William Woof

Enable tagging/debugging of rewards for better debbuging/control from trainer

5

**Is your feature request related to a problem? Please describe.** Managing contributions of different aspects of the reward function is inflexible. Some situations may involve providing the agent with multiple...

Visualized simulation based on ground truth data

7

In the visualised simulations there is the following line `ts_loss_part,velocities=sess.run([mse,P],feed_dict={O:[np.transpose(raw_data[i-1])] ...` which is then fed in to update `estimated_data`. This means the model is predicting based on the GT data,...

Mainly for reference as this library isn't being maintained anymore, but this addresses two issues Adding abbreviations for dimension names https://github.com/harvardnlp/namedtensor/issues/103#issue-464568229 Making it easier to refer to dimensions https://github.com/harvardnlp/namedtensor/issues/102#issuecomment-508701487

Feature Request: Tensor Abbreviations

2

This is actually more of a discussion thread since I have an implementation [here](https://github.com/EndingCredits/namedtensor), but I wanted to discuss the general idea before submitting a PR. Essentially, I propose that...

Tensorflow optimizer

3

Hi, I made a very quick implementation of the Adam optimiser with OMD in Tensorflow which may be useful to people wanting to try this out quickly in Tensorflow. N.B:...

Support for different language model backends/APIs

6

I realise the simple answer is to replace the hostname in the openai module, but it might be nice to add more support for various backends (as in, for the...

Hi, In your paper, one of the important points you make is that your method is O(n) whereas Relation nets are O(n^2). In fact, if we write both architectures in...

Error in install-nvm.sh

2

Hi, I'm not a big docker user, but I ran into installation errors due to a problem in `install-nvm.sh` ``` export NVM_DIR=./.nvm . $NVM_DIR/nvm.sh ``` Gives me an error Replacing...

Taking making inputs from multiple directions

1

Rather than a purely sequential scan through the patches, I feel it would make more sense to use take inputs from multiple patches at once. Obviously you are limited in...

Opportunities for consolidation?

1

I've been working on my own solution for working with retinal scans in python wtih the key idea of providing a common interface to build tooling on top of and...