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.

100
Duration
5s
Input Tokens
480
Output Tokens
147
Cost
$0.00
Context
Input
From the extracted description. Extract all colors found in the clothing
Expected output
{
  "colors": [
    "blue",
    "navy",
    "black",
    "beige",
    "brown",
    "gray"
  ]
}
Model output
{
  "colors": [
    "blue",
    "navy",
    "beige",
    "brown",
    "black",
    "gray"
  ]
}