Yen-Chen Lin

Below are some resources which may improve current implementation 1. Algorithm suggested [here](https://news.ycombinator.com/item?id=11296439) 2. [Prioritized Experience Replay](http://arxiv.org/abs/1511.05952) I will try to implement all these when I got time!

ERROR 403 when running wget to get pre-trained model

2

Hello @prafullasd , I tried runing the following command in README to download the pre-trained mode: ``` wget https://storage.googleapis.com/glow-demo/logs/lsun-bdr-[96/128].tar ``` but it shows: ``` HTTP request sent, awaiting response... 403...

Meaning of endpoint error

1

Hello, I'm trying to understand the meaning of [this function](https://github.com/studywolf/control/blob/master/studywolf_control/controllers/ilqr.py#L169-L183) in your iLQR code. I wonder why there is `np.cos` and `np.sin` involved. BTW, thanks for such a great code.

Fix syntax and dependency error for cifar

4

1. `pip install torchvision==0.4` results in the error: ``` ERROR: Could not find a version that satisfies the requirement torchvision==0.4 (from versions: 0.1.6, 0.1.7, 0.1.8, 0.1.9, 0.2.0, 0.2.1, 0.2.2, 0.2.2.post2,...

I've changed function **fabsf** to **fabs** to avoid error.

Speed Up

3

May I know the exact config (hyperparameters) to get an average speed of 14.2 it/s (on an RTX2080Ti, PyTorch 1.4.0, CUDA 9.2) reported [here](https://github.com/krrish94/nerf-pytorch/issues/6#issuecomment-615818834)? I couldn't get it by simply...

# Description The original implementation saves each attention layer's whole `K`, `V`. Instead, we can discard used `K`, `V` to ensure each GPU only holds the current and the next...