Cập nhật mô hình, xóa các file không cần thiết

Code.ipynb

@@ -6,32 +6,76 @@
    "metadata": {},
    "outputs": [
     {
-     "data": {
-      "application/vnd.jupyter.widget-view+json": {
-       "model_id": "1165efaad02849e4945d9b0f726a33c3",
-       "version_major": 2,
-       "version_minor": 0
-      },
-      "text/plain": [
-       "Saving the dataset (0/7 shards):   0%|          | 0/82560 [00:00<?, ? examples/s]"
-      ]
-     },
-     "metadata": {},
-     "output_type": "display_data"
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Some weights of ViTForImageClassification were not initialized from the model checkpoint at google/vit-base-patch16-224 and are newly initialized because the shapes did not match:\n",
+      "- classifier.bias: found shape torch.Size([1000]) in the checkpoint and torch.Size([1]) in the model instantiated\n",
+      "- classifier.weight: found shape torch.Size([1000, 768]) in the checkpoint and torch.Size([1, 768]) in the model instantiated\n",
+      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Conversion to ONNX completed successfully!\n"
+     ]
     }
    ],
    "source": [
-    "# Method 1: Using datasets library\n",
-    "from datasets import load_dataset\n",
+    "import torch\n",
+    "from transformers import ViTForImageClassification\n",
+    "import torch.nn as nn\n",
+    "\n",
+    "# 1. Định nghĩa lại lớp mô hình (phải giống hệt khi training)\n",
+    "class ViTBinaryClassifier(nn.Module):\n",
+    "    def __init__(self, pretrained_model=\"google/vit-base-patch16-224\", freeze_base=False):\n",
+    "        super().__init__()\n",
+    "        self.vit = ViTForImageClassification.from_pretrained(\n",
+    "            pretrained_model,\n",
+    "            num_labels=1,\n",
+    "            ignore_mismatched_sizes=True\n",
+    "        )\n",
+    "        self.vit.classifier = nn.Sequential(\n",
+    "            nn.Linear(self.vit.config.hidden_size, 256),\n",
+    "            nn.ReLU(),\n",
+    "            nn.Dropout(0.1),\n",
+    "            nn.Linear(256, 1)\n",
+    "        )\n",
+    "        if freeze_base:\n",
+    "            for param in self.vit.vit.parameters():\n",
+    "                param.requires_grad = False\n",
+    "\n",
+    "    def forward(self, pixel_values):\n",
+    "        outputs = self.vit(pixel_values)\n",
+    "        return outputs.logits\n",
     "\n",
-    "# Download and load the dataset\n",
-    "dataset = load_dataset(\"akinsanyaayomide/skin_cancer_dataset_balanced_labels_aug\")\n",
+    "# 2. Khởi tạo và load weights\n",
+    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
+    "model = ViTBinaryClassifier().to(device)\n",
+    "model.load_state_dict(torch.load(r\"D:\\SonCode\\Taosafescan\\vit_binary_classifier_best.pt\", map_location=device))\n",
+    "model.eval()  # Chuyển sang chế độ inference\n",
     "\n",
-    "# Print basic dataset information\n",
-    "print(dataset)\n",
+    "# 3. Tạo dummy input với kích thước phù hợp (batch_size, channels, height, width)\n",
+    "dummy_input = torch.randn(1, 3, 224, 224).to(device)  # Kích thước ảnh 224x224 cho ViT-base\n",
     "\n",
-    "# Example: Access first few samples\n",
-    "print(dataset[\"train\"][:3])"
+    "# 4. Xuất sang ONNX\n",
+    "torch.onnx.export(\n",
+    "    model,\n",
+    "    dummy_input,\n",
+    "    \"vit_binary_classification.onnx\",\n",
+    "    export_params=True,\n",
+    "    opset_version=14,  # Thay đổi tại đây\n",
+    "    do_constant_folding=True,\n",
+    "    input_names=[\"pixel_values\"],\n",
+    "    output_names=[\"logits\"],\n",
+    "    dynamic_axes={\n",
+    "        \"pixel_values\": {0: \"batch_size\"},\n",
+    "        \"logits\": {0: \"batch_size\"}\n",
+    "    }\n",
+    ")\n",
+    "print(\"Conversion to ONNX completed successfully!\")"
    ]
   }
  ],

Code2.ipynb

@@ -1,94 +0,0 @@
-{
- "cells": [
-  {
-   "cell_type": "code",
-   "execution_count": 1,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "Collecting monai\n",
-      "  Downloading monai-1.4.0-py3-none-any.whl.metadata (11 kB)\n",
-      "Requirement already satisfied: numpy<2.0,>=1.24 in c:\\users\\1\\anaconda3\\lib\\site-packages (from monai) (1.26.4)\n",
-      "Requirement already satisfied: torch>=1.9 in c:\\users\\1\\anaconda3\\lib\\site-packages (from monai) (2.6.0+cu126)\n",
-      "Requirement already satisfied: filelock in c:\\users\\1\\anaconda3\\lib\\site-packages (from torch>=1.9->monai) (3.13.1)\n",
-      "Requirement already satisfied: typing-extensions>=4.10.0 in c:\\users\\1\\anaconda3\\lib\\site-packages (from torch>=1.9->monai) (4.11.0)\n",
-      "Requirement already satisfied: networkx in c:\\users\\1\\anaconda3\\lib\\site-packages (from torch>=1.9->monai) (3.3)\n",
-      "Requirement already satisfied: jinja2 in c:\\users\\1\\anaconda3\\lib\\site-packages (from torch>=1.9->monai) (3.1.4)\n",
-      "Requirement already satisfied: fsspec in c:\\users\\1\\anaconda3\\lib\\site-packages (from torch>=1.9->monai) (2024.6.1)\n",
-      "Requirement already satisfied: setuptools in c:\\users\\1\\anaconda3\\lib\\site-packages (from torch>=1.9->monai) (75.1.0)\n",
-      "Requirement already satisfied: sympy==1.13.1 in c:\\users\\1\\anaconda3\\lib\\site-packages (from torch>=1.9->monai) (1.13.1)\n",
-      "Requirement already satisfied: mpmath<1.4,>=1.1.0 in c:\\users\\1\\anaconda3\\lib\\site-packages (from sympy==1.13.1->torch>=1.9->monai) (1.3.0)\n",
-      "Requirement already satisfied: MarkupSafe>=2.0 in c:\\users\\1\\anaconda3\\lib\\site-packages (from jinja2->torch>=1.9->monai) (2.1.3)\n",
-      "Downloading monai-1.4.0-py3-none-any.whl (1.5 MB)\n",
-      "   ---------------------------------------- 0.0/1.5 MB ? eta -:--:--\n",
-      "   ---------------------------------------- 0.0/1.5 MB ? eta -:--:--\n",
-      "   ---------------------------------------- 0.0/1.5 MB ? eta -:--:--\n",
-      "   ------ --------------------------------- 0.3/1.5 MB ? eta -:--:--\n",
-      "   ------------- -------------------------- 0.5/1.5 MB 1.0 MB/s eta 0:00:01\n",
-      "   -------------------- ------------------- 0.8/1.5 MB 1.3 MB/s eta 0:00:01\n",
-      "   -------------------- ------------------- 0.8/1.5 MB 1.3 MB/s eta 0:00:01\n",
-      "   --------------------------- ------------ 1.0/1.5 MB 1.0 MB/s eta 0:00:01\n",
-      "   --------------------------- ------------ 1.0/1.5 MB 1.0 MB/s eta 0:00:01\n",
-      "   ---------------------------------- ----- 1.3/1.5 MB 871.6 kB/s eta 0:00:01\n",
-      "   ---------------------------------- ----- 1.3/1.5 MB 871.6 kB/s eta 0:00:01\n",
-      "   ---------------------------------- ----- 1.3/1.5 MB 871.6 kB/s eta 0:00:01\n",
-      "   ---------------------------------------- 1.5/1.5 MB 695.8 kB/s eta 0:00:00\n",
-      "Installing collected packages: monai\n",
-      "Successfully installed monai-1.4.0\n"
-     ]
-    }
-   ],
-   "source": [
-    "!pip install monai"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "from monai.transforms import (\n",
-    "    Compose,\n",
-    "    LoadImage,\n",
-    "    ScaleIntensity,\n",
-    "    AddChannel\n",
-    ")\n",
-    "\n",
-    "# Define transforms for image preprocessing\n",
-    "transforms = Compose([\n",
-    "    LoadImage(image_only=True),\n",
-    "    AddChannel(),\n",
-    "    ScaleIntensity()\n",
-    "])\n",
-    "\n",
-    "# Apply transforms to your image\n",
-    "image = transforms(image_path)"
-   ]
-  }
- ],
- "metadata": {
-  "kernelspec": {
-   "display_name": "base",
-   "language": "python",
-   "name": "python3"
-  },
-  "language_info": {
-   "codemirror_mode": {
-    "name": "ipython",
-    "version": 3
-   },
-   "file_extension": ".py",
-   "mimetype": "text/x-python",
-   "name": "python",
-   "nbconvert_exporter": "python",
-   "pygments_lexer": "ipython3",
-   "version": "3.12.3"
-  }
- },
- "nbformat": 4,
- "nbformat_minor": 2
-}

Code3.ipynb

@@ -1,103 +0,0 @@
-{
- "cells": [
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "Some weights of ViTForImageClassification were not initialized from the model checkpoint at google/vit-base-patch16-224 and are newly initialized because the shapes did not match:\n",
-      "- classifier.bias: found shape torch.Size([1000]) in the checkpoint and torch.Size([1]) in the model instantiated\n",
-      "- classifier.weight: found shape torch.Size([1000, 768]) in the checkpoint and torch.Size([1, 768]) in the model instantiated\n",
-      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "Conversion to ONNX completed successfully!\n"
-     ]
-    }
-   ],
-   "source": [
-    "import torch\n",
-    "from transformers import ViTForImageClassification\n",
-    "import torch.nn as nn\n",
-    "\n",
-    "# 1. Định nghĩa lại lớp mô hình (phải giống hệt khi training)\n",
-    "class ViTBinaryClassifier(nn.Module):\n",
-    "    def __init__(self, pretrained_model=\"google/vit-base-patch16-224\", freeze_base=False):\n",
-    "        super().__init__()\n",
-    "        self.vit = ViTForImageClassification.from_pretrained(\n",
-    "            pretrained_model,\n",
-    "            num_labels=1,\n",
-    "            ignore_mismatched_sizes=True\n",
-    "        )\n",
-    "        self.vit.classifier = nn.Sequential(\n",
-    "            nn.Linear(self.vit.config.hidden_size, 256),\n",
-    "            nn.ReLU(),\n",
-    "            nn.Dropout(0.1),\n",
-    "            nn.Linear(256, 1)\n",
-    "        )\n",
-    "        if freeze_base:\n",
-    "            for param in self.vit.vit.parameters():\n",
-    "                param.requires_grad = False\n",
-    "\n",
-    "    def forward(self, pixel_values):\n",
-    "        outputs = self.vit(pixel_values)\n",
-    "        return outputs.logits\n",
-    "\n",
-    "# 2. Khởi tạo và load weights\n",
-    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
-    "model = ViTBinaryClassifier().to(device)\n",
-    "model.load_state_dict(torch.load(r\"D:\\SonCode\\Taosafescan\\vit_binary_classifier_best.pt\", map_location=device))\n",
-    "model.eval()  # Chuyển sang chế độ inference\n",
-    "\n",
-    "# 3. Tạo dummy input với kích thước phù hợp (batch_size, channels, height, width)\n",
-    "dummy_input = torch.randn(1, 3, 224, 224).to(device)  # Kích thước ảnh 224x224 cho ViT-base\n",
-    "\n",
-    "# 4. Xuất sang ONNX\n",
-    "torch.onnx.export(\n",
-    "    model,\n",
-    "    dummy_input,\n",
-    "    \"vit_binary_classification.onnx\",\n",
-    "    export_params=True,\n",
-    "    opset_version=14,  # Thay đổi tại đây\n",
-    "    do_constant_folding=True,\n",
-    "    input_names=[\"pixel_values\"],\n",
-    "    output_names=[\"logits\"],\n",
-    "    dynamic_axes={\n",
-    "        \"pixel_values\": {0: \"batch_size\"},\n",
-    "        \"logits\": {0: \"batch_size\"}\n",
-    "    }\n",
-    ")\n",
-    "print(\"Conversion to ONNX completed successfully!\")"
-   ]
-  }
- ],
- "metadata": {
-  "kernelspec": {
-   "display_name": "base",
-   "language": "python",
-   "name": "python3"
-  },
-  "language_info": {
-   "codemirror_mode": {
-    "name": "ipython",
-    "version": 3
-   },
-   "file_extension": ".py",
-   "mimetype": "text/x-python",
-   "name": "python",
-   "nbconvert_exporter": "python",
-   "pygments_lexer": "ipython3",
-   "version": "3.12.3"
-  }
- },
- "nbformat": 4,
- "nbformat_minor": 2
-}