File size: 1,628 Bytes
8fc2b4e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
# Evaluation

defaults:
  - config

hydra:
  run:
    dir: ${root_dir}

mode: val # 'val' or 'test'

# eval settings
agent: cliport
n_demos: 100 # number of val instances
train_demos: 100 # training demos used to train model
n_repeats: 1 # number of repeats
gpu: [0]
save_results: True # write results to json
update_results: False # overwrite existing json results?
checkpoint_type: 'val_missing'
val_on_heldout: True

disp: False
shared_memory: False
eval_task: packing-boxes-pairs-seen-colors # task to evaluate the model on
model_task: ${eval_task} # task the model was trained on (e.g. multi-language-conditioned or packing-boxes-pairs-seen-colors)
type: single # 'single' or 'multi'

# paths
model_dir: ${root_dir}
exp_folder: exps
data_dir: ${root_dir}/data
assets_root: ${root_dir}/cliport/environments/assets/

model_path: ${model_dir}/${exp_folder}/${model_task}-${agent}-n${train_demos}-train/checkpoints/ # path to pre-trained models
train_config: ${model_dir}/${exp_folder}/${model_task}-${agent}-n${train_demos}-train/.hydra/config.yaml # path to train config
save_path: ${model_dir}/${exp_folder}/${eval_task}-${agent}-n${train_demos}-train/checkpoints/ # path to save results
results_path: ${model_dir}/${exp_folder}/${eval_task}-${agent}-n${train_demos}-train/checkpoints/ # path to existing results


# record videos (super slow)
record:
  save_video: False
  save_video_path: ${model_dir}/${exp_folder}/${eval_task}-${agent}-n${train_demos}-train/videos/
  add_text: True
  fps: 20
  video_height: 640
  video_width: 720
  add_task_text: False
  blender_render: False # new: use blender recorder for rendering