update readme and help message etc.

README.md

@@ -165,6 +165,10 @@ The majority of scripts is licensed under ASL 2.0 (including codes from Diffuser
   - Specify the learning rate and dim (rank) for each block.
   - See [Block-wise learning rates in LoRA](./docs/train_network_README-ja.md#階層別学習率) for details (Japanese only).
 
+- An option `--disable_mmap_load_safetensors` is added to disable memory mapping when loading the model's .safetensors in SDXL. PR [#1266](https://github.com/kohya-ss/sd-scripts/pull/1266) Thanks to Zovjsra!
+  - It seems that the model file loading is faster in the WSL environment etc.
+  - Available in `sdxl_train.py`, `sdxl_train_network.py`, `sdxl_train_textual_inversion.py`, and `sdxl_train_control_net_lllite.py`.
+
 - Fixed some bugs when using DeepSpeed. Related [#1247](https://github.com/kohya-ss/sd-scripts/pull/1247)
 
 - SDXL の学習時に Fused optimizer が使えるようになりました。PR [#1259](https://github.com/kohya-ss/sd-scripts/pull/1259) 2kpr 氏に感謝します。
@@ -193,6 +197,10 @@ The majority of scripts is licensed under ASL 2.0 (including codes from Diffuser
   - ブロックごとに学習率および dim (rank) を指定することができます。
   - 詳細は [LoRA の階層別学習率](./docs/train_network_README-ja.md#階層別学習率) をご覧ください。
 
+- SDXL でモデルの .safetensors を読み込む際にメモリマッピングを無効化するオプション `--disable_mmap_load_safetensors` が追加されました。PR [#1266](https://github.com/kohya-ss/sd-scripts/pull/1266) Zovjsra 氏に感謝します。
+  - WSL 環境等でモデルファイルの読み込みが高速化されるようです。
+  - `sdxl_train.py`、`sdxl_train_network.py`、`sdxl_train_textual_inversion.py`、`sdxl_train_control_net_lllite.py` で使用可能です。
+
 - DeepSpeed 使用時のいくつかのバグを修正しました。関連 [#1247](https://github.com/kohya-ss/sd-scripts/pull/1247)
 
 

library/sdxl_model_util.py

@@ -9,8 +9,10 @@ from diffusers import AutoencoderKL, EulerDiscreteScheduler, UNet2DConditionMode
 from library import model_util
 from library import sdxl_original_unet
 from .utils import setup_logging
+
 setup_logging()
 import logging
+
 logger = logging.getLogger(__name__)
 
 VAE_SCALE_FACTOR = 0.13025
@@ -171,8 +173,8 @@ def load_models_from_sdxl_checkpoint(model_version, ckpt_path, map_location, dty
     # Load the state dict
     if model_util.is_safetensors(ckpt_path):
         checkpoint = None
-        if(disable_mmap):
+        if disable_mmap:
-            state_dict = safetensors.torch.load(open(ckpt_path, 'rb').read())
+            state_dict = safetensors.torch.load(open(ckpt_path, "rb").read())
         else:
             try:
                 state_dict = load_file(ckpt_path, device=map_location)

library/sdxl_train_util.py

@@ -5,6 +5,7 @@ from typing import Optional
 
 import torch
 from library.device_utils import init_ipex, clean_memory_on_device
+
 init_ipex()
 
 from accelerate import init_empty_weights
@@ -13,8 +14,10 @@ from transformers import CLIPTokenizer
 from library import model_util, sdxl_model_util, train_util, sdxl_original_unet
 from library.sdxl_lpw_stable_diffusion import SdxlStableDiffusionLongPromptWeightingPipeline
 from .utils import setup_logging
+
 setup_logging()
 import logging
+
 logger = logging.getLogger(__name__)
 
 TOKENIZER1_PATH = "openai/clip-vit-large-patch14"
@@ -44,7 +47,7 @@ def load_target_model(args, accelerator, model_version: str, weight_dtype):
                 weight_dtype,
                 accelerator.device if args.lowram else "cpu",
                 model_dtype,
-                args.disable_mmap_load_safetensors
+                args.disable_mmap_load_safetensors,
             )
 
             # work on low-ram device
@@ -336,6 +339,7 @@ def add_sdxl_training_arguments(parser: argparse.ArgumentParser):
     parser.add_argument(
         "--disable_mmap_load_safetensors",
         action="store_true",
+        help="disable mmap load for safetensors. Speed up model loading in WSL environment / safetensorsのmmapロードを無効にする。WSL環境等でモデル読み込みを高速化できる",
     )