Shivansh Patel

## Motivation and Context Adding [MultiON](https://shivanshpatel35.github.io/multi-ON/) task (NeurIPS 2020). ## How Has This Been Tested Tested locally ## Types of changes - New feature (non-breaking change which adds functionality) ##...

Version mismatch

1

Which python and pytorch version should I use? This repository relies on Python 2.7 and PyTorch 0.2.0 whereas flownet depends on Python 3.6 and PyTorch 0.3.0.

encode_QA does not have specific format for test set. Test set have just image id and questions.

While running the extract_feat file it says - `net.blobs['data'].data[0] = transformer.preprocess('data', image) ValueError: could not broadcast input array from shape (3,448,448) into shape (3,224,224)`

Question about contact force

1

Hi, The code uses `self.contact_forces` to compute the reward. It is initialized [here](https://github.com/ZiwenZhuang/parkour/blob/main/legged_gym/legged_gym/envs/base/legged_robot.py#L586). But I cannot find where it is updated after each simulation step. Can you please point me...