📘 Overview

The PerplexitySampleEvaluator is an operator designed to calculate the perplexity of a given text using a Hugging Face language model. Perplexity is a measurement of how well a probability model predicts a sample; in natural language processing, a lower perplexity score indicates that the text is more fluent and understandable according to the model.

__init__ function

def __init__(self, model_name: str = 'gpt2', device='cuda')

Parameter	Type	Default	Description
model_name	str	'gpt2'	The path or name of the Hugging Face model to be used for calculation.
device	str	'cuda'	The device on which the model will run (e.g., 'cuda' or 'cpu').

Prompt Template Descriptions

Prompt Template Name	Primary Use	Applicable Scenarios	Feature Description

run function

def run(self, storage: DataFlowStorage, input_key: str = 'raw_content', output_key: str = 'PerplexityScore')

Parameter	Type	Default	Description
storage	DataFlowStorage	Required	The DataFlow storage instance for reading and writing data.
input_key	str	'raw_content'	The name of the input column containing the text to be evaluated.
output_key	str	'PerplexityScore'	The name of the output column where the calculated perplexity score will be stored.

🧠 Example Usage

🧾 Default Output Format

Field	Type	Description
raw_content	str	The input text.
PerplexityScore	float	The calculated perplexity score for the input text. Lower is better.

Example Input:

{
  "raw_content": "The quick brown fox jumps over the lazy dog."
}

Example Output:

{
  "raw_content": "The quick brown fox jumps over the lazy dog.",
  "PerplexityScore": 35.82
}