File size: 2,388 Bytes
ca4fc4d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import matplotlib.pyplot as plt
import time
import torch
from torch.utils.data import DataLoader
from torchvision import datasets, transforms
import numpy as np
import tracemalloc

# from Andromeda.model import Andromeda
from Andromeda.model import Andromeda
from Andromeda.utils.stable_adamw import StableAdamWUnfused

torch.manual_seed(0)
if torch.cuda.is_available():
    torch.cuda.manual_seed(0)

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")




import torch.nn.functional as F
from nltk.translate.bleu_score import corpus_bleu
from rouge import Rouge
from sklearn.metrics import f1_score


class AccuracyMetrics:
    def __init__(self):
        self.rouge = Rouge()
    
    def calculate_perplexity(self, model, data_loader):
        model.eval()
        total_loss = 0
        with torch.no_grad():
            for batch in data_loader:
                input_ids, labels = batch
                output = model(input_ids)
                loss = F.cross_entropy(output.view(-1, output.size(-1)), labels.view(-1))
                total_loss += loss.item()
        return torch.exp(torch.tensor(total_loss / len(data_loader)))
    
    def calculate_bleu(self, references, hypotheses):
        return corpus_bleu(references, hypotheses)
    
    def calculate_rouge(self, references, hypotheses):
        scores = self.rouge.get_scores(hypotheses, references, avg=True)
        return scores
    
    def calculate_f1(self, true_labels, pred_labels):
        return f1_score(true_labels, pred_labels, average="weighted")





#mock test dataset
test_dataset = datasets.FakeData(size=1000, transform=transforms.ToTensor())

#model
model = Andromeda(
    num_tokens=50304, 
    dim=1024,
    depth=24,
    dim_head=128,
    heads=8,
    alibi_num_heads=4
)



# Usage:
accuracy_metrics = AccuracyMetrics()

# Calculate Perplexity
perplexity = accuracy_metrics.calculate_perplexity(model, data_loader)
print('Perplexity:', perplexity)

# Calculate BLEU
bleu = accuracy_metrics.calculate_bleu(references, hypotheses)
print('BLEU Score:', bleu)

# Calculate ROUGE
rouge_scores = accuracy_metrics.calculate_rouge(references, hypotheses)
print('ROUGE Scores:', rouge_scores)

# Calculate F1 Score
f1 = accuracy_metrics.calculate_f1(true_labels, pred_labels)
print('F1 Score:', f1)




# Add at the bottom of your file
if __name__ == "__main__":
    AccuracyMetrics()