Cannot use a folder of demonstrations for Imitation Learning

Unity Engine
, ,
1

I am trying to create an agent to navigate a randomly generated maze and have been trying to record multiple demonstrations for my agent (as the maze is rarely ever the same) but when I enter the folder name into the .yml file, it only accepts one demonstration and gives me this error in the cmd prompt when I try enter the name of the folder, which is “Demos”. It does not work when I enter the full path either or any other way I have tried. Please help me :frowning:

This is what I am typing into the command line.
mlagents-learn config/NavigateMaze.yml --run-id=ImitationNavigation --force

And this is the error message I am receiving whenever I try entering the folder path.

File "X:\Unity\Projects\Production Project\venv\lib\site-packages\mlagents\trainers\demo_loader.py", line 214, in load_demonstration
agent_info_action.ParseFromString(data[pos : pos + next_pos])
google.protobuf.message.DecodeError: Error parsing message with type 'communicator_objects.AgentInfoActionPairProto'
2

Apologies, here is the .yml file.

default_settings: null
behaviors:
  Navigate Maze:
    trainer_type: ppo
    hyperparameters:
      batch_size: 1024
      buffer_size: 10240
      learning_rate: 0.0003
      beta: 0.005
      epsilon: 0.2
      lambd: 0.95
      num_epoch: 3
      shared_critic: false
      learning_rate_schedule: linear
      beta_schedule: linear
      epsilon_schedule: linear
    checkpoint_interval: 500000
    network_settings:
      normalize: false
      hidden_units: 128
      num_layers: 2
      vis_encode_type: simple
      memory: null
      goal_conditioning_type: hyper
      deterministic: false
    reward_signals:
      extrinsic:
        gamma: 0.99
        strength: 1.0
      gail:
        strength: 0.5
       demo_path: Demos // I have also tried the full path from the file explorer, put \Demos, /Demos, Demos\*.demo but nothing works
        network_settings:
          normalize: false
          hidden_units: 128
          num_layers: 2
          vis_encode_type: simple
          memory: null
          goal_conditioning_type: hyper
          deterministic: false
    init_path: null
    keep_checkpoints: 5
    even_checkpoints: false
    max_steps: 1000000
    time_horizon: 64
    summary_freq: 50000
    threaded: false
    self_play: null
    behavioral_cloning:
        strength: 0.5
        demo_path: Demos // The same with this too
env_settings:
  env_path: null
  env_args: null
  base_port: 5005
  num_envs: 1
  num_areas: 1
  timeout_wait: 60
  seed: -1
  max_lifetime_restarts: 10
  restarts_rate_limit_n: 1
  restarts_rate_limit_period_s: 60
engine_settings:
  width: 84
  height: 84
  quality_level: 5
  time_scale: 25
  target_frame_rate: -1
  capture_frame_rate: 60
  no_graphics: false
environment_parameters: null
checkpoint_settings:
  run_id: ppo
  initialize_from: null
  load_model: false
  resume: false
  force: false
  train_model: false
  inference: false
  results_dir: results
torch_settings:
  device: null
debug: false