Justus Schock

UI progress bar for computing hashes

4

Hi, I regularly add large directories with many many files to dvc storages. All works fine, but I just don't know how long it takes. Most of the steps are...

[WIP] Transforms API

3

Transforms are love, transforms are live...

Don't use plain forward in python

3

In opposite to what you say in the section > A nn.module can be used on input data in two ways whereas the latter one is commonly used for better...

Incomplete Image Nvidia/PyTorch

2

Expected Behavior ================= A fully working image (just as the original) --- Actual Behavior ================= The size reduced nicely, but it's not working. ``` -> docker run nvcr.io/nvidia/pytorch.slim /etc/shinit_v2: line...

`DeliraConfig({"a": {"b": 2}, "b":2})` should work since the "b" keys are on a different dict-level. However this results in `ValueError: b already in config. Can not overwrite value.`

Introduce function similar to `build_optim_fn`

Training Pipeline Example

2

Right now our experiments only cover how to train networks with different backends. We should add a notebook containing the whole training pipeline (including training, resume, test (and maybe even...

Argument Resolution Refactoring

5

We should definately check where to give what argument (and maybe expose more keyword arguments) in experiment (Possible ways are to pass arguments via keyword or parameter object in `__init__.py`...

`DistributedDataParallel` is reported to be faster for most networks compared to `DataParallel` (Up to 30%)

We should change saving for TF Graoh and TF Eager to [`tf.train.Checkpoint`](https://www.tensorflow.org/api_docs/python/tf/train/Checkpoint)