Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix target-network-frequency in DQN documentation #329

Merged
merged 1 commit into from
Dec 9, 2022
Merged

Fix target-network-frequency in DQN documentation #329

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions docs/rl-algorithms/dqn.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -76,7 +76,7 @@ with the Bellman update target is $y = r + \gamma \, Q^{'}(s', a')$ and the repl
)
```
- `dqn_atari.py` uses `--learning-starts=80000` whereas (Mnih et al., 2015)[^1] (Exntended Data Table 1) uses `--learning-starts=50000`.
- `dqn_atari.py` uses `--target-network-frequency=1000` whereas (Mnih et al., 2015)[^1] (Exntended Data Table 1) uses `--learning-starts=10000`.
- `dqn_atari.py` uses `--target-network-frequency=1000` whereas (Mnih et al., 2015)[^1] (Exntended Data Table 1) uses `--target-network-frequency=10000`.
- `dqn_atari.py` uses `--total-timesteps=10000000` (i.e., 10M timesteps = 40M frames because of frame-skipping) whereas (Mnih et al., 2015)[^1] uses `--total-timesteps=50000000` (i.e., 50M timesteps = 200M frames) (See "Training details" under "METHODS" on page 6 and the related source code [run_gpu#L32](https://github.com/deepmind/dqn/blob/9d9b1d13a2b491d6ebd4d046740c511c662bbe0f/run_gpu#L32), [dqn/train_agent.lua#L81-L82](https://github.com/deepmind/dqn/blob/9d9b1d13a2b491d6ebd4d046740c511c662bbe0f/dqn/train_agent.lua#L81-L82), and [dqn/train_agent.lua#L165-L169](https://github.com/deepmind/dqn/blob/9d9b1d13a2b491d6ebd4d046740c511c662bbe0f/dqn/train_agent.lua#L165-L169)).
- `dqn_atari.py` uses `--end-e=0.01` (the final exploration epsilon) whereas (Mnih et al., 2015)[^1] (Exntended Data Table 1) uses `--end-e=0.1`.
- `dqn_atari.py` uses `--exploration-fraction=0.1` whereas (Mnih et al., 2015)[^1] (Exntended Data Table 1) uses `--exploration-fraction=0.02` (all corresponds to 250000 steps or 1M frames being the frame that epsilon is annealed to `--end-e=0.1` ).
Expand Down Expand Up @@ -336,4 +336,4 @@ Tracked experiments and game play videos:

[^1]:Mnih, V., Kavukcuoglu, K., Silver, D. et al. Human-level control through deep reinforcement learning. Nature 518, 529–533 (2015). https://doi.org/10.1038/nature14236
[^2]:\[Proposal\] Formal API handling of truncation vs termination. https://github.com/openai/gym/issues/2510
[^3]: Hessel, M., Modayil, J., Hasselt, H.V., Schaul, T., Ostrovski, G., Dabney, W., Horgan, D., Piot, B., Azar, M.G., & Silver, D. (2018). Rainbow: Combining Improvements in Deep Reinforcement Learning. AAAI.
[^3]: Hessel, M., Modayil, J., Hasselt, H.V., Schaul, T., Ostrovski, G., Dabney, W., Horgan, D., Piot, B., Azar, M.G., & Silver, D. (2018). Rainbow: Combining Improvements in Deep Reinforcement Learning. AAAI.