refactor(Modify.py): 重构代码并改进模型训练流程

- 重构了代码结构，优化了导入顺序和格式
- 改进了模型训练流程，添加了早停机制和学习率调度器- 增加了模型测试和可视化部分的代码
-优化了量子卷积层和模型的实现
- 调整了训练参数和数据预处理方法

Modify.py

@@ -1,291 +1,508 @@
-# Modify.py
-
-#%% 导入所有需要的包
+#%%
+# 首先我们导入所有需要的包：
 import os
 import random
 
 import numpy as np
 import pandas as pd
-import deepquantum as dq
-import matplotlib.pyplot as plt
 import torch
 import torch.nn as nn
 import torch.optim as optim
 import torchvision.transforms as transforms
-from torchvision.datasets import FashionMNIST
 from tqdm import tqdm
+from sklearn.metrics import roc_auc_score
 from torch.utils.data import DataLoader
-from multiprocessing import freeze_support
+from torchvision.datasets import FashionMNIST
+import deepquantum as dq
+import matplotlib.pyplot as plt
 
-#%% 设置随机种子以保证可复现
 def seed_torch(seed=1024):
+    """
+    Set random seeds for reproducibility.
+
+    Args:
+        seed (int): Random seed number to use. Default is 1024.
+    """
+
     random.seed(seed)
     os.environ['PYTHONHASHSEED'] = str(seed)
     np.random.seed(seed)
     torch.manual_seed(seed)
     torch.cuda.manual_seed(seed)
+
+    # Seed all GPUs with the same seed if using multi-GPU
     torch.cuda.manual_seed_all(seed)
     torch.backends.cudnn.benchmark = False
     torch.backends.cudnn.deterministic = True
 
-#%% 准确率计算函数
+seed_torch(42)  # 使用更常见的随机种子值
+#%%
 def calculate_score(y_true, y_preds):
+    # 将模型预测结果转为概率分布
     preds_prob = torch.softmax(y_preds, dim=1)
+    # 获得预测的类别（概率最高的一类）
     preds_class = torch.argmax(preds_prob, dim=1)
+    # 计算准确率
     correct = (preds_class == y_true).float()
-    return (correct.sum() / len(correct)).cpu().numpy()
+    accuracy = correct.sum() / len(correct)
+    return accuracy.cpu().numpy()
 
-#%% 训练与验证函数
-def train_model(model, criterion, optimizer, scheduler, train_loader, valid_loader, num_epochs, device, save_path):
-    model.to(device)
-    best_acc = 0.0
-    metrics = {'epoch': [], 'train_acc': [], 'valid_acc': [], 'train_loss': [], 'valid_loss': []}
 
-    for epoch in range(1, num_epochs + 1):
-        # --- 训练阶段 ---
+def train_model(model, criterion, optimizer, train_loader, valid_loader, num_epochs, device):
+    """
+    训练和验证模型。
+
+    Args:
+        model (torch.nn.Module): 要训练的模型。
+        criterion (torch.nn.Module): 损失函数。
+        optimizer (torch.optim.Optimizer): 优化器。
+        train_loader (torch.utils.data.DataLoader): 训练数据加载器。
+        valid_loader (torch.utils.data.DataLoader): 验证数据加载器。
+        num_epochs (int): 训练的epoch数。
+
+    Returns:
+        model (torch.nn.Module): 训练后的模型。
+    """
+
     model.train()
-        running_loss, running_acc = 0.0, 0.0
-        for imgs, labels in train_loader:
-            imgs, labels = imgs.to(device), labels.to(device)
+    train_loss_list = []
+    valid_loss_list = []
+    train_acc_list = []
+    valid_acc_list = []
+
+    best_valid_acc = 0.0
+    patience = 10  # 早停耐心值
+    counter = 0   # 计数器
+
+    scheduler = optim.lr_scheduler.ReduceLROnPlateau(optimizer, mode='max', factor=0.5, patience=10)
+
+    with tqdm(total=num_epochs) as pbar:
+        for epoch in range(num_epochs):
+            # 训练阶段
+            train_loss = 0.0
+            train_acc = 0.0
+            for images, labels in train_loader:
+                images = images.to(device)
+                labels = labels.to(device)
                 optimizer.zero_grad()
-            outputs = model(imgs)
+                outputs = model(images)
                 loss = criterion(outputs, labels)
                 loss.backward()
-            torch.nn.utils.clip_grad_norm_(model.parameters(), max_norm=1.0)
                 optimizer.step()
-            running_loss += loss.item()
-            running_acc += calculate_score(labels, outputs)
+                train_loss += loss.item()
+                train_acc += calculate_score(labels, outputs)
 
-        train_loss = running_loss / len(train_loader)
-        train_acc = running_acc / len(train_loader)
-        scheduler.step()
+            train_loss /= len(train_loader)
+            train_acc /= len(train_loader)
 
-        # --- 验证阶段 ---
+            # 验证阶段
             model.eval()
-        val_loss, val_acc = 0.0, 0.0
+            valid_loss = 0.0
+            valid_acc = 0.0
             with torch.no_grad():
-            for imgs, labels in valid_loader:
-                imgs, labels = imgs.to(device), labels.to(device)
-                outputs = model(imgs)
+                for images, labels in valid_loader:
+                    images = images.to(device)
+                    labels = labels.to(device)
+                    outputs = model(images)
                     loss = criterion(outputs, labels)
-                val_loss += loss.item()
-                val_acc += calculate_score(labels, outputs)
+                    valid_loss += loss.item()
+                    valid_acc += calculate_score(labels, outputs)
 
-        valid_loss = val_loss / len(valid_loader)
-        valid_acc = val_acc / len(valid_loader)
+            valid_loss /= len(valid_loader)
+            valid_acc /= len(valid_loader)
 
-        metrics['epoch'].append(epoch)
-        metrics['train_loss'].append(train_loss)
-        metrics['valid_loss'].append(valid_loss)
-        metrics['train_acc'].append(train_acc)
-        metrics['valid_acc'].append(valid_acc)
+            # 学习率调度器更新
+            scheduler.step(valid_acc)
 
-        tqdm.write(f"[{save_path}] Epoch {epoch}/{num_epochs}  "
-                   f"Train Acc: {train_acc:.4f}  Valid Acc: {valid_acc:.4f}")
+            # 早停机制
+            if valid_acc > best_valid_acc:
+                best_valid_acc = valid_acc
+                torch.save(model.state_dict(), './data/notebook2/best_model.pt')
+                counter = 0
+            else:
+                counter += 1
+
+            if counter >= patience:
+                print(f'Early stopping at epoch {epoch+1} due to no improvement in validation accuracy.')
+                break
+
+            pbar.set_description(f"Train loss: {train_loss:.3f}  Valid Acc: {valid_acc:.3f}")
+            pbar.update()
+
+
+            train_loss_list.append(train_loss)
+            valid_loss_list.append(valid_loss)
+            train_acc_list.append(train_acc)
+            valid_acc_list.append(valid_acc)
+
+    # 加载最佳模型权重
+    if os.path.exists('./data/notebook2/best_model.pt'):
+        model.load_state_dict(torch.load('./data/notebook2/best_model.pt'))
+
+    # 修改metrics构建方式，确保各数组长度一致
+    metrics = {
+        'epoch': list(range(1, len(train_loss_list) + 1)),
+        'train_acc': train_acc_list,
+        'valid_acc': valid_acc_list,
+        'train_loss': train_loss_list,
+        'valid_loss': valid_loss_list
+    }
 
-        if valid_acc > best_acc:
-            best_acc = valid_acc
-            torch.save(model.state_dict(), save_path)
 
     return model, metrics
 
-#%% 测试函数
 def test_model(model, test_loader, device):
-    model.to(device).eval()
-    acc = 0.0
+    model.eval()
+    test_acc = 0.0
     with torch.no_grad():
-        for imgs, labels in test_loader:
-            imgs, labels = imgs.to(device), labels.to(device)
-            outputs = model(imgs)
-            acc += calculate_score(labels, outputs)
-    acc /= len(test_loader)
-    print(f"Test Accuracy: {acc:.4f}")
-    return acc
+        for images, labels in test_loader:
+            images = images.to(device)
+            labels = labels.to(device)
+            outputs = model(images)
+            test_acc += calculate_score(labels, outputs)
 
-#%% 定义量子卷积层与模型
-singlegate_list = ['rx','ry','rz','s','t','p','u3']
-doublegate_list = ['rxx','ryy','rzz','swap','cnot','cp','ch','cu','ct','cz']
+        test_acc /= len(test_loader)
+        print(f'Test Acc: {test_acc:.3f}')
+    return test_acc
+#%%
+# 定义图像变换
+trans1 = transforms.Compose([
+    transforms.RandomHorizontalFlip(),  # 随机水平翻转
+    transforms.RandomRotation(10),     # 随机旋转±10度
+    transforms.ColorJitter(brightness=0.2, contrast=0.2),  # 颜色调整
+    transforms.Resize((18, 18)),       # 调整大小为18x18
+    transforms.ToTensor(),           # 转换为张量
+    transforms.Normalize((0.5,), (0.5,))  # 归一化到[-1, 1]
+])
 
+trans2 = transforms.Compose([
+    transforms.RandomHorizontalFlip(),  # 随机水平翻转
+    transforms.RandomRotation(10),     # 随机旋转±10度
+    transforms.ColorJitter(brightness=0.2, contrast=0.2),  # 颜色调整
+    transforms.Resize((16, 16)),       # 调整大小为16x16
+    transforms.ToTensor(),           # 转换为张量
+    transforms.Normalize((0.5,), (0.5,))  # 归一化到[-1, 1]
+])
+train_dataset = FashionMNIST(root='./data/notebook2', train=False, transform=trans1,download=True)
+test_dataset = FashionMNIST(root='./data/notebook2', train=False, transform=trans1,download=True)
+
+# 定义训练集和测试集的比例
+train_ratio = 0.8  # 训练集比例为80%，验证集比例为20%
+valid_ratio = 0.2
+total_samples = len(train_dataset)
+train_size = int(train_ratio * total_samples)
+valid_size = int(valid_ratio * total_samples)
+
+# 分割训练集和测试集
+train_dataset, valid_dataset = torch.utils.data.random_split(train_dataset, [train_size, valid_size])
+
+# 加载随机抽取的训练数据集
+train_loader = DataLoader(train_dataset, batch_size=64, shuffle=True, drop_last=True)
+valid_loader = DataLoader(valid_dataset, batch_size=64, shuffle=False, drop_last=True)
+test_loader = DataLoader(test_dataset, batch_size=64, shuffle=False, drop_last=True)
+#%%
+singlegate_list = ['rx', 'ry', 'rz', 's', 't', 'p', 'u3']
+doublegate_list = ['rxx', 'ryy', 'rzz', 'swap', 'cnot', 'cp', 'ch', 'cu', 'ct', 'cz']
+#%%
+# 随机量子卷积层
 class RandomQuantumConvolutionalLayer(nn.Module):
-    def __init__(self, nqubit, num_circuits, seed=1024):
-        super().__init__()
+    def __init__(self, nqubit, num_circuits, seed:int=1024):
+        super(RandomQuantumConvolutionalLayer, self).__init__()
         random.seed(seed)
         self.nqubit = nqubit
         self.cirs = nn.ModuleList([self.circuit(nqubit) for _ in range(num_circuits)])
+
     def circuit(self, nqubit):
         cir = dq.QubitCircuit(nqubit)
-        cir.rxlayer(encode=True); cir.barrier()
-        for _ in range(3):
-            for i in range(nqubit):
-                getattr(cir, random.choice(singlegate_list))(i)
-            c,t = random.sample(range(nqubit),2)
-            gate = random.choice(doublegate_list)
-            if gate[0] in ['r','s']:
-                getattr(cir, gate)([c,t])
-            else:
-                getattr(cir, gate)(c,t)
+        cir.rxlayer(encode=True)  # 对原论文的量子线路结构并无影响，只是做了一个数据编码的操作
         cir.barrier()
+        for iter in range(3):
+            for i in range(nqubit):
+                singlegate = random.choice(singlegate_list)
+                getattr(cir, singlegate)(i)
+            control_bit, target_bit = random.sample(range(0, nqubit - 1), 2)
+            doublegate = random.choice(doublegate_list)
+            if doublegate[0] in ['r', 's']:
+                getattr(cir, doublegate)([control_bit, target_bit])
+            else:
+                getattr(cir, doublegate)(control_bit, target_bit)
+            cir.barrier()
+
         cir.observable(0)
         return cir
-    def forward(self, x):
-        k,s = 2,2
-        x_unf = x.unfold(2,k,s).unfold(3,k,s)
-        w = (x.shape[-1]-k)//s + 1
-        x_r = x_unf.reshape(-1, self.nqubit)
-        exps = []
-        for cir in self.cirs:
-            cir(x_r)
-            exps.append(cir.expectation())
-        exps = torch.stack(exps,1).reshape(x.size(0), len(self.cirs), w, w)
-        return exps
 
+    def forward(self, x):
+        kernel_size, stride = 2, 2
+        # [64, 1, 18, 18] -> [64, 1, 9, 18, 2] -> [64, 1, 9, 9, 2, 2]
+        x_unflod = x.unfold(2, kernel_size, stride).unfold(3, kernel_size, stride)
+        w = int((x.shape[-1] - kernel_size) / stride + 1)
+        x_reshape = x_unflod.reshape(-1, self.nqubit)
+
+        exps = []
+        for cir in self.cirs: # out_channels
+            cir(x_reshape)
+            exp = cir.expectation()
+            exps.append(exp)
+
+        exps = torch.stack(exps, dim=1)
+        exps = exps.reshape(x.shape[0], 3, w, w)
+        return exps
+#%%
+net = RandomQuantumConvolutionalLayer(nqubit=4, num_circuits=3, seed=1024)
+net.cirs[0].draw()
+#%%
+# 基于随机量子卷积层的混合模型
 class RandomQCCNN(nn.Module):
     def __init__(self):
-        super().__init__()
+        super(RandomQCCNN, self).__init__()
         self.conv = nn.Sequential(
-            RandomQuantumConvolutionalLayer(4,3,seed=1024),
-            nn.ReLU(), nn.MaxPool2d(2,1),
-            nn.Conv2d(3,6,2,1), nn.ReLU(), nn.MaxPool2d(2,1)
+            RandomQuantumConvolutionalLayer(nqubit=4, num_circuits=3, seed=1024),   # num_circuits=3代表我们在quanv1层只用了3个量子卷积核
+            nn.BatchNorm2d(3),  # 添加批量归一化
+            nn.ReLU(),
+            nn.MaxPool2d(kernel_size=2, stride=1),
+            nn.Conv2d(3, 6, kernel_size=2, stride=1),
+            nn.BatchNorm2d(6),  # 添加批量归一化
+            nn.ReLU(),
+            nn.MaxPool2d(kernel_size=2, stride=1)
         )
         self.fc = nn.Sequential(
-            nn.Linear(6*6*6,1024), nn.Dropout(0.4),
-            nn.Linear(1024,10)
+            nn.Linear(6 * 6 * 6, 1024),
+            nn.BatchNorm1d(1024),  # 添加批量归一化
+            nn.Dropout(0.5),  # 增加dropout比例
+            nn.ReLU(),
+            nn.Linear(1024, 10)
         )
-    def forward(self,x):
-        x = self.conv(x)
-        x = x.view(x.size(0),-1)
-        return self.fc(x)
 
+    def forward(self, x):
+        x = self.conv(x)
+        x = x.reshape(x.size(0), -1)
+        x = self.fc(x)
+        return x
+#%%
+# 修改RandomQCCNN模型的训练参数
+num_epochs = 300
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+print(device)
+seed_torch(42)   # 使用相同的随机种子值
+model = RandomQCCNN()
+model.to(device)
+criterion = nn.CrossEntropyLoss()
+optimizer = optim.AdamW(model.parameters(), lr=3e-4, weight_decay=1e-5)  # 使用AdamW优化器和适当的权重衰减
+optim_model, metrics = train_model(model, criterion, optimizer, train_loader, valid_loader, num_epochs, device)
+torch.save(optim_model.state_dict(), './data/notebook2/random_qccnn_weights.pt')   # 保存训练好的模型参数，用于后续的推理或测试
+pd.DataFrame(metrics).to_csv('./data/notebook2/random_qccnn_metrics.csv', index='None')   # 保存模型训练过程，用于后续图标展示
+#%%
+state_dict = torch.load('./data/notebook2/random_qccnn_weights.pt', map_location=device)
+random_qccnn_model = RandomQCCNN()
+random_qccnn_model.load_state_dict(state_dict)
+random_qccnn_model.to(device)
+
+test_acc = test_model(random_qccnn_model, test_loader, device)
+#%%
+data = pd.read_csv('./data/notebook2/random_qccnn_metrics.csv')
+epoch = data['epoch']
+train_loss = data['train_loss']
+valid_loss = data['valid_loss']
+train_acc = data['train_acc']
+valid_acc = data['valid_acc']
+
+# 创建图和Axes对象
+fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(12, 5))
+
+# 绘制训练损失曲线
+ax1.plot(epoch, train_loss, label='Train Loss')
+ax1.plot(epoch, valid_loss, label='Valid Loss')
+ax1.set_title('Training Loss Curve')
+ax1.set_xlabel('Epoch')
+ax1.set_ylabel('Loss')
+ax1.legend()
+
+# 绘制训练准确率曲线
+ax2.plot(epoch, train_acc, label='Train Accuracy')
+ax2.plot(epoch, valid_acc, label='Valid Accuracy')
+ax2.set_title('Training Accuracy Curve')
+ax2.set_xlabel('Epoch')
+ax2.set_ylabel('Accuracy')
+ax2.legend()
+
+plt.show()
+#%%
 class ParameterizedQuantumConvolutionalLayer(nn.Module):
-    def __init__(self,nqubit,num_circuits):
+    def __init__(self, nqubit, num_circuits):
         super().__init__()
         self.nqubit = nqubit
         self.cirs = nn.ModuleList([self.circuit(nqubit) for _ in range(num_circuits)])
-    def circuit(self,nqubit):
+
+    def circuit(self, nqubit):
         cir = dq.QubitCircuit(nqubit)
-        cir.rxlayer(encode=True); cir.barrier()
-        for _ in range(4):
-            cir.rylayer(); cir.cnot_ring(); cir.barrier()
+        cir.rxlayer(encode=True)  #对原论文的量子线路结构并无影响，只是做了一个数据编码的操作
+        cir.barrier()
+        for iter in range(4): #对应原论文中一个量子卷积线路上的深度为4，可控参数一共16个
+            cir.rylayer()
+            cir.cnot_ring()
+            cir.barrier()
+
         cir.observable(0)
         return cir
-    def forward(self,x):
-        k,s = 2,2
-        x_unf = x.unfold(2,k,s).unfold(3,k,s)
-        w = (x.shape[-1]-k)//s +1
-        x_r = x_unf.reshape(-1,self.nqubit)
-        exps = []
-        for cir in self.cirs:
-            cir(x_r); exps.append(cir.expectation())
-        exps = torch.stack(exps,1).reshape(x.size(0),len(self.cirs),w,w)
-        return exps
 
+    def forward(self, x):
+        kernel_size, stride = 2, 2
+        # [64, 1, 18, 18] -> [64, 1, 9, 18, 2] -> [64, 1, 9, 9, 2, 2]
+        x_unflod = x.unfold(2, kernel_size, stride).unfold(3, kernel_size, stride)
+        w = int((x.shape[-1] - kernel_size) / stride + 1)
+        x_reshape = x_unflod.reshape(-1, self.nqubit)
+
+        exps = []
+        for cir in self.cirs: # out_channels
+            cir(x_reshape)
+            exp = cir.expectation()
+            exps.append(exp)
+
+        exps = torch.stack(exps, dim=1)
+        exps = exps.reshape(x.shape[0], 3, w, w)
+        return exps
+#%%
+# 此处我们可视化其中一个量子卷积核的线路结构：
+net = ParameterizedQuantumConvolutionalLayer(nqubit=4, num_circuits=3)
+net.cirs[0].draw()
+#%%
+# QCCNN整体网络架构：
 class QCCNN(nn.Module):
     def __init__(self):
-        super().__init__()
+        super(QCCNN, self).__init__()
         self.conv = nn.Sequential(
-            ParameterizedQuantumConvolutionalLayer(4,3),
-            nn.ReLU(), nn.MaxPool2d(2,1)
+            ParameterizedQuantumConvolutionalLayer(nqubit=4, num_circuits=3),
+            nn.BatchNorm2d(3),  # 添加批量归一化
+            nn.ReLU(),
+            nn.MaxPool2d(kernel_size=2, stride=1)
         )
-        self.fc = nn.Sequential(
-            nn.Linear(8*8*3,128), nn.Dropout(0.4), nn.ReLU(),
-            nn.Linear(128,10)
-        )
-    def forward(self,x):
-        x = self.conv(x); x = x.view(x.size(0),-1)
-        return self.fc(x)
 
-def vgg_block(in_c,out_c,n_convs):
-    layers = [nn.Conv2d(in_c,out_c,3,padding=1), nn.ReLU()]
-    for _ in range(n_convs-1):
-        layers += [nn.Conv2d(out_c,out_c,3,padding=1), nn.ReLU()]
-    layers.append(nn.MaxPool2d(2,2))
+        self.fc = nn.Sequential(
+            nn.Linear(8 * 8 * 3, 128),
+            nn.BatchNorm1d(128),  # 添加批量归一化
+            nn.Dropout(0.5),  # 增加dropout比例
+            nn.ReLU(),
+            nn.Linear(128, 10)
+        )
+
+    def forward(self, x):
+        x = self.conv(x)
+        x = x.reshape(x.size(0), -1)
+        x = self.fc(x)
+        return x
+#%%
+# 修改QCCNN模型的训练参数
+num_epochs = 300
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
+model = QCCNN()
+model.to(device)
+criterion = nn.CrossEntropyLoss()
+optimizer = optim.AdamW(model.parameters(), lr=3e-4, weight_decay=1e-5)  # 使用AdamW优化器和适当的权重衰减
+optim_model, metrics = train_model(model, criterion, optimizer, train_loader, valid_loader, num_epochs, device)
+torch.save(optim_model.state_dict(), './data/notebook2/qccnn_weights.pt')   # 保存训练好的模型参数，用于后续的推理或测试
+pd.DataFrame(metrics).to_csv('./data/notebook2/qccnn_metrics.csv', index='None')   # 保存模型训练过程，用于后续图标展示
+#%%
+state_dict = torch.load('./data/notebook2/qccnn_weights.pt', map_location=device)
+qccnn_model = QCCNN()
+qccnn_model.load_state_dict(state_dict)
+qccnn_model.to(device)
+
+test_acc = test_model(qccnn_model, test_loader, device)
+#%%
+def vgg_block(in_channel,out_channel,num_convs):
+    layers = nn.ModuleList()
+    assert num_convs >= 1
+    layers.append(nn.Conv2d(in_channel,out_channel,kernel_size=3,padding=1))
+    layers.append(nn.ReLU())
+    for _ in range(num_convs-1):
+        layers.append(nn.Conv2d(out_channel,out_channel,kernel_size=3,padding=1))
+        layers.append(nn.ReLU())
+    layers.append(nn.MaxPool2d(kernel_size=2,stride=2))
     return nn.Sequential(*layers)
 
 VGG = nn.Sequential(
-    vgg_block(1,10,3),
-    vgg_block(10,16,3),
+    vgg_block(1, 32, 2),  # 增加通道数和调整卷积层数量
+    vgg_block(32, 64, 2),
     nn.Flatten(),
-    nn.Linear(16*4*4,120), nn.Sigmoid(),
-    nn.Linear(120,84), nn.Sigmoid(),
-    nn.Linear(84,10), nn.Softmax(dim=-1)
+    nn.Linear(64 * 4 * 4, 256),  # 调整全连接层大小
+    nn.BatchNorm1d(256),  # 添加批量归一化
+    nn.ReLU(),
+    nn.Dropout(0.5),  # 增加dropout比例
+    nn.Linear(256, 128),
+    nn.BatchNorm1d(128),  # 添加批量归一化
+    nn.ReLU(),
+    nn.Dropout(0.5),
+    nn.Linear(128, 10),
+    nn.Softmax(dim=-1)
 )
+#%%
+# 修改VGG模型的训练参数
+num_epochs = 300
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 
-#%% 主入口
-if __name__ == '__main__':
-    freeze_support()
+vgg_model = VGG
+vgg_model.to(device)
+criterion = nn.CrossEntropyLoss()
+optimizer = optim.AdamW(vgg_model.parameters(), lr=3e-4, weight_decay=1e-5)  # 使用AdamW优化器和适当的权重衰减
+vgg_model, metrics = train_model(vgg_model, criterion, optimizer, train_loader, valid_loader, num_epochs, device)
+torch.save(vgg_model.state_dict(), './data/notebook2/vgg_weights.pt')   # 保存训练好的模型参数，用于后续的推理或测试
+pd.DataFrame(metrics).to_csv('./data/notebook2/vgg_metrics.csv', index='None')   # 保存模型训练过程，用于后续图标展示
+#%%
+state_dict = torch.load('./data/notebook2/vgg_weights.pt', map_location=device)
+vgg_model = VGG
+vgg_model.load_state_dict(state_dict)
+vgg_model.to(device)
 
-    # 数据增广与加载
-    train_transform = transforms.Compose([
-        transforms.Resize((18, 18)),
-        transforms.RandomRotation(15),
-        transforms.RandomHorizontalFlip(),
-        transforms.RandomVerticalFlip(0.3),
-        transforms.ToTensor(),
-        transforms.Normalize((0.5,), (0.5,))
-    ])
-    eval_transform = transforms.Compose([
-        transforms.Resize((18, 18)),
-        transforms.ToTensor(),
-        transforms.Normalize((0.5,), (0.5,))
-    ])
+vgg_test_acc = test_model(vgg_model, test_loader, device)
+#%%
+vgg_data = pd.read_csv('./data/notebook2/vgg_metrics.csv')
+qccnn_data = pd.read_csv('./data/notebook2/qccnn_metrics.csv')
+vgg_epoch = vgg_data['epoch']
+vgg_train_loss = vgg_data['train_loss']
+vgg_valid_loss = vgg_data['valid_loss']
+vgg_train_acc = vgg_data['train_acc']
+vgg_valid_acc = vgg_data['valid_acc']
 
-    full_train = FashionMNIST(root='./data/notebook2', train=True, transform=train_transform, download=True)
-    test_dataset = FashionMNIST(root='./data/notebook2', train=False, transform=eval_transform, download=True)
-    train_size = int(0.8 * len(full_train))
-    valid_size = len(full_train) - train_size
-    train_ds, valid_ds = torch.utils.data.random_split(full_train, [train_size, valid_size])
-    valid_ds.dataset.transform = eval_transform
+qccnn_epoch = qccnn_data['epoch']
+qccnn_train_loss = qccnn_data['train_loss']
+qccnn_valid_loss = qccnn_data['valid_loss']
+qccnn_train_acc = qccnn_data['train_acc']
+qccnn_valid_acc = qccnn_data['valid_acc']
 
-    batch_size = 128
-    train_loader = DataLoader(train_ds, batch_size=batch_size, shuffle=True,  drop_last=True, num_workers=4)
-    valid_loader = DataLoader(valid_ds, batch_size=batch_size, shuffle=False, drop_last=True, num_workers=4)
-    test_loader  = DataLoader(test_dataset, batch_size=batch_size, shuffle=False, drop_last=False, num_workers=4)
+# 创建图和Axes对象
+fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(12, 5))
 
-    # 三种模型配置
-    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-    models = {
-        'random_qccnn': (RandomQCCNN(), 1e-3, './data/notebook2/random_qccnn_best.pt'),
-        'qccnn':        (QCCNN(),        1e-4, './data/notebook2/qccnn_best.pt'),
-        'vgg':          (VGG,            1e-4, './data/notebook2/vgg_best.pt')
-    }
+# 绘制训练损失曲线
+ax1.plot(vgg_epoch, vgg_train_loss, label='VGG Train Loss')
+ax1.plot(vgg_epoch, vgg_valid_loss, label='VGG Valid Loss')
+ax1.plot(qccnn_epoch, qccnn_train_loss, label='QCCNN Valid Loss')
+ax1.plot(qccnn_epoch, qccnn_valid_loss, label='QCCNN Valid Loss')
+ax1.set_title('Training Loss Curve')
+ax1.set_xlabel('Epoch')
+ax1.set_ylabel('Loss')
+ax1.legend()
 
-    all_metrics = {}
-    for name, (model, lr, save_path) in models.items():
-        seed_torch(1024)
-        model = model.to(device)
-        criterion = nn.CrossEntropyLoss()
-        optimizer = optim.AdamW(model.parameters(), lr=lr, weight_decay=1e-4)
-        scheduler = optim.lr_scheduler.CosineAnnealingLR(optimizer, T_max=50)
+# 绘制训练准确率曲线
+ax2.plot(vgg_epoch, vgg_train_acc, label='VGG Train Accuracy')
+ax2.plot(vgg_epoch, vgg_valid_acc, label='VGG Valid Accuracy')
+ax2.plot(qccnn_epoch, qccnn_train_acc, label='QCCNN Train Accuracy')
+ax2.plot(qccnn_epoch, qccnn_valid_acc, label='QCCNN Valid Accuracy')
+ax2.set_title('Training Accuracy Curve')
+ax2.set_xlabel('Epoch')
+ax2.set_ylabel('Accuracy')
+ax2.legend()
 
-        print(f"\n=== Training {name} ===")
-        _, metrics = train_model(
-            model, criterion, optimizer, scheduler,
-            train_loader, valid_loader,
-            num_epochs=50, device=device, save_path=save_path
-        )
-        all_metrics[name] = metrics
-        pd.DataFrame(metrics).to_csv(f'./data/notebook2/{name}_metrics.csv', index=False)
+plt.show()
+#%%
+# 这里我们对比不同模型之间可训练参数量的区别
 
-    # 测试与可视化
-    plt.figure(figsize=(12,5))
-    for i,(name,metrics) in enumerate(all_metrics.items(),1):
-        model, _, save_path = models[name]
-        best_model = model.to(device)
-        best_model.load_state_dict(torch.load(save_path))
-        print(f"\n--- Testing {name} ---")
-        test_model(best_model, test_loader, device)
+def count_parameters(model):
+    """
+    计算模型的参数数量
+    """
+    return sum(p.numel() for p in model.parameters() if p.requires_grad)
 
-        plt.subplot(1,3,i)
-        plt.plot(metrics['epoch'], metrics['valid_acc'], label=f'{name} Val Acc')
-        plt.xlabel('Epoch'); plt.ylabel('Valid Acc')
-        plt.title(name); plt.legend()
-
-    plt.tight_layout(); plt.show()
-
-    # 参数量统计
-    def count_parameters(m):
-        return sum(p.numel() for p in m.parameters() if p.requires_grad)
-
-    print("\nParameter Counts:")
-    for name,(model,_,_) in models.items():
-        print(f"{name}: {count_parameters(model)}")
+number_params_VGG = count_parameters(VGG)
+number_params_QCCNN = count_parameters(QCCNN())
+print(f'VGG 模型可训练参数量：{number_params_VGG}\t QCCNN模型可训练参数量：{number_params_QCCNN}')

data/notebook1/random_qccnn_metrics.csv

@@ -1,301 +1,33 @@
 ,epoch,train_acc,valid_acc,train_loss,valid_loss
-0,1,0.12575,0.21421370967741934,2.2937231807708742,2.2676862593620055
-1,2,0.248,0.3240927419354839,2.2377592124938963,2.1976155081102924
-2,3,0.354375,0.3815524193548387,2.129876153945923,2.044375258107339
-3,4,0.42625,0.4324596774193548,1.9269471073150635,1.815262294584705
-4,5,0.482125,0.5282258064516129,1.6988583679199218,1.6081839684517152
-5,6,0.528375,0.5151209677419355,1.5167635107040405,1.4654438649454424
-6,7,0.5465,0.5574596774193549,1.3927514476776124,1.3655932180343135
-7,8,0.5705,0.5403225806451613,1.3035813112258912,1.2985444838000881
-8,9,0.586875,0.5902217741935484,1.2359741706848144,1.2315824993195073
-9,10,0.5985,0.5967741935483871,1.1828507437705993,1.1789349471369097
-10,11,0.61075,0.6013104838709677,1.1400936079025268,1.146019495302631
-11,12,0.61425,0.6098790322580645,1.105873393535614,1.1111935800121677
-12,13,0.623,0.6234879032258065,1.076672074317932,1.089227545645929
-13,14,0.63475,0.6159274193548387,1.0523606872558593,1.0597389667264876
-14,15,0.63575,0.5866935483870968,1.0302134299278258,1.0517296521894393
-15,16,0.638625,0.625,1.0127733755111694,1.030857661078053
-16,17,0.644125,0.623991935483871,0.9972539176940918,1.0212982143125227
-17,18,0.647125,0.6491935483870968,0.9847306752204895,1.0031032081573241
-18,19,0.651625,0.6295362903225806,0.9718224520683288,0.9922776433729357
-19,20,0.653375,0.655241935483871,0.9614444556236267,1.004007081831655
-20,21,0.65475,0.639616935483871,0.9525526990890503,0.9718767077692093
-21,22,0.657125,0.6491935483870968,0.944016770362854,0.9612755986952013
-22,23,0.660125,0.6512096774193549,0.9386747822761535,0.9546890701017072
-23,24,0.660125,0.6607862903225806,0.9306546316146851,0.9596057341944787
-24,25,0.665625,0.6466733870967742,0.9239095873832702,0.9440147549875321
-25,26,0.664,0.6537298387096774,0.9177467222213745,0.9464947581291199
-26,27,0.66675,0.6582661290322581,0.912406234741211,0.9296330482729019
-27,28,0.66625,0.6607862903225806,0.9084503087997436,0.9246035218238831
-28,29,0.669375,0.6633064516129032,0.9031522455215454,0.9183507677047483
-29,30,0.66925,0.6622983870967742,0.8971806693077088,0.9155043325116557
-30,31,0.669625,0.6658266129032258,0.8963782095909119,0.9107550824842146
-31,32,0.6715,0.6607862903225806,0.8907575702667236,0.9062397806875168
-32,33,0.672875,0.6592741935483871,0.8865971283912659,0.9067775895518642
-33,34,0.6755,0.6643145161290323,0.8831153435707092,0.8996775977073177
-34,35,0.6745,0.6643145161290323,0.8799522204399108,0.9008925057226612
-35,36,0.67625,0.6738911290322581,0.8765962233543396,0.8957007392760246
-36,37,0.67775,0.6602822580645161,0.8732238512039184,0.8957233851955783
-37,38,0.677,0.6663306451612904,0.8700605201721191,0.8884582346485507
-38,39,0.677625,0.6779233870967742,0.86691774892807,0.8847945544027513
-39,40,0.683875,0.6683467741935484,0.8638643326759339,0.8812634848779247
-40,41,0.6825,0.6703629032258065,0.8619350233078003,0.8736868096936133
-41,42,0.68525,0.6769153225806451,0.8574845008850097,0.8757798306403621
-42,43,0.68325,0.6829637096774194,0.8559755001068116,0.8762649028531967
-43,44,0.681875,0.6794354838709677,0.8526963171958923,0.868664155083318
-44,45,0.6855,0.670866935483871,0.8508440051078796,0.8656267696811307
-45,46,0.687375,0.6738911290322581,0.8477729640007019,0.8647398410304901
-46,47,0.689125,0.6723790322580645,0.844515609741211,0.8638435967506901
-47,48,0.685,0.6779233870967742,0.8429114985466003,0.8618940441839157
-48,49,0.689125,0.6794354838709677,0.8397265086174012,0.8545605559502879
-49,50,0.686625,0.6804435483870968,0.8384793972969056,0.8552838506237153
-50,51,0.691125,0.6774193548387096,0.8345498304367065,0.848295844370319
-51,52,0.687875,0.6859879032258065,0.8335188755989075,0.855028792735069
-52,53,0.691,0.6824596774193549,0.8311837863922119,0.8479280106482967
-53,54,0.692875,0.6859879032258065,0.8279946088790894,0.8404466413682506
-54,55,0.6925,0.6804435483870968,0.8260387935638428,0.84263696593623
-55,56,0.693125,0.6849798387096774,0.8229942922592163,0.8379247784614563
-56,57,0.694375,0.6844758064516129,0.8220300149917602,0.8415298577277891
-57,58,0.696375,0.6880040322580645,0.8190384802818298,0.8321282882844249
-58,59,0.694,0.6875,0.816288417339325,0.8343535277151293
-59,60,0.694875,0.6839717741935484,0.8151030368804931,0.8336188389408973
-60,61,0.69425,0.6854838709677419,0.8126006669998169,0.8352141207264315
-61,62,0.697125,0.6854838709677419,0.8106505064964294,0.8332801768856664
-62,63,0.69975,0.6905241935483871,0.8080810284614564,0.8234034065277346
-63,64,0.70075,0.6844758064516129,0.8052171502113342,0.8192608933294973
-64,65,0.69925,0.6869959677419355,0.8035496191978455,0.815050482749939
-65,66,0.7005,0.6875,0.8014146156311035,0.8252233305285054
-66,67,0.698125,0.6905241935483871,0.7996929430961609,0.8083832244719228
-67,68,0.69975,0.6895161290322581,0.7969387445449829,0.8108954429626465
-68,69,0.702125,0.6869959677419355,0.7955113248825073,0.8145994409438102
-69,70,0.7,0.6935483870967742,0.792340503692627,0.807374858087109
-70,71,0.702625,0.6950604838709677,0.7911754889488221,0.8086352617509903
-71,72,0.703625,0.6985887096774194,0.7885409245491027,0.8008877135092213
-72,73,0.70425,0.702116935483871,0.7873198952674866,0.7998276199063947
-73,74,0.704125,0.6970766129032258,0.7848704538345337,0.7971179581457569
-74,75,0.70625,0.6970766129032258,0.7822935285568238,0.7927621314602513
-75,76,0.70475,0.6960685483870968,0.7788817420005798,0.7937883177111226
-76,77,0.704875,0.6985887096774194,0.7792668890953064,0.7862545732528933
-77,78,0.703625,0.7006048387096774,0.7767630572319031,0.7841528077279368
-78,79,0.710875,0.6980846774193549,0.7729922785758973,0.7877849590393805
-79,80,0.7085,0.7036290322580645,0.7725881729125976,0.7884936524975684
-80,81,0.7145,0.6975806451612904,0.7703448913097382,0.7794140846498551
-81,82,0.707875,0.703125,0.7683381514549256,0.7827766460757102
-82,83,0.71425,0.704133064516129,0.7661359086036682,0.7740850121744217
-83,84,0.716,0.7247983870967742,0.7635970077514649,0.7837481344899824
-84,85,0.714625,0.7046370967741935,0.7620524802207946,0.7760492024883148
-85,86,0.717875,0.7273185483870968,0.7587687454223633,0.7693652645234139
-86,87,0.71925,0.7167338709677419,0.7571495487689972,0.7791498034231125
-87,88,0.723125,0.7061491935483871,0.7559212753772736,0.7657425865050285
-88,89,0.721125,0.7227822580645161,0.7540467400550842,0.7639261926374128
-89,90,0.72775,0.7212701612903226,0.7519344382286072,0.7616109309657928
-90,91,0.72775,0.7283266129032258,0.7487852301597595,0.7575869021877166
-91,92,0.727875,0.7273185483870968,0.7474929718971253,0.7539062634591134
-92,93,0.72625,0.7237903225806451,0.744426561832428,0.7515108796858019
-93,94,0.7285,0.717741935483871,0.7436468710899353,0.7493124450406721
-94,95,0.72875,0.7227822580645161,0.7407840843200684,0.7509921366168607
-95,96,0.731875,0.7232862903225806,0.7383509948253631,0.7504203108049208
-96,97,0.732125,0.7288306451612904,0.7363486151695251,0.7452502135307558
-97,98,0.732875,0.7283266129032258,0.7337991933822632,0.7476337955844018
-98,99,0.735625,0.7303427419354839,0.7317234919071197,0.739826298529102
-99,100,0.733625,0.7283266129032258,0.7287046048641205,0.739542922665996
-100,101,0.73375,0.7147177419354839,0.7274810273647309,0.7344657682603405
-101,102,0.7355,0.7338709677419355,0.7247261893749237,0.7346700814462477
-102,103,0.73225,0.733366935483871,0.7240956840515137,0.7285718571755194
-103,104,0.737,0.7247983870967742,0.720921181678772,0.7279678198599047
-104,105,0.73725,0.7162298387096774,0.7193985497951507,0.7334578460262667
-105,106,0.73975,0.7232862903225806,0.7166938002109527,0.728367522839577
-106,107,0.737875,0.7363911290322581,0.715068681716919,0.7223057727659902
-107,108,0.742625,0.7318548387096774,0.7129860005378723,0.7190377539204013
-108,109,0.74,0.7389112903225806,0.7117807359695435,0.7183994016339702
-109,110,0.739875,0.7384072580645161,0.7105983903408051,0.7164657365891242
-110,111,0.742375,0.7328629032258065,0.7074811918735504,0.7246334398946455
-111,112,0.74,0.7348790322580645,0.7077446887493134,0.7127827925066794
-112,113,0.740125,0.7348790322580645,0.7058010087013245,0.7185006468526779
-113,114,0.74125,0.7379032258064516,0.7062028846740722,0.711994872939202
-114,115,0.741375,0.734375,0.7025900321006775,0.722117660507079
-115,116,0.741875,0.7404233870967742,0.702047369480133,0.7054055179319074
-116,117,0.741,0.7368951612903226,0.7008094470500946,0.7059461897419345
-117,118,0.74,0.7379032258064516,0.7000908722877502,0.7077065225570432
-118,119,0.74275,0.7389112903225806,0.697608702659607,0.7044590307820228
-119,120,0.741125,0.735383064516129,0.6974484593868255,0.705822677381577
-120,121,0.745,0.7363911290322581,0.6940790374279022,0.726842547616651
-121,122,0.741875,0.7409274193548387,0.6941187741756439,0.6984606769777113
-122,123,0.7455,0.7363911290322581,0.6942173736095428,0.7010489721452037
-123,124,0.74125,0.7404233870967742,0.6926323885917663,0.704462626288014
-124,125,0.743875,0.7409274193548387,0.6916290044784545,0.696697513903341
-125,126,0.74375,0.7394153225806451,0.6898484938144683,0.6978444207099176
-126,127,0.743625,0.7399193548387096,0.6888203046321869,0.7016230494745316
-127,128,0.74525,0.7414314516129032,0.6875256526470185,0.6972026151995505
-128,129,0.744125,0.7414314516129032,0.686995727300644,0.6930867991139812
-129,130,0.748,0.7424395161290323,0.6873089451789856,0.6934665374217495
-130,131,0.744,0.7328629032258065,0.6838541362285614,0.7066772426328352
-131,132,0.7435,0.7363911290322581,0.6849607121944428,0.6904300124414505
-132,133,0.746,0.7444556451612904,0.6836790750026703,0.6916862668529633
-133,134,0.746125,0.7419354838709677,0.6825263681411743,0.6903918568165072
-134,135,0.747125,0.7454637096774194,0.6808646347522735,0.6892580957181992
-135,136,0.746625,0.7328629032258065,0.6810887951850891,0.6992948112949249
-136,137,0.747625,0.7399193548387096,0.6801265494823456,0.6914754644516976
-137,138,0.744125,0.7449596774193549,0.6794462163448334,0.686861309313005
-138,139,0.746375,0.7379032258064516,0.6796345655918121,0.6865283789173249
-139,140,0.74825,0.7363911290322581,0.6776200189590454,0.6895612343665092
-140,141,0.748375,0.7368951612903226,0.6769659533500672,0.6895686657198014
-141,142,0.745875,0.7409274193548387,0.6762004556655884,0.684873923178642
-142,143,0.75,0.7429435483870968,0.6766425142288208,0.6841276749487846
-143,144,0.7495,0.7429435483870968,0.6757207989692688,0.6857355977258375
-144,145,0.74775,0.7409274193548387,0.6757489387989044,0.6874895538053205
-145,146,0.75275,0.7358870967741935,0.6723645563125611,0.6919096823661558
-146,147,0.75125,0.7399193548387096,0.672961886882782,0.6845684782151253
-147,148,0.751375,0.7389112903225806,0.672047397851944,0.6911592175883632
-148,149,0.7485,0.7399193548387096,0.6713605234622956,0.6868368666018209
-149,150,0.749625,0.7434475806451613,0.6715062477588654,0.6797055390573317
-150,151,0.752,0.7429435483870968,0.6708727264404297,0.6786900566470239
-151,152,0.749375,0.7404233870967742,0.6705182585716247,0.6893971793113216
-152,153,0.756375,0.7409274193548387,0.6681198358535767,0.6818877074026293
-153,154,0.74875,0.7323588709677419,0.668769829750061,0.6870744218749385
-154,155,0.751375,0.7404233870967742,0.6693622040748596,0.6791631608240066
-155,156,0.7495,0.7384072580645161,0.6677224984169007,0.6791054318028111
-156,157,0.751125,0.7530241935483871,0.6669107344150543,0.6757311301846658
-157,158,0.750375,0.7454637096774194,0.6689134707450867,0.675748364579293
-158,159,0.749875,0.7394153225806451,0.665742442369461,0.6785702628474082
-159,160,0.755125,0.7439516129032258,0.6639340476989746,0.6909028849294109
-160,161,0.754375,0.7484879032258065,0.6655936982631684,0.6800686274805376
-161,162,0.753,0.7444556451612904,0.6626871073246002,0.6847033404534862
-162,163,0.754375,0.7394153225806451,0.6636988418102264,0.678314108041025
-163,164,0.756,0.7439516129032258,0.6631928851604462,0.679127223068668
-164,165,0.7535,0.7449596774193549,0.6635311431884766,0.6755695458381407
-165,166,0.75425,0.7439516129032258,0.6608122355937958,0.6740436525114121
-166,167,0.7535,0.751008064516129,0.6623840556144714,0.6725323296362354
-167,168,0.753375,0.7474798387096774,0.6615988018512726,0.6734585973524279
-168,169,0.7555,0.7404233870967742,0.658470029592514,0.6751049209025598
-169,170,0.75275,0.748991935483871,0.6596595213413239,0.6707770478340888
-170,171,0.75475,0.7464717741935484,0.6594919128417969,0.6760938773232121
-171,172,0.752375,0.7404233870967742,0.6587064244747162,0.6735088594498173
-172,173,0.754875,0.7419354838709677,0.6581688785552978,0.6718735867930997
-173,174,0.755625,0.7444556451612904,0.6594367871284484,0.672063180515843
-174,175,0.75675,0.7484879032258065,0.6572930109500885,0.6681920395743463
-175,176,0.7565,0.7555443548387096,0.6567469036579132,0.666823229482097
-176,177,0.75625,0.7242943548387096,0.6563848433494568,0.69633928614278
-177,178,0.7545,0.7520161290322581,0.6568190479278564,0.6688290171084865
-178,179,0.758125,0.7464717741935484,0.6569823319911957,0.668416089588596
-179,180,0.756125,0.7474798387096774,0.6567688817977906,0.6825112623553122
-180,181,0.75825,0.7384072580645161,0.6541404497623443,0.6793540994967183
-181,182,0.754125,0.7484879032258065,0.6573447947502136,0.6673893765095742
-182,183,0.7575,0.7459677419354839,0.6545488471984864,0.669147877923904
-183,184,0.757625,0.7520161290322581,0.653099508523941,0.6698833357903266
-184,185,0.7575,0.7459677419354839,0.652852609872818,0.6746915771115211
-185,186,0.75775,0.7469758064516129,0.6531596164703369,0.6692096014176646
-186,187,0.75625,0.7429435483870968,0.65095947265625,0.6785491474213139
-187,188,0.759875,0.748991935483871,0.6528274257183075,0.6676096195174802
-188,189,0.75825,0.751008064516129,0.6515383367538452,0.6677034651079485
-189,190,0.761,0.7429435483870968,0.6507666997909546,0.6711848301272239
-190,191,0.756,0.7505040322580645,0.6502130575180054,0.6673150216379473
-191,192,0.755625,0.7520161290322581,0.6508474731445313,0.6693628420752864
-192,193,0.758375,0.7449596774193549,0.6497379775047303,0.6839032778816838
-193,194,0.760125,0.7449596774193549,0.6500401992797852,0.6711471715281087
-194,195,0.7595,0.751008064516129,0.6485152614116668,0.6667885655356992
-195,196,0.761,0.751008064516129,0.6486285464763641,0.6671426277006826
-196,197,0.76,0.7515120967741935,0.6487596340179443,0.6690003919985986
-197,198,0.759375,0.7494959677419355,0.6480154263973236,0.6674725115299225
-198,199,0.761125,0.7494959677419355,0.6471124620437622,0.6651709810380013
-199,200,0.7585,0.7389112903225806,0.6480610625743866,0.6811543289692171
-200,201,0.75975,0.7379032258064516,0.6461314346790313,0.6756981101728254
-201,202,0.759375,0.7459677419354839,0.6457764523029328,0.6665134122294765
-202,203,0.75775,0.7550403225806451,0.6469809007644653,0.661889091614754
-203,204,0.761125,0.7479838709677419,0.6435232808589936,0.6644216324052503
-204,205,0.760375,0.7439516129032258,0.6471346819400787,0.6682568625096352
-205,206,0.762,0.7414314516129032,0.6459955296516419,0.6739240298348088
-206,207,0.75925,0.7540322580645161,0.6436120250225067,0.6613490379625752
-207,208,0.764875,0.7484879032258065,0.6442920634746552,0.665196894638
-208,209,0.7595,0.7439516129032258,0.6450260980129242,0.66756044376281
-209,210,0.762375,0.7525201612903226,0.6450640928745269,0.6618116507607121
-210,211,0.764125,0.7479838709677419,0.6432662625312805,0.6611761902609179
-211,212,0.76175,0.7464717741935484,0.6430184645652771,0.6601478543973738
-212,213,0.7635,0.7540322580645161,0.6424563028812409,0.6675388928382627
-213,214,0.76075,0.7464717741935484,0.6431009001731872,0.6693989961378036
-214,215,0.763375,0.7545362903225806,0.6424487299919128,0.6593958552806608
-215,216,0.759875,0.7550403225806451,0.6421677529811859,0.6666595695480224
-216,217,0.760625,0.7464717741935484,0.6414295537471771,0.6686101180891837
-217,218,0.76525,0.7545362903225806,0.641946748971939,0.663229928862664
-218,219,0.759875,0.751008064516129,0.6410156943798065,0.6790805785886703
-219,220,0.76,0.7484879032258065,0.6424972369670868,0.6616327435739578
-220,221,0.7605,0.751008064516129,0.6389381773471833,0.6645574127474139
-221,222,0.763125,0.7515120967741935,0.6390037200450898,0.6623315830384532
-222,223,0.761,0.7570564516129032,0.6397264943122863,0.6680223576484188
-223,224,0.76425,0.7520161290322581,0.6378453469276428,0.6630678580653283
-224,225,0.763875,0.7520161290322581,0.6382321789264679,0.662065879952523
-225,226,0.762,0.7479838709677419,0.639366191625595,0.661988039170542
-226,227,0.765,0.7525201612903226,0.6383702094554902,0.6641764746558282
-227,228,0.763125,0.751008064516129,0.6380119869709014,0.6607372049362429
-228,229,0.759375,0.7520161290322581,0.6383262400627137,0.6545132811992399
-229,230,0.764375,0.7555443548387096,0.6372324013710022,0.6568373855083219
-230,231,0.761625,0.7484879032258065,0.6373445687294006,0.6696269560244775
-231,232,0.76475,0.7515120967741935,0.6373440618515015,0.6574311083362948
-232,233,0.768,0.7555443548387096,0.6352594494819641,0.6599096617391033
-233,234,0.7635,0.7464717741935484,0.6361170980930329,0.6795821391767071
-234,235,0.76175,0.75,0.6369703621864319,0.6554409640450631
-235,236,0.765,0.7535282258064516,0.6348738882541657,0.6552178427096336
-236,237,0.762625,0.7520161290322581,0.6365966260433197,0.6692352737149885
-237,238,0.7645,0.7449596774193549,0.6336214027404785,0.688425297698667
-238,239,0.76525,0.7560483870967742,0.6345989825725555,0.6607512629801228
-239,240,0.7625,0.7479838709677419,0.6337947535514832,0.65921801136386
-240,241,0.764,0.7515120967741935,0.6340638983249665,0.6573365686401245
-241,242,0.767,0.7535282258064516,0.6342987337112427,0.6604504883289337
-242,243,0.768,0.7444556451612904,0.6332927324771881,0.671097691020658
-243,244,0.76475,0.7520161290322581,0.6348397693634034,0.6558002243118901
-244,245,0.76675,0.7474798387096774,0.632552001953125,0.6722479366487072
-245,246,0.768,0.748991935483871,0.6314238250255585,0.6594607532024384
-246,247,0.764875,0.7444556451612904,0.6325469689369202,0.6605463912410121
-247,248,0.7655,0.7525201612903226,0.6310685338973999,0.6538672639477637
-248,249,0.766,0.7570564516129032,0.6316399626731872,0.6516222578863944
-249,250,0.766125,0.7515120967741935,0.6315728986263275,0.6539699367938503
-250,251,0.764625,0.7479838709677419,0.6300767264366149,0.6660763275238776
-251,252,0.768375,0.7525201612903226,0.6311213612556458,0.6618637481043416
-252,253,0.764375,0.75,0.6319745321273804,0.6553429893908962
-253,254,0.765125,0.7484879032258065,0.6321416351795196,0.6639243306652192
-254,255,0.765125,0.7424395161290323,0.6311585788726807,0.6792801503212221
-255,256,0.766125,0.7575604838709677,0.6310848982334137,0.6525870225121898
-256,257,0.767625,0.7520161290322581,0.6288257777690888,0.6681103860178301
-257,258,0.7655,0.7550403225806451,0.6306543848514556,0.659768590042668
-258,259,0.764,0.7555443548387096,0.6302573113441468,0.6550669179808709
-259,260,0.769125,0.7580645161290323,0.6280254762172699,0.6525000554900016
-260,261,0.76875,0.7540322580645161,0.6282578265666962,0.6584180526195034
-261,262,0.76775,0.7560483870967742,0.6287466604709625,0.6528177463239239
-262,263,0.770375,0.7409274193548387,0.6289410617351532,0.6700733111750695
-263,264,0.76825,0.7515120967741935,0.628407103061676,0.6540230772187633
-264,265,0.7675,0.7520161290322581,0.630276577949524,0.6641475250644069
-265,266,0.767875,0.7530241935483871,0.6273887372016906,0.6583489189224858
-266,267,0.76625,0.7535282258064516,0.6282406325340271,0.6604514862260511
-267,268,0.76925,0.7535282258064516,0.6271086316108704,0.6536620384262454
-268,269,0.76825,0.7469758064516129,0.6269893596172332,0.6587162335072795
-269,270,0.767125,0.7479838709677419,0.6261151025295257,0.6642194390296936
-270,271,0.768,0.7545362903225806,0.62709636926651,0.659077537636603
-271,272,0.767875,0.7575604838709677,0.626769639968872,0.6581989363316567
-272,273,0.768625,0.7570564516129032,0.6237564489841462,0.6504895312170829
-273,274,0.767875,0.7540322580645161,0.6259858434200287,0.6550556219393208
-274,275,0.76925,0.7565524193548387,0.6264045915603638,0.6529571308243659
-275,276,0.768375,0.7605846774193549,0.6264821493625641,0.6480650353816247
-276,277,0.767375,0.7585685483870968,0.6253943295478821,0.6529869091126227
-277,278,0.768,0.7585685483870968,0.6250380325317383,0.6509808676858102
-278,279,0.77225,0.7560483870967742,0.6249497833251954,0.652328065326137
-279,280,0.768375,0.7610887096774194,0.6253566663265229,0.6481388095886477
-280,281,0.769,0.7631048387096774,0.6269658114910126,0.6511666505567489
-281,282,0.77225,0.7615927419354839,0.6219663026332856,0.6503843261349586
-282,283,0.768375,0.7530241935483871,0.6239040396213531,0.6559025366460124
-283,284,0.773125,0.7565524193548387,0.6230207705497741,0.6521498032154576
-284,285,0.77225,0.7575604838709677,0.6224600894451141,0.6498911640336437
-285,286,0.772375,0.7535282258064516,0.6230463716983795,0.6594464038648913
-286,287,0.772125,0.7454637096774194,0.623299633026123,0.6596774737681111
-287,288,0.7705,0.7575604838709677,0.6226675305366516,0.6503997896948168
-288,289,0.771875,0.7615927419354839,0.6228007516860962,0.6582970378860351
-289,290,0.769,0.7610887096774194,0.6228819963932037,0.6468615512694081
-290,291,0.77075,0.7560483870967742,0.6219233012199402,0.6543603664444338
-291,292,0.770125,0.7570564516129032,0.6231648538112641,0.6516289316838787
-292,293,0.772625,0.7555443548387096,0.6210650091171265,0.6526086061231552
-293,294,0.7735,0.7580645161290323,0.6205819368362426,0.650269466061746
-294,295,0.76975,0.7600806451612904,0.6209871962070465,0.6479213949172727
-295,296,0.770125,0.7565524193548387,0.6204864876270294,0.6524337001385228
-296,297,0.76875,0.7464717741935484,0.622816385269165,0.6591307309366041
-297,298,0.771375,0.7585685483870968,0.620723108291626,0.647264729584417
-298,299,0.768625,0.7580645161290323,0.621616400718689,0.6502204921937758
-299,300,0.7725,0.7555443548387096,0.6198663051128388,0.6479197202190277
+0,1,0.59325,0.6854838709677419,1.1784187927246095,0.8956879992638865
+1,2,0.713,0.7127016129032258,0.788702152967453,0.7735376723351017
+2,3,0.755,0.7389112903225806,0.7006268813610077,0.7271041427889178
+3,4,0.763625,0.7273185483870968,0.668900458574295,0.7277685692233424
+4,5,0.76275,0.748991935483871,0.6469288661479949,0.672559670863613
+5,6,0.773625,0.7434475806451613,0.6203210880756378,0.6748099394382969
+6,7,0.771625,0.7494959677419355,0.6234635796546936,0.6763338706185741
+7,8,0.784125,0.7580645161290323,0.5965014040470124,0.6313219945276937
+8,9,0.781,0.7419354838709677,0.5893729448318481,0.6552943125847848
+9,10,0.785375,0.7600806451612904,0.582163923740387,0.6227685501498561
+10,11,0.790625,0.7605846774193549,0.5674120993614197,0.6145914264263646
+11,12,0.7995,0.765625,0.5617314722537995,0.6158498919779255
+12,13,0.7965,0.7883064516129032,0.5522617139816284,0.581030644716755
+13,14,0.803125,0.7817540322580645,0.5367116575241089,0.5911272004727395
+14,15,0.80025,0.7752016129032258,0.5398472018241882,0.5930967292478008
+15,16,0.808125,0.7620967741935484,0.5326456694602967,0.6306540206555398
+16,17,0.8145,0.7772177419354839,0.5097175936698913,0.5843374094655437
+17,18,0.817375,0.7918346774193549,0.5045695571899415,0.5643920071663395
+18,19,0.81175,0.7681451612903226,0.5022978014945984,0.593349628871487
+19,20,0.824375,0.7928427419354839,0.4974315061569214,0.5595071181174247
+20,21,0.818,0.7762096774193549,0.49526858401298524,0.5555580258369446
+21,22,0.815625,0.7862903225806451,0.4976941578388214,0.5627694274148634
+22,23,0.825875,0.7908266129032258,0.4906779823303223,0.5514134312829664
+23,24,0.823875,0.7953629032258065,0.4869117760658264,0.5559701804191836
+24,25,0.823,0.7807459677419355,0.48527450203895567,0.5541293746040713
+25,26,0.824625,0.795866935483871,0.4811685870885849,0.5384060884675672
+26,27,0.827875,0.7938508064516129,0.48096660792827606,0.5493545234203339
+27,28,0.826875,0.7998991935483871,0.477142098903656,0.5478304624557495
+28,29,0.82725,0.7948588709677419,0.47373487854003904,0.547797841410483
+29,30,0.825375,0.7847782258064516,0.47746446084976196,0.5636355328944421
+30,31,0.828875,0.7757056451612904,0.4754366238117218,0.5671144060550197
+31,32,0.8315,0.7908266129032258,0.4769072663784027,0.5523799340571126

data/notebook2/qccnn_metrics.csv

@@ -1,51 +1,49 @@
-epoch,train_acc,valid_acc,train_loss,valid_loss
-1,0.3112916666666667,0.5076444892473119,2.1129255460103353,1.7863053237238238
-2,0.46210416666666665,0.5547715053763441,1.5605127541224162,1.3499009532313193
-3,0.5200833333333333,0.5892137096774194,1.298180630683899,1.1617528520604616
-4,0.5641666666666667,0.6311323924731183,1.161390997727712,1.0419180438082705
-5,0.6055833333333334,0.6598622311827957,1.062113331158956,0.9560088053826363
-6,0.6316666666666667,0.6804435483870968,0.9944041889508565,0.8958990977656457
-7,0.6513958333333333,0.6928763440860215,0.9395476338068645,0.8504296951396491
-8,0.6677708333333333,0.7007728494623656,0.8985106336275737,0.8148919709267155
-9,0.6809375,0.7095934139784946,0.8673048818906148,0.7897684433126962
-10,0.6876666666666666,0.7185819892473119,0.8408271819750468,0.7671139176173877
-11,0.697625,0.7247143817204301,0.8202435002326965,0.7502269674372929
-12,0.7030625,0.7303427419354839,0.8024592914581299,0.7343912464316174
-13,0.7118958333333333,0.7316868279569892,0.7869214735031128,0.7222210591839205
-14,0.7163958333333333,0.7404233870967742,0.7718908907572428,0.7093277560767307
-15,0.7211041666666667,0.7437836021505376,0.7589795832633972,0.6981546148177116
-16,0.724875,0.748067876344086,0.7483940289815267,0.6881407947950465
-17,0.7293958333333334,0.75,0.7373875479698181,0.6788085179944192
-18,0.7345,0.7517641129032258,0.7301613558133443,0.6724013622089099
-19,0.7350416666666667,0.7550403225806451,0.7210039290587107,0.6651292238184201
-20,0.7391458333333333,0.758736559139785,0.711990216255188,0.6584682823509298
-21,0.7433958333333334,0.7608366935483871,0.7047773049672444,0.6513981505106854
-22,0.7442916666666667,0.7634408602150538,0.6997552577654521,0.6454382868864204
-23,0.7481458333333333,0.7653729838709677,0.6937046423753103,0.6401395256160408
-24,0.7498333333333334,0.7683131720430108,0.6873725473880767,0.6349253362865859
-25,0.7529166666666667,0.769573252688172,0.6828897857666015,0.6308066611007977
-26,0.7547291666666667,0.7708333333333334,0.6747500312328338,0.6260690948655528
-27,0.7547291666666667,0.7722614247311828,0.6727884339491527,0.6230032514500362
-28,0.7580208333333334,0.7736055107526881,0.6705719958146413,0.6201616948650729
-29,0.757125,0.7748655913978495,0.6678872625033061,0.6165956912502166
-30,0.76025,0.7759576612903226,0.6641453323364258,0.6141596103227267
-31,0.7609583333333333,0.7767137096774194,0.6619020007451375,0.6114715427480718
-32,0.7618333333333334,0.776377688172043,0.6576849890549977,0.609050533463878
-33,0.7629583333333333,0.7783938172043011,0.6566993356545766,0.6069687149857962
-34,0.7631666666666667,0.7791498655913979,0.6553838003476461,0.6052348261238426
-35,0.7626666666666667,0.7785618279569892,0.6523663277626037,0.6037159392269709
-36,0.7653333333333333,0.780325940860215,0.6526273953914642,0.6023694485105494
-37,0.7639791666666667,0.7810819892473119,0.6524330813884736,0.6012957778669172
-38,0.7673958333333334,0.7799059139784946,0.6500039516290029,0.6002496516191831
-39,0.764875,0.7815860215053764,0.6476710465749105,0.5992967845291219
-40,0.7661041666666667,0.782258064516129,0.6484741485118866,0.5987250285763894
-41,0.7668333333333334,0.7818380376344086,0.64447620677948,0.5979425240588444
-42,0.7662291666666666,0.7817540322580645,0.6458657967249553,0.5976644568545844
-43,0.76675,0.7824260752688172,0.6449048202037811,0.5971484975789183
-44,0.7677916666666667,0.7825100806451613,0.6437487976551056,0.5968063899906733
-45,0.7677083333333333,0.7825940860215054,0.6432626353104909,0.5964810960395361
-46,0.7666041666666666,0.7826780913978495,0.6452286802132925,0.5963560240243071
-47,0.768875,0.7827620967741935,0.6432473388512929,0.5962550809947393
-48,0.7674583333333334,0.7826780913978495,0.6433716455300649,0.5961945524779699
-49,0.7678333333333334,0.7827620967741935,0.6440556212266286,0.596176440036425
-50,0.7677083333333333,0.7825940860215054,0.6438165396849315,0.5961719805835396
+,epoch,train_acc,valid_acc,train_loss,valid_loss
+0,1,0.391375,0.5882056451612904,1.7793689422607422,1.3356874796652025
+1,2,0.617125,0.6365927419354839,1.059556163787842,0.9690603344671188
+2,3,0.663125,0.6577620967741935,0.9102170023918151,0.9078671547674364
+3,4,0.69,0.6819556451612904,0.8537670917510987,0.8649420180628377
+4,5,0.71125,0.7056451612903226,0.8021836678981781,0.8317571128568342
+5,6,0.719375,0.6844758064516129,0.7773117737770081,0.8293971457789021
+6,7,0.733,0.7247983870967742,0.7432277894020081,0.7570091389840649
+7,8,0.727125,0.7051411290322581,0.7392586808204651,0.7708722833664187
+8,9,0.740375,0.7091733870967742,0.716008551120758,0.7547545788749572
+9,10,0.74775,0.7091733870967742,0.6988923935890198,0.7634256799374858
+10,11,0.75025,0.7328629032258065,0.6836859595775604,0.7220065324537216
+11,12,0.75525,0.7313508064516129,0.6790840411186219,0.7320531125991575
+12,13,0.755875,0.734375,0.6686462206840516,0.7067360877990723
+13,14,0.755875,0.7338709677419355,0.659578008890152,0.6924379564100697
+14,15,0.758625,0.7217741935483871,0.6591809678077698,0.7092515037905786
+15,16,0.770125,0.7474798387096774,0.6366513178348542,0.6689942498360911
+16,17,0.76975,0.75,0.64143789935112,0.6781372427940369
+17,18,0.766625,0.7469758064516129,0.6327295961380005,0.6749444469328849
+18,19,0.772125,0.7298387096774194,0.6213552060127259,0.6950428524324971
+19,20,0.771625,0.7560483870967742,0.619529750585556,0.6620656597998834
+20,21,0.77425,0.7459677419354839,0.6170263500213623,0.6944894117693747
+21,22,0.77375,0.7379032258064516,0.610548350572586,0.698592597438443
+22,23,0.774125,0.7494959677419355,0.6073116481304168,0.6649176555295144
+23,24,0.78,0.75,0.601892656326294,0.6502023070089279
+24,25,0.777875,0.7671370967741935,0.5965238115787506,0.6416350141648324
+25,26,0.7875,0.751008064516129,0.584319313287735,0.6557927843063108
+26,27,0.784625,0.7651209677419355,0.5858220131397247,0.6265111680953733
+27,28,0.77975,0.7610887096774194,0.5928270950317382,0.6353864400617538
+28,29,0.780625,0.7520161290322581,0.5824392430782318,0.6569667564284417
+29,30,0.78475,0.7686491935483871,0.5810435285568237,0.6306867743692091
+30,31,0.789,0.7706653225806451,0.5672282783985138,0.6261125274242894
+31,32,0.78525,0.7560483870967742,0.5757509377002716,0.6505500414679127
+32,33,0.792,0.7681451612903226,0.5613697295188904,0.629849144527989
+33,34,0.793875,0.7620967741935484,0.5625830183029175,0.6189906856706066
+34,35,0.791875,0.7681451612903226,0.5602230775356293,0.6212261732547514
+35,36,0.794625,0.7615927419354839,0.5545250961780548,0.619811047469416
+36,37,0.794625,0.7701612903225806,0.5573954427242279,0.6205428954093687
+37,38,0.79425,0.764616935483871,0.5514744529724122,0.628905875067557
+38,39,0.792375,0.7842741935483871,0.5618353080749512,0.5954909113145643
+39,40,0.796375,0.7711693548387096,0.5491654114723206,0.6137347759739045
+40,41,0.799,0.7641129032258065,0.5372960684299469,0.6470560056547965
+41,42,0.800375,0.7671370967741935,0.5395989503860473,0.6211921880322118
+42,43,0.806,0.7671370967741935,0.5370515692234039,0.6075864828401997
+43,44,0.801875,0.7605846774193549,0.5388010408878326,0.5891174308715328
+44,45,0.800875,0.766633064516129,0.539761929512024,0.610026998865989
+45,46,0.802375,0.780241935483871,0.5270701496601105,0.6000283591208919
+46,47,0.799875,0.7772177419354839,0.5320828959941865,0.5864833074231302
+47,48,0.807,0.7837701612903226,0.5309389193058014,0.5761975809451072

data/notebook2/random_qccnn_metrics.csv

@@ -1,51 +1,34 @@
-epoch,train_acc,valid_acc,train_loss,valid_loss
-1,0.6377291666666667,0.7426075268817204,1.0222952149709066,0.6658745436899124
-2,0.7588541666666667,0.7799059139784946,0.6436424539883931,0.5876969707909451
-3,0.7829791666666667,0.7955309139784946,0.5888768948713938,0.5508138095178912
-4,0.7927916666666667,0.8009912634408602,0.5641531114578247,0.5473136597423143
-5,0.7997291666666667,0.8065356182795699,0.5444657148520152,0.515406842834206
-6,0.8043541666666667,0.8140120967741935,0.5316140202681223,0.5052012169873843
-7,0.80975,0.8156922043010753,0.521293937365214,0.4983856466508681
-8,0.8155208333333334,0.8162802419354839,0.5079634555180867,0.49958501611986467
-9,0.8161875,0.8159442204301075,0.5036936206022898,0.4907228536503289
-10,0.816875,0.8160282258064516,0.497530157327652,0.4895895427914076
-11,0.8210416666666667,0.8269489247311828,0.4916797243754069,0.47517218673101036
-12,0.8210416666666667,0.8266129032258065,0.4892559293905894,0.468310099135163
-13,0.8241041666666666,0.8211525537634409,0.4825246704419454,0.4856182368852759
-14,0.8239791666666667,0.8264448924731183,0.48390908201535543,0.46895075997998636
-15,0.8245,0.8230846774193549,0.4795244421164195,0.47176380727880746
-16,0.8270416666666667,0.829133064516129,0.47462198424339297,0.4671747069205007
-17,0.8273958333333333,0.8297211021505376,0.4733834793567657,0.4658442559421703
-18,0.828125,0.8337533602150538,0.47195579528808596,0.46035799896845253
-19,0.8294583333333333,0.829133064516129,0.4668528276284536,0.46208705472689804
-20,0.830625,0.8303931451612904,0.46666145197550457,0.4616392642580053
-21,0.8320416666666667,0.8280409946236559,0.46406093080838523,0.4690516353935324
-22,0.8315833333333333,0.8293010752688172,0.4628266294002533,0.4590829178210228
-23,0.8315416666666666,0.8298051075268817,0.46202420779069264,0.4582436110383721
-24,0.8328541666666667,0.8282090053763441,0.4594616918563843,0.46520241454083433
-25,0.8321458333333334,0.8313172043010753,0.45899707794189454,0.4611524093535639
-26,0.835,0.8311491935483871,0.4561348853111267,0.460902286793596
-27,0.8342708333333333,0.8335013440860215,0.45571692689259846,0.4524733103731627
-28,0.83475,0.831989247311828,0.45314634958902994,0.4535403392648184
-29,0.8353958333333333,0.8333333333333334,0.4508692183494568,0.4509869323622796
-30,0.8358958333333333,0.8314012096774194,0.4508490650653839,0.4559444804345408
-31,0.8379791666666667,0.8347614247311828,0.44810916344324747,0.45132114586009775
-32,0.8369375,0.832997311827957,0.4488534228801727,0.4531173186917459
-33,0.8375208333333334,0.8366935483870968,0.4460577855904897,0.4518213977095901
-34,0.8372708333333333,0.8314852150537635,0.4453533016045888,0.4571616124081355
-35,0.8381458333333334,0.8351814516129032,0.4446527805328369,0.4513627043975297
-36,0.8392916666666667,0.8373655913978495,0.4431237142880758,0.44869983228304056
-37,0.838375,0.8347614247311828,0.4424218458334605,0.4486082660895522
-38,0.8397916666666667,0.8355174731182796,0.4412206415732702,0.44703892129723743
-39,0.8401875,0.8365255376344086,0.4396123299598694,0.4470836206149029
-40,0.8399791666666667,0.8345934139784946,0.43971687642733254,0.4466231196157394
-41,0.8410833333333333,0.8373655913978495,0.43810279977321626,0.4461495735312021
-42,0.8408958333333333,0.8372815860215054,0.4371747978528341,0.4456534408113008
-43,0.8414583333333333,0.8363575268817204,0.43774009283383686,0.4454879440287108
-44,0.841375,0.8358534946236559,0.4363077602386475,0.4460145914426414
-45,0.8408958333333333,0.8363575268817204,0.43665687982241314,0.44518788110825325
-46,0.841875,0.8371135752688172,0.43598757874965666,0.44520101848468985
-47,0.8422083333333333,0.836945564516129,0.4349166991710663,0.44520143988311933
-48,0.8421041666666667,0.8360215053763441,0.43520630804697674,0.44510498354511874
-49,0.8416875,0.836861559139785,0.4354708949327469,0.44491737311886204
-50,0.8408125,0.836861559139785,0.43524290529886883,0.4449239748139535
+,epoch,train_acc,valid_acc,train_loss,valid_loss
+0,1,0.561125,0.6476814516129032,1.271985122203827,0.9911825291572078
+1,2,0.676375,0.6668346774193549,0.9134431419372558,0.8937955517922679
+2,3,0.699125,0.6824596774193549,0.8243759956359863,0.8520721235582905
+3,4,0.716625,0.6995967741935484,0.7818560593128204,0.8005917687569896
+4,5,0.725125,0.7101814516129032,0.745261854171753,0.8037946531849522
+5,6,0.733375,0.7061491935483871,0.7344184167385102,0.7732808570707997
+6,7,0.737875,0.7253024193548387,0.7150477197170257,0.74149090051651
+7,8,0.74325,0.7253024193548387,0.7022454526424408,0.7516527022084882
+8,9,0.742625,0.7298387096774194,0.7041031284332275,0.7247120468847214
+9,10,0.747875,0.7116935483870968,0.6772661633491516,0.7338270487323884
+10,11,0.757875,0.7212701612903226,0.6562292041778565,0.7387931039256435
+11,12,0.761625,0.7328629032258065,0.6542983632087708,0.725050816612859
+12,13,0.75925,0.7318548387096774,0.6493379819393158,0.7001086867624714
+13,14,0.767375,0.7379032258064516,0.6460576868057251,0.6988139988914612
+14,15,0.765875,0.7288306451612904,0.6339491362571716,0.7128441218406923
+15,16,0.763125,0.7368951612903226,0.6262373595237732,0.714297366719092
+16,17,0.76925,0.7227822580645161,0.6279029569625855,0.7181522269402781
+17,18,0.77025,0.7449596774193549,0.6159816448688507,0.6757595616002237
+18,19,0.7745,0.7283266129032258,0.6136245548725128,0.6998546681096477
+19,20,0.775375,0.7469758064516129,0.6000997524261474,0.6749713065162781
+20,21,0.7805,0.748991935483871,0.5928600332736969,0.6656959902855658
+21,22,0.77525,0.7444556451612904,0.599046837568283,0.6857193170055267
+22,23,0.785,0.7384072580645161,0.5875316572189331,0.6785462142959717
+23,24,0.778375,0.765625,0.588378502368927,0.627939272311426
+24,25,0.78575,0.7459677419354839,0.5700427904129028,0.6502643679418871
+25,26,0.7805,0.75,0.5785817315578461,0.6712307555060233
+26,27,0.78675,0.7474798387096774,0.5676946561336518,0.6555941143343526
+27,28,0.787875,0.7505040322580645,0.575938116312027,0.6507430134281036
+28,29,0.789,0.7605846774193549,0.5655779435634612,0.6520830373610219
+29,30,0.790625,0.7434475806451613,0.5578647639751434,0.6789213486256138
+30,31,0.787375,0.7565524193548387,0.5687701859474182,0.649224087115257
+31,32,0.79575,0.7560483870967742,0.5432633152008056,0.6388471222692921
+32,33,0.7895,0.7595766129032258,0.5557173223495483,0.6391933618053314

data/notebook2/vgg_metrics.csv

@@ -1,51 +1,86 @@
-epoch,train_acc,valid_acc,train_loss,valid_loss
-1,0.2535833333333333,0.41952284946236557,2.274448096593221,2.2001695325297694
-2,0.4691875,0.5073084677419355,2.116311640103658,2.051948335862929
-3,0.5270416666666666,0.5266297043010753,2.009436710357666,1.9777893276624783
-4,0.5353541666666667,0.535114247311828,1.9530798486073813,1.9356736265203005
-5,0.5935416666666666,0.6175235215053764,1.9118746633529664,1.891348486305565
-6,0.636625,0.6625504032258065,1.8629151741663614,1.847881397893352
-7,0.686625,0.7302587365591398,1.8228577140172322,1.8057919330494379
-8,0.7415416666666667,0.7511760752688172,1.7803055089314779,1.7647134245082896
-9,0.7542291666666666,0.7582325268817204,1.7502256110509236,1.7423059594246648
-10,0.7607708333333333,0.764616935483871,1.7317150996526083,1.724781782396378
-11,0.7677291666666667,0.7719254032258065,1.7192926041285197,1.713816902970755
-12,0.7704791666666667,0.7731014784946236,1.71117463239034,1.7070557173862253
-13,0.771625,0.7736055107526881,1.7053865385055542,1.7027398668309695
-14,0.7747708333333333,0.7749495967741935,1.7004834445317587,1.6978983327906618
-15,0.7763333333333333,0.7751176075268817,1.6967166414260864,1.6960316857983988
-16,0.7790416666666666,0.7767977150537635,1.6929608987172444,1.692371742699736
-17,0.780125,0.7801579301075269,1.6910814901987712,1.6917471321680213
-18,0.7817916666666667,0.7827620967741935,1.6882368882497152,1.6861866725388395
-19,0.7827291666666667,0.7815860215053764,1.6857908573150635,1.6868788478195027
-20,0.7827083333333333,0.7840221774193549,1.6843910643259685,1.6850647259784002
-21,0.7848541666666666,0.7857862903225806,1.6823169787724812,1.6813292862266622
-22,0.7876666666666666,0.7848622311827957,1.6801685171127319,1.6815461574062225
-23,0.7879791666666667,0.7848622311827957,1.6788572374979656,1.6818229216401295
-24,0.7890833333333334,0.7859543010752689,1.6776897859573365,1.6799169009731663
-25,0.7902708333333334,0.7883904569892473,1.6763870159784953,1.6777271570697907
-26,0.791,0.789986559139785,1.6751194190979004,1.6777111407249206
-27,0.7918541666666666,0.7886424731182796,1.674490571975708,1.6776156912567795
-28,0.7936666666666666,0.7905745967741935,1.6728278172810873,1.6744540147883917
-29,0.793375,0.7876344086021505,1.671903525352478,1.6776354030896259
-30,0.7945833333333333,0.7901545698924731,1.6710031661987306,1.6751063805754467
-31,0.7959166666666667,0.7932627688172043,1.6700964994430543,1.6719800926023913
-32,0.7955833333333333,0.7933467741935484,1.6697869466145834,1.6723043623790945
-33,0.7971458333333333,0.792002688172043,1.668815224647522,1.6722341519530102
-34,0.7974375,0.7931787634408602,1.6682115157445272,1.6723163063808153
-35,0.7979791666666667,0.7947748655913979,1.6676976483662924,1.670254216399244
-36,0.7984583333333334,0.7950268817204301,1.6669070380528768,1.6704239499184392
-37,0.799,0.795866935483871,1.6664897476832072,1.6691849411174815
-38,0.7991875,0.7948588709677419,1.6660230029424032,1.6700230785595473
-39,0.800125,0.795950940860215,1.6655609397888183,1.6688298884258475
-40,0.8006458333333333,0.7956989247311828,1.6652250595092772,1.6690674840763051
-41,0.8006875,0.7962869623655914,1.664971160888672,1.6688041135828982
-42,0.8015416666666667,0.795866935483871,1.6647087678909303,1.668453808753721
-43,0.8013958333333333,0.7962029569892473,1.6643381754557292,1.6682484457569737
-44,0.8016458333333333,0.7965389784946236,1.6641663211186728,1.668392535178892
-45,0.8017708333333333,0.7957829301075269,1.6640429185231527,1.6681856429705055
-46,0.8021041666666666,0.7965389784946236,1.663849822362264,1.6680005634984663
-47,0.8024583333333334,0.7966229838709677,1.663770879427592,1.6679569816076627
-48,0.8025833333333333,0.7962029569892473,1.6636734215418498,1.6680130330465173
-49,0.8024583333333334,0.7970430107526881,1.6635978577931723,1.6679544077124646
-50,0.8025,0.7970430107526881,1.6635685895284016,1.6679527682642783
+,epoch,train_acc,valid_acc,train_loss,valid_loss
+0,1,0.575125,0.7011088709677419,2.031563010215759,1.8535375249001287
+1,2,0.723875,0.7515120967741935,1.7430778923034669,1.713461822079074
+2,3,0.76425,0.7525201612903226,1.6991203842163085,1.7051894049490652
+3,4,0.779875,0.7716733870967742,1.6807003259658813,1.690080665772961
+4,5,0.79525,0.7696572580645161,1.6654855661392212,1.6926762519344207
+5,6,0.80175,0.7888104838709677,1.6598792200088501,1.6711357831954956
+6,7,0.80875,0.7721774193548387,1.6529147624969482,1.6847506146277151
+7,8,0.812375,0.8004032258064516,1.649049828529358,1.6597587523921844
+8,9,0.815375,0.7893145161290323,1.6458229179382324,1.6707303101016628
+9,10,0.825,0.811491935483871,1.6370085287094116,1.6518691432091497
+10,11,0.832375,0.795866935483871,1.630222158432007,1.6616583139665666
+11,12,0.83225,0.8145161290322581,1.6293829317092896,1.6454557026586225
+12,13,0.834875,0.8125,1.6268642024993896,1.6500283018235238
+13,14,0.840875,0.8104838709677419,1.6199358901977539,1.652194688397069
+14,15,0.83975,0.8069556451612904,1.6210605192184449,1.6520756982987927
+15,16,0.84175,0.8004032258064516,1.6195698108673096,1.6582853217278757
+16,17,0.838625,0.8049395161290323,1.6228710346221924,1.656198097813514
+17,18,0.83775,0.8140120967741935,1.6226149969100951,1.6462942131104008
+18,19,0.845375,0.8251008064516129,1.6150365982055663,1.6358921681680987
+19,20,0.85125,0.8296370967741935,1.6097205333709717,1.6333312757553593
+20,21,0.851875,0.8210685483870968,1.609424132347107,1.6388044895664338
+21,22,0.850375,0.8210685483870968,1.6101643466949462,1.6369191561975787
+22,23,0.858125,0.8185483870967742,1.6031983375549317,1.6421872877305554
+23,24,0.857125,0.8361895161290323,1.6038059520721435,1.62519553015309
+24,25,0.854,0.8301411290322581,1.606280044555664,1.6290803609355804
+25,26,0.858375,0.8119959677419355,1.6011180095672608,1.648349738890125
+26,27,0.861375,0.8371975806451613,1.6000273780822754,1.6237776010267195
+27,28,0.8705,0.8351814516129032,1.5907322645187378,1.6240986316434798
+28,29,0.87175,0.828125,1.5896439476013184,1.6316479982868317
+29,30,0.867375,0.8266129032258065,1.5936143741607667,1.632056209348863
+30,31,0.86575,0.8377016129032258,1.5949567575454713,1.6240303862479426
+31,32,0.86975,0.8336693548387096,1.5907692136764526,1.6251203167823054
+32,33,0.86975,0.8397177419354839,1.5919580411911012,1.6214879328204739
+33,34,0.872375,0.8422379032258065,1.5886561880111694,1.6194973222671016
+34,35,0.876,0.8331653225806451,1.585767653465271,1.6247131862948019
+35,36,0.871875,0.8392137096774194,1.5895709590911866,1.6215527211466143
+36,37,0.871125,0.8245967741935484,1.590462643623352,1.6370407227546937
+37,38,0.87325,0.8301411290322581,1.5876847248077393,1.6306157150576193
+38,39,0.8705,0.8240927419354839,1.590844289779663,1.634965923524672
+39,40,0.88125,0.8402217741935484,1.5803495874404907,1.6220719968118975
+40,41,0.881875,0.8462701612903226,1.5788077726364136,1.614263488400367
+41,42,0.87625,0.8487903225806451,1.5843929862976074,1.6131737078389814
+42,43,0.88175,0.842741935483871,1.5789586782455445,1.6184405088424683
+43,44,0.880375,0.8417338709677419,1.5802030544281005,1.6189708786626016
+44,45,0.880375,0.8492943548387096,1.5803771505355835,1.610948174230514
+45,46,0.877625,0.8523185483870968,1.5828620948791503,1.6089561793111986
+46,47,0.882125,0.8442540322580645,1.578727219581604,1.6160847948443504
+47,48,0.872875,0.8392137096774194,1.5874832077026366,1.6228746021947553
+48,49,0.884125,0.8371975806451613,1.5771635084152222,1.6237398655183855
+49,50,0.88975,0.8341733870967742,1.5712118272781372,1.6256768280459988
+50,51,0.876,0.8432459677419355,1.585065812110901,1.6188401445265739
+51,52,0.883,0.844758064516129,1.5785351629257203,1.6169025359615203
+52,53,0.889125,0.8548387096774194,1.5723771095275878,1.607426397262081
+53,54,0.88625,0.8518145161290323,1.5749757404327394,1.6076742141477522
+54,55,0.891125,0.8452620967741935,1.5695797872543336,1.6156867473356185
+55,56,0.891625,0.8497983870967742,1.5691107511520386,1.6118658050414054
+56,57,0.883375,0.8508064516129032,1.577266471862793,1.6087883749315817
+57,58,0.891625,0.8422379032258065,1.5702907581329346,1.6168161553721274
+58,59,0.89375,0.8553427419354839,1.5679762859344482,1.6071604336461713
+59,60,0.885875,0.8482862903225806,1.5749410953521727,1.6108381056016492
+60,61,0.891875,0.8417338709677419,1.5696327953338622,1.6187968907817718
+61,62,0.894,0.8392137096774194,1.5671770153045654,1.6208721822307957
+62,63,0.891125,0.8392137096774194,1.569726734161377,1.6224797733368412
+63,64,0.89325,0.8518145161290323,1.568245210647583,1.606977516605008
+64,65,0.895125,0.8462701612903226,1.5658181638717652,1.6140611633177726
+65,66,0.89175,0.8568548387096774,1.5686321334838866,1.6044510756769488
+66,67,0.896375,0.8412298387096774,1.5650024271011354,1.6208048520549652
+67,68,0.8895,0.8477822580645161,1.571241024017334,1.6136699338113107
+68,69,0.901625,0.8472782258064516,1.5600450325012207,1.612250724146443
+69,70,0.89025,0.8306451612903226,1.5705311574935914,1.630635638390818
+70,71,0.89675,0.8538306451612904,1.5637341051101685,1.6056317244806597
+71,72,0.897,0.8412298387096774,1.5633089447021484,1.6192954970944313
+72,73,0.900375,0.8503024193548387,1.560647201538086,1.6097132775091356
+73,74,0.895625,0.8533266129032258,1.565330421447754,1.6075265484471475
+74,75,0.895125,0.8543346774193549,1.565690894126892,1.6060891266792052
+75,76,0.900625,0.8608870967741935,1.560530616760254,1.6011101891917567
+76,77,0.897125,0.8563508064516129,1.563725378036499,1.6037284789546844
+77,78,0.89175,0.8482862903225806,1.5693257989883422,1.6122698245509979
+78,79,0.90225,0.8487903225806451,1.5589981718063355,1.6111104180735927
+79,80,0.893375,0.8442540322580645,1.5669814805984497,1.6161983090062295
+80,81,0.897125,0.8573588709677419,1.56400607585907,1.6040064134905416
+81,82,0.898,0.8568548387096774,1.5621892538070679,1.6044225615839804
+82,83,0.89075,0.8503024193548387,1.5697740039825439,1.6114465228972896
+83,84,0.900625,0.8598790322580645,1.5606124105453492,1.6002429416102748
+84,85,0.89825,0.8553427419354839,1.5628600664138794,1.603792409743032