Vasileios Kochliaridis

PPOAgent Entropy Regularization, Clipping, GAE are working Incorrectly

3

I have been trying to implement a PPO Agent that solves LunarLander-v2 as in the official example in the github repo: https://github.com/tensorflow/agents/blob/master/tf_agents/agents/ppo/examples/v2/train_eval_clip_agent.py In this example, a PPOClip agent is used....

Support of Inverse Problems (e.g. Lorenz System)

2

Does torchdiffeq support inverse problem solving? For example, can it compute the sigma, rho, beta parameters in the Lorenz system? The solution of the system, with the known parameters (10,...

Allow passing conv_type from ActionDistributionNetwork to EncodingNetwork

1

EncodingNetwork currently allows 1D convolutions. However, the ActionDistributionNetwork class does not allow you to pass CONV_TYPE_1D to the parameter "conv_type" of the encoding network.

Conv1D Option for Networks

5

Hello, I have noticed that EncodingNetwork has set its conv_type to 2d by default, but there is no way to actually change this parameter. Am i missing anything? ``` @gin.configurable...

I am trying to fetch the last N hours trends data of a keyword. To do that, I use the Pytrends package, using the get_historical_interest() function. Even though Google Trends'...

The documentation of PPO describes the training process of PPO as the following: ``` # Build PPO agent ppo_agent = PPOClipAgent(num_epochs=40, ...) # Build Replay Buffer replay_buffer = TFUniformReplayBuffer(data_spec=ppo_agent.collect_data_spec,batch_size=env.batch_size, max_length=1000)...