job_name_lpgftw_seed + '/iterations/task_{}/'.format(t) +
'baseline_0.pickle', 'rb')
baseline_mtl[t] = pickle.load(f)
f.close()
if isinstance(policy_mtl.model.theta, list):
policy_mtl.model.theta = torch.autograd.Variable(torch.zeros(0))
agent_mtl = NPGFTW(e,
policy_mtl,
baseline_mtl,
normalized_step_size=0.1,
seed=SEED,
save_logs=False,
new_col_mode='performance')
mean_test_perf = agent_mtl.test_tasks(test_rollouts=10,
num_cpu=num_cpu,
task_ids=np.array([t]))
forward_transfer_results = {
**forward_transfer_results,
**mean_test_perf
}
result_file = open(job_name_lpgftw_seed + '/start_results.txt', 'w')
result_file.write(str(forward_transfer_results))
result_file.close()
SEED += 10
f = open(job_name_lpgftw_seed + '/trained_mtl_baseline.pickle', 'wb')
pickle.dump(baseline_mtl, f)
f.close()
f = open(job_name_lpgftw_seed + '/trained_mtl_alphas.pickle', 'wb')
pickle.dump(agent_mtl.theta, f)
f.close()
f = open(job_name_lpgftw_seed + '/trained_mtl_grads.pickle', 'wb')
pickle.dump(agent_mtl.grad, f)
f.close()
f = open(job_name_lpgftw_seed + '/trained_mtl_hess.pickle', 'wb')
pickle.dump(agent_mtl.hess, f)
f.close()
f = open(job_name_lpgftw_seed + '/task_order.pickle', 'wb')
pickle.dump(task_order, f)
f.close()
make_multitask_train_plots(loggers=agent_mtl.logger,
keys=['stoc_pol_mean'],
save_loc=job_name_lpgftw_seed + '/logs/')
mean_test_perf = agent_mtl.test_tasks(test_rollouts=10, num_cpu=num_cpu)
result = np.mean(list(mean_test_perf.values()))
print(result)
make_multitask_test_plots(mean_test_perf,
save_loc=job_name_lpgftw_seed + '/')
result_file = open(job_name_lpgftw_seed + '/results.txt', 'w')
result_file.write(str(mean_test_perf))
result_file.close()
SEED += 10
