Datasets:
Update README.md
Browse files
README.md
CHANGED
|
@@ -35,10 +35,10 @@ dataset_info:
|
|
| 35 |
splits:
|
| 36 |
- name: train
|
| 37 |
num_bytes: null
|
| 38 |
-
num_examples:
|
| 39 |
- name: val
|
| 40 |
num_bytes: null
|
| 41 |
-
num_examples:
|
| 42 |
- name: test
|
| 43 |
num_bytes: null
|
| 44 |
num_examples: null
|
|
@@ -60,19 +60,25 @@ train-eval-index:
|
|
| 60 |
|
| 61 |
### Dataset Description
|
| 62 |
|
| 63 |
-
The Minecraft Command Dataset is a JSONL dataset designed for training AI models
|
| 64 |
-
It contains
|
| 65 |
-
Each
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 66 |
|
| 67 |
-
- **Curated by:** Individual project
|
| 68 |
-
- **Shared via:** HuggingFace Hub
|
| 69 |
-
- **Language(s)
|
| 70 |
- **License:** CC0 1.0 Universal
|
| 71 |
|
| 72 |
### Dataset Structure
|
| 73 |
|
| 74 |
-
- JSONL
|
| 75 |
-
- Fields
|
| 76 |
|
| 77 |
```json
|
| 78 |
{
|
|
@@ -81,7 +87,7 @@ Each command entry contains a natural language explanation (`output`) and metada
|
|
| 81 |
"metadata": {
|
| 82 |
"command_type": "<scoreboard/execute/function/give/effect/tp/setblock>",
|
| 83 |
"difficulty": "<1-5, difficulty of the command>",
|
| 84 |
-
"components": ["<List of command components
|
| 85 |
"alternative_descriptions": ["<List of alternative natural language descriptions>"],
|
| 86 |
"player_or_virtual_user": "<Real player (@p/@a/@r/@s/Steve/Alex) or virtual (#Calc/#Bot)>",
|
| 87 |
"coordinates_type": "<absolute/relative, e.g., for ~ ~ ~ usage>",
|
|
@@ -89,4 +95,4 @@ Each command entry contains a natural language explanation (`output`) and metada
|
|
| 89 |
"version": "<large_70k/medium_40k/small_10k/tiny_1k>",
|
| 90 |
"timestamp": "<ISO 8601 timestamp>"
|
| 91 |
}
|
| 92 |
-
}
|
|
|
|
| 35 |
splits:
|
| 36 |
- name: train
|
| 37 |
num_bytes: null
|
| 38 |
+
num_examples: 50000
|
| 39 |
- name: val
|
| 40 |
num_bytes: null
|
| 41 |
+
num_examples: 5000
|
| 42 |
- name: test
|
| 43 |
num_bytes: null
|
| 44 |
num_examples: null
|
|
|
|
| 60 |
|
| 61 |
### Dataset Description
|
| 62 |
|
| 63 |
+
The Minecraft Command Dataset is a JSONL dataset designed for training AI models to understand Minecraft Data Pack commands.
|
| 64 |
+
It contains various command types (`scoreboard`, `execute`, `function`, `give`, `effect`, `tp`, `setblock`) with real and virtual users, items, effects, and blocks.
|
| 65 |
+
Each entry provides:
|
| 66 |
+
|
| 67 |
+
- **Input**: A natural language prompt including the command.
|
| 68 |
+
- **Output**: A non-overlapping natural language description of the command.
|
| 69 |
+
- **Metadata**: Additional structured information for AI understanding, including command type, difficulty, components, alternative descriptions, player type, coordinates type, dataset split, version, and timestamp.
|
| 70 |
+
|
| 71 |
+
This dataset is automatically generated based on randomized Minecraft commands, ensuring rich variation and coverage of gameplay mechanics.
|
| 72 |
|
| 73 |
+
- **Curated by:** Individual project
|
| 74 |
+
- **Shared via:** HuggingFace Hub
|
| 75 |
+
- **Language(s):** English, Korean
|
| 76 |
- **License:** CC0 1.0 Universal
|
| 77 |
|
| 78 |
### Dataset Structure
|
| 79 |
|
| 80 |
+
- **Format:** JSONL (one JSON object per line)
|
| 81 |
+
- **Fields:**
|
| 82 |
|
| 83 |
```json
|
| 84 |
{
|
|
|
|
| 87 |
"metadata": {
|
| 88 |
"command_type": "<scoreboard/execute/function/give/effect/tp/setblock>",
|
| 89 |
"difficulty": "<1-5, difficulty of the command>",
|
| 90 |
+
"components": ["<List of command components>"],
|
| 91 |
"alternative_descriptions": ["<List of alternative natural language descriptions>"],
|
| 92 |
"player_or_virtual_user": "<Real player (@p/@a/@r/@s/Steve/Alex) or virtual (#Calc/#Bot)>",
|
| 93 |
"coordinates_type": "<absolute/relative, e.g., for ~ ~ ~ usage>",
|
|
|
|
| 95 |
"version": "<large_70k/medium_40k/small_10k/tiny_1k>",
|
| 96 |
"timestamp": "<ISO 8601 timestamp>"
|
| 97 |
}
|
| 98 |
+
}
|