Edouard Leurent

Since the reward of a transition from (s, a) depends on the next state s', there is no way around it: you have to step the environment to simulate the...

Hi, > The documentation says: sim freq is the count of visualized frames for each dt = (1/policy_freq) time This is incorrect, I'm sorry if the documentation was misleading, can...

Hi! i) Thanks for letting me know, will fix. ii) > Is this agent trained with the sb3_racetrack_ppo.py script from your repo ? Yes, I used SB3's PPO. > Or...

Hi, I just ran the sb3_highway_dqn.py script at head (20k steps), and here is what I get: logs ``` ---------------------------------- | rollout/ | | | ep_len_mean | 9.75 | |...

I completely agree! It should be quite straightforward to implement, I think, similarly to what is done for friendships.

The script in this project only implements data collection, and all these graph-related methods for layout, node ranking and clustering, etc. are already provided by Gephi . I just thought...

But thanks for the reference, it would also be worth having a full python-based pipeline for generating the final graph, so that the process is fully automated.

Yes, that is because the account is private

Hi @erickhun Yes, that could easily be done by replacing the calls to `api.GetFriends()` and `api.GetFollowers()` in `fetch_users` by a call to [`api.GetLists`](https://python-twitter.readthedocs.io/en/latest/_modules/twitter/api.html#Api.GetLists) followed by a call to [`api.GetListMembers`](https://python-twitter.readthedocs.io/en/latest/twitter.html#twitter.api.Api.GetListMembers)

> @eleurent thanks! Would you consider to merge if I do a PR on that? Of course! > Looks like using the --graph-nodes param would be appropriate? You're right. Actually,...