version 0.1 pub

Author: Lr-2002

scripts/play.sh

@@ -2,7 +2,7 @@
 
 fps=15
 header=0
-reconstruction=0
+reconstruction=1
 save_mode=0
 mode="agent_in_env"
 
@@ -37,4 +37,5 @@ while [ "$1" != "" ]; do
     shift
 done
 
-python src/play.py hydra.run.dir=. hydra.output_subdir=null +mode="${mode}" +fps="${fps}" +header="${header}" +reconstruction="${reconstruction}" +save_mode="${save_mode}"
+python src/play.py hydra.run.dir=. hydra.output_subdir=null train +mode="${mode}" +fps="${fps}" +header="${header}" +reconstruction="${reconstruction}" +save_mode="${save_mode}"
+#python src/play.py env.train.id=homegrid-task hydra.run.dir=. hydra.output_subdir=null +mode="play_in_world_model" +fps="15" +header="1" +reconstruction="1" +save_mode="1"

src/play.py

@@ -40,7 +40,7 @@ def main(cfg: DictConfig):
                                  config=instantiate(cfg.world_model))
         actor_critic = ActorCritic(**cfg.actor_critic, act_vocab_size=test_env.num_actions)
         agent = Agent(tokenizer, world_model, actor_critic).to(device)
-        agent.load(Path('/root/root/iris/outputs/2024-01-17/16-47-18/checkpoints/last.pt'), device)
+        agent.load(Path('/root/iris/outputs/2024-01-20/14-49-13/checkpoints/last.pt'), device)
 
         if cfg.mode == 'play_in_world_model':
             env = WorldModelEnv(tokenizer=agent.tokenizer, world_model=agent.world_model, device=device, env=env_fn())