Edouard Leurent

Hi @OscarHuangWind, The image input and convolutional DQN netwok are already implemented in highway-env and rl-agents, respecrively. Here is a script allowing you to run it (also available as a...

The Grayscale observation relies on a pygame renderer to produce images of the scene, and pygame typically also interacts with I/O like displays and keyboard/mouse, which is probably why it...

Ah, I see. Yes, I agree that your proposal seems feasible: you should be able to copy the GrayscaleObservation class and have an instance created independently of the environment. Then,...

Sorry the late reply, and thanks @KexianShen for jumping in :) For more details, you can look at [this function](https://github.com/eleurent/rl-agents/blob/master/rl_agents/agents/common/models.py#L431) and subsequent classes to see how the config is parsed.

Yes, everything you said is absolutely correct - previous_state is currently only used to render information about the agent's decision-making process. In particular, when we need access to internal information...

Hi @SimoMaestri Very interesting results, thank you for sharing them! I think that although we can observe and describe _what_ these models do, it is often quite difficult to explain...

Actually, it looks like the 2nd policy does look at front vehicles but only when they are very close:    So the information of a possibly imminent collision...

Hi! It is indeed a bit tricky. I have a recommendation: you can start with the [implementation from openai/baselines](https://github.com/openai/baselines/blob/master/baselines/deepq/replay_buffer.py#L71). This is what I did for another project, and it worked...

During training, the agent model is saved through the `save_agent_model` function: https://github.com/eleurent/rl-agents/blob/a290be38351cf29c03779cb6683d831a06b74864/rl_agents/trainer/evaluation.py#L276 There are two cases where this function is used to automatically save the model in `after_some_episodes`: https://github.com/eleurent/rl-agents/blob/a290be38351cf29c03779cb6683d831a06b74864/rl_agents/trainer/evaluation.py#L318 1....

Oh, I see! Then you can just use `agent.save(model_path)` and `agent.load(model_path)`