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.

50
Duration
1m 41s
Input Tokens
469
Output Tokens
578
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",
    "silver",
    "black",
    "gray"
  ]
}