The goal of this assignment is to experiment with imitation learning, including direct behavior cloning and the DAgger algorithm. In lieu of a human demonstrator, demonstrations will be provided via an expert policy that we have trained for you. Your goals will be to set up behavior cloning and DAgger, and compare their performance on a few di erent continuous control tasks from the OpenAI Gym benchmark suite. Turn in your report and code as described in Section 4.
Section 1. Behavioral Cloning
The starter code provides an expert policy for each of the MuJoCo tasks in OpenAI Gym. Fill in the blanks in the code marked with Todo to implement behavioral cloning. A command for running behavioral cloning is given in the Readme le.
The following les have blanks in them and can be read in this order:
scripts/run hw1 behavior cloning.py infrastructure/rl trainer.py
infrastructure/replay buffer.py infrastructure/utils.py
Run behavioral cloning (BC) and report results on two tasks: one task where a behavioral cloning agent achieves at least 30% of the performance of the expert, and one task where it does not. When providing results, report the mean and standard deviation of the return over multiple rollouts in a table, and state which task was used. Be sure to set up a fair comparison, in terms of network size, amount of data, and number of training iterations, and provide these details (and any others you feel are appropriate) in the table caption.
Tip: to speed up run times, the video logging can be disabled by setting –video log freq -1
Experiment with one set of hyperparameter that a ects the performance of the behavioral cloning agent, such as the number of demonstrations, the number of training epochs, the variance of the expert policy, or something that you come up with yourself. For one of the tasks used in the previous question, show a graph of how the BC agent’s performance varies with the value of this hyperparameter, and state the hyperparameter and a brief rationale for why you chose it in the caption for the graph.
Section 2. DAgger
Implement DAgger by lling out all the remaining blanks in the code marked with Todo. A command for running DAgger is provided in the Readme le.
Run DAgger and report results on one task in which DAgger can learn a better policy than behavioral cloning. Report your results in the form of a learning curve, plotting the number of DAgger iterations vs. the policy’s mean return, with error bars to show the standard deviation. Include the performance of the expert policy and the behavioral cloning agent on the same plot. In the caption, state which task you used, and any details regarding network architecture, amount of data, etc. (as in the previous section).
Submitting the PDF Make a PDF report containing: Table 1 for a table of results from Question 1.2, and Figure 1 for Question 1.3. and Figure 2 with results from question 2.2.
You do not need to write anything else in the report, just include the gures with captions as de-scribed in each question above. See the handout at http://rail.eecs.berkeley.edu/deeprlcourse/ static/misc/viz.pdf for notes on how to generate plots.
Submitting the code and experiment runs In order to turn in your code and experiment logs, create a folder that contains the following:
A folder named run logs with at most one folder per environment for either the behav-ioral cloning (part 2, not part 3) or DAgger exercise. These folders can be copied directly from the cs285/data folder. Important: Disable video logging for the runs that you sub-mit, otherwise the les ize will be too large! You can do this by setting the ag –video log freq -1
The cs285 folder with all the .py les, with the same names and directory structure as the original homework repository. Also include any special instructions we need to run it to produce each of your gures or tables (e.g. \run python myassignment.py -sec2q1″ to generate the result for Section 2 Question 1) in the form of a README le.
As an example, the unzipped version of your submission should result in the following le structure.
Make sure that the submit.zip le is below 15MB.
dagger Ant-v2 03-09-2019 16-50-56
policy itr 0.data-00000-of-00001
bc Ant-v2 03-09-2019 16-50-56
policy itr 0.data-00000-of-00001
Turn in your assignment on Gradescope. Upload the zip le with your code and log les to HW1 Code, and upload the PDF of your report to HW1.