Implementation and report on a Reinforcement Learning problem solved by Deep RL.
Link to the Report.
Additionally, the Report is saved in the folder report along with the meeting log and the distribution of the work.
Running these below mentioned scripts creates a new directory inside experiments with algorithm and the
hyperparameters as the directory name, example sarsa_adagrad/ep0.1_be0.2_ga0.3_et0.4/
measurements_sarsa.shmeasurements_sarsa_adagrad.shmeasurements_q_learning.shmeasurements_ex_replay.sh
In addition to this, the hyperparameters are saved in a text file hyperparam.txt and the plots are also saved inside this directory.
Experience replay: in ex_replay.py incremental version is stored. Once the episode is finished we take the batch fromthe database and unfold it backwards recalculating Q values. In Assignement - default.py mini batch version is realized.
