Haresh
The input image is normalized between 0-1, so the output should be a nn.Sigmoid() layer instead of nn.Tanh(). Note that nn.Tanh() may also work, but using nn.Sigmoid() is more appropriate...
In the current implementation, it is possible that there can be bad transitions in the rollout storage. The feedforward generator does not filter out these bad samples, when sampling imitator...
I noticed that the predict reward function uses log(D(.)) - log(1-D(.)) as the reward to update the generator. However, this is the reward function proposed in the AIRL paper which...
Hi, I just had a question about the dataset classes. In some of the classes like the cityscapes dataset, the labels are being mapped to negative keys. Why is that...
 deriving the epistemic uncertainty, from S12 to S13, the negative term disappears for the 2nd term [(beta/(alpha-1))/v]
Hi, Thanks for this repo ! I'm having issues with the robot not being tracked properly by the camera. As a test, I trained a HalfCheetah agent and I visualize...
Translated Index.njk from English to Tamil. Used google translate mostly and found it was pretty accurate, except at few places where I had to manually reword english sentences to make...
Hi, Thanks for making your code accessible!. I was wondering if it was possible you could also share the code to reproduce the gridworld experiments, specifically Fig 13 from your...
Hi, thanks for open-sourcing your code. I noticed that your text and image vectors which you used to compute the logits are not unit normalized vectors. [https://github.com/moein-shariatnia/OpenAI-CLIP/blob/e2c5bb3859d7478752af8c69862f63b1afe4a9cb/modules.py#L68](url) . In this...