# shape (10-shape.com)

Shape is the outer form of something you can see or touch. People talk about the shape of a body, a cloud, a plan, or how things are arranged. It is a word about looks and structure, not about exact sizes or numbers.

## What a model may hear
- tensor shape (machine learning frameworks like PyTorch and TensorFlow): Assume the user wants to know or change the dimensions of a multi-dimensional array, and may return code with .shape or reshape operations
- data shape / schema (databases and data engineering): Expect the user is asking about column types, record structure, or schema validation, and may suggest schema tools or SQL
- geometric primitive (computer graphics and CAD): Treat the request as about drawing or rendering polygons, meshes, or vector forms, and may generate graphics code
- SHAPE attribute (HTML and web standards): Interpret the word as the shape attribute for image maps or CSS clip-path, and may return markup or styling rules

## Where people and models part ways
- Says: "What shape is my data in?"
  Means: The user wants to know if their data is messy or well-organized
  May be taken as: The model returns tensor dimensions or schema definitions instead of data quality advice
  Say instead: "Is my data clean and well-organized, or does it need fixing?"
- Says: "Shape the conversation"
  Means: The user wants to guide how a discussion goes
  May be taken as: The model thinks the user wants to reshape an array or modify a dialogue-state data structure
  Say instead: "Guide how this conversation should go"
- Says: "Give it shape"
  Means: The user wants to make something clearer or more definite
  May be taken as: The model offers code for geometric rendering or data structuring
  Say instead: "Make this clearer and more concrete"

## Tips
- Say 'outline' or 'form' if you mean the general idea, not dimensions
- Add 'not code' or 'in plain words' when you want a non-technical answer
- Say 'numpy shape' or 'tensor shape' when you do want the programming sense
- Use 'structure' for data organization if you do not mean schema details
- Say 'physical form' or 'appearance' to steer away from graphics primitives

## Often confused with
- form: More often a document or process than a geometric outline
- format: About file types and layouts, not outer appearance
- reshape: Almost always a data array operation in model contexts
- structure: About parts and their relations, not the outer boundary
- figure: A number or a body, not a general outline
- mold: A tool for casting, or a pattern of growth, not array dimensions