def _test_state_embed(self, mdnrnn_params, rl_params, use_gpu=False):
env, mdnrnn_trainer, embed_rl_dataset = create_mdnrnn_trainer_and_embed_dataset(
mdnrnn_params, use_gpu)
max_embed_seq_len = mdnrnn_params["run_details"]["seq_len"]
avg_reward_history, _, _, _, _ = run_gym(
rl_params,
use_gpu,
None, # score bar
embed_rl_dataset,
env.env,
mdnrnn_trainer.mdnrnn,
max_embed_seq_len,
)
return avg_reward_history
def _test_state_embed(
mdnrnn_params: OpenAiGymParameters, rl_params: OpenAiGymParameters
):
env, mdnrnn_trainer, embed_rl_dataset = create_mdnrnn_trainer_and_embed_dataset(
mdnrnn_params, rl_params.use_gpu
)
max_embed_seq_len = mdnrnn_params.run_details.seq_len
avg_reward_history, _, _, _, _ = run_gym(
rl_params,
None, # score bar
embed_rl_dataset,
env.env,
mdnrnn_trainer.mdnrnn,
max_embed_seq_len,
)
return avg_reward_history