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
12s
Input Tokens
1272
Output Tokens
639
Cost
$0.00
Context
Input
From the extracted description. Extra all info about the clothing
Expected output
{
  "top_piece": {
    "type": "shirt",
    "color": "blue",
    "size": "M"
  },
  "outerwear": {
    "type": "blazer",
    "color": "navy",
    "size": "L"
  },
  "bottom_piece": {
    "type": "chinos",
    "color": "beige",
    "size": "M"
  },
  "shoes": {
    "type": "brogues",
    "color": "brown",
    "size": "M"
  },
  "watch": {
    "material": "stainless steel",
    "strap_color": "black"
  },
  "hat": {
    "type": "flat cap",
    "color": "gray",
    "material": "wool"
  }
}
Model output
{
  "top_piece": {
    "type": "shirt",
    "color": "medium blue",
    "size": "M",
    "material": "linen"
  },
  "outerwear": {
    "type": "blazer",
    "color": "navy",
    "size": "L",
    "material": "wool"
  },
  "bottom_piece": {
    "type": "chinos",
    "color": "medium beige",
    "size": "M",
    "material": "cotton twill (with stretch)"
  },
  "shoes": {
    "type": "brogues",
    "color": "brown",
    "size": "M",
    "material": "leather"
  },
  "watch": {
    "material": "stainless steel (case), black leather (strap)",
    "strap_color": "black"
  },
  "hat": {
    "type": "flat cap",
    "color": "gray",
    "material": "wool"
  }
}