clinc150

This is a text classification dataset. It is intended for machine learning research and experimentation.

This dataset is obtained via formatting another publicly available data to be compatible with our AutoIntent Library.

Usage

It is intended to be used with our AutoIntent Library:

from autointent import Dataset

clinc150 = Dataset.from_datasets("AutoIntent/clinc150")

Source

This dataset is taken from cmaldona/All-Generalization-OOD-CLINC150 and formatted with our AutoIntent Library:

# define util
from datasets import load_dataset
from autointent import Dataset


def convert_clinc150(clinc150_train, shots_per_intent, oos_intent_name="ood"):
    intent_names = sorted(clinc150_train.unique("labels"))
    oos_intent_id = intent_names.index(oos_intent_name)
    intent_names = intent_names[:oos_intent_id] + intent_names[oos_intent_id + 1 :]
    n_classes = len(intent_names)
    assert n_classes == 150
    name_to_id = dict(zip(intent_names, range(n_classes), strict=False))

    oos_utterances = []
    classwise_utterance_records = [[] for _ in range(n_classes)]
    intents = [
        {
            "id": i,
            "name": name,
        }
        for name, i in name_to_id.items()
    ]

    for batch in clinc150_train.iter(batch_size=16, drop_last_batch=False):
        for txt, name in zip(batch["data"], batch["labels"], strict=False):
            if name == oos_intent_name:
                oos_utterances.append({"utterance": txt})
                continue
            intent_id = name_to_id[name]
            target_list = classwise_utterance_records[intent_id]
            if shots_per_intent is not None and len(target_list) >= shots_per_intent:
                continue
            target_list.append({"utterance": txt, "label": intent_id})

    train_utterances = [rec for lst in classwise_utterance_records for rec in lst]

    return Dataset.from_dict({"intents": intents, "train": train_utterances + oos_utterances})

# load and format
clinc150 = load_dataset("cmaldona/All-Generalization-OOD-CLINC150")
clinc150_converted = convert_clinc150(clinc150["train"], shots_per_intent=None)