Episodic-FewShotLearning-Intent-Classification-PyTorch/samplers.py at main · rishi2002/Episodic-FewShotLearning-Intent-Classification-PyTorch · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
import random

class BalancedSampler:
    def __init__(self, full_dataset, class_list, n_way, k_shot):
        self.class_list = class_list
        self.n_way = n_way
        self.k_shot = k_shot

        dataset = {}
        for label in class_list:
            dataset[label] = full_dataset[label]
        self.dataset = dataset

    def sample(self):
        support = {'texts':[],
                   'labels':[]}
        query = {'texts':[],
                 'labels':[]}
        n_classes = random.sample(self.class_list, self.n_way)
        for label in n_classes:
            k_samples = random.sample(self.dataset[label],2*self.k_shot)
            random.shuffle(k_samples)
            texts_s = k_samples[:self.k_shot]
            texts_q = k_samples[self.k_shot:]
            support['texts'].extend(texts_s)
            support['labels'].extend([label]*self.k_shot)
            query['texts'].extend(texts_q)
            query['labels'].extend([label]*self.k_shot)

        support_final = {}
        query_final = {}

        indices = list(range(len(support['texts'])))
        random.shuffle(indices)
        support_final['texts'] = [support['texts'] for i in indices]
        support_final['labels'] = [support['labels'] for i in indices]

        indices = list(range(len(query['texts'])))
        random.shuffle(indices)
        query_final['texts'] = [query['texts'] for i in indices]
        query_final['labels'] = [query['labels'] for i in indices]


        return support, query


class UnbalancedSampler:
    def __init__(self, full_dataset, class_list, n_way, k_shot):
        self.class_list = class_list

        dataset = {}
        for label in class_list:
            dataset[label] = full_dataset[label]
        self.dataset = dataset

    def get_n_way(self):
        n_way = random.randint(2, 5) #change this line to implement different sampling for n_way
        return n_way

    def get_k_shot(self, sampled_classes):
        k_shot = {}
        for label in sampled_classes:
            k_shot[label] = random.randint(1, 5) #change this line to imlpement different sampling for n_way
        return k_shot

    def sample(self):
        n_way = self.get_n_way()
        sampled_classes = random.sample(self.class_list, n_way)
        support_k_shot = self.get_k_shot(sampled_classes)
        query_k_shot = 5

        support = {'texts':[],
                   'labels':[]}
        query = {'texts':[],
                 'labels':[]}

        for label in sampled_classes:
            k_samples = random.sample(self.dataset[label],query_k_shot+support_k_shot[label])
            random.shuffle(k_samples)
            texts_s = k_samples[:support_k_shot[label]]
            texts_q = k_samples[support_k_shot[label]:]
            support['texts'].extend(texts_s)
            support['labels'].extend([label]*support_k_shot[label])
            query['texts'].extend(texts_q)
            query['labels'].extend([label]*query_k_shot)

        support_final = {}
        query_final = {}

        indices = list(range(len(support['texts'])))
        random.shuffle(indices)
        support_final['texts'] = [support['texts'] for i in indices]
        support_final['labels'] = [support['labels'] for i in indices]

        indices = list(range(len(query['texts'])))
        random.shuffle(indices)
        query_final['texts'] = [query['texts'] for i in indices]
        query_final['labels'] = [query['labels'] for i in indices]


        return support, query