Add executed notebook with outputs (all 33 cells run)

clip_tutorial.ipynb

@@ -65,11 +65,21 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 1,
    "id": "25a6d7c7",
-   "metadata": {},
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:07:45.417657Z",
+     "iopub.status.busy": "2026-05-30T01:07:45.417118Z",
+     "iopub.status.idle": "2026-05-30T01:07:46.121194Z",
+     "shell.execute_reply": "2026-05-30T01:07:46.120053Z"
+    }
+   },
    "outputs": [],
    "source": [
+    "import matplotlib\n",
+    "matplotlib.use(\"Agg\")\n",
+    "\n",
     "# ── Cell 1.1: Visualisasi konsep similarity matrix ──────────────────────────\n",
     "import numpy as np\n",
     "import matplotlib.pyplot as plt\n",
@@ -127,10 +137,40 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 2,
    "id": "1e24cd2b",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:07:46.125861Z",
+     "iopub.status.busy": "2026-05-30T01:07:46.125608Z",
+     "iopub.status.idle": "2026-05-30T01:07:49.587548Z",
+     "shell.execute_reply": "2026-05-30T01:07:49.586502Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Using device: cuda\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/venv/main/lib/python3.12/site-packages/torchvision/datasets/cifar.py:83: VisibleDeprecationWarning: dtype(): align should be passed as Python or NumPy boolean but got `align=0`. Did you mean to pass a tuple to create a subarray type? (Deprecated NumPy 2.4)\n",
+      "  entry = pickle.load(f, encoding=\"latin1\")\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Toy dataset size: 1000 images\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 2.1: Setup dan download CIFAR-10 ───────────────────────────────────\n",
     "import torch\n",
@@ -151,7 +191,7 @@
     "    transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))\n",
     "])\n",
     "\n",
-    "full_train = datasets.CIFAR10(root='./data', train=True, download=True, transform=transform)\n",
+    "full_train = datasets.CIFAR10(root='/workspace/data', train=True, download=True, transform=transform)\n",
     "\n",
     "# Ambil 100 images per class = 1000 total\n",
     "indices_per_class = {c: [] for c in range(10)}\n",
@@ -170,9 +210,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 3,
    "id": "48df6b98",
-   "metadata": {},
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:07:49.592486Z",
+     "iopub.status.busy": "2026-05-30T01:07:49.592039Z",
+     "iopub.status.idle": "2026-05-30T01:07:49.600668Z",
+     "shell.execute_reply": "2026-05-30T01:07:49.599519Z"
+    }
+   },
    "outputs": [],
    "source": [
     "# ── Cell 2.2: Tiny Image Encoder ────────────────────────────────────────────\n",
@@ -222,10 +269,26 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 4,
    "id": "0bd8a6ea",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:07:49.605442Z",
+     "iopub.status.busy": "2026-05-30T01:07:49.605177Z",
+     "iopub.status.idle": "2026-05-30T01:07:49.730158Z",
+     "shell.execute_reply": "2026-05-30T01:07:49.729076Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Vocabulary: {'a': 1, 'airplane': 2, 'automobile': 3, 'bird': 4, 'cat': 5, 'deer': 6, 'dog': 7, 'frog': 8, 'horse': 9, 'of': 10, 'photo': 11, 'ship': 12, 'truck': 13, '<pad>': 0}\n",
+      "Template tokens shape: torch.Size([10, 8])\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 2.4: Simple Tokenizer ───────────────────────────────────────────────\n",
     "templates = [f\"a photo of a {c}\" for c in CIFAR10_CLASSES]\n",
@@ -252,9 +315,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 5,
    "id": "6d6dce03",
-   "metadata": {},
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:07:49.733518Z",
+     "iopub.status.busy": "2026-05-30T01:07:49.733317Z",
+     "iopub.status.idle": "2026-05-30T01:07:49.738357Z",
+     "shell.execute_reply": "2026-05-30T01:07:49.736505Z"
+    }
+   },
    "outputs": [],
    "source": [
     "# ── Cell 2.5: InfoNCE Loss ───────────────────────────────────────────────────\n",
@@ -279,10 +349,25 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 6,
    "id": "45e059f3",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:07:49.742790Z",
+     "iopub.status.busy": "2026-05-30T01:07:49.742553Z",
+     "iopub.status.idle": "2026-05-30T01:07:49.750386Z",
+     "shell.execute_reply": "2026-05-30T01:07:49.749325Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "TinyCLIP parameters: 36,992\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 2.6: TinyCLIP model ────────────────────────────────────────────────\n",
     "class TinyCLIP(nn.Module):\n",
@@ -304,10 +389,58 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 7,
    "id": "f360ddba",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:07:49.754647Z",
+     "iopub.status.busy": "2026-05-30T01:07:49.754364Z",
+     "iopub.status.idle": "2026-05-30T01:07:51.795842Z",
+     "shell.execute_reply": "2026-05-30T01:07:51.793975Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch  1/10 | Loss: 4.0007\n",
+      "Epoch  2/10 | Loss: 3.8239\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch  3/10 | Loss: 3.7374\n",
+      "Epoch  4/10 | Loss: 3.6494\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch  5/10 | Loss: 3.5925\n",
+      "Epoch  6/10 | Loss: 3.5350\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch  7/10 | Loss: 3.4929\n",
+      "Epoch  8/10 | Loss: 3.4293\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch  9/10 | Loss: 3.3991\n",
+      "Epoch 10/10 | Loss: 3.3990\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 2.7: Training loop ──────────────────────────────────────────────────\n",
     "optimizer = torch.optim.Adam(model.parameters(), lr=1e-3)\n",
@@ -348,9 +481,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 8,
    "id": "cb49c0e2",
-   "metadata": {},
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:07:51.800144Z",
+     "iopub.status.busy": "2026-05-30T01:07:51.799672Z",
+     "iopub.status.idle": "2026-05-30T01:07:51.853728Z",
+     "shell.execute_reply": "2026-05-30T01:07:51.852713Z"
+    }
+   },
    "outputs": [],
    "source": [
     "# ── Cell 2.8: Plot loss curve ────────────────────────────────────────────────\n",
@@ -368,9 +508,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 9,
    "id": "7e920a81",
-   "metadata": {},
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:07:51.858456Z",
+     "iopub.status.busy": "2026-05-30T01:07:51.858122Z",
+     "iopub.status.idle": "2026-05-30T01:07:55.968437Z",
+     "shell.execute_reply": "2026-05-30T01:07:55.967679Z"
+    }
+   },
    "outputs": [],
    "source": [
     "# ── Cell 2.9: Visualisasi embedding space (t-SNE) ───────────────────────────\n",
@@ -431,10 +578,273 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 10,
    "id": "1cde344f",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:07:55.972774Z",
+     "iopub.status.busy": "2026-05-30T01:07:55.972129Z",
+     "iopub.status.idle": "2026-05-30T01:08:02.791603Z",
+     "shell.execute_reply": "2026-05-30T01:08:02.790296Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      "  0%|                                               | 0.00/338M [00:00<?, ?iB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      "  0%|                                       | 240k/338M [00:00<02:26, 2.42MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      "  1%|▎                                     | 2.69M/338M [00:00<00:21, 16.0MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      "  4%|█▋                                    | 14.9M/338M [00:00<00:05, 67.0MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      "  8%|███                                   | 27.7M/338M [00:00<00:03, 93.1MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 12%|████▋                                  | 41.0M/338M [00:00<00:02, 110MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 16%|██████▏                                | 54.0M/338M [00:00<00:02, 119MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 20%|███████▋                               | 66.2M/338M [00:00<00:02, 122MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 23%|█████████▏                             | 79.3M/338M [00:00<00:02, 127MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 27%|██████████▋                            | 92.5M/338M [00:00<00:01, 130MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 31%|████████████▌                           | 106M/338M [00:01<00:01, 133MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 35%|██████████████                          | 119M/338M [00:01<00:01, 134MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 39%|███████████████▌                        | 132M/338M [00:01<00:01, 134MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 43%|█████████████████▏                      | 145M/338M [00:01<00:01, 136MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 47%|██████████████████▊                     | 158M/338M [00:01<00:01, 137MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 51%|████████████████████▎                   | 171M/338M [00:01<00:01, 137MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 55%|█████████████████████▊                  | 184M/338M [00:01<00:01, 130MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 58%|███████████████████████▎                | 197M/338M [00:01<00:01, 103MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 62%|████████████████████████▌               | 208M/338M [00:01<00:01, 104MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 65%|█████████████████████████▉              | 219M/338M [00:02<00:01, 107MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 69%|███████████████████████████▍            | 232M/338M [00:02<00:00, 115MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 72%|████████████████████████████▉           | 244M/338M [00:02<00:00, 119MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 76%|██████████████████████████████▎         | 256M/338M [00:02<00:00, 120MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 79%|██████████████████████████████▉        | 268M/338M [00:02<00:00, 90.1MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 83%|████████████████████████████████▎      | 280M/338M [00:02<00:00, 98.8MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 87%|██████████████████████████████████▋     | 293M/338M [00:02<00:00, 108MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 90%|████████████████████████████████████    | 305M/338M [00:02<00:00, 112MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 94%|█████████████████████████████████████▋  | 318M/338M [00:02<00:00, 119MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      " 98%|███████████████████████████████████████▏| 331M/338M [00:03<00:00, 123MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\r",
+      "100%|████████████████████████████████████████| 338M/338M [00:03<00:00, 114MiB/s]"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Model loaded. Input resolution: 224\n",
+      "Embedding dimension: 512\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 3.1: Install dan load CLIP ─────────────────────────────────────────\n",
     "# !pip install git+https://github.com/openai/CLIP.git  # jalankan sekali jika belum\n",
@@ -455,10 +865,32 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 11,
    "id": "51eb6f2e",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:02.795735Z",
+     "iopub.status.busy": "2026-05-30T01:08:02.795482Z",
+     "iopub.status.idle": "2026-05-30T01:08:07.727582Z",
+     "shell.execute_reply": "2026-05-30T01:08:07.726747Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "STL-10 classes: ['airplane', 'bird', 'car', 'cat', 'deer', 'dog', 'horse', 'monkey', 'ship', 'truck']\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "ZS set: 200 | LP test set: 200\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 3.2: Load STL-10 test images ───────────────────────────────────────\n",
     "from torchvision import datasets\n",
@@ -472,7 +904,7 @@
     "              (0.26862954, 0.26130258, 0.27577711))\n",
     "])\n",
     "\n",
-    "stl10_test = datasets.STL10(root='./data', split='test', download=True,\n",
+    "stl10_test = datasets.STL10(root='/workspace/raw_data', split='test', download=True,\n",
     "                             transform=clip_preprocess)\n",
     "\n",
     "STL10_CLASSES = stl10_test.classes\n",
@@ -505,10 +937,33 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 12,
    "id": "0c99135b",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:07.729568Z",
+     "iopub.status.busy": "2026-05-30T01:08:07.729354Z",
+     "iopub.status.idle": "2026-05-30T01:08:08.214303Z",
+     "shell.execute_reply": "2026-05-30T01:08:08.213291Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Extracting embeddings...\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Zero-shot embeddings: (200, 512)\n",
+      "Linear probing test embeddings: (200, 512)\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 3.3: Extract CLIP image embeddings ──────────────────────────────────\n",
     "from torch.utils.data import DataLoader, TensorDataset\n",
@@ -536,10 +991,25 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 13,
    "id": "61608dde",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:08.218031Z",
+     "iopub.status.busy": "2026-05-30T01:08:08.217685Z",
+     "iopub.status.idle": "2026-05-30T01:08:08.258211Z",
+     "shell.execute_reply": "2026-05-30T01:08:08.257368Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Text embeddings: (10, 512)\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 3.4: Extract text embeddings untuk 10 kelas ────────────────────────\n",
     "def get_text_embeddings(class_names, template=\"a photo of a {}\"):\n",
@@ -558,9 +1028,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 14,
    "id": "5a113a9e",
-   "metadata": {},
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:08.261919Z",
+     "iopub.status.busy": "2026-05-30T01:08:08.261731Z",
+     "iopub.status.idle": "2026-05-30T01:08:08.864075Z",
+     "shell.execute_reply": "2026-05-30T01:08:08.863083Z"
+    }
+   },
    "outputs": [],
    "source": [
     "# ── Cell 3.5: Visualisasi t-SNE ─────────────────────────────────────────────\n",
@@ -612,10 +1089,25 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 15,
    "id": "6d54f9b1",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:08.868302Z",
+     "iopub.status.busy": "2026-05-30T01:08:08.867825Z",
+     "iopub.status.idle": "2026-05-30T01:08:08.882139Z",
+     "shell.execute_reply": "2026-05-30T01:08:08.880819Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Zero-shot accuracy: 0.9700 (97.0%)\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 4.1: Zero-shot prediction ──────────────────────────────────────────\n",
     "# zs_embeddings dan text_embeddings sudah dihitung di Section 3\n",
@@ -630,10 +1122,31 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 16,
    "id": "bfb07706",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:08.886646Z",
+     "iopub.status.busy": "2026-05-30T01:08:08.886415Z",
+     "iopub.status.idle": "2026-05-30T01:08:08.943583Z",
+     "shell.execute_reply": "2026-05-30T01:08:08.942419Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Prompt engineering comparison:\n",
+      "--------------------------------------------------\n",
+      "  {}                                       → 97.0%\n",
+      "  a photo of a {}                          → 97.0%\n",
+      "  a photo of a {}, high quality            → 96.5%\n",
+      "  a {} in the wild                         → 96.0%\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 4.2: Pengaruh prompt engineering ────────────────────────────────────\n",
     "templates_to_compare = [\n",
@@ -655,9 +1168,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 17,
    "id": "aa9db826",
-   "metadata": {},
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:08.947319Z",
+     "iopub.status.busy": "2026-05-30T01:08:08.947076Z",
+     "iopub.status.idle": "2026-05-30T01:08:09.015923Z",
+     "shell.execute_reply": "2026-05-30T01:08:09.014911Z"
+    }
+   },
    "outputs": [],
    "source": [
     "# ── Cell 4.3: Confusion matrix ───────────────────────────────────────────────\n",
@@ -676,10 +1196,36 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 18,
    "id": "eb5c282d",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:09.020221Z",
+     "iopub.status.busy": "2026-05-30T01:08:09.019982Z",
+     "iopub.status.idle": "2026-05-30T01:08:09.025774Z",
+     "shell.execute_reply": "2026-05-30T01:08:09.024616Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Per-class accuracy:\n",
+      "  bird       100.0% ████████████████████\n",
+      "  dog        100.0% ████████████████████\n",
+      "  horse      100.0% ████████████████████\n",
+      "  monkey     100.0% ████████████████████\n",
+      "  ship       100.0% ████████████████████\n",
+      "  airplane    95.0% ███████████████████\n",
+      "  car         95.0% ███████████████████\n",
+      "  cat         95.0% ███████████████████\n",
+      "  deer        95.0% ███████████████████\n",
+      "  truck       90.0% ██████████████████\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 4.4: Per-class accuracy ─────────────────────────────────────────────\n",
     "per_class_acc = {}\n",
@@ -716,13 +1262,35 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 19,
    "id": "c57f46c5",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:09.029486Z",
+     "iopub.status.busy": "2026-05-30T01:08:09.029279Z",
+     "iopub.status.idle": "2026-05-30T01:08:15.811089Z",
+     "shell.execute_reply": "2026-05-30T01:08:15.810407Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Extracting train embeddings...\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Train embeddings: (1000, 512)\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 5.1: Extract train embeddings untuk linear probing ─────────────────\n",
-    "stl10_train = datasets.STL10(root='./data', split='train', download=True,\n",
+    "stl10_train = datasets.STL10(root='/workspace/raw_data', split='train', download=True,\n",
     "                              transform=clip_preprocess)\n",
     "\n",
     "# Ambil 100/class = 1000 images untuk training\n",
@@ -746,10 +1314,25 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 20,
    "id": "1dc5eed7",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:15.815676Z",
+     "iopub.status.busy": "2026-05-30T01:08:15.815477Z",
+     "iopub.status.idle": "2026-05-30T01:08:17.308346Z",
+     "shell.execute_reply": "2026-05-30T01:08:17.307666Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Linear probing accuracy: 0.9700 (97.0%)\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 5.2: Train classifier ───────────────────────────────────────────────\n",
     "from sklearn.linear_model import LogisticRegression\n",
@@ -764,10 +1347,32 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 21,
    "id": "9832535f",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:17.320194Z",
+     "iopub.status.busy": "2026-05-30T01:08:17.319219Z",
+     "iopub.status.idle": "2026-05-30T01:08:17.326850Z",
+     "shell.execute_reply": "2026-05-30T01:08:17.325630Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "==================================================\n",
+      "  Perbandingan: Zero-Shot vs Linear Probing\n",
+      "==================================================\n",
+      "  Zero-shot    (0 training images)  :  97.0%\n",
+      "  Linear probe (1.000 labeled imgs) :  97.0%\n",
+      "  Gap: +0.0%\n",
+      "==================================================\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 5.3: Perbandingan zero-shot vs linear probing ───────────────────────\n",
     "print(\"\\n\" + \"=\"*50)\n",
@@ -781,9 +1386,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 22,
    "id": "69fafdc6",
-   "metadata": {},
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:17.329019Z",
+     "iopub.status.busy": "2026-05-30T01:08:17.328815Z",
+     "iopub.status.idle": "2026-05-30T01:08:17.437675Z",
+     "shell.execute_reply": "2026-05-30T01:08:17.436433Z"
+    }
+   },
    "outputs": [],
    "source": [
     "# ── Cell 5.4: Bar chart per-class comparison ─────────────────────────────────\n",
@@ -832,16 +1444,46 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 23,
    "id": "4eac3bd2",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:17.440562Z",
+     "iopub.status.busy": "2026-05-30T01:08:17.440322Z",
+     "iopub.status.idle": "2026-05-30T01:08:17.451847Z",
+     "shell.execute_reply": "2026-05-30T01:08:17.450432Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Retrieval test: 200 queries\n",
+      "         id    image_path                                          caption_0  \\\n",
+      "0  ret_0001  ret_0001.jpg  A couple of people sit outdoors at a table wit...   \n",
+      "1  ret_0002  ret_0002.jpg  A brown and white dog stands outside while it ...   \n",
+      "\n",
+      "                                           caption_1  \\\n",
+      "0     A couple in an embrace looking at each other .   \n",
+      "1  A girl with dark brown hair and eyes in a blue...   \n",
+      "\n",
+      "                                           caption_2  \\\n",
+      "0  A girl 's hands , another person 's feet , and...   \n",
+      "1  A brown and white greyhound wearing a red numb...   \n",
+      "\n",
+      "                                           caption_3  correct_idx  \n",
+      "0       A long haired drummer plays music outdoors .            0  \n",
+      "1  a boy on skateboard is making a jump from a bl...            1  \n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 6.1: Load Flickr8k test data ────────────────────────────────────────\n",
     "import pandas as pd\n",
     "from PIL import Image as PILImage\n",
     "\n",
-    "DATASET_ROOT = '/kaggle/input/clip-pelatnas-p2'\n",
+    "DATASET_ROOT = '/workspace/output/clip-pelatnas-p2'\n",
     "candidates_df = pd.read_csv(f'{DATASET_ROOT}/test/retrieval/candidates.csv')\n",
     "print(f\"Retrieval test: {len(candidates_df)} queries\")\n",
     "print(candidates_df.head(2))\n"
@@ -849,9 +1491,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 24,
    "id": "876b6c60",
-   "metadata": {},
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:17.454380Z",
+     "iopub.status.busy": "2026-05-30T01:08:17.454178Z",
+     "iopub.status.idle": "2026-05-30T01:08:17.459992Z",
+     "shell.execute_reply": "2026-05-30T01:08:17.458850Z"
+    }
+   },
    "outputs": [],
    "source": [
     "# ── Cell 6.2: CLIP retrieval pipeline ────────────────────────────────────────\n",
@@ -877,10 +1526,26 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 25,
    "id": "70ca735c",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:17.462830Z",
+     "iopub.status.busy": "2026-05-30T01:08:17.462361Z",
+     "iopub.status.idle": "2026-05-30T01:08:21.346917Z",
+     "shell.execute_reply": "2026-05-30T01:08:21.345877Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Retrieval accuracy (4-way): 0.9400 (94.0%)\n",
+      "Random baseline: 25.0%\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 6.3: Batch evaluation ────────────────────────────────────────────────\n",
     "base_path = f'{DATASET_ROOT}/test/retrieval/queries/'\n",
@@ -903,9 +1568,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 26,
    "id": "0c01f0d0",
-   "metadata": {},
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:21.351698Z",
+     "iopub.status.busy": "2026-05-30T01:08:21.351462Z",
+     "iopub.status.idle": "2026-05-30T01:08:21.553840Z",
+     "shell.execute_reply": "2026-05-30T01:08:21.552682Z"
+    }
+   },
    "outputs": [],
    "source": [
     "# ── Cell 6.4: Qualitative examples ───────────────────────────────────────────\n",
@@ -966,10 +1638,36 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 27,
    "id": "bd004ef7",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:21.558193Z",
+     "iopub.status.busy": "2026-05-30T01:08:21.557954Z",
+     "iopub.status.idle": "2026-05-30T01:08:21.567638Z",
+     "shell.execute_reply": "2026-05-30T01:08:21.566348Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "MCQA test: 200 questions\n",
+      "          id     image_path  \\\n",
+      "0  mcqa_0001  mcqa_0001.png   \n",
+      "1  mcqa_0002  mcqa_0002.png   \n",
+      "\n",
+      "                                            question choice_A       choice_B  \\\n",
+      "0                    Which continent is highlighted?   Africa  North America   \n",
+      "1  Which property do these four objects have in c...  fragile       stretchy   \n",
+      "\n",
+      "  choice_C choice_D choice_E  num_choices          subject   grade  \n",
+      "0   Europe     Asia      NaN            4   social science  grade7  \n",
+      "1     sour      NaN      NaN            3  natural science  grade5  \n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 7.1: Load ScienceQA test data ──────────────────────────────────────\n",
     "mcqa_df = pd.read_csv(f'{DATASET_ROOT}/test/mcqa/test.csv')\n",
@@ -979,9 +1677,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 28,
    "id": "3fedf9ce",
-   "metadata": {},
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:21.571029Z",
+     "iopub.status.busy": "2026-05-30T01:08:21.570792Z",
+     "iopub.status.idle": "2026-05-30T01:08:21.577861Z",
+     "shell.execute_reply": "2026-05-30T01:08:21.576438Z"
+    }
+   },
    "outputs": [],
    "source": [
     "# ── Cell 7.2: CLIP MCQA pipeline ─────────────────────────────────────────────\n",
@@ -1010,10 +1715,26 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 29,
    "id": "80755731",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:21.580942Z",
+     "iopub.status.busy": "2026-05-30T01:08:21.580701Z",
+     "iopub.status.idle": "2026-05-30T01:08:25.613344Z",
+     "shell.execute_reply": "2026-05-30T01:08:25.612552Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "MCQA predictions done (answer column not available in test set)\n",
+      "Random baseline (~3.5 choices avg): ~28.6%\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 7.3: Batch evaluation ────────────────────────────────────────────────\n",
     "base_path_mcqa = f'{DATASET_ROOT}/test/mcqa/images/'\n",
@@ -1038,10 +1759,28 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 30,
    "id": "0b34a2f3",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:25.615571Z",
+     "iopub.status.busy": "2026-05-30T01:08:25.615332Z",
+     "iopub.status.idle": "2026-05-30T01:08:25.623214Z",
+     "shell.execute_reply": "2026-05-30T01:08:25.622425Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Question count per subject:\n",
+      "  language science          44 questions\n",
+      "  natural science           82 questions\n",
+      "  social science            74 questions\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 7.4: Analysis per subject ───────────────────────────────────────────\n",
     "mcqa_df_copy = mcqa_df.copy()\n",
@@ -1086,9 +1825,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 31,
    "id": "98109b5f",
-   "metadata": {},
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:25.625351Z",
+     "iopub.status.busy": "2026-05-30T01:08:25.625185Z",
+     "iopub.status.idle": "2026-05-30T01:08:25.631943Z",
+     "shell.execute_reply": "2026-05-30T01:08:25.631245Z"
+    }
+   },
    "outputs": [],
    "source": [
     "# ── Cell 8.1: Compile semua prediksi ─────────────────────────────────────────\n",
@@ -1119,10 +1865,42 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 32,
    "id": "0675c9b8",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:25.634244Z",
+     "iopub.status.busy": "2026-05-30T01:08:25.634046Z",
+     "iopub.status.idle": "2026-05-30T01:08:25.647532Z",
+     "shell.execute_reply": "2026-05-30T01:08:25.646412Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Submission preview (2 per task):\n",
+      "     id prediction\n",
+      "zs_0001      horse\n",
+      "zs_0002     monkey\n",
+      "\n",
+      "     id prediction\n",
+      "lp_0001       bird\n",
+      "lp_0002       bird\n",
+      "\n",
+      "      id prediction\n",
+      "ret_0001          0\n",
+      "ret_0002          1\n",
+      "\n",
+      "       id prediction\n",
+      "mcqa_0001          B\n",
+      "mcqa_0002          B\n",
+      "\n",
+      "Saved: submission.csv (800 rows)\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 8.2: Generate submission ────────────────────────────────────────────\n",
     "submission = sample_sub.copy()\n",
@@ -1143,10 +1921,30 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 33,
    "id": "9fcd617f",
-   "metadata": {},
-   "outputs": [],
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2026-05-30T01:08:25.650270Z",
+     "iopub.status.busy": "2026-05-30T01:08:25.649862Z",
+     "iopub.status.idle": "2026-05-30T01:08:25.658680Z",
+     "shell.execute_reply": "2026-05-30T01:08:25.657211Z"
+    }
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Distribusi prediksi per task:\n",
+      "\n",
+      "  Zero-shot (200 items): {'horse': 21, 'monkey': 21, 'dog': 21, 'car': 21, 'ship': 21, 'bird': 20, 'airplane': 19, 'cat': 19, 'deer': 19, 'truck': 18}\n",
+      "  Linear probe (200 items): {'bird': 21, 'deer': 21, 'dog': 21, 'car': 21, 'ship': 20, 'airplane': 20, 'horse': 20, 'monkey': 19, 'truck': 19, 'cat': 18}\n",
+      "  Retrieval (200 items): {1: 60, 3: 58, 0: 46, 2: 36}\n",
+      "  MCQA (200 items): {'A': 77, 'B': 76, 'C': 29, 'D': 17, 'E': 1}\n"
+     ]
+    }
+   ],
    "source": [
     "# ── Cell 8.3: Sanity check ────────────────────────────────────────────────────\n",
     "print(\"Distribusi prediksi per task:\\n\")\n",
@@ -1161,13 +1959,21 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "Python 3",
+   "display_name": "Python3 (main venv)",
    "language": "python",
-   "name": "python3"
+   "name": "main"
   },
   "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
    "name": "python",
-   "version": "3.10.0"
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.12.13"
   }
  },
  "nbformat": 4,