1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
|
- meta: >-
- Same as exp 9, but with longer chains (len 1000 after warm-up removal and splitting).
- Same observations as exp 9 (not regenerated)
- 5 block task, no E0 no-norm ll filter, no-norm in grammar and prod. probs. changed; more generic obligation (1 moved conditions, mostly 'any') + permission norms; 50 observations
- env_name: default
- colour_specific: ['r', 'g', 'b']
- shape_specific: ['square', 'circle', 'triangle']
- target_area: -0.8, 0.7 ; 0.25, 0.99
- true_norm:
- name: default
- exp: ['Norms', ['Obl', ['Moved', ['Colour', 'r'], ['Shape', 'any'], ['Zone', '1'], ['Next-Move', ['Colour', 'any'], ['Shape', 'any']]], ['Zone', '2']], ['Per', ['Action', 'putdown'], ['Colour', 'any'], ['Shape', 'square'], ['PerZone', '3']]]
- num_observations: 100 #Times to generate observations for task
- obs_data_set: 1
- n: 1000 #Length of sequence after discarding warm-up part and splitting in half
- m: 10 #Number of sequences after splitting in half
- rf: 0.6
- rhat_step_size: 50
|