Clean code; Refactor environment to pass screenshot content instead of path

run.py

@@ -50,12 +50,6 @@ logger.addHandler(sdebug_handler)
 
 logger = logging.getLogger("desktopenv.experiment")
 
-# wandb config
-### set your wandb api key here
-# os.environ["WANDB_API_KEY"] = "48ec18fb4da7087238c6d6833eab9907565adbf3"
-# wandb.login(key=os.environ.get("WANDB_API_KEY", None))
-
-
 def config() -> argparse.Namespace:
     parser = argparse.ArgumentParser(
         description="Run end-to-end evaluation on the benchmark"
@@ -153,9 +147,6 @@ def test(
 
     for domain in tqdm(test_all_meta, desc="Domain"):
         for example_id in tqdm(test_all_meta[domain], desc="Example", leave=False):
-            # run = wandb.init(project=f"OSworld-{args.action_space}-{args.observation_type}-{args.model}", group=f"{domain}",
-            #         name=f"{example_id}")
-            # example setting
             config_file = os.path.join(args.test_config_base_dir, f"examples/{domain}/{example_id}.json")
             with open(config_file, "r", encoding="utf-8") as f:
                 example = json.load(f)
@@ -186,19 +177,12 @@ def test(
                                                   scores)
             except Exception as e:
                 logger.error(f"Exception in {domain}/{example_id}: {e}")
-                # wandb.log({"Exception": wandb.Table(data=[[f"Exception in {domain}/{example_id}: {e}"]], columns=["Error"])})
                 env.controller.end_recording(os.path.join(example_result_dir, "recording.mp4"))
                 with open(os.path.join(example_result_dir, "traj.jsonl"), "a") as f:
                     f.write(json.dumps({
                         "Error": f"Time limit exceeded in {domain}/{example_id}"
                     }))
                     f.write("\n")
-            # wandb settings
-            # os.mkdir(os.path.join(wandb.run.dir, "results/"))
-            # for file in os.listdir(example_result_dir):
-            #     # move file to just under the root dir
-            #     os.rename(os.path.join(example_result_dir, file), os.path.join(wandb.run.dir, f"./results/{file}"))
-            # wandb.finish()
 
     env.close()
     logger.info(f"Average score: {sum(scores) / len(scores)}")