Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,107 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
task_categories:
|
| 4 |
+
- reinforcement-learning
|
| 5 |
+
- other
|
| 6 |
+
language:
|
| 7 |
+
- code
|
| 8 |
+
tags:
|
| 9 |
+
- minecraft
|
| 10 |
+
- expert-demonstrations
|
| 11 |
+
- skill-segmentation
|
| 12 |
+
- action-segmentation
|
| 13 |
+
- object-centric
|
| 14 |
+
pretty_name: Minecraft Skill Segmentation Dataset
|
| 15 |
+
size_categories:
|
| 16 |
+
- 1K<n<10K
|
| 17 |
+
---
|
| 18 |
+
# Dataset Card for Minecraft Expert Skill Data
|
| 19 |
+
This dataset consists of expert demonstration trajectories from a Minecraft simulation environment. Each trajectory includes ground-truth skill segmentation annotations, enabling research into action segmentation, skill discovery, imitation learning, and reinforcement learning with temporally-structured data.
|
| 20 |
+
|
| 21 |
+
## Dataset Details
|
| 22 |
+
|
| 23 |
+
### Dataset Description
|
| 24 |
+
|
| 25 |
+
The **Minecraft Skill Segmentation Dataset** contains gameplay trajectories from an expert policy in the Minecraft environment. Each trajectory is labeled with ground-truth skill boundaries and skill identifiers, allowing users to train and evaluate models for temporal segmentation, behavior cloning, and skill-based representation learning.
|
| 26 |
+
|
| 27 |
+
- **Curated by:** [dami2106]
|
| 28 |
+
- **License:** Apache 2.0
|
| 29 |
+
- **Language(s) (NLP):** Not applicable (code/visual)
|
| 30 |
+
|
| 31 |
+
### Dataset Sources
|
| 32 |
+
|
| 33 |
+
- **Repository:** https://github.com/dami2106/Minecraft-Skill-Data
|
| 34 |
+
|
| 35 |
+
## Uses
|
| 36 |
+
|
| 37 |
+
### Direct Use
|
| 38 |
+
|
| 39 |
+
This dataset is designed for use in:
|
| 40 |
+
- Training models to segment long-horizon behaviors into reusable skills.
|
| 41 |
+
- Evaluating action segmentation or hierarchical RL approaches.
|
| 42 |
+
- Studying object-centric or spatially grounded RL methods.
|
| 43 |
+
- Pretraining representations from visual expert data.
|
| 44 |
+
|
| 45 |
+
### Out-of-Scope Use
|
| 46 |
+
|
| 47 |
+
- Language-based tasks (no natural language data is included).
|
| 48 |
+
- Real-world robotics (simulation-only data).
|
| 49 |
+
- Tasks requiring raw image pixels if they are not included in your setup.
|
| 50 |
+
|
| 51 |
+
## Dataset Structure
|
| 52 |
+
|
| 53 |
+
Each data file includes:
|
| 54 |
+
- A sequence of states (e.g., pixel POV observations, PCA features).
|
| 55 |
+
- Skill labels marking where each skill begins and ends.
|
| 56 |
+
|
| 57 |
+
Example structure:
|
| 58 |
+
```json
|
| 59 |
+
{
|
| 60 |
+
"pixel_obs": [...], // Raw visual observations (e.g., RGB frames)
|
| 61 |
+
"pca_features": [...], // Compressed feature vectors (e.g., from CNN or ResNet)
|
| 62 |
+
"groundTruth": [...], // Ground-truth skill segmentation labels
|
| 63 |
+
"mapping": { // Mapping metadata for skill ID -> groundTruth label
|
| 64 |
+
"0": "chop_tree",
|
| 65 |
+
"1": "craft_table",
|
| 66 |
+
"2": "mine_stone",
|
| 67 |
+
...
|
| 68 |
+
}
|
| 69 |
+
}
|
| 70 |
+
```
|
| 71 |
+
|
| 72 |
+
## Dataset Creation
|
| 73 |
+
|
| 74 |
+
### Curation Rationale
|
| 75 |
+
|
| 76 |
+
This dataset was created to support research in skill discovery and temporal abstraction in complex, open-ended environments like Minecraft. The environment supports high-level goals and diverse interactions, making it suitable for testing generalizable skills.
|
| 77 |
+
|
| 78 |
+
### Source Data
|
| 79 |
+
|
| 80 |
+
#### Data Collection and Processing
|
| 81 |
+
|
| 82 |
+
- Expert trajectories were generated using a scripted or trained policy within the Minecraft simulation.
|
| 83 |
+
- Skill labels were added based on environment signals (e.g., changes to inventory, task completions, block state transitions) and verified using heuristics.
|
| 84 |
+
|
| 85 |
+
#### Who are the source data producers?
|
| 86 |
+
|
| 87 |
+
The data was generated programmatically in the Minecraft simulation environment by expert agents using scripted or learned behavior policies.
|
| 88 |
+
|
| 89 |
+
### Annotations
|
| 90 |
+
|
| 91 |
+
#### Annotation process
|
| 92 |
+
|
| 93 |
+
Skill annotations were derived from internal game state events and heuristics related to player intent and task segmentation. Manual inspection was performed to ensure consistency across trajectories.
|
| 94 |
+
|
| 95 |
+
#### Who are the annotators?
|
| 96 |
+
|
| 97 |
+
Automated rule-based annotation systems with developer oversight during dataset development.
|
| 98 |
+
|
| 99 |
+
## Bias, Risks, and Limitations
|
| 100 |
+
|
| 101 |
+
- The dataset is derived from simulation, so its findings may not generalize to real-world robotics or broader RL environments.
|
| 102 |
+
- Skill definitions depend on domain-specific heuristics, which may not reflect all valid strategies.
|
| 103 |
+
- Expert strategies may be biased toward specific pathways (e.g., speedrunning logic).
|
| 104 |
+
|
| 105 |
+
### Recommendations
|
| 106 |
+
|
| 107 |
+
Researchers should evaluate the robustness of learned skills across diverse environments and initial conditions. Segmentations reflect task approximations and should be interpreted within the scope of the simulation constraints.
|