Alfred Increment

icon indicating a website link https://alfredplpl.github.io/en/index.html icon indicating an email [email protected]

icon company CyberAgent, inc icon location Japan icon location Research Scientist. Interests: data science, machine learning, robotics, neuroscience

Results 5 comments of Alfred Increment

Any pre-trained model we can play with?

I have made a simple model. Please play it. [Sample repo.](https://github.com/alfredplpl/pretrained-diffusion-model-pytorch)

Support image-based observation spaces in same way as SB3

I seem this problem is seriously. How should we solve this problem. How we should able to activate VecTransposeImage?

Support image-based observation spaces in same way as SB3

The one of the solution is the following code. The we modify features_extractor_class of ActorCriticPolicy. However, It is not smart. ``` from stable_baselines3.common.torch_layers import BaseFeaturesExtractor class CNNExtractor(BaseFeaturesExtractor): def forward(self, observations:...

how generate custom motion sequence?

The Japanese engineer, peisuke, created the google colab to generate a dense pose video. https://colab.research.google.com/drive/1KjPpZun9EtlEMcFDEo93kFPqbL4ZmEOq?usp=sharing The result is here. https://x.com/peisuke/status/1732066240741671090?s=46&t=aBgVHjAMy0TFw0zYAE90WQ

Support GaLore Optimazer

Sorry. It does not have feasibly. It takes many days.