Upload README.md with huggingface_hub

README.md

@@ -1,36 +1,43 @@
 ---
-dataset_info:
-  features:
-  - name: prompt
-    dtype: json
-  - name: completion
-    dtype: json
-  - name: wire_xml
-    dtype: string
-  - name: num_segments
-    dtype: int8
-  - name: points
-    dtype:
-      array2_d:
-        shape:
-        - null
-        - 3
-        dtype: float32
-  - name: mesh_binary
-    dtype: binary
-  - name: image
-    dtype: image
-  - name: depth_map
-    dtype: image
-  splits:
-  - name: train
-    num_bytes: 1935781248
-    num_examples: 10000
-  download_size: 1670013198
-  dataset_size: 1935781248
-configs:
-- config_name: default
-  data_files:
-  - split: train
-    path: data/train-*
+language: en
+license: mit
+size_categories:
+- 10K
+pretty_name: Random Wires
+tags:
+- synthetic
 ---
+
+## Generation Parameters
+
+| Parameter | Value |
+| --- | --- |
+| seed | 42 |
+| num points | 1000 |
+| test proportion | None |
+| include reasoning | False |
+
+
+## System Prompt
+```
+Please predict the structure of this shape using this XML format:
+
+- Use <wire> as the enclosing tags
+- wire must have a diameter attribute of type float
+- wire has two child elements, <transform> and <segments>
+- transform has two child elements, <pos> and <rot>.
+- <pos> and <rot> represent the position and rotation of the start of the wire respectively.
+- <pos> has x, y, and z attributes in the units shown on the images grid overaly.
+- the location of <pos> is shown by the colored axis on the image.
+- <rot> has x, y, z, and w attributes. They are the quaternion in degrees.
+- <segments> is a mixed list of <bend> and <straight> elements.
+- <segments> must have at least one child element.
+- <segments> is a sequence that starts from the colored axis.
+- <bend> tags have angle, radius, and roll attributes. Angle and roll are in degrees, radius is in world units shown by the grid.
+- <straight> just has a length attribute in grid units.
+- All attributes are floats to 1 decimal place.
+
+Your goal is to match the shape in the image as closely as possible by predicting the correct sequence of <straight> and <bend> tags and choosing the attribute values.
+
+```
+