Conditional Probability Classifiers

This module contains code to build a conditional probability classifier, which is inspired by this paper: https://arxiv.org/pdf/1911.06475.pdf

import pandas as pd

---

source

build_standard_condition_mask

 build_standard_condition_mask (df_labels, label1, label2)

_df_labels=pd.DataFrame({
    'col_1':[0,0,0,1,1,2,2,2],
    'col_2':[0,1,2,3,4,5,6,7]
})
_df_labels

# 0 -> (0,1,2), 1 -> (3,4), 2-> (5,6,7)

	col_1	col_2
0	0	0
1	0	1
2	0	2
3	1	3
4	1	4
5	2	5
6	2	6
7	2	7

print(build_standard_condition_mask(_df_labels,'col_1','col_2'))

tensor([[ True, False, False,  True,  True,  True, False, False, False, False,
         False],
        [False,  True, False, False, False, False,  True,  True, False, False,
         False],
        [False, False,  True, False, False, False, False, False,  True,  True,
          True]])

---

source

RobertaHSCCProbSequenceClassification

 RobertaHSCCProbSequenceClassification (config, size_l1=None,
                                        size_l2=None, standard_mask=None,
                                        layer2concat=4, device=None,
                                        head_class=None,
                                        **head_class_kwargs)

Roberta Conditional Probability Architecture with Hidden-State-Concatenation for Sequence Classification task

	Type	Default	Details
config			HuggingFace model configuration
size_l1	NoneType	None	Number of classes for head 1
size_l2	NoneType	None	Number of classes for head 2
standard_mask	NoneType	None	Mask for conditional probability
layer2concat	int	4	number of hidden layer to concatenate (counting from top)
device	NoneType	None	CPU or GPU
head_class	NoneType	None	The class object of the head. You can use RobertaClassificationHeadCustom as default
head_class_kwargs