Petteri Teikari
Petteri Teikari
> So I think that by writing standalone prediction functions (not part of the classes), it should be straight forward to use the pycox models without any code from torchtuples....
@havakv Sounds good if you have the time for that, and I was initially interested in the basic `LogisticHazard()`, and I pretty much followed that [`04_mnist_dataloaders_cnn.ipynb`](https://nbviewer.jupyter.org/github/havakv/pycox/blob/master/examples/04_mnist_dataloaders_cnn.ipynb) example as I have...
@havakv and if it is possible to make requests with some urgency emphasis. Is there an easy method that allows one to extract the CNN activations from the model? i.e....
@havakv Yes I think that is what I meant more or less https://discuss.pytorch.org/t/how-can-i-extract-intermediate-layer-output-from-loaded-cnn-model/77301. When I have first trained the model, I would be interested in having the intermediate layer activations...
Essentially needed to change three parameters: ```yaml num_class: 400 layout: 'openpose' node_num: 18 ``` I used the following `train_aim.yaml` ```yaml work_dir: ./work_dir/recognition/kinetics/AS_GCN feeder: feeder.feeder.Feeder train_feeder_args: data_path: ../data/kinetics/train_data_joint.npy label_path: ../data/kinetics/train_label.pkl random_move:...
And I found couple of other hard-coded `25`s from your code, and after changing them to a variable containing `18` fixed my training problem
Hey @izmailovpavel , yeah I commented it away and got away with it, but thought of letting you know if you ever update the code :)
> > Probably best then to just switch to [Better BibTeX (BBT)](https://retorque.re/zotero-better-bibtex/)? > > If that works for you, that seems like the simplest solution. Zotero 7 is a major...