Problem type if statement seems to be wrong in text_classification.py?

xdong · November 2, 2022, 9:50pm

I was reading the source code in transformers\pipelines\text_classification.py

huggingface/transformers/blob/main/src/transformers/pipelines/text_classification.py

import warnings
from typing import Dict

import numpy as np

from ..utils import ExplicitEnum, add_end_docstrings, is_tf_available, is_torch_available
from .base import PIPELINE_INIT_ARGS, GenericTensor, Pipeline


if is_tf_available():
    from ..models.auto.modeling_tf_auto import TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING

if is_torch_available():
    from ..models.auto.modeling_auto import MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING


def sigmoid(_outputs):
    return 1.0 / (1.0 + np.exp(-_outputs))

This file has been truncated. show original

Line 175-178

            if self.model.config.problem_type == "multi_label_classification" or self.model.config.num_labels == 1:
                function_to_apply = ClassificationFunction.SIGMOID
            elif self.model.config.problem_type == "single_label_classification" or self.model.config.num_labels > 1:
                function_to_apply = ClassificationFunction.SOFTMAX

Altough when I was debugging the code, it seems problem_type is empty and it is the num_lables to decide which function to apply, but should it be SIGMOID for single label classification and SOFTMAX for multi-label?

Topic		Replies	Views
Different outputs when using pipeline Intermediate	2	1221	July 20, 2023
Sentiment Analysis Pipeline on single label function_to_apply not working 🤗Transformers	1	1006	March 17, 2022
Default `problem_type` Beginners	2	3192	May 25, 2022
Huggingface classification struggling with prediction 🤗Transformers	0	828	April 5, 2022
How to use Transformer XL for sequence classification? 🤗Transformers	2	590	October 6, 2021

Problem type if statement seems to be wrong in text_classification.py?

Related topics