Data Normalization

Data processing and normalization tasks evaluate structured output from messy prose and different structures. This capability is essential for catalogue and product pipelines where data needs to be extracted from unstructured text and formatted consistently. Models must understand the desired output format and extract relevant information accurately.

0
Duration
21s
Input Tokens
3628
Output Tokens
65
Cost
$0.00
Context
Input
From the extracted description about an outfit. Extract info about the the bottom piece of clothing
Expected output
{
  "type": "chinos",
  "color": "beige",
  "size": "M"
}
Model output
{
  "type": "error",
  "color": "red",
  "size": "empty"
}