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
Input Tokens
0
Output Tokens
0
Duration
0
Cost
0
Context
Input
From the extracted description about an outfit. Extract info about the top piece of clothing (not the outerwear)
Expected output
{
  "type": "shirt",
  "color": "blue",
  "size": "M"
}
Model output
{
  "type": "linen shirt",
  "color": "medium blue",
  "size": "M"
}