Error while evaluating LPR agent

[lakshitadodeja]

Hi,

I have trained the LPR agent using the provided implementation and I am trying to evaluate the model using run_agent.py. It seems like the agent is terminating after one step. Here is the error log -

  File "/home/dodeja/Documents/bench/ARM/run_agent.py", line 150, in <module>
    app.run(main)
  File "/home/dodeja/miniconda3/envs/rlbench/lib/python3.9/site-packages/absl/app.py", line 308, in run
    _run_main(main, args)
  File "/home/dodeja/miniconda3/envs/rlbench/lib/python3.9/site-packages/absl/app.py", line 254, in _run_main
    sys.exit(main(argv))
  File "/home/dodeja/Documents/bench/ARM/run_agent.py", line 146, in main
    visualise(FLAGS.logdir, FLAGS.task, FLAGS.method, FLAGS.clipdir)
  File "/home/dodeja/Documents/bench/ARM/run_agent.py", line 132, in visualise
    _save_clips(clips, '%s_%s_%s' % (method, task, ep), clipdir)
  File "/home/dodeja/Documents/bench/ARM/run_agent.py", line 33, in _save_clips
    final_clip = concatenate_videoclips(clips)
  File "/home/dodeja/miniconda3/envs/rlbench/lib/python3.9/site-packages/moviepy/video/compositing/concatenate.py", line 75, in concatenate_videoclips
    w = max(r[0] for r in sizes)
ValueError: max() arg is an empty sequence

I tried to print out the observation information in the act_result variable -

ACT result  {'attention_coordinate_layer_0': array([-0.20625001,  0.09375   ,  1.19375   ], dtype=float32), 'front_pixel_coord': [34.0, 69.0], 'attention_coordinate_layer_1': array([-0.20000002,  0.03749999,  1.1625    ], dtype=float32), 'trans_action_indicies': array([1, 9, 9, 8, 3, 5], dtype=int32), 'rot_grip_action_indicies': array([ 4,  6, 35,  0]), 'trajectory': array([0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
       0., 0., 0., 0., 0.])}

The run agent is working fine when I am trying to evaluate my Behaviour cloning agent and C2FArm agent