height) ```...