search

mwhittaker / deeprl_project

Deep RL Final Project
1 stars 1 forks source link

Add plotting for PPO run #16

Open vlad17 opened 6 years ago

vlad17 commented 6 years ago

Deliverable: plot.py similar to what we had in the assignments that takes a log directory generated by PPO and spits out a learning rate curve.

This is going to involve transition from our in-class plotter plot.py.zip to the one from the openai logger, which should be pretty similar.

Deliverables: we should be able to do something like

mpirun -np 8 python run_atari_ppo.py --someflag --logdir=/tmp/1
mpirun -np 8 python run_atari_ppo.py --anotherflag --logdir=/tmp/2
python plot.py /tmp/1 /tmp/2 --value average_return --legend "someflag" "anotherflag" --outfile exp1.pdf
SamvitJ commented 6 years ago

Deliverable: short learning curve for video PPO, trained on only a couple of training steps

vlad17 commented 6 years ago

Blocked on #30 (to see if CPU is faster)