ComprehendItPredictor

Predictor for knowledgator/comprehend_it models

This predictor is specifically build for knowledgator/comprehend_it models. For more details about models, see:

knowledgator/comprehend_it-base · Hugging Facehuggingface

About use cases and more, see:

The new milestone in zero-shot capabilities (it’s not Generative AI).Medium

Subclass of TransformersPipeline.

Module: implementation.predictors

Methods and properties

Main methods and properties

---

__init__

Arguments:

• cfg (Optional[TransformersPipelineConfig], optional): Configuration for predictor. If value equals to None, default ComprehendItPredictorConfig configuration will be used. Defaults to None.
• input_class (Type[Input], optional): Class for input validation. Defaults to ComprehendItPredictorInput.
• output_class (Type[Output], optional): Class for output validation. Defaults to ComprehendItPredictorOutput.
• name (Optional[str], optional): Name for identification. If equals to None, class name will be used. Defaults to None.

---

---

---

ComprehendItPredictorConfig

Prebuild configuration that describes default parameters for knowledgator/comprehend_it models pipeline. Subclass of TransformersPipelineConfig. It contains all parameters described in TransformersPipelineConfig but overwrites some default values that described here.

---

__init__

Arguments:

• task (Optional[str], optional): Pipeline task. Defaults to "zero-shot-classification".

• model (Optional[Union[str, PreTrainedModel, TFPreTrainedModel]], optional): Model that will be used. Defaults to "knowledgator/comprehend_it-base".
• kwargs (Optional[Dict[str, Any]], optional): Extra parameters. Defaults to

{
    "batch_size": 8
}

• ...

For other parameters, see TransformersPipelineConfig.

---

---

---

ComprehendItPredictorInput

Subclass of IOModel.

---

__init__

Arguments:

• sequences (str): Text to classify.

• candidate_labels (List[str]): Classification labels to use.

---

---

---

ComprehendItPredictorOutput

Subclass of IOModel.

---

__init__

Arguments:

• sequence (str): Text used for classification.

• labels (List[str]): Classified labels.

• scores (List[float]): Classification scores.

---

---