Upload prepare-blip-images.ipynb

prepare-blip-images.ipynb

@@ -0,0 +1,317 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "11397531",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import json\n",
+    "from glob import glob\n",
+    "from tqdm import tqdm\n",
+    "import os\n",
+    "import shutil"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "639a635c",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "macaw-multimodal/synthetic-multi-images-multi-audio-relationship.jsonl\r\n",
+      "macaw-multimodal/synthetic-multi-images-relationship.jsonl\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "!ls macaw-multimodal/synthetic-*images*.jsonl"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "230fda70",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "['macaw-multimodal/synthetic-multi-images-relationship.jsonl',\n",
+       " 'macaw-multimodal/synthetic-multi-images-multi-audio-relationship.jsonl']"
+      ]
+     },
+     "execution_count": 5,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "files = glob('macaw-multimodal/synthetic-*images*.jsonl')\n",
+    "files"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "dbdd6fd6",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!mkdir filtered-blip-images"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "id": "fc2f5947",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "100000it [00:01, 88185.31it/s]\n",
+      "59400it [00:01, 52003.73it/s]\n"
+     ]
+    }
+   ],
+   "source": [
+    "images = []\n",
+    "\n",
+    "for f in files:\n",
+    "    with open(f) as fopen:\n",
+    "        for l in tqdm(fopen):\n",
+    "            l = json.loads(l)\n",
+    "            for f_ in l['filename']:\n",
+    "                if f_.endswith('.jpg'):\n",
+    "                    images.append(f_)\n",
+    "#             break\n",
+    "#             img = l['image']\n",
+    "#             f_img = f'train2014/COCO_train2014_{img}'\n",
+    "#             if not os.path.exists(f_img):\n",
+    "#                 continue\n",
+    "#             shutil.copyfile(f_img, os.path.join('filtered-llava-images', img))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "id": "75ff4044",
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "259400"
+      ]
+     },
+     "execution_count": 10,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "len(images)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "id": "dda18075",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "100%|██████████| 238135/238135 [17:04<00:00, 232.47it/s] \n"
+     ]
+    }
+   ],
+   "source": [
+    "for img in tqdm(list(set(images))):\n",
+    "    f = os.path.join('blip', img)\n",
+    "    folder = f.split('/')[1]\n",
+    "    os.makedirs(os.path.join('filtered-blip-images', folder), exist_ok = True)\n",
+    "    shutil.copyfile(f, os.path.join('filtered-blip-images', img))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "id": "bbc499fe",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "12G\tfiltered-blip-images\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "!du -hs filtered-blip-images"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "id": "29021779",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!~/7zz -v5g a filtered-blip-images.7z filtered-blip-images > /dev/null"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "id": "ac2eb144",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from huggingface_hub import HfApi\n",
+    "api = HfApi()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 23,
+   "id": "34db3d8d",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "filtered-blip-images.7z.003\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "35049f783ad84ea0b66c3c45bd2718d9",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "filtered-blip-images.7z.003:   0%|          | 0.00/747M [00:00<?, ?B/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "filtered-blip-images.7z.001\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "ceca06d30dc442f8a1da0f5ee10c8431",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "filtered-blip-images.7z.001:   0%|          | 0.00/5.37G [00:00<?, ?B/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "filtered-blip-images.7z.002\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "8fbb602e8f494ec680fda22bfafcad2f",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "filtered-blip-images.7z.002:   0%|          | 0.00/5.37G [00:00<?, ?B/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "for f in glob('*.7z.*'):\n",
+    "    print(f)\n",
+    "    api.upload_file(\n",
+    "        path_or_fileobj=f,\n",
+    "        path_in_repo=f,\n",
+    "        repo_id='mesolitica/translated-LLaVA-Pretrain',\n",
+    "        repo_type='dataset',\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 24,
+   "id": "f4916183",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!rm *.7z.*"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 25,
+   "id": "f29eaac3",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "27G\tblip\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "!du -hs blip"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "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": "3.10.12"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}