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.

Duration

Input Tokens

1272

Output Tokens

297

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"
  },
  "outerwear": {
    "type": "blazer",
    "color": "navy",
    "size": "L"
  },
  "bottom_piece": {
    "type": "chinos",
    "color": "medium beige",
    "size": "M"
  },
  "shoes": {
    "type": "brogues",
    "color": "brown",
    "size": "M"
  },
  "watch": {
    "material": "silver stainless steel",
    "strap_color": "black"
  },
  "hat": {
    "type": "flat cap",
    "color": "gray",
    "material": "wool"
  }
}