Marco Pleines

icon indicating an email [email protected]

icon location Germany icon location PostDoc at TU Dortmund. Researching memory capabilities of Deep Reinforcement Learning agents.

Results 63 comments of Marco Pleines

Contributing PPO + Transformer-XL

@vwxyzjn I finally resolved the issue. Only one linear layer was supposed to be between the CNN and the transformer blocks. Quiet surprising that this additional layer hampered performance a...

Introduce Segment Anything 2

Hi @jeanchristopheruel ! Glad that you made this PR! I'm currently playing around with it and I'm experiencing quiet long durations during interaction. From clicking to receiving the mask takes...

Has anyone attempted to deploy SAM-Hq on CVAT (following the deployment of SAM)

Hi @superNCS did you ever dealt with this again? I'm about do this.