Delete civit_dataset_to_hf_dataset.ipynb

civit_dataset_to_hf_dataset.ipynb

@@ -1 +0,0 @@
-{"cells":[{"cell_type":"code","source":["# ────────────────────────────────────────────────────────────────\n","#   Convert image (.jpg/.jpeg) + .txt pairs → HuggingFace Dataset\n","#   Files are directly in /content/ (000.jpg + 000.txt, etc.)\n","# ────────────────────────────────────────────────────────────────\n","\n","# 1. Mount Google Drive (to save the dataset)\n","from google.colab import drive\n","drive.mount('/content/drive')\n","\n","# 2. Imports\n","import os\n","import glob\n","from pathlib import Path\n","from PIL import Image\n","from datasets import Dataset, Image as HFImage\n","from tqdm.auto import tqdm\n","\n","# 3. Paths\n","ZIP_PATH    = \"/content/kaggleset.zip\"\n","DRIVE_SAVE  = \"/content/drive/MyDrive/hf_datasets/kaggleset_arrow\"\n","\n","# ─── Unzip if folder doesn't seem to exist yet ──────────────────────\n","if len(glob.glob(\"/content/*.jpg\")) + len(glob.glob(\"/content/*.jpeg\")) < 5:\n","    print(\"Unzipping dataset directly to /content/ ...\")\n","    !unzip -q -o \"{ZIP_PATH}\" -d /content/\n","else:\n","    print(\"Files already appear to be in /content/ → skipping unzip\")\n","\n","# ─── Collect all images (support both .jpg and .jpeg) ───────────────\n","img_patterns = [\"/content/*.jpg\", \"/content/*.jpeg\"]\n","img_files = []\n","for pat in img_patterns:\n","    img_files.extend(glob.glob(pat))\n","\n","img_files = sorted(img_files)   # 000.jpg, 001.jpg, ...\n","\n","print(f\"Found {len(img_files)} images\")\n","\n","# ─── Build list of dicts ────────────────────────────────────────────\n","data = []\n","\n","for img_path in tqdm(img_files, desc=\"Loading image-text pairs\"):\n","    stem = Path(img_path).stem                  # \"000\", \"001\", ...\n","    txt_path = f\"/content/{stem}.txt\"\n","\n","    if not os.path.isfile(txt_path):\n","        print(f\"Missing text file: {txt_path}\")\n","        continue\n","\n","    with open(txt_path, encoding=\"utf-8\") as f:\n","        text = f.read().strip()\n","\n","    data.append({\n","        \"image\": img_path,     # local path → HF Image feature will handle loading\n","        \"text\":  text\n","    })\n","\n","print(f\"\\nSuccessfully prepared {len(data)} examples\")\n","\n","# ─── Create HuggingFace Dataset ─────────────────────────────────────\n","dataset = Dataset.from_list(data)\n","\n","# Tell HF to treat \"image\" column as image files\n","dataset = dataset.cast_column(\"image\", HFImage())\n","\n","print(\"\\nDataset info:\")\n","print(dataset)\n","print(\"\\nFeatures:\")\n","print(dataset.features)\n","\n","# ─── Optional: decode images now (smaller files, uses RAM) ──────────\n","# Uncomment if you prefer to embed images instead of keeping paths\n","\n","# def load_and_encode(ex):\n","#     img = Image.open(ex[\"image\"]).convert(\"RGB\")\n","#     # img = img.resize((512, 512))  # ← optional resize\n","#     ex[\"image\"] = img\n","#     return ex\n","\n","# dataset = dataset.map(load_and_encode, num_proc=2, desc=\"Encoding images\")\n","\n","# ─── Save to Drive in Arrow format ──────────────────────────────────\n","print(f\"\\nSaving to: {DRIVE_SAVE}\")\n","\n","os.makedirs(DRIVE_SAVE, exist_ok=True)\n","dataset.save_to_disk(DRIVE_SAVE)\n","\n","print(\"\\nDone!\")\n","print(\"You can load it later with:\")\n","print(f\"   from datasets import load_from_disk\")\n","print(f\"   ds = load_from_disk('{DRIVE_SAVE}')\")\n","print(\"   ds[0]['image']         # shows image in notebook\")\n","print(\"   print(ds[0]['text'])\")"],"metadata":{"id":"pJCZVks5EG8W"},"execution_count":null,"outputs":[]}],"metadata":{"colab":{"provenance":[{"file_id":"https://huggingface.co/datasets/codeShare/lora-training-data/blob/main/Drive to WebP.ipynb","timestamp":1768857760851},{"file_id":"https://huggingface.co/datasets/codeShare/lora-training-data/blob/main/Drive to WebP.ipynb","timestamp":1763646205520},{"file_id":"https://huggingface.co/datasets/codeShare/lora-training-data/blob/main/Drive to WebP.ipynb","timestamp":1760993725927},{"file_id":"https://huggingface.co/datasets/codeShare/lora-training-data/blob/main/YT-playlist-to-mp3.ipynb","timestamp":1760450712160},{"file_id":"https://huggingface.co/datasets/codeShare/lora-training-data/blob/main/YT-playlist-to-mp3.ipynb","timestamp":1756712618300},{"file_id":"https://huggingface.co/codeShare/JupyterNotebooks/blob/main/YT-playlist-to-mp3.ipynb","timestamp":1747490904984},{"file_id":"https://huggingface.co/codeShare/JupyterNotebooks/blob/main/YT-playlist-to-mp3.ipynb","timestamp":1740037333374},{"file_id":"https://huggingface.co/codeShare/JupyterNotebooks/blob/main/YT-playlist-to-mp3.ipynb","timestamp":1736477078136},{"file_id":"https://huggingface.co/codeShare/JupyterNotebooks/blob/main/YT-playlist-to-mp3.ipynb","timestamp":1725365086834}]},"kernelspec":{"display_name":"Python 3","name":"python3"},"language_info":{"name":"python"}},"nbformat":4,"nbformat_minor":0}