-
## Abstract
This is a proposal to reduce latency caused by extra-long segments caused by variable GOP sizes in bad network conditions. We can force a fixed small segment-size during segmenting th…
-
Hi y'all! Thank you for providing Dopamine as it is such an awesome resource. I am looking for the PER algorithm itself, but I cannot find it in this repo, and what I see is the Rainbow agent which al…
-
Decide upon an agent architecture. Possible options are:
- DQNs
- Policy Gradient Methods
- Actor Critic Methods
Choose a few options, train them and choose a best policy for the final agent.
-
Submitted by @hcatlin on Spectrum
> So, let's say I launch on Booster and I have users signing up and they are posting content. I can see the content on the site and I'm like "YAY! Good work me! I …
-
Hi, is it possible to use PrioritizedReplayBuffer just by itself in my own DQN code? I would like to be able to append my own tuple of tensors to the replay buffer and sample from the buffer. From wha…
-
I want to use prioritized experience replay as a replay method and batch selection, instead of random selection. As mentioned in [this link](https://jaromiru.com/2016/11/07/lets-make-a-dqn-double-lear…
-
A laundry list of potential future work:
- Cross Entropy Method
- Heuristic preprocessing - using rules to override agent policies
- Auxilary loss functions - state or reward predictions
- DQN e…
-
The composition of Rainbow contains the following:
1.Double DQN
2.Prioritized Experience Replay
3.Dueling Network
4.Noisy Network
5.Categorical DQN
6.N-step Learning
Actually, only 2, 5, and 6 …
-
-
This is a long-living issue for the Atlas design team to post summaries of their work to share with the community