Gregor Kobsik

Results 11 issues of Gregor Kobsik

Hello, I'm working with 3D shapes with Transformers and currently looking closer at the PolyGen paper. In [Colab](https://colab.research.google.com/github/deepmind/deepmind-research/blob/master/polygen/sample-pretrained.ipynb) a pre-trained model is given. I would like to get some quantitative...

Could you provide some additional information about the memory consumption using your Graph Transformer? You state, that sparse-attention favors both computation time and memory consumption, but do not provide actual...

It is only possible to crop inner borders of the images. a attribute, which defines the crop behaviour would allow to have also outer border crop. e.g. `python print.py -border_crop...

**Description:** The provided `conf.pth`-files seem inconsistent with the provided dataset (cmp. Example below). The dataloader `PartGraphShapesDataset` uses the provided argument `data_path` for `self.data_dir`. Later, in the function `__getitem__(...)` the geometric...

Why did you comment out the structure decoding? Is it not the idea of the paper to decode the geometry AND the structure independently? https://github.com/IGLICT/DSG-Net/blob/6c6d117525686fcca14b3efe477dfdb37737b55a/code/model_dsmnet_full.py#L648-L649 https://github.com/IGLICT/DSG-Net/blob/6c6d117525686fcca14b3efe477dfdb37737b55a/code/model_dsmnet_full.py#L655 https://github.com/IGLICT/DSG-Net/blob/6c6d117525686fcca14b3efe477dfdb37737b55a/code/model_dsmnet_full.py#L702

When loading an object with the provided dataloaders I get different number of values for `part_boxes`. Using `PartGraphShapesDataset` from `datav1.py` the Tensor has 10 values, when using `PartGraphShapesDataset` from `data_snv1.py`...

Hi, I just wanted to try out another model, in particular the `exp_vae_chair-full` model. Thus I updated the evaluation script: ``` CUDA_VISIBLE_DEVICES=0 python eval_recon_sn.py \ --exp_name 'exp_vae_chair-full' \ --test_dataset 'test_no_other_less_than_10_parts.txt'...

**Description:** I suppose, that the code in `datav1.py` has a bug in line 572. The returned semantic labels are a concatenation of all parent node labels down to the leaf...

After running the script `code\scripts\eval_vae_chair.sh` I managed to encode and decode some shapes. Unfortunately, all reconstructed models `recon.json` have the parameter "id" set always to 0. Upon inspecting the code...

Hello all, I just tried to run you code, but unfortunately I could not. I created a new fresh environment and installed all dependencies & packages. Next, I downloaded the...