transfer_simplevae3.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "c15deb04-94a0-4073-a174-adcd22af10b8",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "The config attributes {'block_out_channels': [128, 128, 256, 512, 512], 'force_upcast': False} were passed to AsymmetricAutoencoderKL, but are not expected and will be ignored. Please verify your config.json configuration file.\n",
      "The config attributes {'block_out_channels': [128, 128, 256, 512, 512], 'force_upcast': False} were passed to AsymmetricAutoencoderKL, but are not expected and will be ignored. Please verify your config.json configuration file.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "✅ Создана новая модель: <class 'diffusers.models.autoencoders.autoencoder_asym_kl.AsymmetricAutoencoderKL'>\n",
      "\n",
      "--- Перенос весов ---\n",
      "✅ Готово. Модель сохранена в vae9\n"
     ]
    }
   ],
   "source": [
    "from diffusers.models import AsymmetricAutoencoderKL\n",
    "import torch\n",
    "from tqdm import tqdm\n",
    "\n",
    "# ---- Конфиг новой модели ----\n",
    "config = {\n",
    "    \"_class_name\": \"AsymmetricAutoencoderKL\",\n",
    "    \"act_fn\": \"silu\",\n",
    "    \"in_channels\": 3,\n",
    "    \"out_channels\": 3,\n",
    "    \"scaling_factor\": 1.0,\n",
    "    \"norm_num_groups\": 32,\n",
    "    \"down_block_out_channels\": [128, 256, 512, 512],\n",
    "    \"down_block_types\": [\n",
    "        \"DownEncoderBlock2D\",\n",
    "        \"DownEncoderBlock2D\",\n",
    "        \"DownEncoderBlock2D\",\n",
    "        \"DownEncoderBlock2D\",\n",
    "    ],\n",
    "    \"latent_channels\": 16,\n",
    "    \"up_block_out_channels\": [128,128, 256, 512, 512],\n",
    "    \"up_block_types\": [\n",
    "        \"UpDecoderBlock2D\",\n",
    "        \"UpDecoderBlock2D\",\n",
    "        \"UpDecoderBlock2D\",\n",
    "        \"UpDecoderBlock2D\",\n",
    "        \"UpDecoderBlock2D\",\n",
    "    ],\n",
    "}\n",
    "\n",
    "# ---- Создание пустой асимметричной модели ----\n",
    "vae = AsymmetricAutoencoderKL(\n",
    "    act_fn=config[\"act_fn\"],\n",
    "    down_block_out_channels=config[\"down_block_out_channels\"],\n",
    "    down_block_types=config[\"down_block_types\"],\n",
    "    latent_channels=config[\"latent_channels\"],\n",
    "    up_block_out_channels=config[\"up_block_out_channels\"],\n",
    "    up_block_types=config[\"up_block_types\"],\n",
    "    in_channels=config[\"in_channels\"],\n",
    "    out_channels=config[\"out_channels\"],\n",
    "    scaling_factor=config[\"scaling_factor\"],\n",
    "    norm_num_groups=config[\"norm_num_groups\"],\n",
    "    layers_per_down_block=2,\n",
    "    layers_per_up_block=3,\n",
    "    sample_size=256\n",
    ")\n",
    "\n",
    "vae.save_pretrained(\"asymmetric_vae_empty\")\n",
    "print(\"✅ Создана новая модель:\", type(vae))\n",
    "\n",
    "def transfer_weights(old_path, new_path, save_path=\"vae_final\", device=\"cuda\", dtype=torch.float32):\n",
    "    old_vae = AsymmetricAutoencoderKL.from_pretrained(old_path).to(device, dtype=dtype)\n",
    "    new_vae = AsymmetricAutoencoderKL.from_pretrained(new_path).to(device, dtype=dtype)\n",
    "\n",
    "    old_sd = old_vae.state_dict()\n",
    "    new_sd = new_vae.state_dict()\n",
    "\n",
    "    print(\"\\n--- Перенос весов ---\")\n",
    "    \n",
    "    # 1. Переносим всё, что совпадает по именам и формам (Энкодер, конволоции и т.д.)\n",
    "    # Это покроет Encoder полностью, так как там down_blocks не менялись\n",
    "    for k, v in old_sd.items():\n",
    "        if k in new_sd and v.shape == new_sd[k].shape:\n",
    "            new_sd[k] = v.clone()\n",
    "\n",
    "    # 2. Переносим блоки декодера (Up-blocks)\n",
    "    # Старый: [0, 1, 2, 3, 4] -> Новой: [0, 1, 2, 3]\n",
    "    # Нам нужно перенести глубокие слои (0, 1, 2), а последний 3-й взять из 4-го старого\n",
    "    \n",
    "    # Сопоставление индексов: старый блок -> новый блок\n",
    "    # Мы берем 0->0, 1->1, 2->2 и 4->3 (пропуская лишний блок 128->128)\n",
    "    mapping = {0: 0, 1: 1, 2: 2, 3: 3, 4: 4} \n",
    "    \n",
    "    for old_idx, new_idx in mapping.items():\n",
    "        old_prefix = f\"decoder.up_blocks.{old_idx}.\"\n",
    "        new_prefix = f\"decoder.up_blocks.{new_idx}.\"\n",
    "        \n",
    "        for k, v in old_sd.items():\n",
    "            if k.startswith(old_prefix):\n",
    "                new_key = k.replace(old_prefix, new_prefix)\n",
    "                if new_key in new_sd and v.shape == new_sd[new_key].shape:\n",
    "                    new_sd[new_key] = v.clone()\n",
    "                else:\n",
    "                    print(f\"⚠️ Пропущен слой или не совпал размер: {new_key}\")\n",
    "\n",
    "    # Загрузка\n",
    "    new_vae.load_state_dict(new_sd, strict=False)\n",
    "    new_vae.save_pretrained(save_path)\n",
    "    print(f\"✅ Готово. Модель сохранена в {save_path}\")\n",
    "\n",
    "# Запуск\n",
    "transfer_weights(\"vae8\", \"asymmetric_vae_empty\", save_path=\"vae9\")\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "59fcafb9-6d89-49b4-8362-b4891f591687",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.12.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}