Update README.md
Browse files
README.md
CHANGED
@@ -13,37 +13,37 @@ Commonsene Object Affordance Task [COAT]
|
|
13 |
</p>
|
14 |
|
15 |
### Experimental Setup:
|
16 |
-
- Task List: [tasks](https://
|
17 |
-
- Object List: [objects](https://
|
18 |
-
- Utility List[^1]: [utilities](https://
|
19 |
- Variables Used:
|
20 |
```temperature```, ```mass```, ```material```, ```already-in-use```, ```condition```
|
21 |
|
22 |
### Utility Level Pruning:
|
23 |
This gives us ```Utility```to``Object`` mappings also called ```utility objects```
|
24 |
-
- GT Object-Utility Mappings : [utility-mappings](https://
|
25 |
|
26 |
### Task-u(Utility Level):
|
27 |
Here we evaluate models on their ability to prune out appropriate objects on the basis of Utility.
|
28 |
-
- GT (Utility)-(Object) Mappings: [utility-objects](https://
|
29 |
- Task-u Dataset: [4 Variations](https://drive.google.com/drive/folders/1JJSIicKGp0a7ThsenKl0XWKsTtPL_b5z?usp=sharing)
|
30 |
|
31 |
### Task-0(Context Level):
|
32 |
Here we evaluate models on their ability to prune out appropriate objects on the basis of Context. This gives us ```(Task,Utility)```to``Object`` mappings also called ```context objects```
|
33 |
-
- GT (Task-Utility)-(Object) Mappings: [context-objects](https://
|
34 |
- Task-0 Dataset: [4 Variations](https://drive.google.com/drive/folders/1reH0JHhPM_tFzDMcAaJF0PycFMixfIbo?usp=sharing)
|
35 |
|
36 |
### Task-1(Physical State Level):
|
37 |
Here we evaluate models on their ability to prune out the ```ideal``` configuration when presented with a number of ```context object``` configurations. (Something that is pretty obvious to humans)
|
38 |
-
- All Possible Common Configurations: [possible configurations](https://
|
39 |
-
- Ideal Configurations: [ideal configurations](https://
|
40 |
-
- Commonsense Common Occurence Variables: [common variables values](https://
|
41 |
- Task-1 Dataset: [12 Variations](https://drive.google.com/drive/folders/1reH0JHhPM_tFzDMcAaJF0PycFMixfIbo?usp=sharing)
|
42 |
|
43 |
### Task-2(Physical State Level):
|
44 |
Here we evaluate models on their ability to prune out the most appropriate```sub-optimal``` configuration when presented with a number of sub-optimal configurations of ```context objects```. (Something that is pretty obvious to humans)
|
45 |
-
- Suboptimal Configurations: [suboptimal configurations](https://
|
46 |
-
- Human Preference Material Order: [material preference](https://
|
47 |
- Task-2 Dataset: [14 Variations](https://drive.google.com/drive/folders/1reH0JHhPM_tFzDMcAaJF0PycFMixfIbo?usp=sharing)
|
48 |
---------------------------------------------------------------------------------------------------------------
|
49 |
|
|
|
13 |
</p>
|
14 |
|
15 |
### Experimental Setup:
|
16 |
+
- Task List: [tasks](https://huggingface.co/datasets/Ayush8120/commonsense-embodied-affordance/blob/main/tasks.json)
|
17 |
+
- Object List: [objects](https://huggingface.co/datasets/Ayush8120/commonsense-embodied-affordance/blob/main/concepts.json)
|
18 |
+
- Utility List[^1]: [utilities](https://huggingface.co/datasets/Ayush8120/commonsense-embodied-affordance/blob/main/concepts.json)
|
19 |
- Variables Used:
|
20 |
```temperature```, ```mass```, ```material```, ```already-in-use```, ```condition```
|
21 |
|
22 |
### Utility Level Pruning:
|
23 |
This gives us ```Utility```to``Object`` mappings also called ```utility objects```
|
24 |
+
- GT Object-Utility Mappings : [utility-mappings](https://huggingface.co/datasets/Ayush8120/commonsense-embodied-affordance/blob/main/objects.json)
|
25 |
|
26 |
### Task-u(Utility Level):
|
27 |
Here we evaluate models on their ability to prune out appropriate objects on the basis of Utility.
|
28 |
+
- GT (Utility)-(Object) Mappings: [utility-objects](https://huggingface.co/datasets/Ayush8120/commonsense-embodied-affordance/blob/main/objects.json)
|
29 |
- Task-u Dataset: [4 Variations](https://drive.google.com/drive/folders/1JJSIicKGp0a7ThsenKl0XWKsTtPL_b5z?usp=sharing)
|
30 |
|
31 |
### Task-0(Context Level):
|
32 |
Here we evaluate models on their ability to prune out appropriate objects on the basis of Context. This gives us ```(Task,Utility)```to``Object`` mappings also called ```context objects```
|
33 |
+
- GT (Task-Utility)-(Object) Mappings: [context-objects](https://huggingface.co/datasets/Ayush8120/commonsense-embodied-affordance/blob/main/oracle.json)
|
34 |
- Task-0 Dataset: [4 Variations](https://drive.google.com/drive/folders/1reH0JHhPM_tFzDMcAaJF0PycFMixfIbo?usp=sharing)
|
35 |
|
36 |
### Task-1(Physical State Level):
|
37 |
Here we evaluate models on their ability to prune out the ```ideal``` configuration when presented with a number of ```context object``` configurations. (Something that is pretty obvious to humans)
|
38 |
+
- All Possible Common Configurations: [possible configurations](https://huggingface.co/datasets/Ayush8120/commonsense-embodied-affordance/blob/main/task-1/possible_configurations_v1.json)
|
39 |
+
- Ideal Configurations: [ideal configurations](https://huggingface.co/datasets/Ayush8120/commonsense-embodied-affordance/blob/main/task-1/pouch_config_oracle.json)
|
40 |
+
- Commonsense Common Occurence Variables: [common variables values](https://huggingface.co/datasets/Ayush8120/commonsense-embodied-affordance/blob/main/task-1/common_var_responses.json)
|
41 |
- Task-1 Dataset: [12 Variations](https://drive.google.com/drive/folders/1reH0JHhPM_tFzDMcAaJF0PycFMixfIbo?usp=sharing)
|
42 |
|
43 |
### Task-2(Physical State Level):
|
44 |
Here we evaluate models on their ability to prune out the most appropriate```sub-optimal``` configuration when presented with a number of sub-optimal configurations of ```context objects```. (Something that is pretty obvious to humans)
|
45 |
+
- Suboptimal Configurations: [suboptimal configurations](https://huggingface.co/datasets/Ayush8120/commonsense-embodied-affordance/blob/main/task-2/pouch_suboptimal.json)
|
46 |
+
- Human Preference Material Order: [material preference](https://huggingface.co/datasets/Ayush8120/commonsense-embodied-affordance/blob/main/task-2/material_preference.json)
|
47 |
- Task-2 Dataset: [14 Variations](https://drive.google.com/drive/folders/1reH0JHhPM_tFzDMcAaJF0PycFMixfIbo?usp=sharing)
|
48 |
---------------------------------------------------------------------------------------------------------------
|
49 |
|