Danijar Hafner: We introduce Dreamer, an RL agent that solves long-horizon tasks from images purely by latent imagination inside a world model. Dreamer improves over existing methods across 20 tasks.
Thread 👇 https://t.co/K5DnooVIUH
4 replies, 351 likes
hardmaru: Dream to Control: Learning Behaviors by Latent Imagination
The agent learns a latent world model via interactions, and backprops thru imagined latent trajectories of this model to learn useful behaviors. @danijarh et al.
code https://danijar.com/project/dreamer/ https://t.co/67ePHW3Z2T
4 replies, 194 likes
roadrunner01: Dream to Control: Learning Behaviors by Latent Imagination
project page: https://danijar.com/project/dreamer/
videos: https://dreamrl.github.io/ https://t.co/bezVZkB7lE
1 replies, 169 likes
Danijar Hafner: Proud to share our blog post on Dreamer, our latest scalable RL agent. Dreamer learns a world model from images & efficiently finds long-term behaviors by backprop through imagined states🚀
Videos https://dreamrl.github.io/ https://t.co/tUnq37eW89
1 replies, 110 likes
Animesh Garg: Nice work on model learning from @danijarh continuing his line of work on planet
Planet ... Dreamer....hmm 😁
1 replies, 13 likes
Shane Gu: Yet another excellent work on visual model-based RL from @danijarh. Excellent results showing how different design choices (e.g. gradient, GAE, contrastive vs reconstruction) lead to changes in performance.
0 replies, 7 likes
Nikolai Rozanov: I had a dream... https://arxiv.org/pdf/1912.01603.pdf
Muzero, now dreaming, pretty cool work - thanks for continuing to be original.
0 replies, 3 likes
Brundage Bot: Dream to Control: Learning Behaviors by Latent Imagination. Danijar Hafner, Timothy Lillicrap, Jimmy Ba, and Mohammad Norouzi http://arxiv.org/abs/1912.01603
1 replies, 1 likes
Found on Dec 04 2019 at https://arxiv.org/pdf/1912.01603.pdf