TransHP/engine.py at main · WangWenhao0716/TransHP · 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
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
# Copyright (c) 2015-present, Facebook, Inc.
# All rights reserved.
"""
Train and eval functions used in main.py
"""
import math
import sys
from typing import Iterable, Optional
import numpy as np
import pandas as pd

import torch

from timm.data import Mixup
from timm.utils import accuracy, ModelEma

from losses import DistillationLoss, HiLoss
import utils

import gc

def one_hot(x_1, x_2, num_classes, on_value_1, on_value_2, off_value=0.1, device='cuda'):
    x_1 = x_1.long().view(-1, 1)
    x_2 = x_2.long().view(-1, 1)
    #print(x_1 == x_2)
    return torch.full(
        (x_1.size()[0], num_classes), \
        off_value/num_classes, \
        device=device).scatter_\
        (1, x_1, on_value_1 - off_value/2 + off_value/num_classes).scatter_\
        (1, x_2, on_value_2 - off_value/2 + off_value/num_classes)*(x_1 != x_2) + \
        torch.full(
        (x_1.size()[0], num_classes), \
        off_value/num_classes, \
        device=device).scatter_\
        (1, x_1, on_value_1 + on_value_2 - off_value + off_value/num_classes)*(x_1 == x_2)

def train_one_epoch(model: torch.nn.Module, criterion: HiLoss,
                    data_loader: Iterable, optimizer: torch.optim.Optimizer,
                    device: torch.device, epoch: int, loss_scaler, max_norm: float = 0,
                    model_ema: Optional[ModelEma] = None, mixup_fn: Optional[Mixup] = None,
                    set_training_mode=True, args = None):
    model.train(set_training_mode)
    metric_logger = utils.MetricLogger(delimiter="  ")
    metric_logger.add_meter('lr', utils.SmoothedValue(window_size=1, fmt='{value:.6f}'))
    header = 'Epoch: [{}]'.format(epoch)
    print_freq = 10

    idx_to_class_dict = np.load('idx_to_class_dict.npy', allow_pickle=True).item()
    classes = np.array(list(idx_to_class_dict.values()))
    #df_hi_label = pd.read_csv('hi_label.csv')
    df_hi_label_idx = pd.read_csv('hi_label_idx.csv', index_col = 0)

    nb_classes_list = [2, 2, 2, 4, 9, 16, 25, 49, 90, 170, 406, 1000]

    for samples, targets in metric_logger.log_every(data_loader, print_freq, header):
        select_classes = classes[targets]
        targets_ls = list(targets.numpy())
        targets_list = df_hi_label_idx.loc[targets_ls].T.values.tolist()[::-1] + [targets_ls]
        '''
        for t in range(len(targets_list)):
            print("The %d layer"%t)
            print(targets_list[t])
        '''
        #masks_list = np.int64(np.array(targets_list) != -1)#.tolist()


        #targets_list = list(targets_list*(targets_list != -1)) # after getting masks, clean -1 to avoid cls error occuring.

        samples = samples.to(device, non_blocking=True)

        assert len(targets_list) == 12
        for i in range(len(targets_list)):
            #targets_list[i] = targets_list[i].to(device, non_blocking=True)
            #print(i, targets_list[i])
            for j in range(len(targets_list[i])):
                if targets_list[i][j] == -1:
                    targets_list[i][j] = nb_classes_list[i]
            #print(i, targets_list[i])

        #exit()
        targets_list = torch.Tensor(targets_list).type(torch.int64).to(device, non_blocking=True)


        with torch.cuda.amp.autocast():
            outputs_list = model(samples, targets_list)
            loss, loss_0, loss_1, loss_2, loss_3, loss_4, loss_5, loss_6, \
            loss_7, loss_8, loss_9, loss_10, loss_11 \
            = criterion(outputs_list, targets_list)

        loss_value = loss.item()
        loss_value_0 = loss_0.item()
        loss_value_1 = loss_1.item()
        loss_value_2 = loss_2.item()
        loss_value_3 = loss_3.item()
        loss_value_4 = loss_4.item()
        loss_value_5 = loss_5.item()
        loss_value_6 = loss_6.item()
        loss_value_7 = loss_7.item()
        loss_value_8 = loss_8.item()
        loss_value_9 = loss_9.item()
        loss_value_10 = loss_10.item()
        loss_value_11 = loss_11.item()

        if not math.isfinite(loss_value):
            print("Loss is {}, stopping training".format(loss_value))
            sys.exit(1)

        optimizer.zero_grad()

        # this attribute is added by timm on one optimizer (adahessian)
        # Wait for update

        is_second_order = hasattr(optimizer, 'is_second_order') and optimizer.is_second_order
        loss_scaler(loss, optimizer, clip_grad=max_norm,
                    parameters=model.parameters(), create_graph=is_second_order)

        torch.cuda.synchronize()
        if model_ema is not None:
            model_ema.update(model)

        metric_logger.update(loss=loss_value)
        metric_logger.update(loss_0=loss_value_0)
        metric_logger.update(loss_1=loss_value_1)
        metric_logger.update(loss_2=loss_value_2)
        metric_logger.update(loss_3=loss_value_3)
        metric_logger.update(loss_4=loss_value_4)
        metric_logger.update(loss_5=loss_value_5)
        metric_logger.update(loss_6=loss_value_6)
        metric_logger.update(loss_7=loss_value_7)
        metric_logger.update(loss_8=loss_value_8)
        metric_logger.update(loss_9=loss_value_9)
        metric_logger.update(loss_10=loss_value_10)
        metric_logger.update(loss_11=loss_value_11)
        metric_logger.update(lr=optimizer.param_groups[0]["lr"])
    # gather the stats from all processes
    metric_logger.synchronize_between_processes()
    print("Averaged stats:", metric_logger)
    return {k: meter.global_avg for k, meter in metric_logger.meters.items()}


@torch.no_grad()
def evaluate(data_loader, model, device):
    criterion = torch.nn.CrossEntropyLoss()

    metric_logger = utils.MetricLogger(delimiter="  ")
    header = 'Test:'

    # switch to evaluation mode
    model.eval()

    for images, target in metric_logger.log_every(data_loader, 10, header):
        images = images.to(device, non_blocking=True)
        target = target.to(device, non_blocking=True)

        # compute output
        with torch.cuda.amp.autocast():
            output = model(images, '_')
            loss = criterion(output, target)

        acc1, acc5 = accuracy(output, target, topk=(1, 5))

        batch_size = images.shape[0]
        metric_logger.update(loss=loss.item())
        metric_logger.meters['acc1'].update(acc1.item(), n=batch_size)
        metric_logger.meters['acc5'].update(acc5.item(), n=batch_size)
    # gather the stats from all processes
    metric_logger.synchronize_between_processes()
    print('* Acc@1 {top1.global_avg:.3f} Acc@5 {top5.global_avg:.3f} loss {losses.global_avg:.3f}'
          .format(top1=metric_logger.acc1, top5=metric_logger.acc5, losses=metric_logger.loss))

    return {k: meter.global_avg for k, meter in metric_logger.meters.items()}