EpicPinkPenguin commited on
Commit
15ffc90
·
verified ·
1 Parent(s): 72853e9

Update data instances and data fields

Browse files
Files changed (1) hide show
  1. README.md +5 -5
README.md CHANGED
@@ -536,11 +536,10 @@ The PPO RL agent was trained for 25M steps on each environment and obtained the
536
 
537
  ## Dataset Structure
538
  ### Data Instances
539
- Each data instance represents a single step consisting of tuples of the form (observation, action, reward, done, truncated) = (o_t, a_t, r_{t+1}, done_{t+1}, trunc_{t+1}).
540
 
541
  ```json
542
  {'action': 1,
543
- 'done': False,
544
  'observation': [[[0, 166, 253],
545
  [0, 174, 255],
546
  [0, 170, 251],
@@ -568,15 +567,16 @@ Each data instance represents a single step consisting of tuples of the form (ob
568
  [0, 0, 37],
569
  [0, 0, 39]]],
570
  'reward': 0.0,
 
571
  'truncated': False}
572
  ```
573
 
574
  ### Data Fields
575
  - `observation`: The current RGB observation from the environment.
576
  - `action`: The action predicted by the agent for the current observation.
577
- - `reward`: The received reward from stepping the environment with the current action.
578
- - `done`: If the new observation is the start of a new episode. Obtained after stepping the environment with the current action.
579
- - `truncated`: If the new observation is the start of a new episode due to truncation. Obtained after stepping the environment with the current action.
580
 
581
  ### Data Splits
582
  The dataset is divided into a `train` (90%) and `test` (10%) split. Each environment-dataset has in sum 10M steps (data points).
 
536
 
537
  ## Dataset Structure
538
  ### Data Instances
539
+ Each data instance represents a single step consisting of tuples of the form (observation, action, reward, done, truncated) = (o_t, a_t, r_t, terminated_t, truncated_t).
540
 
541
  ```json
542
  {'action': 1,
 
543
  'observation': [[[0, 166, 253],
544
  [0, 174, 255],
545
  [0, 170, 251],
 
567
  [0, 0, 37],
568
  [0, 0, 39]]],
569
  'reward': 0.0,
570
+ 'terminated': False,
571
  'truncated': False}
572
  ```
573
 
574
  ### Data Fields
575
  - `observation`: The current RGB observation from the environment.
576
  - `action`: The action predicted by the agent for the current observation.
577
+ - `reward`: The received reward for the current observation.
578
+ - `terminated`: If the episode has terminated with the current observation.
579
+ - `truncated`: If the episode is truncated with the current observation.
580
 
581
  ### Data Splits
582
  The dataset is divided into a `train` (90%) and `test` (10%) split. Each environment-dataset has in sum 10M steps (data points).