os.path.join(this_dir, 'msp_no_sim', 'path_{}.png'.format(it)))
learnable_anchor_vals = optimizer.step(
Container({'w': learnable_anchor_vals}), grads)['w']
it += 1
sim.execute_motion(joint_query_vals)
sim.close()
unset_framework()
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument(
'--non_interactive',
action='store_true',
help='whether to run the demo in non-interactive mode.')
parser.add_argument(
'--no_sim',
action='store_true',
help=
'whether to run the demo without attempt to use the PyRep simulator.')
parser.add_argument(
'--framework',
type=str,
default=None,
help=
'which framework to use. Chooses a random framework if unspecified.')
parsed_args = parser.parse_args()
framework = None if parsed_args.framework is None else get_framework_from_str(
parsed_args.framework)
main(not parsed_args.non_interactive, not parsed_args.no_sim, framework)
print('\nIterations: {} Mean Score: {}\n'.format(iteration + 1, np.mean(scores)))
scores.clear()
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--no_visuals', action='store_true',
help='whether to run the demo without rendering images.')
parser.add_argument('--env', default='CartPole',
choices=['CartPole', 'Pendulum', 'MountainCar', 'Reacher', 'Swimmer'])
parser.add_argument('--framework', type=str, default=None,
choices=['jax', 'tensorflow', 'torch', 'mxnd', 'numpy'])
parser.add_argument('--steps', type=int, default=100)
parser.add_argument('--iters', type=int, default=10000)
parser.add_argument('--lr', type=float, default=0.001)
parser.add_argument('--seed', type=int, default=0)
parser.add_argument('--log_freq', type=int, default=100)
parser.add_argument('--vis_freq', type=int, default=1000)
parsed_args = parser.parse_args()
if parsed_args.framework is None:
framework = choose_random_framework(excluded=['numpy'])
else:
framework = get_framework_from_str(parsed_args.framework)
if parsed_args.framework == 'numpy':
raise Exception('Invalid framework selection. Numpy does not support auto-differentiation.\n'
'This demo involves gradient-based optimization, and so auto-diff is required.\n'
'Please choose a different backend framework.')
print('\nTraining for {} iterations.\n'.format(parsed_args.iters))
main(parsed_args.env, parsed_args.steps, parsed_args.iters, parsed_args.lr, parsed_args.seed,
parsed_args.log_freq, parsed_args.vis_freq, not parsed_args.no_visuals, framework)
