fix: observation filtering, action clipping, log_std clip, reward discount, bigger MLP

scripts/train.py

@@ -50,7 +50,7 @@ if __name__ == "__main__":
     run_name = f"{args.exp_name}__seed_{args.seed}__{git_hash}__{int(time.time())}"
 
     if args.run_dir is None:
-        run_dir = f"/data/gent/465/vsc46589/runs/{run_name}"
+        run_dir = f"runs/{run_name}"
     else:
         run_dir = args.run_dir