kohya_model_util.py

# mostly from https://github.com/kohya-ss/sd-scripts/blob/main/library/model_util.py
# I am infinitely grateful to @kohya-ss for their amazing work in this field.
# This version is updated to handle the latest version of the diffusers library.
import json
# v1: split from train_db_fixed.py.
# v2: support safetensors

import math
import os
import re

import torch
from transformers import CLIPTextModel, CLIPTokenizer, CLIPTextConfig, logging
from diffusers import AutoencoderKL, DDIMScheduler, StableDiffusionPipeline, UNet2DConditionModel
from safetensors.torch import load_file, save_file
from collections import OrderedDict

# DiffUsers版StableDiffusionのモデルパラメータ
NUM_TRAIN_TIMESTEPS = 1000
BETA_START = 0.00085
BETA_END = 0.0120

UNET_PARAMS_MODEL_CHANNELS = 320
UNET_PARAMS_CHANNEL_MULT = [1, 2, 4, 4]
UNET_PARAMS_ATTENTION_RESOLUTIONS = [4, 2, 1]
UNET_PARAMS_IMAGE_SIZE = 64  # fixed from old invalid value `32`
UNET_PARAMS_IN_CHANNELS = 4
UNET_PARAMS_OUT_CHANNELS = 4
UNET_PARAMS_NUM_RES_BLOCKS = 2
UNET_PARAMS_CONTEXT_DIM = 768
UNET_PARAMS_NUM_HEADS = 8
# UNET_PARAMS_USE_LINEAR_PROJECTION = False

VAE_PARAMS_Z_CHANNELS = 4
VAE_PARAMS_RESOLUTION = 256
VAE_PARAMS_IN_CHANNELS = 3
VAE_PARAMS_OUT_CH = 3
VAE_PARAMS_CH = 128
VAE_PARAMS_CH_MULT = [1, 2, 4, 4]
VAE_PARAMS_NUM_RES_BLOCKS = 2

# V2
V2_UNET_PARAMS_ATTENTION_HEAD_DIM = [5, 10, 20, 20]
V2_UNET_PARAMS_CONTEXT_DIM = 1024
# V2_UNET_PARAMS_USE_LINEAR_PROJECTION = True

# Diffusersの設定を読み込むための参照モデル
DIFFUSERS_REF_MODEL_ID_V1 = "runwayml/stable-diffusion-v1-5"
DIFFUSERS_REF_MODEL_ID_V2 = "stabilityai/stable-diffusion-2-1"


# region StableDiffusion->Diffusersの変換コード
# convert_original_stable_diffusion_to_diffusers をコピーして修正している（ASL 2.0）


def shave_segments(path, n_shave_prefix_segments=1):
    """
    Removes segments. Positive values shave the first segments, negative shave the last segments.
    """
    if n_shave_prefix_segments >= 0:
        return ".".join(path.split(".")[n_shave_prefix_segments:])
    else:
        return ".".join(path.split(".")[:n_shave_prefix_segments])


def renew_resnet_paths(old_list, n_shave_prefix_segments=0):
    """
    Updates paths inside resnets to the new naming scheme (local renaming)
    """
    mapping = []
    for old_item in old_list:
        new_item = old_item.replace("in_layers.0", "norm1")
        new_item = new_item.replace("in_layers.2", "conv1")

        new_item = new_item.replace("out_layers.0", "norm2")
        new_item = new_item.replace("out_layers.3", "conv2")

        new_item = new_item.replace("emb_layers.1", "time_emb_proj")
        new_item = new_item.replace("skip_connection", "conv_shortcut")

        new_item = shave_segments(new_item, n_shave_prefix_segments=n_shave_prefix_segments)

        mapping.append({"old": old_item, "new": new_item})

    return mapping


def renew_vae_resnet_paths(old_list, n_shave_prefix_segments=0):
    """
    Updates paths inside resnets to the new naming scheme (local renaming)
    """
    mapping = []
    for old_item in old_list:
        new_item = old_item

        new_item = new_item.replace("nin_shortcut", "conv_shortcut")
        new_item = shave_segments(new_item, n_shave_prefix_segments=n_shave_prefix_segments)

        mapping.append({"old": old_item, "new": new_item})

    return mapping


def renew_attention_paths(old_list, n_shave_prefix_segments=0):
    """
    Updates paths inside attentions to the new naming scheme (local renaming)
    """
    mapping = []
    for old_item in old_list:
        new_item = old_item

        #         new_item = new_item.replace('norm.weight', 'group_norm.weight')
        #         new_item = new_item.replace('norm.bias', 'group_norm.bias')

        #         new_item = new_item.replace('proj_out.weight', 'proj_attn.weight')
        #         new_item = new_item.replace('proj_out.bias', 'proj_attn.bias')

        #         new_item = shave_segments(new_item, n_shave_prefix_segments=n_shave_prefix_segments)

        mapping.append({"old": old_item, "new": new_item})

    return mapping


def renew_vae_attention_paths(old_list, n_shave_prefix_segments=0):
    """
    Updates paths inside attentions to the new naming scheme (local renaming)
    """
    mapping = []
    for old_item in old_list:
        new_item = old_item

        # updated for latest diffusers
        new_item = new_item.replace("norm.weight", "group_norm.weight")
        new_item = new_item.replace("norm.bias", "group_norm.bias")

        new_item = new_item.replace("q.weight", "to_q.weight")
        new_item = new_item.replace("q.bias", "to_q.bias")

        new_item = new_item.replace("k.weight", "to_k.weight")
        new_item = new_item.replace("k.bias", "to_k.bias")

        new_item = new_item.replace("v.weight", "to_v.weight")
        new_item = new_item.replace("v.bias", "to_v.bias")

        new_item = new_item.replace("proj_out.weight", "to_out.0.weight")
        new_item = new_item.replace("proj_out.bias", "to_out.0.bias")

        new_item = shave_segments(new_item, n_shave_prefix_segments=n_shave_prefix_segments)

        mapping.append({"old": old_item, "new": new_item})

    return mapping


def assign_to_checkpoint(
        paths, checkpoint, old_checkpoint, attention_paths_to_split=None, additional_replacements=None, config=None
):
    """
    This does the final conversion step: take locally converted weights and apply a global renaming
    to them. It splits attention layers, and takes into account additional replacements
    that may arise.

    Assigns the weights to the new checkpoint.
    """
    assert isinstance(paths, list), "Paths should be a list of dicts containing 'old' and 'new' keys."

    # Splits the attention layers into three variables.
    if attention_paths_to_split is not None:
        for path, path_map in attention_paths_to_split.items():
            old_tensor = old_checkpoint[path]
            channels = old_tensor.shape[0] // 3

            target_shape = (-1, channels) if len(old_tensor.shape) == 3 else (-1)

            num_heads = old_tensor.shape[0] // config["num_head_channels"] // 3

            old_tensor = old_tensor.reshape((num_heads, 3 * channels // num_heads) + old_tensor.shape[1:])
            query, key, value = old_tensor.split(channels // num_heads, dim=1)

            checkpoint[path_map["query"]] = query.reshape(target_shape)
            checkpoint[path_map["key"]] = key.reshape(target_shape)
            checkpoint[path_map["value"]] = value.reshape(target_shape)

    for path in paths:
        new_path = path["new"]

        # These have already been assigned
        if attention_paths_to_split is not None and new_path in attention_paths_to_split:
            continue

        # Global renaming happens here
        new_path = new_path.replace("middle_block.0", "mid_block.resnets.0")
        new_path = new_path.replace("middle_block.1", "mid_block.attentions.0")
        new_path = new_path.replace("middle_block.2", "mid_block.resnets.1")

        if additional_replacements is not None:
            for replacement in additional_replacements:
                new_path = new_path.replace(replacement["old"], replacement["new"])

        # proj_attn.weight has to be converted from conv 1D to linear
        is_attn_weight = "proj_attn.weight" in new_path or ("attentions" in new_path and "to_" in new_path)
        shape = old_checkpoint[path["old"]].shape
        if is_attn_weight and len(shape) == 3:
            checkpoint[new_path] = old_checkpoint[path["old"]][:, :, 0]
        elif is_attn_weight and len(shape) == 4:
            checkpoint[new_path] = old_checkpoint[path["old"]][:, :, 0, 0]
        else:
            checkpoint[new_path] = old_checkpoint[path["old"]]


def conv_attn_to_linear(checkpoint):
    keys = list(checkpoint.keys())
    attn_keys = ["query.weight", "key.weight", "value.weight"]
    for key in keys:
        if ".".join(key.split(".")[-2:]) in attn_keys:
            if checkpoint[key].ndim > 2:
                checkpoint[key] = checkpoint[key][:, :, 0, 0]
        elif "proj_attn.weight" in key:
            if checkpoint[key].ndim > 2:
                checkpoint[key] = checkpoint[key][:, :, 0]


def linear_transformer_to_conv(checkpoint):
    keys = list(checkpoint.keys())
    tf_keys = ["proj_in.weight", "proj_out.weight"]
    for key in keys:
        if ".".join(key.split(".")[-2:]) in tf_keys:
            if checkpoint[key].ndim == 2:
                checkpoint[key] = checkpoint[key].unsqueeze(2).unsqueeze(2)


def convert_ldm_unet_checkpoint(v2, checkpoint, config):
    mapping = {}
    """
    Takes a state dict and a config, and returns a converted checkpoint.
    """

    # extract state_dict for UNet
    unet_state_dict = {}
    unet_key = "model.diffusion_model."
    keys = list(checkpoint.keys())
    for key in keys:
        if key.startswith(unet_key):
            unet_state_dict[key.replace(unet_key, "")] = checkpoint.pop(key)

    new_checkpoint = {}

    new_checkpoint["time_embedding.linear_1.weight"] = unet_state_dict["time_embed.0.weight"]
    new_checkpoint["time_embedding.linear_1.bias"] = unet_state_dict["time_embed.0.bias"]
    new_checkpoint["time_embedding.linear_2.weight"] = unet_state_dict["time_embed.2.weight"]
    new_checkpoint["time_embedding.linear_2.bias"] = unet_state_dict["time_embed.2.bias"]

    new_checkpoint["conv_in.weight"] = unet_state_dict["input_blocks.0.0.weight"]
    new_checkpoint["conv_in.bias"] = unet_state_dict["input_blocks.0.0.bias"]

    new_checkpoint["conv_norm_out.weight"] = unet_state_dict["out.0.weight"]
    new_checkpoint["conv_norm_out.bias"] = unet_state_dict["out.0.bias"]
    new_checkpoint["conv_out.weight"] = unet_state_dict["out.2.weight"]
    new_checkpoint["conv_out.bias"] = unet_state_dict["out.2.bias"]

    # Retrieves the keys for the input blocks only
    num_input_blocks = len({".".join(layer.split(".")[:2]) for layer in unet_state_dict if "input_blocks" in layer})
    input_blocks = {
        layer_id: [key for key in unet_state_dict if f"input_blocks.{layer_id}." in key] for layer_id in
        range(num_input_blocks)
    }

    # Retrieves the keys for the middle blocks only
    num_middle_blocks = len({".".join(layer.split(".")[:2]) for layer in unet_state_dict if "middle_block" in layer})
    middle_blocks = {
        layer_id: [key for key in unet_state_dict if f"middle_block.{layer_id}." in key] for layer_id in
        range(num_middle_blocks)
    }

    # Retrieves the keys for the output blocks only
    num_output_blocks = len({".".join(layer.split(".")[:2]) for layer in unet_state_dict if "output_blocks" in layer})
    output_blocks = {
        layer_id: [key for key in unet_state_dict if f"output_blocks.{layer_id}." in key] for layer_id in
        range(num_output_blocks)
    }

    for i in range(1, num_input_blocks):
        block_id = (i - 1) // (config["layers_per_block"] + 1)
        layer_in_block_id = (i - 1) % (config["layers_per_block"] + 1)

        resnets = [key for key in input_blocks[i] if
                   f"input_blocks.{i}.0" in key and f"input_blocks.{i}.0.op" not in key]
        attentions = [key for key in input_blocks[i] if f"input_blocks.{i}.1" in key]

        if f"input_blocks.{i}.0.op.weight" in unet_state_dict:
            new_checkpoint[f"down_blocks.{block_id}.downsamplers.0.conv.weight"] = unet_state_dict.pop(
                f"input_blocks.{i}.0.op.weight"
            )
            mapping[f'input_blocks.{i}.0.op.weight'] = f"down_blocks.{block_id}.downsamplers.0.conv.weight"
            new_checkpoint[f"down_blocks.{block_id}.downsamplers.0.conv.bias"] = unet_state_dict.pop(
                f"input_blocks.{i}.0.op.bias")
            mapping[f'input_blocks.{i}.0.op.bias'] = f"down_blocks.{block_id}.downsamplers.0.conv.bias"

        paths = renew_resnet_paths(resnets)
        meta_path = {"old": f"input_blocks.{i}.0", "new": f"down_blocks.{block_id}.resnets.{layer_in_block_id}"}
        assign_to_checkpoint(paths, new_checkpoint, unet_state_dict, additional_replacements=[meta_path], config=config)

        if len(attentions):
            paths = renew_attention_paths(attentions)
            meta_path = {"old": f"input_blocks.{i}.1", "new": f"down_blocks.{block_id}.attentions.{layer_in_block_id}"}
            assign_to_checkpoint(paths, new_checkpoint, unet_state_dict, additional_replacements=[meta_path],
                                 config=config)

    resnet_0 = middle_blocks[0]
    attentions = middle_blocks[1]
    resnet_1 = middle_blocks[2]

    resnet_0_paths = renew_resnet_paths(resnet_0)
    assign_to_checkpoint(resnet_0_paths, new_checkpoint, unet_state_dict, config=config)

    resnet_1_paths = renew_resnet_paths(resnet_1)
    assign_to_checkpoint(resnet_1_paths, new_checkpoint, unet_state_dict, config=config)

    attentions_paths = renew_attention_paths(attentions)
    meta_path = {"old": "middle_block.1", "new": "mid_block.attentions.0"}
    assign_to_checkpoint(attentions_paths, new_checkpoint, unet_state_dict, additional_replacements=[meta_path],
                         config=config)

    for i in range(num_output_blocks):
        block_id = i // (config["layers_per_block"] + 1)
        layer_in_block_id = i % (config["layers_per_block"] + 1)
        output_block_layers = [shave_segments(name, 2) for name in output_blocks[i]]
        output_block_list = {}

        for layer in output_block_layers:
            layer_id, layer_name = layer.split(".")[0], shave_segments(layer, 1)
            if layer_id in output_block_list:
                output_block_list[layer_id].append(layer_name)
            else:
                output_block_list[layer_id] = [layer_name]

        if len(output_block_list) > 1:
            resnets = [key for key in output_blocks[i] if f"output_blocks.{i}.0" in key]
            attentions = [key for key in output_blocks[i] if f"output_blocks.{i}.1" in key]

            resnet_0_paths = renew_resnet_paths(resnets)
            paths = renew_resnet_paths(resnets)

            meta_path = {"old": f"output_blocks.{i}.0", "new": f"up_blocks.{block_id}.resnets.{layer_in_block_id}"}
            assign_to_checkpoint(paths, new_checkpoint, unet_state_dict, additional_replacements=[meta_path],
                                 config=config)

            # オリジナル：
            # if ["conv.weight", "conv.bias"] in output_block_list.values():
            #   index = list(output_block_list.values()).index(["conv.weight", "conv.bias"])

            # biasとweightの順番に依存しないようにする：もっといいやり方がありそうだが
            for l in output_block_list.values():
                l.sort()

            if ["conv.bias", "conv.weight"] in output_block_list.values():
                index = list(output_block_list.values()).index(["conv.bias", "conv.weight"])
                new_checkpoint[f"up_blocks.{block_id}.upsamplers.0.conv.bias"] = unet_state_dict[
                    f"output_blocks.{i}.{index}.conv.bias"
                ]
                new_checkpoint[f"up_blocks.{block_id}.upsamplers.0.conv.weight"] = unet_state_dict[
                    f"output_blocks.{i}.{index}.conv.weight"
                ]

                # Clear attentions as they have been attributed above.
                if len(attentions) == 2:
                    attentions = []

            if len(attentions):
                paths = renew_attention_paths(attentions)
                meta_path = {
                    "old": f"output_blocks.{i}.1",
                    "new": f"up_blocks.{block_id}.attentions.{layer_in_block_id}",
                }
                assign_to_checkpoint(paths, new_checkpoint, unet_state_dict, additional_replacements=[meta_path],
                                     config=config)
        else:
            resnet_0_paths = renew_resnet_paths(output_block_layers, n_shave_prefix_segments=1)
            for path in resnet_0_paths:
                old_path = ".".join(["output_blocks", str(i), path["old"]])
                new_path = ".".join(["up_blocks", str(block_id), "resnets", str(layer_in_block_id), path["new"]])

                new_checkpoint[new_path] = unet_state_dict[old_path]

    # SDのv2では1*1のconv2dがlinearに変わっている
    # 誤って Diffusers 側を conv2d のままにしてしまったので、変換必要
    if v2 and not config.get('use_linear_projection', False):
        linear_transformer_to_conv(new_checkpoint)

    # print("mapping: ", json.dumps(mapping, indent=4))
    return new_checkpoint


# ldm key: diffusers key
vae_ldm_to_diffusers_dict = {
    "decoder.conv_in.bias": "decoder.conv_in.bias",
    "decoder.conv_in.weight": "decoder.conv_in.weight",
    "decoder.conv_out.bias": "decoder.conv_out.bias",
    "decoder.conv_out.weight": "decoder.conv_out.weight",
    "decoder.mid.attn_1.k.bias": "decoder.mid_block.attentions.0.to_k.bias",
    "decoder.mid.attn_1.k.weight": "decoder.mid_block.attentions.0.to_k.weight",
    "decoder.mid.attn_1.norm.bias": "decoder.mid_block.attentions.0.group_norm.bias",
    "decoder.mid.attn_1.norm.weight": "decoder.mid_block.attentions.0.group_norm.weight",
    "decoder.mid.attn_1.proj_out.bias": "decoder.mid_block.attentions.0.to_out.0.bias",
    "decoder.mid.attn_1.proj_out.weight": "decoder.mid_block.attentions.0.to_out.0.weight",
    "decoder.mid.attn_1.q.bias": "decoder.mid_block.attentions.0.to_q.bias",
    "decoder.mid.attn_1.q.weight": "decoder.mid_block.attentions.0.to_q.weight",
    "decoder.mid.attn_1.v.bias": "decoder.mid_block.attentions.0.to_v.bias",
    "decoder.mid.attn_1.v.weight": "decoder.mid_block.attentions.0.to_v.weight",
    "decoder.mid.block_1.conv1.bias": "decoder.mid_block.resnets.0.conv1.bias",
    "decoder.mid.block_1.conv1.weight": "decoder.mid_block.resnets.0.conv1.weight",
    "decoder.mid.block_1.conv2.bias": "decoder.mid_block.resnets.0.conv2.bias",
    "decoder.mid.block_1.conv2.weight": "decoder.mid_block.resnets.0.conv2.weight",
    "decoder.mid.block_1.norm1.bias": "decoder.mid_block.resnets.0.norm1.bias",
    "decoder.mid.block_1.norm1.weight": "decoder.mid_block.resnets.0.norm1.weight",
    "decoder.mid.block_1.norm2.bias": "decoder.mid_block.resnets.0.norm2.bias",
    "decoder.mid.block_1.norm2.weight": "decoder.mid_block.resnets.0.norm2.weight",
    "decoder.mid.block_2.conv1.bias": "decoder.mid_block.resnets.1.conv1.bias",
    "decoder.mid.block_2.conv1.weight": "decoder.mid_block.resnets.1.conv1.weight",
    "decoder.mid.block_2.conv2.bias": "decoder.mid_block.resnets.1.conv2.bias",
    "decoder.mid.block_2.conv2.weight": "decoder.mid_block.resnets.1.conv2.weight",
    "decoder.mid.block_2.norm1.bias": "decoder.mid_block.resnets.1.norm1.bias",
    "decoder.mid.block_2.norm1.weight": "decoder.mid_block.resnets.1.norm1.weight",
    "decoder.mid.block_2.norm2.bias": "decoder.mid_block.resnets.1.norm2.bias",
    "decoder.mid.block_2.norm2.weight": "decoder.mid_block.resnets.1.norm2.weight",
    "decoder.norm_out.bias": "decoder.conv_norm_out.bias",
    "decoder.norm_out.weight": "decoder.conv_norm_out.weight",
    "decoder.up.0.block.0.conv1.bias": "decoder.up_blocks.3.resnets.0.conv1.bias",
    "decoder.up.0.block.0.conv1.weight": "decoder.up_blocks.3.resnets.0.conv1.weight",
    "decoder.up.0.block.0.conv2.bias": "decoder.up_blocks.3.resnets.0.conv2.bias",
    "decoder.up.0.block.0.conv2.weight": "decoder.up_blocks.3.resnets.0.conv2.weight",
    "decoder.up.0.block.0.nin_shortcut.bias": "decoder.up_blocks.3.resnets.0.conv_shortcut.bias",
    "decoder.up.0.block.0.nin_shortcut.weight": "decoder.up_blocks.3.resnets.0.conv_shortcut.weight",
    "decoder.up.0.block.0.norm1.bias": "decoder.up_blocks.3.resnets.0.norm1.bias",
    "decoder.up.0.block.0.norm1.weight": "decoder.up_blocks.3.resnets.0.norm1.weight",
    "decoder.up.0.block.0.norm2.bias": "decoder.up_blocks.3.resnets.0.norm2.bias",
    "decoder.up.0.block.0.norm2.weight": "decoder.up_blocks.3.resnets.0.norm2.weight",
    "decoder.up.0.block.1.conv1.bias": "decoder.up_blocks.3.resnets.1.conv1.bias",
    "decoder.up.0.block.1.conv1.weight": "decoder.up_blocks.3.resnets.1.conv1.weight",
    "decoder.up.0.block.1.conv2.bias": "decoder.up_blocks.3.resnets.1.conv2.bias",
    "decoder.up.0.block.1.conv2.weight": "decoder.up_blocks.3.resnets.1.conv2.weight",
    "decoder.up.0.block.1.norm1.bias": "decoder.up_blocks.3.resnets.1.norm1.bias",
    "decoder.up.0.block.1.norm1.weight": "decoder.up_blocks.3.resnets.1.norm1.weight",
    "decoder.up.0.block.1.norm2.bias": "decoder.up_blocks.3.resnets.1.norm2.bias",
    "decoder.up.0.block.1.norm2.weight": "decoder.up_blocks.3.resnets.1.norm2.weight",
    "decoder.up.0.block.2.conv1.bias": "decoder.up_blocks.3.resnets.2.conv1.bias",
    "decoder.up.0.block.2.conv1.weight": "decoder.up_blocks.3.resnets.2.conv1.weight",
    "decoder.up.0.block.2.conv2.bias": "decoder.up_blocks.3.resnets.2.conv2.bias",
    "decoder.up.0.block.2.conv2.weight": "decoder.up_blocks.3.resnets.2.conv2.weight",
    "decoder.up.0.block.2.norm1.bias": "decoder.up_blocks.3.resnets.2.norm1.bias",
    "decoder.up.0.block.2.norm1.weight": "decoder.up_blocks.3.resnets.2.norm1.weight",
    "decoder.up.0.block.2.norm2.bias": "decoder.up_blocks.3.resnets.2.norm2.bias",
    "decoder.up.0.block.2.norm2.weight": "decoder.up_blocks.3.resnets.2.norm2.weight",
    "decoder.up.1.block.0.conv1.bias": "decoder.up_blocks.2.resnets.0.conv1.bias",
    "decoder.up.1.block.0.conv1.weight": "decoder.up_blocks.2.resnets.0.conv1.weight",
    "decoder.up.1.block.0.conv2.bias": "decoder.up_blocks.2.resnets.0.conv2.bias",
    "decoder.up.1.block.0.conv2.weight": "decoder.up_blocks.2.resnets.0.conv2.weight",
    "decoder.up.1.block.0.nin_shortcut.bias": "decoder.up_blocks.2.resnets.0.conv_shortcut.bias",
    "decoder.up.1.block.0.nin_shortcut.weight": "decoder.up_blocks.2.resnets.0.conv_shortcut.weight",
    "decoder.up.1.block.0.norm1.bias": "decoder.up_blocks.2.resnets.0.norm1.bias",
    "decoder.up.1.block.0.norm1.weight": "decoder.up_blocks.2.resnets.0.norm1.weight",
    "decoder.up.1.block.0.norm2.bias": "decoder.up_blocks.2.resnets.0.norm2.bias",
    "decoder.up.1.block.0.norm2.weight": "decoder.up_blocks.2.resnets.0.norm2.weight",
    "decoder.up.1.block.1.conv1.bias": "decoder.up_blocks.2.resnets.1.conv1.bias",
    "decoder.up.1.block.1.conv1.weight": "decoder.up_blocks.2.resnets.1.conv1.weight",
    "decoder.up.1.block.1.conv2.bias": "decoder.up_blocks.2.resnets.1.conv2.bias",
    "decoder.up.1.block.1.conv2.weight": "decoder.up_blocks.2.resnets.1.conv2.weight",
    "decoder.up.1.block.1.norm1.bias": "decoder.up_blocks.2.resnets.1.norm1.bias",
    "decoder.up.1.block.1.norm1.weight": "decoder.up_blocks.2.resnets.1.norm1.weight",
    "decoder.up.1.block.1.norm2.bias": "decoder.up_blocks.2.resnets.1.norm2.bias",
    "decoder.up.1.block.1.norm2.weight": "decoder.up_blocks.2.resnets.1.norm2.weight",
    "decoder.up.1.block.2.conv1.bias": "decoder.up_blocks.2.resnets.2.conv1.bias",
    "decoder.up.1.block.2.conv1.weight": "decoder.up_blocks.2.resnets.2.conv1.weight",
    "decoder.up.1.block.2.conv2.bias": "decoder.up_blocks.2.resnets.2.conv2.bias",
    "decoder.up.1.block.2.conv2.weight": "decoder.up_blocks.2.resnets.2.conv2.weight",
    "decoder.up.1.block.2.norm1.bias": "decoder.up_blocks.2.resnets.2.norm1.bias",
    "decoder.up.1.block.2.norm1.weight": "decoder.up_blocks.2.resnets.2.norm1.weight",
    "decoder.up.1.block.2.norm2.bias": "decoder.up_blocks.2.resnets.2.norm2.bias",
    "decoder.up.1.block.2.norm2.weight": "decoder.up_blocks.2.resnets.2.norm2.weight",
    "decoder.up.1.upsample.conv.bias": "decoder.up_blocks.2.upsamplers.0.conv.bias",
    "decoder.up.1.upsample.conv.weight": "decoder.up_blocks.2.upsamplers.0.conv.weight",
    "decoder.up.2.block.0.conv1.bias": "decoder.up_blocks.1.resnets.0.conv1.bias",
    "decoder.up.2.block.0.conv1.weight": "decoder.up_blocks.1.resnets.0.conv1.weight",
    "decoder.up.2.block.0.conv2.bias": "decoder.up_blocks.1.resnets.0.conv2.bias",
    "decoder.up.2.block.0.conv2.weight": "decoder.up_blocks.1.resnets.0.conv2.weight",
    "decoder.up.2.block.0.norm1.bias": "decoder.up_blocks.1.resnets.0.norm1.bias",
    "decoder.up.2.block.0.norm1.weight": "decoder.up_blocks.1.resnets.0.norm1.weight",
    "decoder.up.2.block.0.norm2.bias": "decoder.up_blocks.1.resnets.0.norm2.bias",
    "decoder.up.2.block.0.norm2.weight": "decoder.up_blocks.1.resnets.0.norm2.weight",
    "decoder.up.2.block.1.conv1.bias": "decoder.up_blocks.1.resnets.1.conv1.bias",
    "decoder.up.2.block.1.conv1.weight": "decoder.up_blocks.1.resnets.1.conv1.weight",
    "decoder.up.2.block.1.conv2.bias": "decoder.up_blocks.1.resnets.1.conv2.bias",
    "decoder.up.2.block.1.conv2.weight": "decoder.up_blocks.1.resnets.1.conv2.weight",
    "decoder.up.2.block.1.norm1.bias": "decoder.up_blocks.1.resnets.1.norm1.bias",
    "decoder.up.2.block.1.norm1.weight": "decoder.up_blocks.1.resnets.1.norm1.weight",
    "decoder.up.2.block.1.norm2.bias": "decoder.up_blocks.1.resnets.1.norm2.bias",
    "decoder.up.2.block.1.norm2.weight": "decoder.up_blocks.1.resnets.1.norm2.weight",
    "decoder.up.2.block.2.conv1.bias": "decoder.up_blocks.1.resnets.2.conv1.bias",
    "decoder.up.2.block.2.conv1.weight": "decoder.up_blocks.1.resnets.2.conv1.weight",
    "decoder.up.2.block.2.conv2.bias": "decoder.up_blocks.1.resnets.2.conv2.bias",
    "decoder.up.2.block.2.conv2.weight": "decoder.up_blocks.1.resnets.2.conv2.weight",
    "decoder.up.2.block.2.norm1.bias": "decoder.up_blocks.1.resnets.2.norm1.bias",
    "decoder.up.2.block.2.norm1.weight": "decoder.up_blocks.1.resnets.2.norm1.weight",
    "decoder.up.2.block.2.norm2.bias": "decoder.up_blocks.1.resnets.2.norm2.bias",
    "decoder.up.2.block.2.norm2.weight": "decoder.up_blocks.1.resnets.2.norm2.weight",
    "decoder.up.2.upsample.conv.bias": "decoder.up_blocks.1.upsamplers.0.conv.bias",
    "decoder.up.2.upsample.conv.weight": "decoder.up_blocks.1.upsamplers.0.conv.weight",
    "decoder.up.3.block.0.conv1.bias": "decoder.up_blocks.0.resnets.0.conv1.bias",
    "decoder.up.3.block.0.conv1.weight": "decoder.up_blocks.0.resnets.0.conv1.weight",
    "decoder.up.3.block.0.conv2.bias": "decoder.up_blocks.0.resnets.0.conv2.bias",
    "decoder.up.3.block.0.conv2.weight": "decoder.up_blocks.0.resnets.0.conv2.weight",
    "decoder.up.3.block.0.norm1.bias": "decoder.up_blocks.0.resnets.0.norm1.bias",
    "decoder.up.3.block.0.norm1.weight": "decoder.up_blocks.0.resnets.0.norm1.weight",
    "decoder.up.3.block.0.norm2.bias": "decoder.up_blocks.0.resnets.0.norm2.bias",
    "decoder.up.3.block.0.norm2.weight": "decoder.up_blocks.0.resnets.0.norm2.weight",
    "decoder.up.3.block.1.conv1.bias": "decoder.up_blocks.0.resnets.1.conv1.bias",
    "decoder.up.3.block.1.conv1.weight": "decoder.up_blocks.0.resnets.1.conv1.weight",
    "decoder.up.3.block.1.conv2.bias": "decoder.up_blocks.0.resnets.1.conv2.bias",
    "decoder.up.3.block.1.conv2.weight": "decoder.up_blocks.0.resnets.1.conv2.weight",
    "decoder.up.3.block.1.norm1.bias": "decoder.up_blocks.0.resnets.1.norm1.bias",
    "decoder.up.3.block.1.norm1.weight": "decoder.up_blocks.0.resnets.1.norm1.weight",
    "decoder.up.3.block.1.norm2.bias": "decoder.up_blocks.0.resnets.1.norm2.bias",
    "decoder.up.3.block.1.norm2.weight": "decoder.up_blocks.0.resnets.1.norm2.weight",
    "decoder.up.3.block.2.conv1.bias": "decoder.up_blocks.0.resnets.2.conv1.bias",
    "decoder.up.3.block.2.conv1.weight": "decoder.up_blocks.0.resnets.2.conv1.weight",
    "decoder.up.3.block.2.conv2.bias": "decoder.up_blocks.0.resnets.2.conv2.bias",
    "decoder.up.3.block.2.conv2.weight": "decoder.up_blocks.0.resnets.2.conv2.weight",
    "decoder.up.3.block.2.norm1.bias": "decoder.up_blocks.0.resnets.2.norm1.bias",
    "decoder.up.3.block.2.norm1.weight": "decoder.up_blocks.0.resnets.2.norm1.weight",
    "decoder.up.3.block.2.norm2.bias": "decoder.up_blocks.0.resnets.2.norm2.bias",
    "decoder.up.3.block.2.norm2.weight": "decoder.up_blocks.0.resnets.2.norm2.weight",
    "decoder.up.3.upsample.conv.bias": "decoder.up_blocks.0.upsamplers.0.conv.bias",
    "decoder.up.3.upsample.conv.weight": "decoder.up_blocks.0.upsamplers.0.conv.weight",
    "encoder.conv_in.bias": "encoder.conv_in.bias",
    "encoder.conv_in.weight": "encoder.conv_in.weight",
    "encoder.conv_out.bias": "encoder.conv_out.bias",
    "encoder.conv_out.weight": "encoder.conv_out.weight",
    "encoder.down.0.block.0.conv1.bias": "encoder.down_blocks.0.resnets.0.conv1.bias",
    "encoder.down.0.block.0.conv1.weight": "encoder.down_blocks.0.resnets.0.conv1.weight",
    "encoder.down.0.block.0.conv2.bias": "encoder.down_blocks.0.resnets.0.conv2.bias",
    "encoder.down.0.block.0.conv2.weight": "encoder.down_blocks.0.resnets.0.conv2.weight",
    "encoder.down.0.block.0.norm1.bias": "encoder.down_blocks.0.resnets.0.norm1.bias",
    "encoder.down.0.block.0.norm1.weight": "encoder.down_blocks.0.resnets.0.norm1.weight",
    "encoder.down.0.block.0.norm2.bias": "encoder.down_blocks.0.resnets.0.norm2.bias",
    "encoder.down.0.block.0.norm2.weight": "encoder.down_blocks.0.resnets.0.norm2.weight",
    "encoder.down.0.block.1.conv1.bias": "encoder.down_blocks.0.resnets.1.conv1.bias",
    "encoder.down.0.block.1.conv1.weight": "encoder.down_blocks.0.resnets.1.conv1.weight",
    "encoder.down.0.block.1.conv2.bias": "encoder.down_blocks.0.resnets.1.conv2.bias",
    "encoder.down.0.block.1.conv2.weight": "encoder.down_blocks.0.resnets.1.conv2.weight",
    "encoder.down.0.block.1.norm1.bias": "encoder.down_blocks.0.resnets.1.norm1.bias",
    "encoder.down.0.block.1.norm1.weight": "encoder.down_blocks.0.resnets.1.norm1.weight",
    "encoder.down.0.block.1.norm2.bias": "encoder.down_blocks.0.resnets.1.norm2.bias",
    "encoder.down.0.block.1.norm2.weight": "encoder.down_blocks.0.resnets.1.norm2.weight",
    "encoder.down.0.downsample.conv.bias": "encoder.down_blocks.0.downsamplers.0.conv.bias",
    "encoder.down.0.downsample.conv.weight": "encoder.down_blocks.0.downsamplers.0.conv.weight",
    "encoder.down.1.block.0.conv1.bias": "encoder.down_blocks.1.resnets.0.conv1.bias",
    "encoder.down.1.block.0.conv1.weight": "encoder.down_blocks.1.resnets.0.conv1.weight",
    "encoder.down.1.block.0.conv2.bias": "encoder.down_blocks.1.resnets.0.conv2.bias",
    "encoder.down.1.block.0.conv2.weight": "encoder.down_blocks.1.resnets.0.conv2.weight",
    "encoder.down.1.block.0.nin_shortcut.bias": "encoder.down_blocks.1.resnets.0.conv_shortcut.bias",
    "encoder.down.1.block.0.nin_shortcut.weight": "encoder.down_blocks.1.resnets.0.conv_shortcut.weight",
    "encoder.down.1.block.0.norm1.bias": "encoder.down_blocks.1.resnets.0.norm1.bias",
    "encoder.down.1.block.0.norm1.weight": "encoder.down_blocks.1.resnets.0.norm1.weight",
    "encoder.down.1.block.0.norm2.bias": "encoder.down_blocks.1.resnets.0.norm2.bias",
    "encoder.down.1.block.0.norm2.weight": "encoder.down_blocks.1.resnets.0.norm2.weight",
    "encoder.down.1.block.1.conv1.bias": "encoder.down_blocks.1.resnets.1.conv1.bias",
    "encoder.down.1.block.1.conv1.weight": "encoder.down_blocks.1.resnets.1.conv1.weight",
    "encoder.down.1.block.1.conv2.bias": "encoder.down_blocks.1.resnets.1.conv2.bias",
    "encoder.down.1.block.1.conv2.weight": "encoder.down_blocks.1.resnets.1.conv2.weight",
    "encoder.down.1.block.1.norm1.bias": "encoder.down_blocks.1.resnets.1.norm1.bias",
    "encoder.down.1.block.1.norm1.weight": "encoder.down_blocks.1.resnets.1.norm1.weight",
    "encoder.down.1.block.1.norm2.bias": "encoder.down_blocks.1.resnets.1.norm2.bias",
    "encoder.down.1.block.1.norm2.weight": "encoder.down_blocks.1.resnets.1.norm2.weight",
    "encoder.down.1.downsample.conv.bias": "encoder.down_blocks.1.downsamplers.0.conv.bias",
    "encoder.down.1.downsample.conv.weight": "encoder.down_blocks.1.downsamplers.0.conv.weight",
    "encoder.down.2.block.0.conv1.bias": "encoder.down_blocks.2.resnets.0.conv1.bias",
    "encoder.down.2.block.0.conv1.weight": "encoder.down_blocks.2.resnets.0.conv1.weight",
    "encoder.down.2.block.0.conv2.bias": "encoder.down_blocks.2.resnets.0.conv2.bias",
    "encoder.down.2.block.0.conv2.weight": "encoder.down_blocks.2.resnets.0.conv2.weight",
    "encoder.down.2.block.0.nin_shortcut.bias": "encoder.down_blocks.2.resnets.0.conv_shortcut.bias",
    "encoder.down.2.block.0.nin_shortcut.weight": "encoder.down_blocks.2.resnets.0.conv_shortcut.weight",
    "encoder.down.2.block.0.norm1.bias": "encoder.down_blocks.2.resnets.0.norm1.bias",
    "encoder.down.2.block.0.norm1.weight": "encoder.down_blocks.2.resnets.0.norm1.weight",
    "encoder.down.2.block.0.norm2.bias": "encoder.down_blocks.2.resnets.0.norm2.bias",
    "encoder.down.2.block.0.norm2.weight": "encoder.down_blocks.2.resnets.0.norm2.weight",
    "encoder.down.2.block.1.conv1.bias": "encoder.down_blocks.2.resnets.1.conv1.bias",
    "encoder.down.2.block.1.conv1.weight": "encoder.down_blocks.2.resnets.1.conv1.weight",
    "encoder.down.2.block.1.conv2.bias": "encoder.down_blocks.2.resnets.1.conv2.bias",
    "encoder.down.2.block.1.conv2.weight": "encoder.down_blocks.2.resnets.1.conv2.weight",
    "encoder.down.2.block.1.norm1.bias": "encoder.down_blocks.2.resnets.1.norm1.bias",
    "encoder.down.2.block.1.norm1.weight": "encoder.down_blocks.2.resnets.1.norm1.weight",
    "encoder.down.2.block.1.norm2.bias": "encoder.down_blocks.2.resnets.1.norm2.bias",
    "encoder.down.2.block.1.norm2.weight": "encoder.down_blocks.2.resnets.1.norm2.weight",
    "encoder.down.2.downsample.conv.bias": "encoder.down_blocks.2.downsamplers.0.conv.bias",
    "encoder.down.2.downsample.conv.weight": "encoder.down_blocks.2.downsamplers.0.conv.weight",
    "encoder.down.3.block.0.conv1.bias": "encoder.down_blocks.3.resnets.0.conv1.bias",
    "encoder.down.3.block.0.conv1.weight": "encoder.down_blocks.3.resnets.0.conv1.weight",
    "encoder.down.3.block.0.conv2.bias": "encoder.down_blocks.3.resnets.0.conv2.bias",
    "encoder.down.3.block.0.conv2.weight": "encoder.down_blocks.3.resnets.0.conv2.weight",
    "encoder.down.3.block.0.norm1.bias": "encoder.down_blocks.3.resnets.0.norm1.bias",
    "encoder.down.3.block.0.norm1.weight": "encoder.down_blocks.3.resnets.0.norm1.weight",
    "encoder.down.3.block.0.norm2.bias": "encoder.down_blocks.3.resnets.0.norm2.bias",
    "encoder.down.3.block.0.norm2.weight": "encoder.down_blocks.3.resnets.0.norm2.weight",
    "encoder.down.3.block.1.conv1.bias": "encoder.down_blocks.3.resnets.1.conv1.bias",
    "encoder.down.3.block.1.conv1.weight": "encoder.down_blocks.3.resnets.1.conv1.weight",
    "encoder.down.3.block.1.conv2.bias": "encoder.down_blocks.3.resnets.1.conv2.bias",
    "encoder.down.3.block.1.conv2.weight": "encoder.down_blocks.3.resnets.1.conv2.weight",
    "encoder.down.3.block.1.norm1.bias": "encoder.down_blocks.3.resnets.1.norm1.bias",
    "encoder.down.3.block.1.norm1.weight": "encoder.down_blocks.3.resnets.1.norm1.weight",
    "encoder.down.3.block.1.norm2.bias": "encoder.down_blocks.3.resnets.1.norm2.bias",
    "encoder.down.3.block.1.norm2.weight": "encoder.down_blocks.3.resnets.1.norm2.weight",
    "encoder.mid.attn_1.k.bias": "encoder.mid_block.attentions.0.to_k.bias",
    "encoder.mid.attn_1.k.weight": "encoder.mid_block.attentions.0.to_k.weight",
    "encoder.mid.attn_1.norm.bias": "encoder.mid_block.attentions.0.group_norm.bias",
    "encoder.mid.attn_1.norm.weight": "encoder.mid_block.attentions.0.group_norm.weight",
    "encoder.mid.attn_1.proj_out.bias": "encoder.mid_block.attentions.0.to_out.0.bias",
    "encoder.mid.attn_1.proj_out.weight": "encoder.mid_block.attentions.0.to_out.0.weight",
    "encoder.mid.attn_1.q.bias": "encoder.mid_block.attentions.0.to_q.bias",
    "encoder.mid.attn_1.q.weight": "encoder.mid_block.attentions.0.to_q.weight",
    "encoder.mid.attn_1.v.bias": "encoder.mid_block.attentions.0.to_v.bias",
    "encoder.mid.attn_1.v.weight": "encoder.mid_block.attentions.0.to_v.weight",
    "encoder.mid.block_1.conv1.bias": "encoder.mid_block.resnets.0.conv1.bias",
    "encoder.mid.block_1.conv1.weight": "encoder.mid_block.resnets.0.conv1.weight",
    "encoder.mid.block_1.conv2.bias": "encoder.mid_block.resnets.0.conv2.bias",
    "encoder.mid.block_1.conv2.weight": "encoder.mid_block.resnets.0.conv2.weight",
    "encoder.mid.block_1.norm1.bias": "encoder.mid_block.resnets.0.norm1.bias",
    "encoder.mid.block_1.norm1.weight": "encoder.mid_block.resnets.0.norm1.weight",
    "encoder.mid.block_1.norm2.bias": "encoder.mid_block.resnets.0.norm2.bias",
    "encoder.mid.block_1.norm2.weight": "encoder.mid_block.resnets.0.norm2.weight",
    "encoder.mid.block_2.conv1.bias": "encoder.mid_block.resnets.1.conv1.bias",
    "encoder.mid.block_2.conv1.weight": "encoder.mid_block.resnets.1.conv1.weight",
    "encoder.mid.block_2.conv2.bias": "encoder.mid_block.resnets.1.conv2.bias",
    "encoder.mid.block_2.conv2.weight": "encoder.mid_block.resnets.1.conv2.weight",
    "encoder.mid.block_2.norm1.bias": "encoder.mid_block.resnets.1.norm1.bias",
    "encoder.mid.block_2.norm1.weight": "encoder.mid_block.resnets.1.norm1.weight",
    "encoder.mid.block_2.norm2.bias": "encoder.mid_block.resnets.1.norm2.bias",
    "encoder.mid.block_2.norm2.weight": "encoder.mid_block.resnets.1.norm2.weight",
    "encoder.norm_out.bias": "encoder.conv_norm_out.bias",
    "encoder.norm_out.weight": "encoder.conv_norm_out.weight",
    "post_quant_conv.bias": "post_quant_conv.bias",
    "post_quant_conv.weight": "post_quant_conv.weight",
    "quant_conv.bias": "quant_conv.bias",
    "quant_conv.weight": "quant_conv.weight"
}


def get_diffusers_vae_key_from_ldm_key(target_ldm_key, i=None):
    for ldm_key, diffusers_key in vae_ldm_to_diffusers_dict.items():
        if i is not None:
            ldm_key = ldm_key.replace("{i}", str(i))
            diffusers_key = diffusers_key.replace("{i}", str(i))
        if ldm_key == target_ldm_key:
            return diffusers_key

    if ldm_key in vae_ldm_to_diffusers_dict:
        return vae_ldm_to_diffusers_dict[ldm_key]
    else:
        return None

# def get_ldm_vae_key_from_diffusers_key(target_diffusers_key):
#     for ldm_key, diffusers_key in vae_ldm_to_diffusers_dict.items():
#         if diffusers_key == target_diffusers_key:
#             return ldm_key
#     return None

def get_ldm_vae_key_from_diffusers_key(target_diffusers_key):
    for ldm_key, diffusers_key in vae_ldm_to_diffusers_dict.items():
        if "{" in diffusers_key:  # if we have a placeholder
            # escape special characters in the key, and replace the placeholder with a regex group
            pattern = re.escape(diffusers_key).replace("\\{i\\}", "(\\d+)")
            match = re.match(pattern, target_diffusers_key)
            if match:  # if we found a match
                return ldm_key.format(i=match.group(1))
        elif diffusers_key == target_diffusers_key:
            return ldm_key
    return None


vae_keys_squished_on_diffusers = [
    "decoder.mid_block.attentions.0.to_k.weight",
    "decoder.mid_block.attentions.0.to_out.0.weight",
    "decoder.mid_block.attentions.0.to_q.weight",
    "decoder.mid_block.attentions.0.to_v.weight",
    "encoder.mid_block.attentions.0.to_k.weight",
    "encoder.mid_block.attentions.0.to_out.0.weight",
    "encoder.mid_block.attentions.0.to_q.weight",
    "encoder.mid_block.attentions.0.to_v.weight"
]

def convert_diffusers_back_to_ldm(diffusers_vae):
    new_state_dict = OrderedDict()
    diffusers_state_dict = diffusers_vae.state_dict()
    for key, value in diffusers_state_dict.items():
        val_to_save = value
        if key in vae_keys_squished_on_diffusers:
            val_to_save = value.clone()
            # (512, 512) diffusers and (512, 512, 1, 1) ldm
            val_to_save = val_to_save.unsqueeze(-1).unsqueeze(-1)
        ldm_key = get_ldm_vae_key_from_diffusers_key(key)
        if ldm_key is not None:
            new_state_dict[ldm_key] = val_to_save
        else:
            # for now add current key
            new_state_dict[key] = val_to_save
    return new_state_dict


def convert_ldm_vae_checkpoint(checkpoint, config):
    mapping = {}
    # extract state dict for VAE
    vae_state_dict = {}
    vae_key = "first_stage_model."
    keys = list(checkpoint.keys())
    for key in keys:
        if key.startswith(vae_key):
            vae_state_dict[key.replace(vae_key, "")] = checkpoint.get(key)
    # if len(vae_state_dict) == 0:
    #   # 渡されたcheckpointは.ckptから読み込んだcheckpointではなくvaeのstate_dict
    #   vae_state_dict = checkpoint

    new_checkpoint = {}

    # for key in list(vae_state_dict.keys()):
    #     diffusers_key = get_diffusers_vae_key_from_ldm_key(key)
    #     if diffusers_key is not None:
    #         new_checkpoint[diffusers_key] = vae_state_dict[key]

    new_checkpoint["encoder.conv_in.weight"] = vae_state_dict["encoder.conv_in.weight"]
    new_checkpoint["encoder.conv_in.bias"] = vae_state_dict["encoder.conv_in.bias"]
    new_checkpoint["encoder.conv_out.weight"] = vae_state_dict["encoder.conv_out.weight"]
    new_checkpoint["encoder.conv_out.bias"] = vae_state_dict["encoder.conv_out.bias"]
    new_checkpoint["encoder.conv_norm_out.weight"] = vae_state_dict["encoder.norm_out.weight"]
    new_checkpoint["encoder.conv_norm_out.bias"] = vae_state_dict["encoder.norm_out.bias"]

    new_checkpoint["decoder.conv_in.weight"] = vae_state_dict["decoder.conv_in.weight"]
    new_checkpoint["decoder.conv_in.bias"] = vae_state_dict["decoder.conv_in.bias"]
    new_checkpoint["decoder.conv_out.weight"] = vae_state_dict["decoder.conv_out.weight"]
    new_checkpoint["decoder.conv_out.bias"] = vae_state_dict["decoder.conv_out.bias"]
    new_checkpoint["decoder.conv_norm_out.weight"] = vae_state_dict["decoder.norm_out.weight"]
    new_checkpoint["decoder.conv_norm_out.bias"] = vae_state_dict["decoder.norm_out.bias"]

    new_checkpoint["quant_conv.weight"] = vae_state_dict["quant_conv.weight"]
    new_checkpoint["quant_conv.bias"] = vae_state_dict["quant_conv.bias"]
    new_checkpoint["post_quant_conv.weight"] = vae_state_dict["post_quant_conv.weight"]
    new_checkpoint["post_quant_conv.bias"] = vae_state_dict["post_quant_conv.bias"]

    # Retrieves the keys for the encoder down blocks only
    num_down_blocks = len({".".join(layer.split(".")[:3]) for layer in vae_state_dict if "encoder.down" in layer})
    down_blocks = {layer_id: [key for key in vae_state_dict if f"down.{layer_id}" in key] for layer_id in
                   range(num_down_blocks)}

    # Retrieves the keys for the decoder up blocks only
    num_up_blocks = len({".".join(layer.split(".")[:3]) for layer in vae_state_dict if "decoder.up" in layer})
    up_blocks = {layer_id: [key for key in vae_state_dict if f"up.{layer_id}" in key] for layer_id in
                 range(num_up_blocks)}

    for i in range(num_down_blocks):
        resnets = [key for key in down_blocks[i] if f"down.{i}" in key and f"down.{i}.downsample" not in key]

        if f"encoder.down.{i}.downsample.conv.weight" in vae_state_dict:
            new_checkpoint[f"encoder.down_blocks.{i}.downsamplers.0.conv.weight"] = vae_state_dict.pop(
                f"encoder.down.{i}.downsample.conv.weight"
            )
            mapping[f"encoder.down.{i}.downsample.conv.weight"] = f"encoder.down_blocks.{i}.downsamplers.0.conv.weight"
            new_checkpoint[f"encoder.down_blocks.{i}.downsamplers.0.conv.bias"] = vae_state_dict.pop(
                f"encoder.down.{i}.downsample.conv.bias"
            )
            mapping[f"encoder.down.{i}.downsample.conv.bias"] = f"encoder.down_blocks.{i}.downsamplers.0.conv.bias"

        paths = renew_vae_resnet_paths(resnets)
        meta_path = {"old": f"down.{i}.block", "new": f"down_blocks.{i}.resnets"}
        assign_to_checkpoint(paths, new_checkpoint, vae_state_dict, additional_replacements=[meta_path], config=config)

    mid_resnets = [key for key in vae_state_dict if "encoder.mid.block" in key]
    num_mid_res_blocks = 2
    for i in range(1, num_mid_res_blocks + 1):
        resnets = [key for key in mid_resnets if f"encoder.mid.block_{i}" in key]

        paths = renew_vae_resnet_paths(resnets)
        meta_path = {"old": f"mid.block_{i}", "new": f"mid_block.resnets.{i - 1}"}
        assign_to_checkpoint(paths, new_checkpoint, vae_state_dict, additional_replacements=[meta_path], config=config)

    mid_attentions = [key for key in vae_state_dict if "encoder.mid.attn" in key]
    paths = renew_vae_attention_paths(mid_attentions)
    meta_path = {"old": "mid.attn_1", "new": "mid_block.attentions.0"}
    assign_to_checkpoint(paths, new_checkpoint, vae_state_dict, additional_replacements=[meta_path], config=config)
    conv_attn_to_linear(new_checkpoint)

    for i in range(num_up_blocks):
        block_id = num_up_blocks - 1 - i
        resnets = [key for key in up_blocks[block_id] if
                   f"up.{block_id}" in key and f"up.{block_id}.upsample" not in key]

        if f"decoder.up.{block_id}.upsample.conv.weight" in vae_state_dict:
            new_checkpoint[f"decoder.up_blocks.{i}.upsamplers.0.conv.weight"] = vae_state_dict[
                f"decoder.up.{block_id}.upsample.conv.weight"
            ]
            mapping[f"decoder.up.{block_id}.upsample.conv.weight"] = f"decoder.up_blocks.{i}.upsamplers.0.conv.weight"
            new_checkpoint[f"decoder.up_blocks.{i}.upsamplers.0.conv.bias"] = vae_state_dict[
                f"decoder.up.{block_id}.upsample.conv.bias"
            ]
            mapping[f"decoder.up.{block_id}.upsample.conv.bias"] = f"decoder.up_blocks.{i}.upsamplers.0.conv.bias"

        paths = renew_vae_resnet_paths(resnets)
        meta_path = {"old": f"up.{block_id}.block", "new": f"up_blocks.{i}.resnets"}
        assign_to_checkpoint(paths, new_checkpoint, vae_state_dict, additional_replacements=[meta_path], config=config)

    mid_resnets = [key for key in vae_state_dict if "decoder.mid.block" in key]
    num_mid_res_blocks = 2
    for i in range(1, num_mid_res_blocks + 1):
        resnets = [key for key in mid_resnets if f"decoder.mid.block_{i}" in key]

        paths = renew_vae_resnet_paths(resnets)
        meta_path = {"old": f"mid.block_{i}", "new": f"mid_block.resnets.{i - 1}"}
        assign_to_checkpoint(paths, new_checkpoint, vae_state_dict, additional_replacements=[meta_path], config=config)

    mid_attentions = [key for key in vae_state_dict if "decoder.mid.attn" in key]
    paths = renew_vae_attention_paths(mid_attentions)
    meta_path = {"old": "mid.attn_1", "new": "mid_block.attentions.0"}
    assign_to_checkpoint(paths, new_checkpoint, vae_state_dict, additional_replacements=[meta_path], config=config)
    conv_attn_to_linear(new_checkpoint)
    return new_checkpoint


def create_unet_diffusers_config(v2, use_linear_projection_in_v2=False):
    """
    Creates a config for the diffusers based on the config of the LDM model.
    """
    # unet_params = original_config.model.params.unet_config.params

    block_out_channels = [UNET_PARAMS_MODEL_CHANNELS * mult for mult in UNET_PARAMS_CHANNEL_MULT]

    down_block_types = []
    resolution = 1
    for i in range(len(block_out_channels)):
        block_type = "CrossAttnDownBlock2D" if resolution in UNET_PARAMS_ATTENTION_RESOLUTIONS else "DownBlock2D"
        down_block_types.append(block_type)
        if i != len(block_out_channels) - 1:
            resolution *= 2

    up_block_types = []
    for i in range(len(block_out_channels)):
        block_type = "CrossAttnUpBlock2D" if resolution in UNET_PARAMS_ATTENTION_RESOLUTIONS else "UpBlock2D"
        up_block_types.append(block_type)
        resolution //= 2

    config = dict(
        sample_size=UNET_PARAMS_IMAGE_SIZE,
        in_channels=UNET_PARAMS_IN_CHANNELS,
        out_channels=UNET_PARAMS_OUT_CHANNELS,
        down_block_types=tuple(down_block_types),
        up_block_types=tuple(up_block_types),
        block_out_channels=tuple(block_out_channels),
        layers_per_block=UNET_PARAMS_NUM_RES_BLOCKS,
        cross_attention_dim=UNET_PARAMS_CONTEXT_DIM if not v2 else V2_UNET_PARAMS_CONTEXT_DIM,
        attention_head_dim=UNET_PARAMS_NUM_HEADS if not v2 else V2_UNET_PARAMS_ATTENTION_HEAD_DIM,
        # use_linear_projection=UNET_PARAMS_USE_LINEAR_PROJECTION if not v2 else V2_UNET_PARAMS_USE_LINEAR_PROJECTION,
    )
    if v2 and use_linear_projection_in_v2:
        config["use_linear_projection"] = True

    return config


def create_vae_diffusers_config():
    """
    Creates a config for the diffusers based on the config of the LDM model.
    """
    # vae_params = original_config.model.params.first_stage_config.params.ddconfig
    # _ = original_config.model.params.first_stage_config.params.embed_dim
    block_out_channels = [VAE_PARAMS_CH * mult for mult in VAE_PARAMS_CH_MULT]
    down_block_types = ["DownEncoderBlock2D"] * len(block_out_channels)
    up_block_types = ["UpDecoderBlock2D"] * len(block_out_channels)

    config = dict(
        sample_size=VAE_PARAMS_RESOLUTION,
        in_channels=VAE_PARAMS_IN_CHANNELS,
        out_channels=VAE_PARAMS_OUT_CH,
        down_block_types=tuple(down_block_types),
        up_block_types=tuple(up_block_types),
        block_out_channels=tuple(block_out_channels),
        latent_channels=VAE_PARAMS_Z_CHANNELS,
        layers_per_block=VAE_PARAMS_NUM_RES_BLOCKS,
    )
    return config


def convert_ldm_clip_checkpoint_v1(checkpoint):
    keys = list(checkpoint.keys())
    text_model_dict = {}
    for key in keys:
        if key.startswith("cond_stage_model.transformer"):
            text_model_dict[key[len("cond_stage_model.transformer."):]] = checkpoint[key]
    # support checkpoint without position_ids (invalid checkpoint)
    if "text_model.embeddings.position_ids" not in text_model_dict:
        text_model_dict["text_model.embeddings.position_ids"] = torch.arange(77).unsqueeze(0)  # 77 is the max length of the text
    return text_model_dict


def convert_ldm_clip_checkpoint_v2(checkpoint, max_length):
    # 嫌になるくらい違うぞ！
    def convert_key(key):
        if not key.startswith("cond_stage_model"):
            return None

        # common conversion
        key = key.replace("cond_stage_model.model.transformer.", "text_model.encoder.")
        key = key.replace("cond_stage_model.model.", "text_model.")

        if "resblocks" in key:
            # resblocks conversion
            key = key.replace(".resblocks.", ".layers.")
            if ".ln_" in key:
                key = key.replace(".ln_", ".layer_norm")
            elif ".mlp." in key:
                key = key.replace(".c_fc.", ".fc1.")
                key = key.replace(".c_proj.", ".fc2.")
            elif ".attn.out_proj" in key:
                key = key.replace(".attn.out_proj.", ".self_attn.out_proj.")
            elif ".attn.in_proj" in key:
                key = None  # 特殊なので後で処理する
            else:
                raise ValueError(f"unexpected key in SD: {key}")
        elif ".positional_embedding" in key:
            key = key.replace(".positional_embedding", ".embeddings.position_embedding.weight")
        elif ".text_projection" in key:
            key = None  # 使われない???
        elif ".logit_scale" in key:
            key = None  # 使われない???
        elif ".token_embedding" in key:
            key = key.replace(".token_embedding.weight", ".embeddings.token_embedding.weight")
        elif ".ln_final" in key:
            key = key.replace(".ln_final", ".final_layer_norm")
        return key

    keys = list(checkpoint.keys())
    new_sd = {}
    for key in keys:
        # remove resblocks 23
        if ".resblocks.23." in key:
            continue
        new_key = convert_key(key)
        if new_key is None:
            continue
        new_sd[new_key] = checkpoint[key]

    # attnの変換
    for key in keys:
        if ".resblocks.23." in key:
            continue
        if ".resblocks" in key and ".attn.in_proj_" in key:
            # 三つに分割
            values = torch.chunk(checkpoint[key], 3)

            key_suffix = ".weight" if "weight" in key else ".bias"
            key_pfx = key.replace("cond_stage_model.model.transformer.resblocks.", "text_model.encoder.layers.")
            key_pfx = key_pfx.replace("_weight", "")
            key_pfx = key_pfx.replace("_bias", "")
            key_pfx = key_pfx.replace(".attn.in_proj", ".self_attn.")
            new_sd[key_pfx + "q_proj" + key_suffix] = values[0]
            new_sd[key_pfx + "k_proj" + key_suffix] = values[1]
            new_sd[key_pfx + "v_proj" + key_suffix] = values[2]

    # rename or add position_ids
    ANOTHER_POSITION_IDS_KEY = "text_model.encoder.text_model.embeddings.position_ids"
    if ANOTHER_POSITION_IDS_KEY in new_sd:
        # waifu diffusion v1.4
        position_ids = new_sd[ANOTHER_POSITION_IDS_KEY]
        del new_sd[ANOTHER_POSITION_IDS_KEY]
    else:
        position_ids = torch.Tensor([list(range(max_length))]).to(torch.int64)

    new_sd["text_model.embeddings.position_ids"] = position_ids
    return new_sd


# endregion


# region Diffusers->StableDiffusion の変換コード
# convert_diffusers_to_original_stable_diffusion をコピーして修正している（ASL 2.0）


def conv_transformer_to_linear(checkpoint):
    keys = list(checkpoint.keys())
    tf_keys = ["proj_in.weight", "proj_out.weight"]
    for key in keys:
        if ".".join(key.split(".")[-2:]) in tf_keys:
            if checkpoint[key].ndim > 2:
                checkpoint[key] = checkpoint[key][:, :, 0, 0]


def convert_unet_state_dict_to_sd(v2, unet_state_dict):
    unet_conversion_map = [
        # (stable-diffusion, HF Diffusers)
        ("time_embed.0.weight", "time_embedding.linear_1.weight"),
        ("time_embed.0.bias", "time_embedding.linear_1.bias"),
        ("time_embed.2.weight", "time_embedding.linear_2.weight"),
        ("time_embed.2.bias", "time_embedding.linear_2.bias"),
        ("input_blocks.0.0.weight", "conv_in.weight"),
        ("input_blocks.0.0.bias", "conv_in.bias"),
        ("out.0.weight", "conv_norm_out.weight"),
        ("out.0.bias", "conv_norm_out.bias"),
        ("out.2.weight", "conv_out.weight"),
        ("out.2.bias", "conv_out.bias"),
    ]

    unet_conversion_map_resnet = [
        # (stable-diffusion, HF Diffusers)
        ("in_layers.0", "norm1"),
        ("in_layers.2", "conv1"),
        ("out_layers.0", "norm2"),
        ("out_layers.3", "conv2"),
        ("emb_layers.1", "time_emb_proj"),
        ("skip_connection", "conv_shortcut"),
    ]

    unet_conversion_map_layer = []
    for i in range(4):
        # loop over downblocks/upblocks

        for j in range(2):
            # loop over resnets/attentions for downblocks
            hf_down_res_prefix = f"down_blocks.{i}.resnets.{j}."
            sd_down_res_prefix = f"input_blocks.{3 * i + j + 1}.0."
            unet_conversion_map_layer.append((sd_down_res_prefix, hf_down_res_prefix))

            if i < 3:
                # no attention layers in down_blocks.3
                hf_down_atn_prefix = f"down_blocks.{i}.attentions.{j}."
                sd_down_atn_prefix = f"input_blocks.{3 * i + j + 1}.1."
                unet_conversion_map_layer.append((sd_down_atn_prefix, hf_down_atn_prefix))

        for j in range(3):
            # loop over resnets/attentions for upblocks
            hf_up_res_prefix = f"up_blocks.{i}.resnets.{j}."
            sd_up_res_prefix = f"output_blocks.{3 * i + j}.0."
            unet_conversion_map_layer.append((sd_up_res_prefix, hf_up_res_prefix))

            if i > 0:
                # no attention layers in up_blocks.0
                hf_up_atn_prefix = f"up_blocks.{i}.attentions.{j}."
                sd_up_atn_prefix = f"output_blocks.{3 * i + j}.1."
                unet_conversion_map_layer.append((sd_up_atn_prefix, hf_up_atn_prefix))

        if i < 3:
            # no downsample in down_blocks.3
            hf_downsample_prefix = f"down_blocks.{i}.downsamplers.0.conv."
            sd_downsample_prefix = f"input_blocks.{3 * (i + 1)}.0.op."
            unet_conversion_map_layer.append((sd_downsample_prefix, hf_downsample_prefix))

            # no upsample in up_blocks.3
            hf_upsample_prefix = f"up_blocks.{i}.upsamplers.0."
            sd_upsample_prefix = f"output_blocks.{3 * i + 2}.{1 if i == 0 else 2}."
            unet_conversion_map_layer.append((sd_upsample_prefix, hf_upsample_prefix))

    hf_mid_atn_prefix = "mid_block.attentions.0."
    sd_mid_atn_prefix = "middle_block.1."
    unet_conversion_map_layer.append((sd_mid_atn_prefix, hf_mid_atn_prefix))

    for j in range(2):
        hf_mid_res_prefix = f"mid_block.resnets.{j}."
        sd_mid_res_prefix = f"middle_block.{2 * j}."
        unet_conversion_map_layer.append((sd_mid_res_prefix, hf_mid_res_prefix))

    # buyer beware: this is a *brittle* function,
    # and correct output requires that all of these pieces interact in
    # the exact order in which I have arranged them.
    mapping = {k: k for k in unet_state_dict.keys()}
    for sd_name, hf_name in unet_conversion_map:
        mapping[hf_name] = sd_name
    for k, v in mapping.items():
        if "resnets" in k:
            for sd_part, hf_part in unet_conversion_map_resnet:
                v = v.replace(hf_part, sd_part)
            mapping[k] = v
    for k, v in mapping.items():
        for sd_part, hf_part in unet_conversion_map_layer:
            v = v.replace(hf_part, sd_part)
        mapping[k] = v
    new_state_dict = {v: unet_state_dict[k] for k, v in mapping.items()}

    if v2:
        conv_transformer_to_linear(new_state_dict)

    return new_state_dict


# ================#
# VAE Conversion #
# ================#


def reshape_weight_for_sd(w):
    # convert HF linear weights to SD conv2d weights
    return w.reshape(*w.shape, 1, 1)


def convert_vae_state_dict(vae_state_dict):
    vae_conversion_map = [
        # (stable-diffusion, HF Diffusers)
        ("nin_shortcut", "conv_shortcut"),
        ("norm_out", "conv_norm_out"),
        ("mid.attn_1.", "mid_block.attentions.0."),
    ]

    for i in range(4):
        # down_blocks have two resnets
        for j in range(2):
            hf_down_prefix = f"encoder.down_blocks.{i}.resnets.{j}."
            sd_down_prefix = f"encoder.down.{i}.block.{j}."
            vae_conversion_map.append((sd_down_prefix, hf_down_prefix))

        if i < 3:
            hf_downsample_prefix = f"down_blocks.{i}.downsamplers.0."
            sd_downsample_prefix = f"down.{i}.downsample."
            vae_conversion_map.append((sd_downsample_prefix, hf_downsample_prefix))

            hf_upsample_prefix = f"up_blocks.{i}.upsamplers.0."
            sd_upsample_prefix = f"up.{3 - i}.upsample."
            vae_conversion_map.append((sd_upsample_prefix, hf_upsample_prefix))

        # up_blocks have three resnets
        # also, up blocks in hf are numbered in reverse from sd
        for j in range(3):
            hf_up_prefix = f"decoder.up_blocks.{i}.resnets.{j}."
            sd_up_prefix = f"decoder.up.{3 - i}.block.{j}."
            vae_conversion_map.append((sd_up_prefix, hf_up_prefix))

    # this part accounts for mid blocks in both the encoder and the decoder
    for i in range(2):
        hf_mid_res_prefix = f"mid_block.resnets.{i}."
        sd_mid_res_prefix = f"mid.block_{i + 1}."
        vae_conversion_map.append((sd_mid_res_prefix, hf_mid_res_prefix))

    vae_conversion_map_attn = [
        # (stable-diffusion, HF Diffusers)
        ("norm.", "group_norm."),
        ("q.", "query."),
        ("k.", "key."),
        ("v.", "value."),
        ("proj_out.", "proj_attn."),
    ]

    mapping = {k: k for k in vae_state_dict.keys()}
    for k, v in mapping.items():
        for sd_part, hf_part in vae_conversion_map:
            v = v.replace(hf_part, sd_part)
        mapping[k] = v
    for k, v in mapping.items():
        if "attentions" in k:
            for sd_part, hf_part in vae_conversion_map_attn:
                v = v.replace(hf_part, sd_part)
            mapping[k] = v
    new_state_dict = {v: vae_state_dict[k] for k, v in mapping.items()}
    weights_to_convert = ["q", "k", "v", "proj_out"]
    for k, v in new_state_dict.items():
        for weight_name in weights_to_convert:
            if f"mid.attn_1.{weight_name}.weight" in k:
                # print(f"Reshaping {k} for SD format")
                new_state_dict[k] = reshape_weight_for_sd(v)

    return new_state_dict


# endregion

# region 自作のモデル読み書きなど


def is_safetensors(path):
    return os.path.splitext(path)[1].lower() == ".safetensors"


def load_checkpoint_with_text_encoder_conversion(ckpt_path, device="cpu"):
    # text encoderの格納形式が違うモデルに対応する ('text_model'がない)
    TEXT_ENCODER_KEY_REPLACEMENTS = [
        ("cond_stage_model.transformer.embeddings.", "cond_stage_model.transformer.text_model.embeddings."),
        ("cond_stage_model.transformer.encoder.", "cond_stage_model.transformer.text_model.encoder."),
        ("cond_stage_model.transformer.final_layer_norm.", "cond_stage_model.transformer.text_model.final_layer_norm."),
    ]

    if is_safetensors(ckpt_path):
        checkpoint = None
        state_dict = load_file(ckpt_path)  # , device) # may causes error
    else:
        checkpoint = torch.load(ckpt_path, map_location=device)
        if "state_dict" in checkpoint:
            state_dict = checkpoint["state_dict"]
        else:
            state_dict = checkpoint
            checkpoint = None

    key_reps = []
    for rep_from, rep_to in TEXT_ENCODER_KEY_REPLACEMENTS:
        for key in state_dict.keys():
            if key.startswith(rep_from):
                new_key = rep_to + key[len(rep_from):]
                key_reps.append((key, new_key))

    for key, new_key in key_reps:
        state_dict[new_key] = state_dict[key]
        del state_dict[key]

    return checkpoint, state_dict


# TODO dtype指定の動作が怪しいので確認する text_encoderを指定形式で作れるか未確認
def load_models_from_stable_diffusion_checkpoint(v2, ckpt_path, device="cpu", dtype=None,
                                                 unet_use_linear_projection_in_v2=False):
    _, state_dict = load_checkpoint_with_text_encoder_conversion(ckpt_path, device)

    # Convert the UNet2DConditionModel model.
    unet_config = create_unet_diffusers_config(v2, unet_use_linear_projection_in_v2)
    converted_unet_checkpoint = convert_ldm_unet_checkpoint(v2, state_dict, unet_config)

    unet = UNet2DConditionModel(**unet_config).to(device)
    info = unet.load_state_dict(converted_unet_checkpoint)
    print("loading u-net:", info)

    # Convert the VAE model.
    vae_config = create_vae_diffusers_config()
    converted_vae_checkpoint = convert_ldm_vae_checkpoint(state_dict, vae_config)

    vae = AutoencoderKL(**vae_config).to(device)
    info = vae.load_state_dict(converted_vae_checkpoint)
    print("loading vae:", info)

    # convert text_model
    if v2:
        converted_text_encoder_checkpoint = convert_ldm_clip_checkpoint_v2(state_dict, 77)
        cfg = CLIPTextConfig(
            vocab_size=49408,
            hidden_size=1024,
            intermediate_size=4096,
            num_hidden_layers=23,
            num_attention_heads=16,
            max_position_embeddings=77,
            hidden_act="gelu",
            layer_norm_eps=1e-05,
            dropout=0.0,
            attention_dropout=0.0,
            initializer_range=0.02,
            initializer_factor=1.0,
            pad_token_id=1,
            bos_token_id=0,
            eos_token_id=2,
            model_type="clip_text_model",
            projection_dim=512,
            torch_dtype="float32",
            transformers_version="4.25.0.dev0",
        )
        text_model = CLIPTextModel._from_config(cfg)
        info = text_model.load_state_dict(converted_text_encoder_checkpoint)
    else:
        converted_text_encoder_checkpoint = convert_ldm_clip_checkpoint_v1(state_dict)

        logging.set_verbosity_error()  # don't show annoying warning
        text_model = CLIPTextModel.from_pretrained("openai/clip-vit-large-patch14").to(device)
        logging.set_verbosity_warning()

        # latest transformers doesnt have position ids. Do we remove it?
        if "text_model.embeddings.position_ids" not in text_model.state_dict():
            del converted_text_encoder_checkpoint["text_model.embeddings.position_ids"]

        info = text_model.load_state_dict(converted_text_encoder_checkpoint)
    print("loading text encoder:", info)

    return text_model, vae, unet


def convert_text_encoder_state_dict_to_sd_v2(checkpoint, make_dummy_weights=False):
    def convert_key(key):
        # position_idsの除去
        if ".position_ids" in key:
            return None

        # common
        key = key.replace("text_model.encoder.", "transformer.")
        key = key.replace("text_model.", "")
        if "layers" in key:
            # resblocks conversion
            key = key.replace(".layers.", ".resblocks.")
            if ".layer_norm" in key:
                key = key.replace(".layer_norm", ".ln_")
            elif ".mlp." in key:
                key = key.replace(".fc1.", ".c_fc.")
                key = key.replace(".fc2.", ".c_proj.")
            elif ".self_attn.out_proj" in key:
                key = key.replace(".self_attn.out_proj.", ".attn.out_proj.")
            elif ".self_attn." in key:
                key = None  # 特殊なので後で処理する
            else:
                raise ValueError(f"unexpected key in DiffUsers model: {key}")
        elif ".position_embedding" in key:
            key = key.replace("embeddings.position_embedding.weight", "positional_embedding")
        elif ".token_embedding" in key:
            key = key.replace("embeddings.token_embedding.weight", "token_embedding.weight")
        elif "final_layer_norm" in key:
            key = key.replace("final_layer_norm", "ln_final")
        return key

    keys = list(checkpoint.keys())
    new_sd = {}
    for key in keys:
        new_key = convert_key(key)
        if new_key is None:
            continue
        new_sd[new_key] = checkpoint[key]

    # attnの変換
    for key in keys:
        if "layers" in key and "q_proj" in key:
            # 三つを結合
            key_q = key
            key_k = key.replace("q_proj", "k_proj")
            key_v = key.replace("q_proj", "v_proj")

            value_q = checkpoint[key_q]
            value_k = checkpoint[key_k]
            value_v = checkpoint[key_v]
            value = torch.cat([value_q, value_k, value_v])

            new_key = key.replace("text_model.encoder.layers.", "transformer.resblocks.")
            new_key = new_key.replace(".self_attn.q_proj.", ".attn.in_proj_")
            new_sd[new_key] = value

    # 最後の層などを捏造するか
    if make_dummy_weights:
        print("make dummy weights for resblock.23, text_projection and logit scale.")
        keys = list(new_sd.keys())
        for key in keys:
            if key.startswith("transformer.resblocks.22."):
                new_sd[key.replace(".22.", ".23.")] = new_sd[key].clone()  # copyしないとsafetensorsの保存で落ちる

        # Diffusersに含まれない重みを作っておく
        new_sd["text_projection"] = torch.ones((1024, 1024), dtype=new_sd[keys[0]].dtype, device=new_sd[keys[0]].device)
        new_sd["logit_scale"] = torch.tensor(1)

    return new_sd


def save_stable_diffusion_checkpoint(v2, output_file, text_encoder, unet, ckpt_path, epochs, steps, save_dtype=None,
                                     vae=None):
    if ckpt_path is not None:
        # epoch/stepを参照する。またVAEがメモリ上にないときなど、もう一度VAEを含めて読み込む
        checkpoint, state_dict = load_checkpoint_with_text_encoder_conversion(ckpt_path)
        if checkpoint is None:  # safetensors または state_dictのckpt
            checkpoint = {}
            strict = False
        else:
            strict = True
        if "state_dict" in state_dict:
            del state_dict["state_dict"]
    else:
        # 新しく作る
        assert vae is not None, "VAE is required to save a checkpoint without a given checkpoint"
        checkpoint = {}
        state_dict = {}
        strict = False

    def update_sd(prefix, sd):
        for k, v in sd.items():
            key = prefix + k
            assert not strict or key in state_dict, f"Illegal key in save SD: {key}"
            if save_dtype is not None:
                v = v.detach().clone().to("cpu").to(save_dtype)
            state_dict[key] = v

    # Convert the UNet model
    unet_state_dict = convert_unet_state_dict_to_sd(v2, unet.state_dict())
    update_sd("model.diffusion_model.", unet_state_dict)

    # Convert the text encoder model
    if v2:
        make_dummy = ckpt_path is None  # 参照元のcheckpointがない場合は最後の層を前の層から複製して作るなどダミーの重みを入れる
        text_enc_dict = convert_text_encoder_state_dict_to_sd_v2(text_encoder.state_dict(), make_dummy)
        update_sd("cond_stage_model.model.", text_enc_dict)
    else:
        text_enc_dict = text_encoder.state_dict()
        update_sd("cond_stage_model.transformer.", text_enc_dict)

    # Convert the VAE
    if vae is not None:
        vae_dict = convert_vae_state_dict(vae.state_dict())
        update_sd("first_stage_model.", vae_dict)

    # Put together new checkpoint
    key_count = len(state_dict.keys())
    new_ckpt = {"state_dict": state_dict}

    # epoch and global_step are sometimes not int
    try:
        if "epoch" in checkpoint:
            epochs += checkpoint["epoch"]
        if "global_step" in checkpoint:
            steps += checkpoint["global_step"]
    except:
        pass

    new_ckpt["epoch"] = epochs
    new_ckpt["global_step"] = steps

    if is_safetensors(output_file):
        # TODO Tensor以外のdictの値を削除したほうがいいか
        save_file(state_dict, output_file)
    else:
        torch.save(new_ckpt, output_file)

    return key_count


def save_diffusers_checkpoint(v2, output_dir, text_encoder, unet, pretrained_model_name_or_path, vae=None,
                              use_safetensors=False):
    if pretrained_model_name_or_path is None:
        # load default settings for v1/v2
        if v2:
            pretrained_model_name_or_path = DIFFUSERS_REF_MODEL_ID_V2
        else:
            pretrained_model_name_or_path = DIFFUSERS_REF_MODEL_ID_V1

    scheduler = DDIMScheduler.from_pretrained(pretrained_model_name_or_path, subfolder="scheduler")
    tokenizer = CLIPTokenizer.from_pretrained(pretrained_model_name_or_path, subfolder="tokenizer")
    if vae is None:
        vae = AutoencoderKL.from_pretrained(pretrained_model_name_or_path, subfolder="vae")

    pipeline = StableDiffusionPipeline(
        unet=unet,
        text_encoder=text_encoder,
        vae=vae,
        scheduler=scheduler,
        tokenizer=tokenizer,
        safety_checker=None,
        feature_extractor=None,
        requires_safety_checker=None,
    )
    pipeline.save_pretrained(output_dir, safe_serialization=use_safetensors)


VAE_PREFIX = "first_stage_model."


def load_vae(vae_id, dtype):
    print(f"load VAE: {vae_id}")
    if os.path.isdir(vae_id) or not os.path.isfile(vae_id):
        # Diffusers local/remote
        try:
            vae = AutoencoderKL.from_pretrained(vae_id, subfolder=None, torch_dtype=dtype)
        except EnvironmentError as e:
            print(f"exception occurs in loading vae: {e}")
            print("retry with subfolder='vae'")
            vae = AutoencoderKL.from_pretrained(vae_id, subfolder="vae", torch_dtype=dtype)
        return vae

    # local
    vae_config = create_vae_diffusers_config()

    if vae_id.endswith(".bin"):
        # SD 1.5 VAE on Huggingface
        converted_vae_checkpoint = torch.load(vae_id, map_location="cpu")
    else:
        # StableDiffusion
        vae_model = load_file(vae_id, "cpu") if is_safetensors(vae_id) else torch.load(vae_id, map_location="cpu")
        vae_sd = vae_model["state_dict"] if "state_dict" in vae_model else vae_model

        # vae only or full model
        full_model = False
        for vae_key in vae_sd:
            if vae_key.startswith(VAE_PREFIX):
                full_model = True
                break
        if not full_model:
            sd = {}
            for key, value in vae_sd.items():
                sd[VAE_PREFIX + key] = value
            vae_sd = sd
            del sd

        # Convert the VAE model.
        converted_vae_checkpoint = convert_ldm_vae_checkpoint(vae_sd, vae_config)

    vae = AutoencoderKL(**vae_config)
    vae.load_state_dict(converted_vae_checkpoint)
    return vae


# endregion


def make_bucket_resolutions(max_reso, min_size=256, max_size=1024, divisible=64):
    max_width, max_height = max_reso
    max_area = (max_width // divisible) * (max_height // divisible)

    resos = set()

    size = int(math.sqrt(max_area)) * divisible
    resos.add((size, size))

    size = min_size
    while size <= max_size:
        width = size
        height = min(max_size, (max_area // (width // divisible)) * divisible)
        resos.add((width, height))
        resos.add((height, width))

        # # make additional resos
        # if width >= height and width - divisible >= min_size:
        #   resos.add((width - divisible, height))
        #   resos.add((height, width - divisible))
        # if height >= width and height - divisible >= min_size:
        #   resos.add((width, height - divisible))
        #   resos.add((height - divisible, width))

        size += divisible

    resos = list(resos)
    resos.sort()
    return resos


if __name__ == "__main__":
    resos = make_bucket_resolutions((512, 768))
    print(len(resos))
    print(resos)
    aspect_ratios = [w / h for w, h in resos]
    print(aspect_ratios)

    ars = set()
    for ar in aspect_ratios:
        if ar in ars:
            print("error! duplicate ar:", ar)
        ars.add(ar)