Implementation:Run llama Llama index LabelledTextDataset

Overview

This module defines data structures for simple text classification datasets in LlamaIndex. It provides LabelledSimpleDataset for storing labelled text examples with associated class labels, along with corresponding prediction and example classes. The dataset uses an LLM as its predictor type, unlike the RAG dataset which uses a query engine.

Source file: llama-index-core/llama_index/core/llama_dataset/simple.py (141 lines)

Class Hierarchy

BaseLlamaExamplePrediction
  └── SimpleExamplePrediction

BaseLlamaDataExample
  └── LabelledSimpleDataExample

BaseLlamaPredictionDataset
  └── SimplePredictionDataset

BaseLlamaDataset[LLM]
  └── LabelledSimpleDataset

SimpleExamplePrediction

class SimpleExamplePrediction(BaseLlamaExamplePrediction):

Stores the predicted output for a single text classification example.

Field	Type	Default	Description
label	str	""	The predicted label that can be compared to a reference ground-truth label

The class_name property returns "SimpleExamplePrediction".

SimplePredictionDataset

class SimplePredictionDataset(BaseLlamaPredictionDataset):

A collection of SimpleExamplePrediction instances.

to_pandas

def to_pandas(self) -> Any:

Converts predictions to a pandas DataFrame with a single column:

• label -- the predicted labels

Validates that all predictions match the _prediction_type using assertions. Raises ImportError if pandas is not installed.

The class_name property returns "SimplePredictionDataset".

LabelledSimpleDataExample

class LabelledSimpleDataExample(BaseLlamaDataExample):

Represents a single labelled text classification example.

Field	Type	Default	Description
reference_label	str	""	The ground-truth class label
text	str	""	The text body of the example
text_by	Optional[CreatedBy]	None	Indicates what generated the text (human or AI with model name)

The class_name property returns "LabelledSimpleDataExample".

LabelledSimpleDataset

class LabelledSimpleDataset(BaseLlamaDataset[LLM]):

The primary dataset class for text classification, parameterized with LLM as the predictor type.

to_pandas

def to_pandas(self) -> Any:

Converts the dataset to a pandas DataFrame with columns:

• reference_label -- ground-truth labels
• text -- text content
• text_by -- creation attribution (stringified)

Validates that all examples match _example_type, raising ValueError if not.

_construct_prediction_dataset

def _construct_prediction_dataset(
    self, predictions: Sequence[SimpleExamplePrediction]
) -> SimplePredictionDataset:

Factory method that wraps a sequence of predictions into a SimplePredictionDataset.

_apredict_example

async def _apredict_example(
    self,
    predictor: LLM,
    example: BaseLlamaDataExample,
    sleep_time_in_seconds: int,
) -> BaseLlamaExamplePrediction:

Raises NotImplementedError. This async prediction method has not yet been implemented.

_predict_example

def _predict_example(
    self,
    predictor: LLM,
    example: BaseLlamaDataExample,
    sleep_time_in_seconds: int = 0,
) -> BaseLlamaExamplePrediction:

Raises NotImplementedError. This synchronous prediction method has not yet been implemented.

Dependencies

• llama_index.core.llama_dataset.base -- provides BaseLlamaDataExample, BaseLlamaDataset, CreatedBy, BaseLlamaExamplePrediction, BaseLlamaPredictionDataset
• llama_index.core.llms.LLM -- used as the predictor type parameter
• llama_index.core.bridge.pydantic.Field -- Pydantic field descriptors