From 9e65b9dd432a26aa57fc2c1409ed6b96bf649041 Mon Sep 17 00:00:00 2001
From: smthem <2102561739@qq.com>
Date: Fri, 11 Apr 2025 11:32:17 +0800
Subject: [PATCH] init
---
PuLID/flux/util.py | 6 +-
README.md | 35 +-
Storydiffusion_node.py | 197 +-
UNO/LICENSE | 201 +
UNO/README.md | 131 +
UNO/app.py | 148 +
UNO/assets/cat_cafe.png | Bin 0 -> 1029762 bytes
UNO/assets/clock.png | Bin 0 -> 1897795 bytes
UNO/assets/crystal_ball.png | Bin 0 -> 1035058 bytes
UNO/assets/cup.png | Bin 0 -> 377963 bytes
UNO/assets/examples/0_0.json | 22 +
UNO/assets/examples/0_0.png | Bin 0 -> 501883 bytes
UNO/assets/examples/1_0.json | 23 +
UNO/assets/examples/1_0.png | Bin 0 -> 583130 bytes
UNO/assets/examples/1one2one/config.json | 8 +
UNO/assets/examples/1one2one/ref1.jpg | Bin 0 -> 139451 bytes
UNO/assets/examples/1one2one/result.png | Bin 0 -> 287640 bytes
UNO/assets/examples/2_0.json | 23 +
UNO/assets/examples/2_0.png | Bin 0 -> 304299 bytes
UNO/assets/examples/2one2one/config.json | 8 +
UNO/assets/examples/2one2one/ref1.png | Bin 0 -> 1006887 bytes
UNO/assets/examples/2one2one/result.png | Bin 0 -> 345627 bytes
UNO/assets/examples/3two2one/config.json | 9 +
UNO/assets/examples/3two2one/ref1.png | Bin 0 -> 760957 bytes
UNO/assets/examples/3two2one/ref2.png | Bin 0 -> 1035058 bytes
UNO/assets/examples/3two2one/result.png | Bin 0 -> 292270 bytes
UNO/assets/examples/4two2one/config.json | 9 +
UNO/assets/examples/4two2one/ref1.png | Bin 0 -> 1029762 bytes
UNO/assets/examples/4two2one/ref2.png | Bin 0 -> 377963 bytes
UNO/assets/examples/4two2one/result.png | Bin 0 -> 165940 bytes
UNO/assets/examples/5many2one/config.json | 10 +
UNO/assets/examples/5many2one/ref1.png | Bin 0 -> 1102955 bytes
UNO/assets/examples/5many2one/ref2.png | Bin 0 -> 725600 bytes
UNO/assets/examples/5many2one/ref3.png | Bin 0 -> 714736 bytes
UNO/assets/examples/5many2one/result.png | Bin 0 -> 358515 bytes
UNO/assets/examples/6t2i/config.json | 7 +
UNO/assets/examples/6t2i/result.png | Bin 0 -> 326423 bytes
UNO/assets/figurine.png | Bin 0 -> 760957 bytes
UNO/assets/logo.png | Bin 0 -> 57492 bytes
UNO/assets/simplecase.jpeg | Bin 0 -> 646283 bytes
UNO/assets/simplecase.jpg | Bin 0 -> 646283 bytes
UNO/assets/teaser.jpg | Bin 0 -> 708037 bytes
UNO/config/deepspeed/zero2_config.json | 15 +
UNO/config/deepspeed/zero3_config.json | 33 +
UNO/datasets/dreambench_multiip.json | 5252 +++++++++++++++++
UNO/datasets/dreambench_singleip.json | 4502 ++++++++++++++
UNO/inference.py | 125 +
UNO/requirements.txt | 12 +
UNO/train.py | 482 ++
UNO/uno/dataset/uno.py | 132 +
UNO/uno/flux/math.py | 45 +
UNO/uno/flux/model.py | 223 +
UNO/uno/flux/modules/autoencoder.py | 327 +
UNO/uno/flux/modules/conditioner.py | 53 +
UNO/uno/flux/modules/layers.py | 436 ++
UNO/uno/flux/pipeline.py | 293 +
UNO/uno/flux/pipeline_origin.py | 304 +
UNO/uno/flux/sampling.py | 388 ++
UNO/uno/flux/util.py | 495 ++
UNO/uno/utils/convert_yaml_to_args_file.py | 34 +
.../flux_dev_quantization_map.json | 1 +
images/UNO.jpg | Bin 0 -> 478036 bytes
.../__pycache__/__init__.cpython-311.pyc | Bin 195 -> 0 bytes
...atglm_256_ipadapter_FaceID.cpython-311.pyc | Bin 48997 -> 0 bytes
model_loader_utils.py | 58 +
example.json => storydiffusion_workflow.json | 1449 +++--
utils/pipeline_img.py | 4 +-
67 files changed, 15046 insertions(+), 454 deletions(-)
create mode 100644 UNO/LICENSE
create mode 100644 UNO/README.md
create mode 100644 UNO/app.py
create mode 100644 UNO/assets/cat_cafe.png
create mode 100644 UNO/assets/clock.png
create mode 100644 UNO/assets/crystal_ball.png
create mode 100644 UNO/assets/cup.png
create mode 100644 UNO/assets/examples/0_0.json
create mode 100644 UNO/assets/examples/0_0.png
create mode 100644 UNO/assets/examples/1_0.json
create mode 100644 UNO/assets/examples/1_0.png
create mode 100644 UNO/assets/examples/1one2one/config.json
create mode 100644 UNO/assets/examples/1one2one/ref1.jpg
create mode 100644 UNO/assets/examples/1one2one/result.png
create mode 100644 UNO/assets/examples/2_0.json
create mode 100644 UNO/assets/examples/2_0.png
create mode 100644 UNO/assets/examples/2one2one/config.json
create mode 100644 UNO/assets/examples/2one2one/ref1.png
create mode 100644 UNO/assets/examples/2one2one/result.png
create mode 100644 UNO/assets/examples/3two2one/config.json
create mode 100644 UNO/assets/examples/3two2one/ref1.png
create mode 100644 UNO/assets/examples/3two2one/ref2.png
create mode 100644 UNO/assets/examples/3two2one/result.png
create mode 100644 UNO/assets/examples/4two2one/config.json
create mode 100644 UNO/assets/examples/4two2one/ref1.png
create mode 100644 UNO/assets/examples/4two2one/ref2.png
create mode 100644 UNO/assets/examples/4two2one/result.png
create mode 100644 UNO/assets/examples/5many2one/config.json
create mode 100644 UNO/assets/examples/5many2one/ref1.png
create mode 100644 UNO/assets/examples/5many2one/ref2.png
create mode 100644 UNO/assets/examples/5many2one/ref3.png
create mode 100644 UNO/assets/examples/5many2one/result.png
create mode 100644 UNO/assets/examples/6t2i/config.json
create mode 100644 UNO/assets/examples/6t2i/result.png
create mode 100644 UNO/assets/figurine.png
create mode 100644 UNO/assets/logo.png
create mode 100644 UNO/assets/simplecase.jpeg
create mode 100644 UNO/assets/simplecase.jpg
create mode 100644 UNO/assets/teaser.jpg
create mode 100644 UNO/config/deepspeed/zero2_config.json
create mode 100644 UNO/config/deepspeed/zero3_config.json
create mode 100644 UNO/datasets/dreambench_multiip.json
create mode 100644 UNO/datasets/dreambench_singleip.json
create mode 100644 UNO/inference.py
create mode 100644 UNO/requirements.txt
create mode 100644 UNO/train.py
create mode 100644 UNO/uno/dataset/uno.py
create mode 100644 UNO/uno/flux/math.py
create mode 100644 UNO/uno/flux/model.py
create mode 100644 UNO/uno/flux/modules/autoencoder.py
create mode 100644 UNO/uno/flux/modules/conditioner.py
create mode 100644 UNO/uno/flux/modules/layers.py
create mode 100644 UNO/uno/flux/pipeline.py
create mode 100644 UNO/uno/flux/pipeline_origin.py
create mode 100644 UNO/uno/flux/sampling.py
create mode 100644 UNO/uno/flux/util.py
create mode 100644 UNO/uno/utils/convert_yaml_to_args_file.py
create mode 100644 config/flux-dev-fp8/flux_dev_quantization_map.json
create mode 100644 images/UNO.jpg
delete mode 100644 kolors/pipelines/__pycache__/__init__.cpython-311.pyc
delete mode 100644 kolors/pipelines/__pycache__/pipeline_stable_diffusion_xl_chatglm_256_ipadapter_FaceID.cpython-311.pyc
rename example.json => storydiffusion_workflow.json (92%)
diff --git a/PuLID/flux/util.py b/PuLID/flux/util.py
index 1e08edb..1968c38 100644
--- a/PuLID/flux/util.py
+++ b/PuLID/flux/util.py
@@ -201,9 +201,9 @@ def load_flow_model_quintized(name: str, ckpt_path,quantized_mode,aggressive_off
import logging
sd= load_torch_file(ckpt_path)
diffusion_model_prefix = model_detection.unet_prefix_from_state_dict(sd)
- weight_dtype = weight_dtype( sd, diffusion_model_prefix)
- logging.info("loading weight_dtype is {}".format(weight_dtype))
- if weight_dtype is None:
+ weight_dtype_ = weight_dtype( sd, diffusion_model_prefix)
+ logging.info("loading weight_dtype is {}".format(weight_dtype_))
+ if weight_dtype_ is None:
if str(torch.__version__).split(".")[-1]=="4.1" and "xlab" in ckpt_path.lower():#"XLabs-AI/flux-dev-fp8" need high env torch 24.1
json_path = os.path.join(folder_paths.base_path, "custom_nodes", "ComfyUI_StoryDiffusion", "utils","flux_dev_quantization_map.json")
else: # Kijai/flux-fp8,Shakker-Labs/AWPortrait-FL
diff --git a/README.md b/README.md
index 50873d9..85455e4 100644
--- a/README.md
+++ b/README.md
@@ -3,14 +3,18 @@
Using StoryDiffusion and other methods to make storys in ComfyUI
----
-* Origin from [StoryDiffusion](https://github.com/HVision-NKU/StoryDiffusion) [MS-Diffusion](https://github.com/MS-Diffusion/MS-Diffusion),[StoryMaker](https://github.com/RedAIGC/StoryMaker)，[Consistory](https://github.com/NVlabs/consistory),[Kolor](https://github.com/Kwai-Kolors/Kolors),[Pulid](https://github.com/ToTheBeginning/PuLID),[Flux](https://github.com/black-forest-labs/flux),[photomaker](https://github.com/TencentARC/PhotoMaker),[IP-Adapter](https://github.com/tencent-ailab/IP-Adapter) [InfiniteYou](https://github.com/bytedance/InfiniteYou)
+* Origin from [StoryDiffusion](https://github.com/HVision-NKU/StoryDiffusion) [MS-Diffusion](https://github.com/MS-Diffusion/MS-Diffusion),[StoryMaker](https://github.com/RedAIGC/StoryMaker)，[Consistory](https://github.com/NVlabs/consistory),[Kolor](https://github.com/Kwai-Kolors/Kolors),[Pulid](https://github.com/ToTheBeginning/PuLID),[Flux](https://github.com/black-forest-labs/flux),[photomaker](https://github.com/TencentARC/PhotoMaker),[IP-Adapter](https://github.com/tencent-ailab/IP-Adapter), [InfiniteYou](https://github.com/bytedance/InfiniteYou), [UNO](https://github.com/bytedance/UNO)
+
## Updates:
-* 2025/04/06
+* 2025/04/11
+* Add UNO support，Only the single FLUX model (27G) and UNO's Lora are needed. Please enable FP8 quantization and use storydiffusionw_flowjson workflow testing ，fix a bug，
+* 新增UNO支持，只需要单体FLUX模型(27G)和UNO的lora，请开启fp8量化和使用storydiffusion_workflow.json工作流测试,修复tokens过长的bug;
+
+* previous
* Add infinite svdq v0.2 support,it'work well when your svdq update v0.2，[download wheel](https://huggingface.co/mit-han-lab/nunchaku/tree/main) 更新 svdq v0.2的支持，infinite工作正常，[轮子](https://huggingface.co/mit-han-lab/nunchaku/tree/main)下载地址。
-* 2025/04/05
-* 1.修改了模型加载的流程，更新到V2版本，如果你喜欢旧的，可以下载V1.0版本的,2.请使用example.json，它集成了主要的工作流;3.剔除掉一些过时的功能;
-* 1.Modified the model loading process.Update to V2 version, If you like the old one, you can download version 1.0，2.Please use 'example.json', which integrates the main workflow，3.Remove some outdated features;
+* 1.修改了模型加载的流程，更新到V2版本，如果你喜欢旧的，可以下载V1.0版本的,2.请使用storydiffusion_workflow.json，它集成了主要的工作流;3.剔除掉一些过时的功能;
+* 1.Modified the model loading process.Update to V2 version, If you like the old one, you can download version 1.0，2.Please use 'storydiffusion_workflow.json', which integrates the main workflow，3.Remove some outdated features;
1.Installation
@@ -165,6 +169,13 @@ or
download svdquant repo from [here](https://huggingface.co/mit-han-lab/svdq-fp4-flux.1-dev/tree/main) and fill local path in 'easyfunction_lite' node's 'select_method'
+**3.7 UNO mode**
+download lora [dit_lora.safetensor](https://huggingface.co/bytedance-research/UNO/tree/main),use fp8,if Vram <24.
+```
+├── ComfyUI/models/
+| ├── diffusion_models/flux1-dev.safetensors #
+| ├── loras/dit_lora.safetensors #
+```
4 Example
----
@@ -206,7 +217,10 @@ or
* svdq，升级到v.2工作正常
-**4.8 comfyUI classic（comfyUI经典模式，可以接任意适配CF的流程，主要是方便使用多角色的clip）**
+**4.8 UNO**
+ 
+
+**4.9 comfyUI classic（comfyUI经典模式，可以接任意适配CF的流程，主要是方便使用多角色的clip）**
* any mode SD1.5 SDXL SD3.5 FLUX...
@@ -306,3 +320,12 @@ svdquant
```
[GGUF](https://github.com/city96/ComfyUI-GGUF)
[FLUX LICENSE](https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)
+
+```
+@article{wu2025less,
+ title={Less-to-More Generalization: Unlocking More Controllability by In-Context Generation},
+ author={Wu, Shaojin and Huang, Mengqi and Wu, Wenxu and Cheng, Yufeng and Ding, Fei and He, Qian},
+ journal={arXiv preprint arXiv:2504.02160},
+ year={2025}
+}
+```
\ No newline at end of file
diff --git a/Storydiffusion_node.py b/Storydiffusion_node.py
index dbbac96..0401d7a 100644
--- a/Storydiffusion_node.py
+++ b/Storydiffusion_node.py
@@ -11,9 +11,10 @@ import folder_paths
from comfy.model_management import total_vram
import comfy
import latent_preview
+import torchvision.transforms.functional as TVF
from .utils.utils import get_comic
from .model_loader_utils import (phi2narry,Loader_storydiffusion,Loader_Flux_Pulid,load_pipeline_consistory,Loader_KOLOR,replicate_data_by_indices,glm_single_encode,
- get_float,gc_cleanup,tensor_to_image,photomaker_clip,tensortopil_list,tensortopil_list_upscale,
+ get_float,gc_cleanup,tensor_to_image,photomaker_clip,Loader_UNO,tensortopil_list_upscale,tensortopil_list,
narry_list_pil,pre_text2infer,cf_clip,get_phrases_idx_cf,get_eot_idx_cf,get_ms_phrase_emb,get_extra_function,photomaker_clip_v2,adjust_indices,
get_scheduler,apply_style_positive,fitter_cf_model_type,Infer_MSdiffusion,Loader_story_maker,Loader_InfiniteYou,
nomarl_upscale,SAMPLER_NAMES,SCHEDULER_NAMES,lora_lightning_list)
@@ -412,8 +413,7 @@ class Pre_Translate_prompt:
def translate_prompt(self, scene_prompts, keep_character_name):
captions = scene_prompts.splitlines()
if not keep_character_name:
- captions = [caption.split(")", 1)[-1] if ")" in caption else caption
- for caption in captions] # del character
+ captions = [caption.split(")", 1)[-1] if ")" in caption else caption for caption in captions] # del character
else:
captions = [caption.replace("(", "").replace(")", "") if "(" or ")" in caption else caption
for caption in captions] # del ()
@@ -434,7 +434,7 @@ class EasyFunction_Lite:
def INPUT_TYPES(s):
return {"required": {"extra_repo": ("STRING", { "default": ""}),
"checkpoints": (["none"] + folder_paths.get_filename_list("diffusion_models")+folder_paths.get_filename_list("gguf")+folder_paths.get_filename_list("clip"),),
- "clip_vision": (["none"] + folder_paths.get_filename_list("clip_vision"),),
+ "clip_vision": (["none"] + folder_paths.get_filename_list("clip_vision")+folder_paths.get_filename_list("loras"),),
"function_mode": (["none","tag", "clip","mask","infinite",],),
"select_method":("STRING",{ "default":""}),
"temperature": (
@@ -448,7 +448,7 @@ class EasyFunction_Lite:
def easy_function_main(self, extra_repo,checkpoints,clip_vision,function_mode,select_method,temperature):
use_gguf,use_unet=False,False
- model_path,clip=None,None
+ model_path,clip,lora_path=None,None,None
if checkpoints != "none":
if checkpoints.endswith(".gguf"):
@@ -461,7 +461,12 @@ class EasyFunction_Lite:
use_unet=True
if clip_vision != "none":
- clip_vision_path = folder_paths.get_full_path("clip_vision", clip_vision)
+ if "clip_vision" in clip_vision:
+ clip_vision_path = folder_paths.get_full_path("clip_vision", clip_vision)
+ else:
+ lora_path = folder_paths.get_full_path("loras", clip_vision)
+ clip_vision_path = None
+
else:
clip_vision_path=None
if function_mode=="tag":
@@ -478,24 +483,27 @@ class EasyFunction_Lite:
use_svdq=True if "svdq" in select_method else False
- pipe={"model":model,"extra_repo":extra_repo,"model_path":model_path,"use_svdq":use_svdq,"use_gguf":use_gguf,"use_unet":use_unet,"extra_easy":function_mode,"select_method":select_method,"clip_vision_path":clip_vision_path}
+ pipe={"model":model,"extra_repo":extra_repo,"model_path":model_path,"use_svdq":use_svdq,"use_gguf":use_gguf,"lora_ckpt_path":lora_path,
+ "use_unet":use_unet,"extra_easy":function_mode,"select_method":select_method,"clip_vision_path":clip_vision_path}
return (pipe,clip)
class StoryDiffusion_Apply:
@classmethod
def INPUT_TYPES(s):
- return {"required": {
- "model": ("MODEL",),
- "vae": ("VAE",),
- "infer_mode": (["story", "classic","flux_pulid","infiniteyou","story_maker","story_and_maker","consistory","kolor_face","msdiffusion" ],),
- "photomake_ckpt": (["none"] + [i for i in folder_paths.get_filename_list("photomaker") if "v1" in i or "v2" in i],),
- "ipadapter_ckpt": (["none"] + folder_paths.get_filename_list("photomaker"),),
- "quantize_mode": ([ "fp8", "nf4","fp16", ],),
- "lora_scale": ("FLOAT", {"default": 0.8, "min": 0.1, "max": 1.0, "step": 0.1}),
- "extra_funtion":("STRING", {"default": ""}),
- },
- "optional":{"CLIP_VISION": ("CLIP_VISION",),
+ return {"required":
+ {
+ "model": ("MODEL",),
+ "vae": ("VAE",),
+ "infer_mode": (["story", "classic","flux_pulid","infiniteyou","uno","story_maker","story_and_maker","consistory","kolor_face","msdiffusion" ],),
+ "photomake_ckpt": (["none"] + [i for i in folder_paths.get_filename_list("photomaker") if "v1" in i or "v2" in i],),
+ "ipadapter_ckpt": (["none"] + folder_paths.get_filename_list("photomaker"),),
+ "quantize_mode": ([ "fp8", "nf4","fp16", ],),
+ "lora_scale": ("FLOAT", {"default": 0.8, "min": 0.1, "max": 1.0, "step": 0.1}),
+ "extra_funtion":("STRING", {"default": ""}),
+ },
+ "optional":{
+ "CLIP_VISION": ("CLIP_VISION",),
}
}
@@ -512,7 +520,7 @@ class StoryDiffusion_Apply:
CLIP_VISION=kwargs.get("CLIP_VISION")
-
+
unet_type=torch.float16 #use for sdxl
image_proj_model=None
@@ -526,6 +534,8 @@ class StoryDiffusion_Apply:
clip_vision_path=None
repo=None
+
+ save_quantezed=True if "save" in extra_funtion and quantize_mode=="fp8" else False
print(f"infer model type is {cf_model_type}")
@@ -543,27 +553,24 @@ class StoryDiffusion_Apply:
if infer_mode in ["story_maker" ,"story_and_maker"] and not CLIP_VISION and ipadapter_ckpt_path is None:
raise "story_maker need a clipvison H model,mask.bin"
-
+ # check vram only using in flux pulid or UNO
+ if total_vram > 45000.0:
+ aggressive_offload = False
+ offload = False
+ elif 17000.0 < total_vram < 45000.0:
+ aggressive_offload = False
+ offload = True
+ else:
+ aggressive_offload = True
+ offload = True
+
+ logging.info(f"total_vram is {total_vram},aggressive_offload is {aggressive_offload},offload is {offload}")
if infer_mode in["story", "story_and_maker","msdiffusion"]:# mix mode,use maker or ms to make 2 roles in on image
model = Loader_storydiffusion(model,photomake_ckpt_path,vae)
elif infer_mode =="story_maker":
model = Loader_story_maker(model,ipadapter_ckpt_path,vae,False,lora_scale)
elif infer_mode == "flux_pulid":
-
- # check vram only using in flux pulid
- if total_vram > 45000.0:
- aggressive_offload = False
- offload = False
- elif 17000.0 < total_vram < 45000.0:
- aggressive_offload = False
- offload = True
- else:
- aggressive_offload = True
- offload = True
-
- logging.info(f"total_vram is {total_vram},aggressive_offload is {aggressive_offload},offload is {offload}")
-
from .PuLID.app_flux import get_models
if isinstance(model,dict):
ckpt_path=model.get("model_path")
@@ -585,9 +592,10 @@ class StoryDiffusion_Apply:
if repo is None:
raise "EasyFunction_Lite node extra_repo must fill kolor repo"
model = Loader_KOLOR(repo,clip_vision_path,ipadapter_ckpt_path)
+ elif infer_mode == "uno":
+ model = Loader_UNO(model,offload,quantize_mode,save_quantezed,lora_rank=512)
else: # can not choice a mode
print("infer use comfyui classic mode")
-
story_img=True if photomake_ckpt_path and infer_mode in["story","story_maker","story_and_maker","msdiffusion"] else False
model_=model if infer_mode=="flux_pulid" or story_img else None
@@ -692,6 +700,22 @@ class StoryDiffusion_CLIPTextEncode:
role_key_list=replicate_data_by_indices(role_list, index_char_1_list, index_char_2_list)
role_key_list=[i for i in role_key_list if i is not None]
+ if len(role_list)>1: #重新整理prompt排序，便于后续emb和ID的对应
+ nc_dual_list=[]
+ for i in nc_indexs:
+ nc_dual_list.append(i)
+ for i in positions_index_dual:
+ nc_dual_list.append(i)
+ adjusted_a = adjust_indices(index_char_1_list, nc_dual_list)
+ adjusted_b = adjust_indices(index_char_2_list, nc_dual_list)
+ a_list = [only_role_list[i] for i in adjusted_a if 0 <= i < len(only_role_list)]
+ b_list = [only_role_list[i] for i in adjusted_b if 0 <= i < len(only_role_list)]
+
+ inf_list_split=[a_list,b_list]
+ else:
+ inf_list_split=only_role_list
+
+
# if prompts_dual and infer_mode not in ["story_and_maker","story_maker","msdiffusion"]:
# raise "only support prompts_dual role in story maker and msdiffusion"
@@ -710,10 +734,10 @@ class StoryDiffusion_CLIPTextEncode:
if infer_mode=="msdiffusion" and photomake_ckpt_path and not img2img_mode:
raise "if use msdiffusion txt2img mode,can not use photomake_ckpt_path"
+ uno_pe="d" #Literal['d', 'h', 'w', 'o']
+
+ only_role_emb_ne,x_1_refs_dict,image_emb=None,None,None
-
- only_role_emb_ne=None
- image_emb = None
if img2img_mode:
if infer_mode == "story_and_maker" or infer_mode == "story_maker":
k1, _, _, _ = image.size()
@@ -737,8 +761,53 @@ class StoryDiffusion_CLIPTextEncode:
pass
elif infer_mode == "infiniteyou":
pass
+ elif infer_mode == "uno":
+ from .UNO.uno.flux.sampling import get_noise
+ from .UNO.uno.flux.pipeline import preprocess_ref
+
+ ref_pil_list=tensortopil_list(image) #原方法需要图片缩放，先转回pil
+
+ if control_image is not None:
+ control_pil_list=tensortopil_list(control_image)
+ else:
+ control_pil_list=None
+
+ if control_pil_list is None:# 无控制图时,默认时单图模式,x_1_refs用key取值
+ x_1_refs_dict={}
+ for key ,role_pil,prompts in zip(role_list,ref_pil_list,inf_list_split):
+ role_tensor=phi2narry(preprocess_ref(role_pil, 512))#单图默认512
+ x_1_refs_dict[key]=[[vae.encode(role_tensor[:,:,:,:3]).to(torch.bfloat16)]*len(prompts)] # {key:[[1],[2],key2:[[3],[4]]}
+ else:
+ if len(control_pil_list) != len(only_role_list):
+ raise "control image must same size as prompts,多少句话就多少张图"
+ else:
+ if len(role_list)>1:
+ nc_dual_list=[]
+ for i in nc_indexs:
+ nc_dual_list.append(i)
+ for i in positions_index_dual:
+ nc_dual_list.append(i)
+ adjusted_a = adjust_indices(index_char_1_list, nc_dual_list)
+ adjusted_b = adjust_indices(index_char_2_list, nc_dual_list)
+ a_list = [control_pil_list[i] for i in adjusted_a if 0 <= i < len(control_pil_list)]
+ b_list = [control_pil_list[i] for i in adjusted_b if 0 <= i < len(control_pil_list)]
+
+ control_pil_list=[a_list,b_list]
+ else:
+ control_pil_list=[control_pil_list]
+ x_1_refs_dict={}
+ for key ,role_pil,control_pil in zip(role_list,ref_pil_list,control_pil_list):
+ mix_list=[]
+ role_tensor=phi2narry(preprocess_ref(role_pil, 320) )#多图默认320
+ for c in (control_pil):
+
+ c_tensor=phi2narry(preprocess_ref(c, 320))#多图默认320
+ mix_list.append([vae.encode(role_tensor[:,:,:,:3]).to(torch.bfloat16),vae.encode(c_tensor[:,:,:,:3]).to(torch.bfloat16)])
+ x_1_refs_dict[key]=mix_list #{key:[[1,2],[3,4]]}
else:
pass
+ else:
+ pass
# pre insight face model and emb
if img2img_mode:
@@ -775,23 +844,6 @@ class StoryDiffusion_CLIPTextEncode:
image_list = [nomarl_upscale(img, width, height) for img in img_list]
for index, key in enumerate(role_list):
input_id_images_dict[key]=image_list[index]
-
-
-
- if len(role_list)>1: #重新整理prompt排序，便于后续emb和ID的对应
- nc_dual_list=[]
- for i in nc_indexs:
- nc_dual_list.append(i)
- for i in positions_index_dual:
- nc_dual_list.append(i)
- adjusted_a = adjust_indices(index_char_1_list, nc_dual_list)
- adjusted_b = adjust_indices(index_char_2_list, nc_dual_list)
- a_list = [only_role_list[i] for i in adjusted_a if 0 <= i < len(only_role_list)]
- b_list = [only_role_list[i] for i in adjusted_b if 0 <= i < len(only_role_list)]
-
- inf_list_split=[a_list,b_list]
- else:
- inf_list_split=only_role_list
#print("inf_list_split",inf_list_split)
# pre clip txt emb
@@ -817,8 +869,9 @@ class StoryDiffusion_CLIPTextEncode:
for key ,prompts in zip(role_list,inf_list_split):
ip_emb,noise_,inp_n=[],[],[]
- seed_random = random.randint(0, MAX_SEED) #pulid的emb需要随机数
+
for p,n in zip(prompts,[neg_text]*len(prompts)):
+ seed_random = random.randint(0, MAX_SEED) #pulid 和uno的emb需要随机数
inp,inp_neg,x=get_emb_flux_pulid(t5_,clip_,if_repo,seed_random,p,n,width,height,num_steps=20,guidance=3.5,device=device)
ip_emb.append(inp)
inp_n.append(inp_neg)
@@ -826,6 +879,20 @@ class StoryDiffusion_CLIPTextEncode:
only_role_emb[key]=ip_emb
noise_x[key]=noise_
inp_neg_list[key]=inp_n
+ elif infer_mode == "uno":
+ only_role_emb={}
+ from .UNO.uno.flux.sampling import prepare_multi_ip_wrapper
+ for key ,prompts in zip(role_list,inf_list_split):
+ ip_emb=[]
+
+ for p,x_1 in zip(prompts,x_1_refs_dict[key]):
+ seed_random = random.randint(0, MAX_SEED) #pulid 和uno的emb需要随机数
+ uno_x = get_noise(1, height, width, device=device,dtype=torch.bfloat16, seed=seed_random)
+ inp = prepare_multi_ip_wrapper(clip,img=uno_x,prompt=p, ref_imgs=x_1, pe=uno_pe)
+ ip_emb.append(inp)
+ only_role_emb[key]=ip_emb
+
+
elif infer_mode=="kolor_face":
from .kolors.models.tokenization_chatglm import ChatGLMTokenizer
tokenizer = ChatGLMTokenizer.from_pretrained(os.path.join(switch.get("repo"),'text_encoder'))
@@ -927,6 +994,9 @@ class StoryDiffusion_CLIPTextEncode:
elif infer_mode=="flux_pulid":
postive_dict= {"role": only_role_emb, "nc": None, "daul": daul_emb} #不支持NC
negative = [inp_neg_list,noise_x]
+ elif infer_mode=="uno":
+ postive_dict= {"role": only_role_emb, "nc": None, "daul": daul_emb} #TODO
+ negative=None
elif infer_mode=="kolor_face":
postive_dict = {"role": only_role_emb, "nc": nc_emb, "daul": daul_emb}
negative = only_role_emb_ne[0]
@@ -1615,6 +1685,21 @@ class StoryDiffusion_KSampler:
out = {}
out["samples"] = torch.cat(samples_list, dim=0)
return (out,)
+ elif infer_mode=="uno":
+ samples_list = []
+ for key in role_list:
+ for index,emb in enumerate(only_role_emb[key]):
+ #print(emb)
+ samples = model(width=width,
+ height=height,guidance=cfg,
+ num_steps=steps,
+ inp_cond=emb,
+ ) # torch.Size([1, 4, 64, 64])
+
+ samples_list.append(samples)
+ out = {}
+ out["samples"] = torch.cat(samples_list, dim=0)
+ return (out,)
else: #none:
return
diff --git a/UNO/LICENSE b/UNO/LICENSE
new file mode 100644
index 0000000..f49a4e1
--- /dev/null
+++ b/UNO/LICENSE
@@ -0,0 +1,201 @@
+ Apache License
+ Version 2.0, January 2004
+ http://www.apache.org/licenses/
+
+ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+ 1. Definitions.
+
+ "License" shall mean the terms and conditions for use, reproduction,
+ and distribution as defined by Sections 1 through 9 of this document.
+
+ "Licensor" shall mean the copyright owner or entity authorized by
+ the copyright owner that is granting the License.
+
+ "Legal Entity" shall mean the union of the acting entity and all
+ other entities that control, are controlled by, or are under common
+ control with that entity. For the purposes of this definition,
+ "control" means (i) the power, direct or indirect, to cause the
+ direction or management of such entity, whether by contract or
+ otherwise, or (ii) ownership of fifty percent (50%) or more of the
+ outstanding shares, or (iii) beneficial ownership of such entity.
+
+ "You" (or "Your") shall mean an individual or Legal Entity
+ exercising permissions granted by this License.
+
+ "Source" form shall mean the preferred form for making modifications,
+ including but not limited to software source code, documentation
+ source, and configuration files.
+
+ "Object" form shall mean any form resulting from mechanical
+ transformation or translation of a Source form, including but
+ not limited to compiled object code, generated documentation,
+ and conversions to other media types.
+
+ "Work" shall mean the work of authorship, whether in Source or
+ Object form, made available under the License, as indicated by a
+ copyright notice that is included in or attached to the work
+ (an example is provided in the Appendix below).
+
+ "Derivative Works" shall mean any work, whether in Source or Object
+ form, that is based on (or derived from) the Work and for which the
+ editorial revisions, annotations, elaborations, or other modifications
+ represent, as a whole, an original work of authorship. For the purposes
+ of this License, Derivative Works shall not include works that remain
+ separable from, or merely link (or bind by name) to the interfaces of,
+ the Work and Derivative Works thereof.
+
+ "Contribution" shall mean any work of authorship, including
+ the original version of the Work and any modifications or additions
+ to that Work or Derivative Works thereof, that is intentionally
+ submitted to Licensor for inclusion in the Work by the copyright owner
+ or by an individual or Legal Entity authorized to submit on behalf of
+ the copyright owner. For the purposes of this definition, "submitted"
+ means any form of electronic, verbal, or written communication sent
+ to the Licensor or its representatives, including but not limited to
+ communication on electronic mailing lists, source code control systems,
+ and issue tracking systems that are managed by, or on behalf of, the
+ Licensor for the purpose of discussing and improving the Work, but
+ excluding communication that is conspicuously marked or otherwise
+ designated in writing by the copyright owner as "Not a Contribution."
+
+ "Contributor" shall mean Licensor and any individual or Legal Entity
+ on behalf of whom a Contribution has been received by Licensor and
+ subsequently incorporated within the Work.
+
+ 2. Grant of Copyright License. Subject to the terms and conditions of
+ this License, each Contributor hereby grants to You a perpetual,
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+ copyright license to reproduce, prepare Derivative Works of,
+ publicly display, publicly perform, sublicense, and distribute the
+ Work and such Derivative Works in Source or Object form.
+
+ 3. Grant of Patent License. Subject to the terms and conditions of
+ this License, each Contributor hereby grants to You a perpetual,
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+ (except as stated in this section) patent license to make, have made,
+ use, offer to sell, sell, import, and otherwise transfer the Work,
+ where such license applies only to those patent claims licensable
+ by such Contributor that are necessarily infringed by their
+ Contribution(s) alone or by combination of their Contribution(s)
+ with the Work to which such Contribution(s) was submitted. If You
+ institute patent litigation against any entity (including a
+ cross-claim or counterclaim in a lawsuit) alleging that the Work
+ or a Contribution incorporated within the Work constitutes direct
+ or contributory patent infringement, then any patent licenses
+ granted to You under this License for that Work shall terminate
+ as of the date such litigation is filed.
+
+ 4. Redistribution. You may reproduce and distribute copies of the
+ Work or Derivative Works thereof in any medium, with or without
+ modifications, and in Source or Object form, provided that You
+ meet the following conditions:
+
+ (a) You must give any other recipients of the Work or
+ Derivative Works a copy of this License; and
+
+ (b) You must cause any modified files to carry prominent notices
+ stating that You changed the files; and
+
+ (c) You must retain, in the Source form of any Derivative Works
+ that You distribute, all copyright, patent, trademark, and
+ attribution notices from the Source form of the Work,
+ excluding those notices that do not pertain to any part of
+ the Derivative Works; and
+
+ (d) If the Work includes a "NOTICE" text file as part of its
+ distribution, then any Derivative Works that You distribute must
+ include a readable copy of the attribution notices contained
+ within such NOTICE file, excluding those notices that do not
+ pertain to any part of the Derivative Works, in at least one
+ of the following places: within a NOTICE text file distributed
+ as part of the Derivative Works; within the Source form or
+ documentation, if provided along with the Derivative Works; or,
+ within a display generated by the Derivative Works, if and
+ wherever such third-party notices normally appear. The contents
+ of the NOTICE file are for informational purposes only and
+ do not modify the License. You may add Your own attribution
+ notices within Derivative Works that You distribute, alongside
+ or as an addendum to the NOTICE text from the Work, provided
+ that such additional attribution notices cannot be construed
+ as modifying the License.
+
+ You may add Your own copyright statement to Your modifications and
+ may provide additional or different license terms and conditions
+ for use, reproduction, or distribution of Your modifications, or
+ for any such Derivative Works as a whole, provided Your use,
+ reproduction, and distribution of the Work otherwise complies with
+ the conditions stated in this License.
+
+ 5. Submission of Contributions. Unless You explicitly state otherwise,
+ any Contribution intentionally submitted for inclusion in the Work
+ by You to the Licensor shall be under the terms and conditions of
+ this License, without any additional terms or conditions.
+ Notwithstanding the above, nothing herein shall supersede or modify
+ the terms of any separate license agreement you may have executed
+ with Licensor regarding such Contributions.
+
+ 6. Trademarks. This License does not grant permission to use the trade
+ names, trademarks, service marks, or product names of the Licensor,
+ except as required for reasonable and customary use in describing the
+ origin of the Work and reproducing the content of the NOTICE file.
+
+ 7. Disclaimer of Warranty. Unless required by applicable law or
+ agreed to in writing, Licensor provides the Work (and each
+ Contributor provides its Contributions) on an "AS IS" BASIS,
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+ implied, including, without limitation, any warranties or conditions
+ of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+ PARTICULAR PURPOSE. You are solely responsible for determining the
+ appropriateness of using or redistributing the Work and assume any
+ risks associated with Your exercise of permissions under this License.
+
+ 8. Limitation of Liability. In no event and under no legal theory,
+ whether in tort (including negligence), contract, or otherwise,
+ unless required by applicable law (such as deliberate and grossly
+ negligent acts) or agreed to in writing, shall any Contributor be
+ liable to You for damages, including any direct, indirect, special,
+ incidental, or consequential damages of any character arising as a
+ result of this License or out of the use or inability to use the
+ Work (including but not limited to damages for loss of goodwill,
+ work stoppage, computer failure or malfunction, or any and all
+ other commercial damages or losses), even if such Contributor
+ has been advised of the possibility of such damages.
+
+ 9. Accepting Warranty or Additional Liability. While redistributing
+ the Work or Derivative Works thereof, You may choose to offer,
+ and charge a fee for, acceptance of support, warranty, indemnity,
+ or other liability obligations and/or rights consistent with this
+ License. However, in accepting such obligations, You may act only
+ on Your own behalf and on Your sole responsibility, not on behalf
+ of any other Contributor, and only if You agree to indemnify,
+ defend, and hold each Contributor harmless for any liability
+ incurred by, or claims asserted against, such Contributor by reason
+ of your accepting any such warranty or additional liability.
+
+ END OF TERMS AND CONDITIONS
+
+ APPENDIX: How to apply the Apache License to your work.
+
+ To apply the Apache License to your work, attach the following
+ boilerplate notice, with the fields enclosed by brackets "[]"
+ replaced with your own identifying information. (Don't include
+ the brackets!) The text should be enclosed in the appropriate
+ comment syntax for the file format. We also recommend that a
+ file or class name and description of purpose be included on the
+ same "printed page" as the copyright notice for easier
+ identification within third-party archives.
+
+ Copyright [yyyy] [name of copyright owner]
+
+ Licensed under the Apache License, Version 2.0 (the "License");
+ you may not use this file except in compliance with the License.
+ You may obtain a copy of the License at
+
+ http://www.apache.org/licenses/LICENSE-2.0
+
+ Unless required by applicable law or agreed to in writing, software
+ distributed under the License is distributed on an "AS IS" BASIS,
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ See the License for the specific language governing permissions and
+ limitations under the License.
\ No newline at end of file
diff --git a/UNO/README.md b/UNO/README.md
new file mode 100644
index 0000000..05376cd
--- /dev/null
+++ b/UNO/README.md
@@ -0,0 +1,131 @@
+
+ 
+ Less-to-More Generalization: Unlocking More Controllability by In-Context Generation
+
+
+
+
+
+
+
+
+
+> Shaojin Wu, Mengqi Huang*, Wenxu Wu, Yufeng Cheng, Fei Ding+, Qian He
+>Intelligent Creation Team, ByteDance
+
+
+
+
+
+## 🔥 News
+- [04/2025] 🔥 The [demo](https://huggingface.co/spaces/bytedance-research/UNO-FLUX) of UNO is released.
+- [04/2025] 🔥 The [training code](https://github.com/bytedance/UNO), [inference code](https://github.com/bytedance/UNO), and [model](https://huggingface.co/bytedance-research/UNO) of UNO are released.
+- [04/2025] 🔥 The [project page](https://bytedance.github.io/UNO) of UNO is created.
+- [04/2025] 🔥 The arXiv [paper](https://arxiv.org/abs/2504.02160) of UNO is released.
+
+## 📖 Introduction
+In this study, we propose a highly-consistent data synthesis pipeline to tackle this challenge. This pipeline harnesses the intrinsic in-context generation capabilities of diffusion transformers and generates high-consistency multi-subject paired data. Additionally, we introduce UNO, which consists of progressive cross-modal alignment and universal rotary position embedding. It is a multi-image conditioned subject-to-image model iteratively trained from a text-to-image model. Extensive experiments show that our method can achieve high consistency while ensuring controllability in both single-subject and multi-subject driven generation.
+
+
+## ⚡️ Quick Start
+
+### 🔧 Requirements and Installation
+
+Install the requirements
+```bash
+## create a virtual environment with python >= 3.10 <= 3.12, like
+# python -m venv uno_env
+# source uno_env/bin/activate
+# then install
+pip install -r requirements.txt
+```
+
+then download checkpoints in one of the three ways:
+1. Directly run the inference scripts, the checkpoints will be downloaded automatically by the `hf_hub_download` function in the code to your `$HF_HOME`(the default value is `~/.cache/huggingface`).
+2. use `huggingface-cli download ` to download `black-forest-labs/FLUX.1-dev`, `xlabs-ai/xflux_text_encoders`, `openai/clip-vit-large-patch14`, `bytedance-research/UNO`, then run the inference scripts.
+3. use `huggingface-cli download --local-dir ` to download all the checkpoints menthioned in 2. to the directories your want. Then set the environment variable `AE`, `FLUX`, `T5`, `CLIP`, `LORA` to the corresponding paths. Finally, run the inference scripts.
+
+### 🌟 Gradio Demo
+
+```bash
+python app.py
+```
+
+
+### ✍️ Inference
+Start from the examples below to explore and spark your creativity. ✨
+```bash
+python inference.py --prompt "A clock on the beach is under a red sun umbrella" --image_paths "assets/clock.png" --width 704 --height 704
+python inference.py --prompt "The figurine is in the crystal ball" --image_paths "assets/figurine.png" "assets/crystal_ball.png" --width 704 --height 704
+python inference.py --prompt "The logo is printed on the cup" --image_paths "assets/cat_cafe.png" "assets/cup.png" --width 704 --height 704
+```
+
+Optional prepreration: If you want to test the inference on dreambench at the first time, you should clone the submodule `dreambench` to download the dataset.
+
+```bash
+git submodule update --init
+```
+Then running the following scripts:
+```bash
+# evaluated on dreambench
+## for single-subject
+python inference.py --eval_json_path ./datasets/dreambench_singleip.json
+## for multi-subject
+python inference.py --eval_json_path ./datasets/dreambench_multiip.json
+```
+
+
+
+### 🚄 Training
+
+```bash
+accelerate launch train.py
+```
+
+
+### 📌 Tips and Notes
+We integrate single-subject and multi-subject generation within a unified model. For single-subject scenarios, the longest side of the reference image is set to 512 by default, while for multi-subject scenarios, it is set to 320. UNO demonstrates remarkable flexibility across various aspect ratios, thanks to its training on a multi-scale dataset. Despite being trained within 512 buckets, it can handle higher resolutions, including 512, 568, and 704, among others.
+
+UNO excels in subject-driven generation but has room for improvement in generalization due to dataset constraints. We are actively developing an enhanced model—stay tuned for updates. Your feedback is valuable, so please feel free to share any suggestions.
+
+## 🎨 Application Scenarios
+
+
+
+
+## 📄 Disclaimer
+
+We open-source this project for academic research. The vast majority of images
+used in this project are either generated or licensed. If you have any concerns,
+please contact us, and we will promptly remove any inappropriate content.
+Our code is released under the Apache 2.0 License,, while our models are under
+the CC BY-NC 4.0 License. Any models related to FLUX.1-dev
+base model must adhere to the original licensing terms.
+
This research aims to advance the field of generative AI. Users are free to
+create images using this tool, provided they comply with local laws and exercise
+responsible usage. The developers are not liable for any misuse of the tool by users.
+
+## 🚀 Updates
+For the purpose of fostering research and the open-source community, we plan to open-source the entire project, encompassing training, inference, weights, etc. Thank you for your patience and support! 🌟
+- [x] Release github repo.
+- [x] Release inference code.
+- [x] Release training code.
+- [x] Release model checkpoints.
+- [x] Release arXiv paper.
+- [x] Release huggingface space demo.
+- [ ] Release in-context data generation pipelines.
+
+## Citation
+If UNO is helpful, please help to ⭐ the repo.
+
+If you find this project useful for your research, please consider citing our paper:
+```bibtex
+@article{wu2025less,
+ title={Less-to-More Generalization: Unlocking More Controllability by In-Context Generation},
+ author={Wu, Shaojin and Huang, Mengqi and Wu, Wenxu and Cheng, Yufeng and Ding, Fei and He, Qian},
+ journal={arXiv preprint arXiv:2504.02160},
+ year={2025}
+}
+```
\ No newline at end of file
diff --git a/UNO/app.py b/UNO/app.py
new file mode 100644
index 0000000..e854859
--- /dev/null
+++ b/UNO/app.py
@@ -0,0 +1,148 @@
+# Copyright (c) 2025 Bytedance Ltd. and/or its affiliates. All rights reserved.
+
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+
+# http://www.apache.org/licenses/LICENSE-2.0
+
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import dataclasses
+import json
+from pathlib import Path
+
+import gradio as gr
+import torch
+
+from .uno.flux.pipeline import UNOPipeline
+
+
+def get_examples(examples_dir: str = "assets/examples") -> list:
+ examples = Path(examples_dir)
+ ans = []
+ for example in examples.iterdir():
+ if not example.is_dir():
+ continue
+ with open(example / "config.json") as f:
+ example_dict = json.load(f)
+
+
+ example_list = []
+
+ example_list.append(example_dict["useage"]) # case for
+ example_list.append(example_dict["prompt"]) # prompt
+
+ for key in ["image_ref1", "image_ref2", "image_ref3", "image_ref4"]:
+ if key in example_dict:
+ example_list.append(str(example / example_dict[key]))
+ else:
+ example_list.append(None)
+
+ example_list.append(example_dict["seed"])
+
+ ans.append(example_list)
+ return ans
+
+
+def create_demo(
+ model_type: str,
+ device: str = "cuda" if torch.cuda.is_available() else "cpu",
+ offload: bool = False,
+):
+ pipeline = UNOPipeline(model_type, device, offload, only_lora=True, lora_rank=512)
+
+ badges_text = r"""
+
+ """.strip()
+
+ with gr.Blocks() as demo:
+ gr.Markdown(f"# UNO by UNO team")
+ gr.Markdown(badges_text)
+ with gr.Row():
+ with gr.Column():
+ prompt = gr.Textbox(label="Prompt", value="handsome woman in the city")
+ with gr.Row():
+ image_prompt1 = gr.Image(label="Ref Img1", visible=True, interactive=True, type="pil")
+ image_prompt2 = gr.Image(label="Ref Img2", visible=True, interactive=True, type="pil")
+ image_prompt3 = gr.Image(label="Ref Img3", visible=True, interactive=True, type="pil")
+ image_prompt4 = gr.Image(label="Ref img4", visible=True, interactive=True, type="pil")
+
+ with gr.Row():
+ with gr.Column():
+ width = gr.Slider(512, 2048, 512, step=16, label="Gneration Width")
+ height = gr.Slider(512, 2048, 512, step=16, label="Gneration Height")
+ with gr.Column():
+ gr.Markdown("📌 The model trained on 512x512 resolution.\n")
+ gr.Markdown(
+ "The size closer to 512 is more stable,"
+ " and the higher size gives a better visual effect but is less stable"
+ )
+
+ with gr.Accordion("Advanced Options", open=False):
+ with gr.Row():
+ num_steps = gr.Slider(1, 50, 25, step=1, label="Number of steps")
+ guidance = gr.Slider(1.0, 5.0, 4.0, step=0.1, label="Guidance", interactive=True)
+ seed = gr.Number(-1, label="Seed (-1 for random)")
+
+ generate_btn = gr.Button("Generate")
+
+ with gr.Column():
+ output_image = gr.Image(label="Generated Image")
+ download_btn = gr.File(label="Download full-resolution", type="filepath", interactive=False)
+
+
+ inputs = [
+ prompt, width, height, guidance, num_steps,
+ seed, image_prompt1, image_prompt2, image_prompt3, image_prompt4
+ ]
+ generate_btn.click(
+ fn=pipeline.gradio_generate,
+ inputs=inputs,
+ outputs=[output_image, download_btn],
+ )
+
+ example_text = gr.Text("", visible=False, label="Case For:")
+ examples = get_examples("./assets/examples")
+
+ gr.Examples(
+ examples=examples,
+ inputs=[
+ example_text, prompt,
+ image_prompt1, image_prompt2, image_prompt3, image_prompt4,
+ seed, output_image
+ ],
+ )
+
+ return demo
+
+if __name__ == "__main__":
+ from typing import Literal
+
+ from transformers import HfArgumentParser
+
+ @dataclasses.dataclass
+ class AppArgs:
+ name: Literal["flux-dev", "flux-dev-fp8", "flux-schnell"] = "flux-dev"
+ device: Literal["cuda", "cpu"] = "cuda" if torch.cuda.is_available() else "cpu"
+ offload: bool = dataclasses.field(
+ default=False,
+ metadata={"help": "If True, sequantial offload the models(ae, dit, text encoder) to CPU if not used."}
+ )
+ port: int = 7860
+
+ parser = HfArgumentParser([AppArgs])
+ args_tuple = parser.parse_args_into_dataclasses() # type: tuple[AppArgs]
+ args = args_tuple[0]
+
+ demo = create_demo(args.name, args.device, args.offload)
+ demo.launch(server_port=args.port)
diff --git a/UNO/assets/cat_cafe.png b/UNO/assets/cat_cafe.png
new file mode 100644
index 0000000000000000000000000000000000000000..287fe789de9047b72979b1d4f5c8fe13551c3b70
GIT binary patch
literal 1029762
zcmd3NbC4&^)@Iw7wr!i!wr$(CZBN^_?P>kmwr#t6+P(AMd-wZxV|V}Gs;D^UWIoS9
zWkyC+RaT^eoH#rTHVhCD5WJ*>h!PMG=)W!~kiRXvA{GjN1+cS{xDZh7H16470aX7qo7$N=8M;{7+x-K#GOJF4S^#v-ia}fjr0+F*+(Qwg_mEktFx1~2U
zu{ScM_po*NCvhMi5AMIJt*MJ4p@*%Foin!wFY&(++<*0dni+@*{{?Ze<|WpURUj0$
zcQPepr)Q*RB<6!5BqZc_cV(39<=S=c%C;!urh^e!&lcmEyS6i+p#9VHZ`!}=`gb^TK#HY;SA(m&Et~*_7a6_&+lIPxd?v|J?EaWd!~`y#H$b
zJHC7{JPiLeSNLEuA_dxkfCPXfMFdqmfK|34jSfeD*=
zLEYTu>FnwGI`>(4y|_5=iqlea_4Rc#X=VwwCDTa=-gRm^ta6cv=}bd1o`+R;BXy_KqH3
zS$TgydG&EC*z`nPN%;PFeYhj4xVgF6^Y{0^*}w3`f;iYWj_d0B`Mtfp?dL}dK}u7k
zhWKZtvWKruCr|D~zt@j<`@6k!uJiaN`1$x5g+D?4%QNic^Ha5IVQXq)V&c_jJ`xhOP4nc#LkudNx)|W&+GHRz8o?-7g=>G?%>!8~
zDG?^g+&PTDjazxz&$S`={?^ZJN2Cb*a7Cpyhii9m^?CJhcX#(^#~+dlE?#dCD8G-)*=dlwG
zz#`+P=SvnH=Fs!Imo{uE_@V@Sm5s+>cV$*{X>04Zb#{O|1n80x#O>YJ(4?KE^PtIl
zRk53oPXm(NCWwQ@*X>69NtyaedC2+sfsgl-B4#N)ciiLC(*;d;-pPXH>-UFGNgzlT
z;to&H`~Hjl4Ue)6*}E+LUO)(y|Hq5IJqE1ZB(=B;a7ZfWruEmIuHJiGVz5JKy1$>^
z!&vnhA?)s5^ZVsvF8=nm9%qb@o%H;J@bm#tU8>lx?N*m4edpT
zZ>V^%6Q9FjCHRYgWG)@j?id$ch+Tdh?8>Im3-UnWwp}I?-Ms0>K{QEwJ6H}y_CiUN
zO%ev|qrs1rpNJjYdDo;=q5EOv`a3(P{3AIkp8oD?S&8VEgFQbVlWO)yO};x9NnPT31*3gtS)!xa=2&!J+f?dB#`#
zB@i_L9$oy-cg?;{<@1;?>T!RTz~`%GHa0MiPoxfis9Mw@GasQb#cwe9{&|OF+uqI@
zTHgYE*7<*1Gw+47zn@HVD0L5o4Pf##2>?q7;
z2FluYid}f>vI*$d5lUL9kZ=qfrA9}d{UPw?2}hs2bo`|2o?Xe$X`#mi-Biv$nG;*v
zv%Qmh8zA|)xKM|9;wsS)b4q%{3JTiX+zT>J*YZZGwFkV5-Z%c_tH_@f)!Nj~A_N3s
z{=qn@it5Ow9u)37K&R3B@;6P1sd$$mvC>(^XkV
zFQFojk|vf)!pkPBuW_ZjEVK8(6Jq1aAsd66~a_HKk
zlo${ru8|?=po(nm?K!+A_=9+++{Sf=qnNO|ILme(ZHOxq+f}uptU?Mru45+F267(|
zpB?uw8gKB^2|3a>$l)%;dbJ96Pk>|s862VsxlQx?>R|8MMx^g*`-t|LS=gf8Sc0O1
zjqu>>4HU0K&MUlT!UraAZZt6%WvZ6HP8Kf){M7w(d=&$jgzIFc)Y#&8mm4&$ze>io
zcm@#^-3|2>OrjP4c`H)DuX$zPaLh52+Mby*YaL%~FM~QF>!!5fie~H@=XXmz)bPOGCbqwdEls{mM4aSH90dLTp}TOgZi<(=Ph4J}aAG$Jqf^cO9&%ua9HC8?MAA
zB3*%Se$??OtyU5#KH)sP?cD_OQRXZ+<$L$NEdU)M83fIrw>Ll@LpVjJ-~Lz=soa!P
zwQ_<{;A!q-Ks&s4eslo8ot5=7>^PP#FXS^_nEc?Wg*37@B6w4A0QE49aKCYv%&%o?
zzq|}TeRA`Cxw_(VF26RIk%7Rz8u5XeAAUAJgCL{;gCeGMg_>>e`$IurhRch0$fXj|
z3~A&UROlGY8r*Bi6;RJpP@rYXFh>?3fo13hio1Zb
zg@TXcW6~rifh7V}-`@T%399YJ{RB4}q-DmtQKDlk<05H9DtGScu>E2sje&RPbBi~M
zVTJ+mPdv+7CzmEHZWHpAkN+Pt*S)Xl(`OYM)Qn&s%m6-TxUMX*Q1p*P6NvfXuAMx`H
z!_;h5awb60>kwkig(Fa!B$!`B>DZLQ*#o_=5zOJ|CoOWmSK{t=P8LrpvBZy!vV{X|
zCRl_aqF-Jgz<^=S+VWbA5Va_b5+VJYyALfn#cPT6%|6jri`tBoJO-c;&toX>#af(h
ze~ku48UBTUY6_Cw{xtzNPM$D6(kWi<{6r1fdcVDw08z
z-IkzE9gjKBpp~0O!0I%%VjQmu8Qe#gi}p`T7WRY*bjk5E6OaN!yhbd|O0nRx5%Ynd
zJT5!&AYn9ZJ+>w`%+uLl7^--e^A6&Gr7~D*S$@#!=h>yDDQCeypuziONxrAp>Xx#^
zBSEP=T<0M^y2no^G%V?=!a*xnC+#ArpF}G4J5bFWcl1=2(NdMh_3}l1$%F-ltQr#(
z2jw0<{+@WC94~$G=r_YY>U7g+1Q3C@>Tnvv^S<~
z{)uI)l31Jf*WZ|6)wPCVqE}HYZ>%~C&le{%6skekpjE==Xl%vSTkJ*pf4YNi#1TGW
z+2M8T4u*WV#FeM}T&oO=;}QEk)PDt%Z)o8MKV7e{SHx*kfZCZNdaf3yB}vQxU$9_X
zpW8YF;rUZksom5W=ah_30?l#$`w)H{D--9#((~~sLaAG3UXm!`Pg!`AyYgUS5nUf(px+^0_l&dvYMzT!(M5NF>x&E3X
zOy_^uC?os+`qRrBdf>llULmk=gbQt)RTS0e>0uCa7bZMx!5!=!fgOP>_cPl(PG@K#
zFSJD#QD3c(U7QO-hvTz6a=!QMd;GCggf!g>j`&05LRI(uRK^BhIJ^CIaPu9Tq7eT+fcNf9P6U-Y
zJ8_DU&h@Fb@j}7|w;SwCXO~kpfJ>izPr5A?UxVU(T
zi`KOcqI}*pmu_Po5Wq*YZQ)Z#a*6Vq`RJ7Q*BC9Z~E4;+9cE}a;$2(DMwpQwmA#oD`F
z4pqZ2MU3^+6zcVj!crofgtx(I+*1)`;^Xr7cOzbW@FQLGp`hChpC#n
zjx!F;8q!sPv3JAy=oO?0ku7}0KZvMw;M9=q32A?F6YP%NF+LMht$DE59~piQXa)&e
zn2g<};toFtO)ptZ7hx97ZF3vaNXCW$k$O*~Q;?|R^Q9yBt5ta?TIp1WeF`g%NLXqV
zKCVx9xaQN$l|3jYs{5W(f3SQ3zal#U3;g-rDyBNW|+D%?)Zn_QG#@+t<-~J!-xI&FSo1Qa`8kH!wy;;
z1Mw12j*y{UBgS&?+V?TTYGDii?;^|9J`51>{MX+IX8@8L{E@6S?g`LlL)zxmWs3f;
zrPi=D(V#}CHw7O#`3fcNZd#KMu&toj_EMT7hnH{dSo-IHHE!Sa!4H#2I1?@f{!|^3
z{MylMV82>EW}!N+nj@i7xmYtNo6r}?80QtKU3Z&Kg%Xj3N-v)e+b*V1OzjsH#%Llh
zX3of}4M;dpQw09#n(jwD{c&Kj`$v%7hw!S>JCQ%Qv6#y$%fY$#5#|Y2zcf0NM`#8S
z!V9eBalpJSDAYADm>}?|AikO$acp-tq532QgoDczh4jt`thhZsf;CC6%S(QpYETzg
zI8Rm%h9ZLi4Qr=b!8h{kDqQ<)Vz;>t@{}p3SX$%DWEdW*5brTapNv3WAA`f*sRQ2zwTN#|k29DotK;uRX!;UAb%fc-Ea83*Wxz9D
zp;^MO)*>1t4rVW8K?M;DpS@wphqSZs%z(;JkF6HPn_7}$i=QJE_m)4%*T_093B_O2
z(xrreV-z;S!+ql@|L~K>0I*~h6N?e@bx`3{x*}h5yn_I(^VXxnUnnD~?&bw&)lD99
z0hoY}gWZ1gje$J4+(B>r`f_(_WYh`9*Y>AkrdTsO0mg}5Kr0aY22cLV*?mOC&(`xl
z-^BwY5|sPdK*+d*O{NzS
zJ3=vE-3~?P0VZL{6KJIrbkS-?j=R@-Q?64KFTteEDQ=dDPT};BJ)+5
z^!j;naR5McEC>nhDJf#8Y0nI+0xmyvH*OS@AUV)l#FC@6%#zC-Xa-}uKyf?xd~YN0
zAAoK$G*HQ2OC>X(K|b%nCuol=32%HG8=tzgXu$otx8f{k@cFDCPhsD6mq_@C;pE2F
z8|AMWUpI94Jr_iPMf%$b{3?
zC?;`q%LjUGM&0vwS%{RI?
zmEaLGlLocUwTDWwTm~xq%8?NQw(Kvw^vbh|^$SoxY1GbDYr-c__4!U62d}TEz_3w>YO|ERz>jJWphLi!kx@spvg6uoz{vv<_iP1z`
zpNe4Dn+u(xqd8$S>)~$qb9KbtOA|7uc_69vy)LF%7~Tf$Xx>m7^tU_u#IB8bQQs*f
zKab=-%j}sZx&C`+=S^f?8Dwq)Tb+(wPA{UZrb3?3s@N(=S!w;Brr%=dWScK2cyoB-
z2WT$G<+8j1;R;;P2&P|nxqG)h=1a@Fqh0}POu{j>YZv4gohC8(Y?b+ik-lNmtX1-b
zvmdY5BS7)t@ho*;ImX`j;n2=)MvIttd%ce^MH1N$4=cm)2q@Iq47xd+(!S2|`aAr7
zilNr#62tVbKiF}#+p&DIUyrX|Roh@>Y>$nO&f1dUI9w6W=;472R6!9|v1Mq5&}y6Z
zXO|WlcX@KnS>l;v!IxO2CUVSS{>B?$$~-y*oW1O^>S;B>*|&h)z}W<0c*ho|7TT%U
z6MMof6EgLxuEW9FzU&6tQ}yVkEDkaDrzQ3ll@b-|l0YmwwVGQskwQzcUih%j|k!*>mp<^!{95^f0sq&$ZpV@SJbHkIHRY-H*r=+
ztCCMykGe5E*zgvH2IPIe?eRLa?he|G-r(q`2n*&>?>?7jJt&=*ofR66ePaW9cnH
z%S{Zi&5IyN(EL2lLHE5_x;}cUEPbtCvfdSw%2R9
zlDpyaIEaTSrNUYsf9K&fy-}rfO}o>N6?Wz&*(GGvgTNt|>WGP}(G;r5rrE$$8Q_gn
z6oHjZ1A15y`K)nnwA%v7lK?s!RYf|7Imd?j2AC(9d_Vw0h__+Bi}Zwp%0(jDfRb~O
ztmKIj#)w>k>!Lov9e1SE@qbsg23iP@BtwLPtA(vJBnPb)PFbzV_O$Rb2f^SljZ-8P
zbneLzDA#hNo#Oy7O4??h?4aWD>)WD<+m&b>;oz3*QkhNus+*M@GhGJ2cO(nMF%2R%
zk&&k0$GPqjbkw&7Qj3UFY|sEPNqCZqvTJEi7aOm|{a&+V+SWt}1?y;(kv=)fr>ekZ
zq?&$^F?w}0bExtqTmcqZwQ0{7tVjBupg6MME8f2wAf~RLJ5rpV^2c?~>bi$qe*_5`
zH-)2v%*Y;DYnsv_Pz=HSNM=Y)wg!)HvJO4g_J9!n{=;G|PSYQq8Fz&Ha~DZu`5S|-okKBJ;=uBY#V#}KJo9r_TF`zE4uDri+f@{K)xZg(NSL0Uu1o329*;^GGxDLya)AodwPV5(pSa-NIY*-OfAmXI*ja9CUK
zg91&)+33iGIIW_wQ{Y)_?PGo3=DOOFhcvU%!0PnKKdj;jF(ouMH#`{QH&DknCTwml
zUr{i5`gSeW0Wc1%xdst;8wcPBa9UYbS3eFJ8Nteu^sYMGXH&EpLvKqV0xNgjDl&UO?bB76u~9Z&DvYGJTPGDNasjJyu}{natmGqMxln;-T$YzH
zMPCUgVigLFRC$2mgj(n}NTKQ@b&GnAc&;5qyLB9HEy%`V7YjOD{rf~#+~&9BlI*cy
z!fJ=z$gYx3k#sh40aI2qXhbL32mL%3LD>1kbO3hJ_PcfN3yfYy^sPYP7rsK3fnU)|
zU&EO~kn$Rz?P5r|OAroZAI`tv24e2Fr41H+%8_Eq)w>YLPud0G1;W+21tt}dTIiW>
z+-~`gV1_KB1x@{Z%+*aI6jB(iHER@~`d2g@mHX-`r+=xBp-JI>4TD<708jb=%x|z9
z*N1?PXa!Q!nZBz?jAUvOh*k&YEFIw!8YHr|PGp@^UM
zOvs`-vg^_CRBnYk3)XuGK6wd}tLHo9$;B;UI
z;8aa;dn@L+MVT6pv7Ha|QzH9T_kN)ske<_i95N=k25_~YPb
zZ%!AgP4y?l@zFEM*=S0?T^rw`efN?ps-9X@^mEs32TPolhsp-mPYK6X9s{kciicepF0vWyJGimuH)=*(*-HS2Rz}$qd<_qr*fFlA%9Pi8~(bmXsPW
zbnq|_TPiSNv}Ch~jYp8RBE39hP|*9AJv~hHvmk)m+4d;OW+7AxcXTDqyZK-U8|jT|
zU#lQ!RqnYIJZZYmAKyb&ay6FyjKh@xGiyC79K~yXV-aFuA$3BgBw({@N%Toha`j@c
zqaqmg1tyMjBM~FCvmv1lElIm1HzlX#Yef~~Rzcick(;n>sjNqKGRb}dT19P0oz^4j
z>(OAkb&EiAY5-{qP`S26AEYccV7M}ZbHNy`@^DMbZaS?#a9Du>g`Q?n63Aj}0F<&Rd
z=A`zT7<i4
z$MG>7W}5=$bR(GuTw0W*v-7>ZZ58_LCS*hz?V4V^I_agSBVLb7nAldW#nomSlHX7%b
z@RC*l|BK-g3DWV8#S(BCdixOCiOC2}lGJ)$ncK!5Bmb%~k=#F2>fXjpv1!ef9k{GK
zX_0VxBdnh9qs-3rsT3C`{Syx&eH1v2En&Og^yOj*r(_CCq=@_1%e*g$aQ&4d(;nRf2(|!G{aI;S(1T^y5zX_)1ke`O5WXjb9hLEtB?j_L>C
zq!LqYcCz~mi9y|g_rWJ$_8AWj6XU-+PM{y%m9;qm&qpYIi*JinK!Azy+uuyj_r#SO
z7}t4+@hL&`gk%~ki=|C$3<`vtyE^cDvl+?EumUhv4(3@?$sGbuqHC3MP$4$VoH1D8
z#!k6XqR?S9J-z6pt!G-1oE06c`Q4;72Mx9SGrYxY%ocq
z4Q=c@x>1T-td@VYv(aH`(bA
zKr6s}tYw;y75Kp)0hooLa}sIi&eN7Z6}j7Rd(Ba0Y98u&k@{B+)Kt)IFxI(#o&1yx
zDn$JUscVD1M8L__O$Lr`Xk**5rR`%=4XX|>!b#EMl~dF^rgK=J`xgITE6A-waS{^-
z+^zD;IA06s9H+KjByW0lOD7OVcb3|c)~&c0uNJSvF2)l$7BI2wGf>^=Qy7=Q1MzP>
zX*pDDwodGPTPPz2VivH5t3(myzP~S{o1z8zQ_3o0pAIY#BK>U5M=hXs4PhFunS$Yc
zx|V+&=BA9SJh|opdE4AX3Gbb;U;T=j5kd@L$5n~8c;L9Gjwbzer6H^Un*GSXm)D9Jh?wsO7AiP-;uB+5T}?ziokXBTz)X$*
zWC#W*(X5p?<-BsDa3zj((%XM0Q&G(b3>tyiGP=^A7u4h;H
zze@!IRq)pz{VzB7#!7W*Gy&z1tOWOVH?diZsy*eu8(lz=W4=RIfO&&r-J*y;z?4yn
z;kx03Qb_Az%|BMwFjX_HcAKX0U7~rjJ+yvt3n6sdz6IU^V(d`gp6}jXmmi^EUOr4m
zbckS6ro?L0Y+q*;)UnNq!^1sTVu#TrMN7l^02B|F$Vo!@l}yTx(A
zvjxnct7>*KJM`=NrdkM3nxsXBc+iVU@-ZnjqYZ7MS&Ib9aQv=tsj0P>*gnJ|Xkj`+
zH(zGFtKUk(mHO0`+742tVT9#gK3-N6uZ9CeIdnR55?$D$5w?)`&t4;wWKA?k
z4QB7>4-ds61RN;7jjQj#wn1%wzP<#HS4Yk|Ft^F3w_Easw`5dC{n`~Des)!S;nZ^$
z*>BgY%Lm+1Va-kV&g4e6d3Y{71ZCLt7rmaly{yP#rny}@H$s^Fi09v|MBBSq1(t1g
zeo!jvVGnP(OFS^WfPNqBi@X@4<7``*Q19wHr)HF=L3u>Mwlp+0(M1hxk%JkdWL@KW
zH>0#V5u|E=)i2su^asdbj^g~nq)S1n7Zk;wfl*EZS*nE>mJ)oTjqBXwE`1#&2O>Rd
zS^|>i?Vj2aND5oBh~GeWofXPNAc~)^mTU)So%qOzJcu#Tl{XRHLel0+`3?Le+%mnl
zBxfSqT~`*%zX&5OOzNiV@9RPOOFuk=zIxsK&CpRI4jtzhDD}j!Y4)=AIjQiFH)ckm
z2rlk-fcXQcKd)yfF)T^JNxm6L%_@xmHVAgK$u*O%B?rGeB?
zgc&QNF<*3JeWW;2nfNnIWxI9qFu!_UZ9$9O-tq|@t%=lt4e4zCxSrJt7e_x6+I;09
z@bCKH`su{VTyxi*_&c+nR)h1blRsd*wVRM~*YGB{i$b*VjYnKWgTsAKN`Q6d&?4TF
zMDP|&ey$wJoENre?(JR~%t!y({C4?D`ixl=ASNx;wMEsTvnNgTH0m@@BeyJAu};T`
z(|Uq53W#JS)mFV8jW*QGj(h!S~v-4^7nwCREI;ED(R5OO6f
zS>ZvNPp5+RdQ{7r?sJrv39R#I3s$@fmjkmN$wi2}xDsFwo2BhZ
zGlNoSsjK9MwHF99gWr}kfrMcuPm*=(8}2=|&16)>5TEYMPbz3ZdseGM5&^v4z@^X)
zfU_W6a%=@D?6Q?86n{~zpqhEO3}O4wi5}?zy#vYfX%1ir2skj#G+0aoCFQ#ZiI43(
zV2za?H=8n--eH8x^+Iyw47wCypAP?>oaGON;Co_1&DJw4S2{2~RPd}LHXsF`1i
zXq6AT46>SqTZ*b*F>w1TE1)hPhEYidEoxVybcx>FJYKbsYrXm@lQ!l!byuBA;Zpn4
z2Hfs3k1cJS^kZ|Z-~ltL?*RI$Zha{-UCbjt+nEc85yd<+%9Fy)Zc2`i05QQ3F
z=x(hP8>k=Z+Xz`&!QM4mQ)KOtL$txM_LIq3n4l*;HaK%e!<^@#qR6$m&vvjR(
zhHqHUK$8a;x_Qy-rN;B!BmB8+l0)Y*LX|YJ=cy`K7pX%vlTU*Wm`7>ixU#Pe1fbZq`nb3g}dhLc})?;tqc1O*l!-d!})q|d*jcCNn-G(=-U7(SO
z^%Qb^q^j)gck{n5U!=_>#t6{Gm+5qcdyhJN&H{eAg&+>hLv5Bbs8_ZwbLrEXRq0ZD
z!)wN}Xuu4d3={4)2vuVXw@*W5E}UGYTvdA2-5Bl|AvitO(+J!WhP2Iv4a&6mIy3o5
zjfgQ5A=)+Mx8BuXx?Yn2c?BF)H38io!%;xiH+PiI(bOFu*fwf9>BLoEH_yvhfMVB5p~H
zf+`>vWlf?o`;4>$+pprpyQ#dNO~=4OGlJ`ry9|uEX^HQk3c-E%;h75AL7m7jN0^2)
z8@GR`QhQn3wB|H|3o*gv)~bl7p~aNY{E0v6((@@>&>GMN%XfyQz^AJ7eV=P_4C%7KwunUi#%5Zt@S6S9Q!Cy1PK~39A29l58wCiuLFNIgVXksYiIZ-OkHu*
zwQWevq=iH5ZSY4$cSWl#<17stMGoYk0JxnZ3}00oW9rOpezsdoBuKvq_NAnvGV_G%
zC}&*ZVM8^QHv8XH9j%tfIrL`r11)Zk%5^dvvcn(NNHi&Pb*%zua?FEJ_u1?!E`xAk
zOHD8833X<0?_#DEJ!74;?69HECpJaZU$MB-_X2~%jAKN6|bC~AL_
z&*wXl>DI`>o3M)1Whj$0>^m5&o`SP8q4SS~O(d6*%^O<|kC%iT$hf%Nh54vj|LEr`
zA7b-xpP+EhbzHhUZar>b<2F&!;s{?bU7|fkV|qVUO#d4_TYip|}9YLxR!%n3cMRQ`PhBpt<}t
zeswz)?4qdB7IaIe=Ptsk`JHOL^m6UTDKLlKJkB>D6}WXKotE2
z63W>iS3+mcFRRR(3j+zk5wj}=%P|Gz>y?25#M7VNhvt?(gL@?-tOgKi&=yS4XnC^B
z_-s0g$#wLvdKIzaogIe3|>SdW$d1@j`IIL-h6x@|_2VR}u}=8CpUa**$Oi
zbn*OU^5@Ihh!!Y}A>2qT?~z#Zf~S}@%QvmUKUQfssbfwYaQ8a4&bJ%qM2yTkVWYz>
zcpYftRY$Kl`@4-0A#KeyU`+bPERAUqtyI{@wGK*e?ohH3riTe_%FsL@J>q5%xn7}E
za9@iM?cb!+Crh&J3Z#_NqDw<&H-hBBs9&jf>$75
zFH3`0cU@HGK&x){Yg*I$jgOJtu~9+D6sieJ)CANt<2s7GI*7NqKquia=n^*#$@!45
zz%i|m_ot>E8fH}hwE7pDenC)6`_C;8Z5lTz;>D5#|6{}hmV^w=IS(*Wu4>R*AmF|P
z_*_{opNrWu;F*X&qKSsollLE$>TH-a-Ke!4OGSs*l@93teT4+gA?)@SV(}0`58zv|UIYJR9A(pfI_Pmm*pclmrFo+{l_ec#uU0$8p
zY02fJ^?$P1?}&P@UvD7M{5iyl=x$j}o%d*L@JpR)e#NOmhD5Y`w-7{^v+8uZyM2+k
zN-qYGN+y|ai{0f9Zxy#`0cNmbl?K-dCo^!*H}B;Ok6@h8gl|sH&x`N((zD%{ByB-6
z2XvtX&?zW?V{&>C*fXM-4_`cSuyV*fL1rH4&C>|}6G);f8b7wp%Izu`*dRrzB|otNXGT8qMWaR=I0(uDxB*2$E^4_U3{$hqZjQf+V~CWxI)@*&c6>4
zxUjVCu7Rw4(z*bmLc5g}lyPJ`JxKi#Y9ceUyq$6IFjquJul!$H()zF3aYqSSL~GVB
zc(a}0mqprE(YPE5m5GM@
zQR(7C;Fw2tc%qSb-Yk7ea>mq*G4cecP*)y8M+;z33@*Rn7}!@7s9Y!o@7^<`lHn?_
z)+Q_IUc532jXEaKCsJ5;F$0}@Xn|s4>*0d+ZK8jC@p{z3hVB6M8928)^woEF=w<(?+#=T=h(IV^+XzK1v1>q0VnG1C
z%_~;`t7oi53ArpfGa}mm#6lk4CyOnUh+|Bso5)zHnGh?>n^@b`aZ@+|V?8FwQ`d5$mh{
zyg(>*Pj}}L2V{Oc>+?TSWg_0svX>!Yk%XG2UXgg;t6*aD9Yz<|%aBz|RDwUNoHw}j
z(jw|GrJ38AUce&r`+weN+M*7FRf)X~mLVSr$q19}>{IRt)6IySDkV@0C?V`ldd?xj
zE!@ox9tNBxN5pN4WrQWPyIQ17OiLkZcL{*gG4}d><5l{9Ug+~!zRrz@3U{=wLd2Mp
z6`enj`xpc>Q@Fr;NcROaA0EADvcp$=8g{+4{{f-hvaQ-IxMIICqed^7Fx_5T70S7M
z);*nT-myw-;I&y?D?Gegny91jj{cN6>bVF`jR^zlV<*#1B`r6@5`da~!*8WB&rq?J
z0vc__SX2ept}Pd$lS9Lf;${leGfV$dGwM!2bRZw_M?htX{m-4rEnG39{mAy!SeYX*
zkX{!_ncJFkseFqR?Tc?7urcU#sEgV@oD)uAI*imlj_-eZ?DVg)a&hML$
zL-lpbXtm&V$;1id=81UN{^*d6Bl9JYba1Qmw34Ns6Uf0DL^Y08i`NCZ!uxmxOfjV<
zD4XlSw%>|&c(;AKadQ5k%Hl{
zo7@yunq_cgscx
zq~Ed_6SgG5XXB{rYWu)ZVwLUaHY7TLx1;lCd;j+bx1$i`_(?MnGe$OlJpzFB0#lYW
zt>pIR{d8_LPW(Ax_mVv5(GdjYqwdHN6WdjTCun%snAy>Ttul1GXS{D(E_W6Wb3Ck-
zl}|`Mtk<1}SR0D_<`lD1(HU|l{9I3$m~>vTP8evXO2gnb
zzqPHIsBFxLB8AGbhg(;@t&K)X_ECYUdT7SDI!U6mKu?Uk-W6y0CGrbe2NwzX`|DbU
z2Pvdu3TDc9Tc7WPxvwl^SQgZJhqd#9B6
zAiKvhcMC#Ic<$Bw0lmoMFs^sneK$EVPdlX}rQz)=-!#AXE;FpjqvSYp&@PbpqfTgr
zIBY|#xkYm(WuM!d2*@|KlD?g?`c~dm@G1?#hddFt*I=Nswo$wbRSz2kwJ90-^V
zHsX#k7IlKpA5HRV@5`3HVgwhSlKt&M%1(4RVM*7x7U(3ON^@6b`jMwhSklVdR~zQF
z7FOrrgCpVmHCiGyLuwnumYttkbS{D4-F~*~y5;+*Uz3-?4#Qa*yi*fDUg;rNWTE@>
zKoHTvp-_E;g;oCWv<*sZeM(m;onA(#6l88Av`zDczgxmy!$bI!>Otdd(cbOP*-y#!nHffutj6}}W%VAfh?utKwAX~&eR77%MJnqB99!wrNvKb17cPHw8X1mBGiCqfD+61&(zmI3i
zQ&>0Cmb3U?cn4eQu^OJgzZp;MykLUEkWH|ak~NcJe=JML$yvAknMX5QEAo&?dHOz0
zSt{d#_0mLBILi@6qtWxPWngd`)fn!%NACPhkKe`rCKe86N7GwSp$Tgsw;(YDH2QXa
zdB-uZCv7>VRO*qu@B3I_?>KYsZwXg5Pgr`cDF9>7FB`%>v%hC^czO2m8UL-&Dm-CN{C&?(=s&zH
zr^#dY=Oe};9yM5Ri>P`@4XZLo3S*_-)YhJ^9yS=z&GRGUmd_c5OKQ}jc*aDpccGai
zTYQaBpFxKK5zt3v5@KL=yE
zzO(Agc-V%zF5`F?GB4ot$0=WAqBlPdxT}W=F{aU38`Y1xp2`+`u#ADFi#?wDQLI;y
zwDmj;u08N9(!LEW>dhlp#9V+Tr~z->Ml}k~oTBtMh+G#;MEX6z1)iJM#zF{tWf6RQBc8^guA9_*%TIu)GJ?KA3
zQ8QhB*J
zko;r1o-tgfg3j7jDpU;4kzuc`#jP&LtT@b+;=5DMZn}0BF8=m!#aS-l0z)XO1*S{f`vB)Ir92o=roM2_A`!<3PG-jvHHWHI9D@=dk~EH^
zzsn=GDm^hPdlW0~klgCV>H-h#oo?1u29ei=aZ@9y0C8r>a>T!s6R@I0n-!T-bk3^jV5rRQT-19EqIfXSw-?*5(H}b-)VTo_-9+Q2u{bk>Xe|_sAqm3vyfXTIbm~@{;AFDBA
zKCNOP_o&=>1JZn`W6tIB!_DXhlQIl$H}|&4Fr^Ol7amZCE%PjwHc_R+-
z%nC+htiTqnvy1}qj-Nrr8TL1n%YLl6@T%Y@;qZ?AGNw+8g^a1(iwW@j6K~T6;3ZwP
zXP<^MHA^B2Z5Cm#182|Q`pT}56Rp57>5SWZW6fD1;}#{j;jj^!p~=`~&k52{rna0D
zm~p44K_*oFghSEC-%KF-QIMoC+Bx>$x41|$4fV9?uH_tGb!dyr7{d%}vE@c6lGDr$@)cZ1GE43mH^3|v8G5b=WKB5mtL%Qe6
z`iBW;eS~3emLd@h&-UdTG)=AmgWa*m7^e
z&xdTaidLg1L%p&asple
za9IhL(!e_psm`mqp-@A$B-Jq0Mm8Qy0pldbzC?N9oqk%qU!0_t%H6_TM5X__Mv_88
zlT(!gDuId)!8!{wLRsD6aL=zlQBTGhyAP9Er;2YT>+mO3Q14d|`fUKQ9b*j?X41c`
zI}1X}nEXDy{yYRIe#UWKK+9+N4W1hmv&m>?2qp_F!;HMV$ux*+RqXGeqX6p<+tVb)
zh8#O>m@7P#!!Rl(hXw(Ra`aErIgM-=8B1{H&$S(GYre<_|9DPZM!*^I)$lgYMXigr
z?7!KezOz6|e}N3xp?_@4JCL`mY~<>NlJk#3KoXgeR(>Ar;UZX{o9!cEB6zG1&SZ4q
zTuJ3c`$bYP{VDv0apSNGE~DUKd?tU`wm~+Q$cq*UV1WWZS7e2532%M(h>@^yRpcpf
z55Nx1FVt5HHIxA@1t!5r%Bq&C+<-U!gGDRk5vmhZG0kx@Bg>cHpgTL-)mW}*is}SocQtM2Ovn_!2)Yx+&
z0|M=)(~UX4(E8WpkU@od^Mq0U;icn-g-N-k$GDYGCiBp#c%SI75@QE{Pt(*ljx3sT
zEp)-4#SkEIG8n8xcyq`i$7(5er6m41u{j_S%Dn@DWENiF@xa9`p3(XaE4Nz`Z&cLl
zkTXDQk*PV@EU(r<;?uw=|LKJT>xiZO8LmnG@ATu{QdHy&+3)?MPtBn%{-DjE+*E2}
zRJm?mu!UdMvT6|n&eh~{FKc%)9fsJnQO!+)>l;)|0U50vuM)ojU76mp2%7y40lX%C!+tN|`?lEHj
z7-r4nM?)APoWB<5q&bB^o_$uV;kuTlC~Yna$&BMQkp#v@^|0GvSt-M9gN^BQm>^;G
z19?$a!yd$27!SQYoRd%&|7pmc4f<)_RWf*UQvNJ9u?>-u$hG~IVVQ0-q++Z44W=`A
zPe96kksd)NE($J`fEH>7M1~ZjuF0i*+mh4vN|TUv#M4EpbNm~=|l#|!i@}6L@w1QW=(RDG)!sC2Pa0gq6#IcIWDvCrx2Wd_BAKaU;p|S%``w&
z8#+o0ex7_rIa;Zz*67<3NcPgB@9PwYZ1^!AnGgmbO>@@;c>A}kopYHX5ubK-2VRi@3Wc`9^L~d
z)g|T(u;Rd)oY^NOSwxNJMkr!!=yvKo&zY9hb~K9?{kUdqBqoZc)rvDXHT19Ap;gf^
zoK#ww*$#la{EB{(m^|ZK+(bV#gBBjyjutSQZ*)P$v~Gp2b0|P3-7%I*Cyo%(Fj%__+ZMEN}R98_*hI_7v?6$-wAL-O@G=WZHY5TkGl54^wzvvw=t*=mrG}Z%9*fpo=$Jd$uw=pqrQXF
zh`|zg>8b7<23CV}6*vqe`orRRlpMdGm0hP1bT{&~971PM1VixC))S_LnnQ!2?MA8R((w>=#M2$B9ZkH&i
zq4LK}TKOCROJ~ivCiEmemW~Sc=I7OgLYYGD$$#eYq@kOl`3{-F5pz1%t9@lEiD1=;
zWUigvQ7!#!1d#}J%z}|3A>l;uaN4!dWN}LhyUfZEqeN0U5nqqg;}){p!z@ZxE|OhEw*Orl;9cmasy1y
z0EGo(a=j?huvaBh^`eZLVX|Rv`ow=YELoLOE@{7^8#}K!kn3vIF3ybt*XeT%4b%l*
z@RMXMNA;N%Z19p6mimrho`cYwTM&)vq)=_zZZLdnMNAD8K})61-C%h1U+KS*Q1i8^X(%LZveqn6^F-9)z?FZpNK
z3_b6g9Z{o(fnE+c>rcnkm%WLlIz*V9afi1RcCo2gUvgH(6YMZS6!4JlZ6}f?7o`+p
zJ|c8v`^^~QT{%2BoM~0f9S}PB0#@}|YQcoTkhjjI1wB0ljQmwYcC@62j#j)0Ji3(d
z`qmEAacU&$ZRr^gix8~bTLz-N{iTyx8QtHaczXJ!Ek5B%ImJk6&A&=;uF&9lbwzwS
z#d#i69w=p@i=Sx{{|>&XRuup(!ZrPlDpnO>TC3eA&inKUYK4s~$u)Ebl|YVv2;3$&
z51u70+BdVF=Et&;q=S$N9QR(jYKKEjo32b@vnC5ne{eYKn*4`P)z{5#kq
z>tFt?CCuAKsf$VL=xBgp(=W4-HUM-k_chst1+0?wU1k|bZ&_ObZft=+sA8Hb-OLdO
z9c^+^zkl)zVYdkls>WYDT2R4|U2oPv+d5r_T|bmYke=jg9W+iGRPT|oe3_rc~
z@L-&Jcjw)`M(?bSf~3)>xthPW!klNd{t;&g!=~7=nsN4+$?d9W@)s~Km{?li93~}&
zmRikGBrjlg#WNL$sQptdO=Dj}t;D+-MsV|p&1_3cg+v15;SH&5=O~5P-#mQsOWuLX
zauwEbR_?%*^AuHCCUf0#XFKHFrvJ$cM53GI-LRC0GbpO8eC*T8BYC{48E7@c8EL{o
zge8iBmuCcW8yum+zct(v`k{o;74S&Tey_v9l&GrWa!3LAvtbDZCc}w8i4_%iXx@0u
zcEeHocf6ra*)R!`bh~_x8x%{)bwrfxPzgiOD283boJsy;%%||1D;E>{x?6!T`s_gf
z2dFKrj@)s=#N~`UKg>HVp$8(*KEfO;Uh;~zJf!{hRPAUerQ+gT6*iE7ii{4|sLVjlSH74wLB_!&5I1u~iwDkZy30#2u2x%B
zo!%}~wMw7@RihT!bG
zCZ_bqv*wn%cxh9AHn%TD$~5{zXjUdBf=41~S=K9!^3QJbBK(GoZ@HGoTQJMrfz6<&
z#XVh>;u_H-F`^
zWfFRXM;1#7^J$V9rmUFdb7sxyZoCiV;cr(s9(!tP!fCuXsj}FaZyA;^F;(F2xQ9R3
zJT&TJ!4KP1APK-@O|MOidf~(d9&l8oqukyDEEgRP@XjHz<}?vXS6a2`(t;VHg`|Ff
z*$C%n0$x8}`vh}CN~E-Hmm09yvs;4YA99p)?9RIEs?>SZnWSFyp?s{R$q9MuC`L$4
zXd8!^*uP3dX%dICFV=;v$xP7?w%Yf4KUJVJL9>(MfmBJCP>vn;O^En%=a=7zr+L^1
zXQL>p)OOvl1~au*@UMUU(e2(9$H2v$aIT)*k!ULjt3=2|-)s2+lsSB$U=NfFsjM%(Xk@hcRVi%iFX6mR5
zHpY3mx%FeD4Ug_09!W@QPevHg(sv&*ox|gA%Xp~kFQk9c(LWxTmq0n(A9=nWN&%P&
zOCD#^+fZ_|2eUH`x9IuA8y<+=COn%2US6Mn*i-}65Dmj9iU+L;Xc+&SrnZ+6$kXxG
zZP6hC8DcNs;Ln-6uRay*8;^CM<$01Cs+VAR#<)iMcZ~0i2lhj7-YY7cI1+>|AjiMh
zvYSnfEg7-dS1$?#f3m_PK0?$xfG^Su15ZJG`>FC2b~BO1Ic#MqlNFnJ-T(JL|9d@Y
ztd$B@5kvb36>kXA)T9t47)xVljDmo7kVR%JtsL(qHLoI`TVNV}Qvp+ij+`rhV0<>o
zjJKL&+LGy7TZ1#@c#~GPTUCfRn_=N0xi6%PPbGXnl>;FM>TKq$%5gsLf_yvDP&a_$
z3G;NRu&1@@#+swM2H!|%2qjxPdP2J;m}zi5jsViFs!>aAmjNkz*>+ZvS8fvI>uq!>
z{)TRe*r*f?+k+8Jnt1p!I%yJj;qHHPHwn7Pi(R=XL_!+GQamqIy6qSJ6QvE1%3v2vKF)zN>X
zqN4afwF(a*$dIvSgJb1KVH!*S*t@^+v7UFkKt}D2X$H>-=H|F1`}pMwJpQ>}3IJ~~
z&p;$vnmgc!(1jAL0$W#E9_VUd8dRx@;oExm@rc7W*c4_+rhHGOdcY9sMUr}^zw%Ke
z*u->|@fruz@naUSChlfbK&WLeY?|+MK~AB
zL$mgyu(dor*>+DB2a=tbL#~Mp-C53thh3sRAKeghO_ST4r#|x$O%oSdFec=La)_bm
zTCM42+jPQSmCVx4I75xFy4=oS{V?S;Z=L}8pp7@EBXr5j8F*+Zgki6pN7cc*Sj@|w
zTgUeC@)bz_Ih=NKgQeZ{Lu4iJYrd_P?U}ANnG9GLd-41MroMPUSo76E`F*-%NTdhd
z-_M_39XNALf|)kQIX7xUTUmY5!-j1}$B)%<*u2Bbsl-7>Im9ncXUdJo@BqW4rJT)s
zGi=R!mCH*Q$oIzu70$)vs3~!))=x$nX(4d5BD*h=Y&gW;gvp3Z%=~RK2kz}*Y8|LR
zZa8mZG0e1^oEX@{s#=<7P*t^tZrB)$wC?DMkV4WqJE+bkmf+h>*S;4LHHRJ_&>l7n
zKtj3(E8beT_Fcf?A!8QY8#U~;${Eenog=EmCy2qpj~xA%&*w|76^;Yo`SOr=e5093
zV-B?rXqS1{_{XsBfW{WN!n)qkKZJ|Z?YcCI%N2^>Fosq&?uA|NSkx;S7ZSlMS|%4l
zdkjdnMp>*X1|3Ml27lf%sghtBb3V?64YiK8`s~x8k&(-3E$1NDyJmjwte<*HWBB2#
z_u%#K_(*5#J{oZyao~jV}I|
z3Y8`c03)l3riAf^j=PyjFc1=|P!F;n{KT}&R$;M68{!G{INp&qO9`{v_~XkffIX)L
zAJ{+s@pnw3?Z72sBlgBXc$G+TaVMlkEec(06?@`Hwl(1E4Hs4vVjhGquhSXRySDljS?;A>5oo$P2HcA%1x=~sIgYA)e
zO#?^q8Y@%3Tn{BhFRm7seGNn}$*dMA}{DUgNBf-0_?rpby
zWh==VG~DFfy~nQ1h!(K7Iv5DCk-2rH{hn;2F&a+Z$f@iAFC(t!XqYa>mg_v@t2V1yzxA460ju4-
zG~KYQo|e)^6$+3oQGwqZjvAuntX0pcN8wD>%z%anMwYB5
zDpJB6BUF)&O}Y}B0$eK&>=p=)u#^~cG=!R2E_fU|_c!sLD=Et~wX(uzNWq*|0ay01
zVRm$J8S9k#Fr(0gR?4OAnHxwKJ?Y9z8!0NDg?l7H*_eB9Z_HeSiWvR(a1tU0sT;B3Pc3BDP|vy#=P
zR&KcyvsuTAR=vunMUkcgnJ@RoOjeu21?g9lS-hc`*Qj$^xdm_|3_XigC{J39@ILy6
z4w)e7Nar~@WDsG%nX)ztcny83$-bjWrr|EdV_U?q(ZH8z7{#79XX=gkUz<9-_+eZ{
zd48aR$5L4Pn#R$$S1p9_iQ))V$XC)1zDIQ0(pOhrD1kx3g+Dr?hk90j(^L24z#o6s
zo3qR;#}H0vYnZ<_0ir6+1!_GK+My3R|~O{vsoQOBgj^g
zh~W?`aJeYL2A#$kfE&!}r9uYjws-F*VB@-%6zT=r}ShX3cUe=-ADI&w>K
znQ2qT_Y&ji4)R#(vn-C{oa7B;XaCmtXd*jUpQlg{JamxoTqfV?Ap`(z9ZI<7$apL^
zz}tFRf^@h5yyv|j`dp;v#Z$+gyp@O%uQGaMknKysiOOBAv=fCb9CYqHO}1fLg(|H_
zE{o>jb11(%wYrHeK1Zx1!nuRU^;@aCgpQd1_1e=k6gYBLPVJB^UFSe)aAfIY4c+b#
ze>$q!wyYuQdn`OnxYcT_V1QFeB&OGqqx+P9X3_$mRnaHvmaI%a+j#E9`FVT^Kywxx
zoQJ>)#aILm6;!^uUFI*roPh{5stbCVsrekEt3+OyuwJUjJB*E_(-CgR_N^S;!ZtWv
z_QJngYSjw55iK)O4;UrRiyVNa^AH&u2YzkE5j->nvV95D3h-J7hiaGM3O8&gi#3xL
zzoN0M*RqGWG#*>_lsJ(7KBgcep{Nk6(H%cWL2VXHguEg*kf3M%46qXL7%*!oInHQu
z5T;uf7b@M3qiqumoo8Vt|AB^(tIuHoZ11|{HyyQoef>4m^Tsw^rR@|B9t7WTC>(Rt
z##@3qX$q$zv;W~Fe9Mnp0$y$-=c7wFbD_#$1*%JJE$ASHkTh*<&S^}5sB*xIRuyu1
z;GfDJN|x5fAbqG6<8s$nZ>Ox$g#_dPH`)fnX$e-aRiI=#&&5}N7@{R
zU?BR1WzL5yn3o4N7be^Dgb`J1(+rc~JAPzr5*h>^;F
zU_dR=LUehYLJ-FwXTh>lTDf>2eFi+Yj6^xsIIzoP8F#w2QFw#4bnb`W8*Y!(O+R|R
z33?hEd{*+tw((-b)^}cJ`FUZG1Li;r9*Tiw!-Q07fhlJT
z*1Z3*uWUB-EO>T*(@G)sztLhi*(cS)U$_f+qRwKH0{rwG%^a(R&-O|1
znC%_inp3Sr+QO`3*(bjW9RR{KE{+)`FcaT-miI+&(?qwb*4^Q8OaEGhV)`3xfle=I
zQdpO;O`L(nvzYLo{9X2|u(^HTee_yMLyhx;*IBASebELpw2rg>paNzM`P9aMWggda
zo1$>C*K7)keHXuSm{&&3?=%3{)W7^RW|BerSX!UzN-Ez78OVKP|uoOq9OM0_(e
z=@X6cx|!@)1^t%1z`Go|z{MLvU~A&v<557*6Kr(ba_PdwJ`90?P_+q(;)VhXro@5I
z9S4*Q+Y<&@uC}H*v#fEgOt+bKelSxL2gfu<6ytP6moz}t$``FtVRwGKCy7GrQHi*!
z#}JwLi40W<$+QHfb-Aq?|5uRmEJjw^f~|Vm;9xwmYv1hXvzM*@h9s4Hf$5J`9~}0v
z$#EyB|C2o!sLtVIg7U7Gsbx?s52;rZpGO%#8S!E8-M&8TpvuuQ48k*L0H6sa465=i
z*;;&UF*tDn9IZw$-&O^xtr@aciKQ>~lwxIOz=%%&@U$}Wgdy99Fx}A3+M9x$6HH%{8yc~o@FV<
z#I_wyb1mnr_9Ee+mBChrvOOaWZ56hWfO+I3j#M_!m2||emOP8J6&gspd!cfuaU`td
zZ(m`v+zN3Xw!yhlQ1uF~e7YTQ*n=mHEzC+yjpK{q0j!F*(2sqtqwKV8Y{25Q>Jl5k
zRs?5Mn3`FWNk;Mct^Eb|Q(tye`}RD$IXz$*eMZi9mlp2tU9*L=h3diCI1&jwxBxHo
zFzJ2Yv@pSCl*=Eo`QQHgugP&aa6by#@pXR5=crOI;%wrjzaK`6kbz8Vh-Lnf%IhN~
zY=(#f2s-==%mqH2X=3hc$oAy7xO12cgR^cV)fb$XwTBTq41jm{g@8A!4r0l+bs
z$E{~jZ2d){MZ~dI^Yz{gduav(QnbMAR|w#&su105UanMyMKB;+f0_>*tct-45LLb+
z%Z4-~3Uye;V&d1C4T24cWaks1?*ERs7N+$p%^;Pa4r`EeJ5W$fyv-a-?YK*(1%rUv
z`3D$rhVwp^{K<#7_lytl+h=@Wr-yCDX~d)s6&mT8;BCSfznSx5PW(L}+zWJPz7)U|
z_5h{Oi+UZu3&#>t0<*{;
z7j&~!GZujsX}u`Z~oI@f_PONPKTf)aUe&RAvp&|{?O+bORW+lGSKY7
z9V^9v`9{GzM-*q12xNn^!OhFF#}+2R7d=NlA6TMpU=GiefPZ-G6`w63z+4>Nfa~Q`
zD=-d7y!^QXjyRmR!s&C7*9Xs()$Nuzt8E+CWC?h_IghREl4+bJw#jp3Y+Zy-l&)Hq
zw1raa$Bd2*sSahozRN(Wa?|6qn6)E<@+2Nf;S&lwe9BD@r2YY~oIh}44%(&1=
zE2QVB`)TND`tcnp7w-aw9IjY)RVQHpTUttR4#^5KG{-Ju)fHhg%*C3@4(4NTWEA5j
znbhLW`xbOG_;cZ8>j)B+c!=@FRp58n>~`^B#u0wQn%CzCW68eLTL{6t&eUzQxOSUu
zEAIuQ^G3rI01{wUoE6Zfkbsep@vNL^!YY2%3!&)FmidA0H<35?jSqCR+-VYK`xl2*
zdg_G%LL|uXahBF!sT?-?1OPG}PPR{&2^?p8^RNNH*+3bVi>1PP7FQE(bi
zs>0H=opsZ(1I=rN5RLBR>|Usi6YPnCmMX-tYsq8z**-`s+D+j!$C=aEF)Bi6ELE`N
z>2>8Uzpu+ws}>7hE{2ddksrD7Ca}oWrtAdUwC((TVbbZBMF#6?HUCkILnfXk24bA9
z^0Za79#b{f7yv^p@H;5Vs|ebH96Z@wLLD9g2Hs>s!7g5s#go>2c(ShgKAqEVZNn`j
zAb|}M*@HWw)dsLxk5eLE@L=s#f2zNZxiB@3as`?5FcSlMkamw%Td!(t284R7_WcDL
zf$B=c&yAN~m_oxGZrn)l@B@8A{5y!Mib*tpF}V-})VPpJqiUZm*&1qFQx==RlXCi7uy_zJyjtZ8j1c=fwfkaLVwajU)aFcUwQss@(>lc;EEi#wUr
z-~&IvVjz$6JtV)M;hTo@dc^#a$<}h6gNIy;_eThtmK9Ftdebs;v8~WLL_xwlCJjBT
zCX)dpXJw>%wf|0Vu?_^077V%83z?9hzG*t&u8LFl;+WLv2kmD7kouhA@b!
z;m{gZ+jfR5>O~hYJPaGTmRk;xB8Fk*ZbxIso{SZNa;6IIq;L*@+u=j&v%#^^Dx@R#
zUYtXk1_E|=kdz>p+%&PyHq4|MQ~)I7@y9r=3ee-^|8y>^XbV`6PBW{~;XMox(ga7x
z)S{(Lh<)6jq_t(-qCsf*qdlpH
zV>Ie&wnshk1SD`L$RIqttIma0vE?CCt3f1!UeTI$bX%g(DpTIIO(AER=+-nzy#(KH
zlcY4_c_fxdHRu%GxpNkS-Xn9S6fQy_*^=FD8cKj3eR@#lbxzqoKamH9E;>>?a}8UV}>sDiN_wMjtZ45o2dd{2p9_b
z)nM5ATN7J>AQ`hy0jhqP+)dT60j)+QQ_lzshId_eYx~?xPDOWdAf!o;On`ZTo>21m
zX4m@G2;D)3LuFF01Rg!vd!Lr6^CqjE8{x0@VxW3HdZA$zwN4iKTD9n9BHMTjX;>rf
zeaF_uhD%(Cu^xvnC%+&5Yh-9P)bxQ?@P0;1a$3BV4R-1)JR7;!t#kLDVDLW9_dXnz
zZxV$ME}IzPQ{R08YDXlEGOjIN6~?qg`7US9@L;;V(wWU;ftC7`q7a1OtTAO#A_vzq
zg2R-0d>`^m>g;=C48VS(zumoF1j>mNK~?HruoAhL4GSWcdY7PkdmEPQpHncJYtfNi
ziL_#de;&Kq3Mr~*>iI!w5@3}0-=kDzdsU6C_$}-dy9xipUI<{TDEf^S8-e!X?si)u
z4lHSd-;ztCBkzoa6|e@$4-&*`h_s#0>RMeQ%ypD`a^Kr}gV|{N=Rf}=MP=-Pbu+hX
z9|^YWDslz3zHA-&EGKW6cf~?nX{_UZZoHpG=dRbf3saWp
zj^%b=mbk#Ilk}NI2INp%2}oSvV6z>YR#`ftDi~)}1@q^hpT2#mA8WbrPoL9qSM|>}
z3=K0Y)P$|7>e69$M9Z>AFgS%Um<2Z4-frE>jHM+4{hss<I_K;TD+*6(u(17NsWj(b^xfcHuZYVlc&oDS9pZ4j8Of_&>O9LwUF8L
zvw-JwyC8N@0bFKonVYl%o8z=j0I7LBv(HJ{vK;P(Z*oI|5jWB1Wgohno{3ndt1Xlu
z;ut}Yaw)(XN!KWcALp)XiDn~&Kg`ay86fVLT@60l7U$tn!7T|W;pTp&E@z(R6d@z)iD)SR~PbWHpk(ii9Ow)v*aBPRrVZ7
zjh<^`VP=r%9I*Kfn==?C>=TcGbw@NCgu7D#QBsJo1ZGv(l=~$aR~Jd7QA#li53@)P
z+r_0Js}%|>TxX$03x76uHn;yfu8IfD=&az-|~JR_q;+UX8GDD`j2L(Y%x1~96HDqh(YH;4JmuW9BIlJm9_DAMhld<6sj
znt%O%FWnsP
zA)|Lj@*n@>->~6|k@YFM9&GS2F^18)r|TqbVyO?KiNd`+&*lw%+4}vc(&T7MsT2D8
z*D>W5S>*Pq97#<$F$j<@7Cy>sh(M=DGBZ-THAUAzTP-sf7+{_-gMV;UmzU8J;IDQr
zsT&?J7qHp|Jnuj9G*^#9yggsjXyDsdZ+p(qyETb0Ce%vEG$&0F2yUwcgjPqD+Zs+=
zT2zq>!LPY%CHWMuQJ$vD>I>Nk2Fqt3;%2_GwVt@>B@FOTf>6-6EK}2ts&UTg&xK&x
ztH`!j1}rw|Zh#S!3Bln^8X<_Celi|kjeM_9@|0pz!4NuJz^@L^bgnxr6Lr`A4=&C4
z;9%|sCr4BUbX8n9)bevvIjKqSK%UI_(fr*V3&ID?1pDFrd&tb`V&C!Y%XgEVb*uE)
zo`Hg8X3
zt=l>4+9s!2St>&t{0$^K_j0<=aYE(W#JJ_cw6MLJfdPyNArz)$Hhytr{H~~<->=MrO59-eDW3M
z-#@aw<~*A97UD4zQ*5caeF&B8*i9}*UtnZfi|u-@rhc*DIk}ocKlLz+!m~Rl)%Ob>
zhg5C!qkRTIZiPW+rcIsgtaO$R&*FUDZDLx#E523ph3tfI#49&#P{DXtj))ZzGHmUc}H}ot~Is3$y=X&eu
zj;~O@{_?q`s*>eBE+TABq0w!94Ce*Iq&-!FQ3t1qXR*sq@Fgci|3FgoPsZ1|v$eYX
zvRI|}G=tRA*pimjx>y8TmFC9I#=XSg0W@iGA0hC)brH^_`wzpm_vkxteF|bsQI=?K
z7!#fxK&EKkuCGY10DLBKIzlsI2xko@4#0YL-5-QP3M5Uv5>8%C6Fu+GB+gcgiA&iT
z2&3XV5a6C=ZW1&p$_gC0e&i5WA9g49bNLy6?M~)@Ii{bEjOyoiQBD4UPMHQemlj3o
zCT*Ja%uJV$A(juGiPDBzCrs@sb#@%5&QN5XKFZr0^#N$z-oC%L$ZNhCm4MV=(jDo?
zfU}wyI273S#=|BLHlS8`ZIG!8a7whD4s&?P`j}S=H<7LhmHcL)XQt=RK3{Lh
zHO3M5R}83oqr{tl1(k{&$G)E@0cW1$CnDV)E%93dC7ZOgy&lFaNoNIPSc1C1YxL{{!6wDs_5rB}vA^^Kk4*c_{
zPf2oPL>?$cVJ(-%IX6{T2xyeqVA1TRxTe9itIseVGpXxrEegaVCGJ=y~`soPM6OV18Vg;n1f5%i=9@1bNJ`=G&S|J&Q1u?k1mAa#e
z8KyKJFAk`XY+KOj!dWecDuX|k3wLO4wXEel#6O&iy`h$K)k9J7>=W{5rd+6e^dH(>
zAQ3*;sdW%a7X&nv=*L(8%-D`9*A8IIDynfhg;q%JTCEK0p*R2>1QPIIv@(RTJpLXk
z)3dUX_8N8GwAR5tYW4lJx7+@mr)!1Nag3u7D`py$xFsEr
zwM<^ZXmrt-Civ~sBHPgxcjlzrK?>%nwDd<6QX>4vW}c-`QZ(3yaNHG+t_l2IL(^2P
zunyXMi1Ep=A3viZIr}O8AmH(^&UOKvkE?$G!4xjyNXyEH6<@nvYBk9)66S}`E@DmY
z^mwW)Vj`?L9q;@2V&~70pN}MzI*adhaEM{IN?iDP7wRrfo;HKgxU(QRj&$7?&Hgk~
z#U5~-XAq`1xyQD!iWxdO+uGC|jpSfCe>tBX)>b61nCSB-ByiF7K|}q%xLO>(cuTTU
z1a}?aA^*B>`1%P3!cjidV+6u0JdG2=y^jW1kWG1thsefFuUqfrhK~Ozo&jH**^T>`f;2iYVtzA)=AeiTW~Y
z+3f$&KA}pjFauN){3LL9D`(vT{F6l~dG+uA{@-J7VO9{YpT6T
z8C|$e4-$Wvyk?wL*OD`uVtf~v<$4K5rd*#zE=MA>ybW6hvN$Y~M&X14ZcOs;j_
zIJX?#YEEd}@a2SClp9(uRO_UmQ%ly9GRs(M>{kN1U_{6wGOB=S0%p>L{El2^9MO=e
zBQR>IE)&1%V~$Wx_q3j}?=UV0IpJs`{~Sq0@Q)vUhu=gIr)X()5=MWa+=Blmw^z6B
zZEvFPQ62Of!-z+sLvgk#R9rieD_2hlqSQ5t1|_C8x&D{0EncwKEdvn|kZdk`tPJL#
zd2cKayEu813lyfyZ-!;Pcobu6$Kpw3_~XTQYu(r_$B{cB!n$%MpRGHG$(DIwUixGn@Hfvkpf-`
zgwDxooz2lEq&Bt5vo^zj^7I}igfUma2=JIFZtl&CXbzt7tVhoP9C*vfpWUIpql)-)
z_X0?I()vW=PV?{ESjqk3th!0{1un)du{irY8xZLMoTSNq`@o1TVE{nC
z=&U1P21G|5jCRo^tko*&kxb@&LD)cX4wQgXnGXfZN+czy1`pDWc!mbLM-zyA>5r@e~Xy0loIZprPtr8Yjoqp14K;@%QiJVCE-J
z3>OmE>a{f}_@4mxub^q`G#A
zVTwZqc`Ba3H~*)X1v3xIlWE88@XSk3zsgyQm;*^4iYR4lVuWG-kLOq*7o9fAA@|M-K#g7zM0hI9JPpR50(TM&g_bK
zU6<-aXc4u4vZOjA(pC(W>lqYz6b;bh|AeMMXC^n*_!1D<+=D|(vmUG^Y}~?s?2#45
zsQ_jwBguCkKUi;Sj0^+qm^#W;$P|%9$npM=xQX>aHf#MoTMGZX8$KiBs5?ZN#cry7K(BtxH$yrk>9#CkiX(#&
z5BM(F8%JpWDAu}}{Ud|*s^Yrh64R@Wuhz}8sP^ZjU{2`SO>U^9{YbJ#em$w;{n_8c
z!%drD%2pqPz+)k4Pd@}F|A*M@0ZogilA&X{hPP={?4U9*^$pE2L^*iC&WME2
z^`n(}k#dC&7j)PE^D$sFqTJZ!(!QgXd>Wv)$pPj5P{^jEp8ye(#^*O~HG=9p(m61M
zVhS3#{e$aSycVD3sXVM}2_*YG23MYKy8uuByVx#@n3LR8Pb>Tn%Vfn1K7Raol0*oc
z1d=`Ma1D&mU~^bif|OWY0~H!v)C6z$GxM9O)_UZPaWF(AMI)aoUSA8@G6VE$a9o$5
zkMZ7RsKl5L@DU@W!Q(=e_(C|1J6QnF0&|{-Uog32M-n!t0%4*~eOkvmx_aqud6tiP
z_XtH8EPo^C$XJq5hX=H{F_Qq(ghfYLPPs~->S=2Mswk8#fjl3aRHz^=R3nyKVI>C#
z@f@dUMIOu13hdMv_T4^RxbA1*>3dtyi>&=tu4j3=rnYoq
z2=dvix__Zgk6I)Z&ssuC@_O3ix4PY}QPq(_il{9~UX*a>F;Bmt1z=2xd+s{OimcT_
z8xb9+_Y+oyUh}cgS!5$6ov3eH<(OL);RRIK_NiiTc;Kf{GS^vXhRWEF3%76$3~d^n
zg%q8j9kn&0u7i%091Q>t=kB;X$A3Vgzc{>KZZS*FOJ9N~^Em&~5U{1($P3gjz~>8o
zxi-sV+v#_&i%&(FiZmBt?8iOXwpMtf%!n2$f4`@0x1tg|uQu}`&oj31K<<{?9nJ=^EUr~y
znS@B%mn{oN7?%(vtHpmJA2Sl^mtQ|;V3H?lIt!eiC)Hw<$1czXCt4h=Bc
zDb9>+okx?p!JwzRGmM4*{HxBw?yOTs9Vm;lDd>`wM1p41gaw@Dyp2*$n69up8gLm4
zp5GrMnJ&AceJntDQ4SCa*;w*-3iHLQl&%WM)07LqARTv04pXgJBhlS8n6F*~p&5mx{AFdj2liaur+B3~*sk50Uh;S}oOs~?FPDVxhIZ(w
zaHtUjjFFkH=k@c8R@k2U)7W%85t2XN^&xkQ?xJl|1Fj!5{O3Ra0mAMe%gSDxYp3A1
zUti?T_7cmghgTWP2N(>E7&>$}9WomTVJ<4^>Lx)p
z%@b;F<%UsifzLzG?=G~3o*d?cp3)WR7)l{}AmzL1stoc^n{u>U7mH4rCGYmlYxIL`
zH(UU`~g}I-`1L8~iOV8JOw!`Da
zZ{B-m--ZLDlO`<({Pyz?u^39$gDLVX!u-}Q
zx=BEjYU%LoTSB1idRs@>7sR>n*r~c0cT-0=#7CPb>Rn_N^+qnR_vMEb;n$prI6r^&
zdL3`v6)ax9`|!c;O|4t71#jddq!)6@N^37lzViaavA!23Q2l^VebTPec9%Q>kj&9v
z@0^NDY6};=nqM2=+_%S-QYh<8m*p`406+jqL_t*h4`RtaXGNlg&uIR5{o>8fZ$$Ur
z;K>V6EwnOgokI~oa`Wr!di8HIse+%u+pn)I`sWKF#&0&i+>m~wSO@Hr%|CsnNo*ky
zA>sSia;!ApI-_7E(sBC|K1|?CwEF#bkAMi~_2|rl)iD*h8a>*&x=ttqmf$ZunVDdL
zW%YOmEbj-eIrBd4%=wd7GM2AB-Kz!5n7%S#^{xHjwmKu#yIDRJpZ_%9(D$;gM^9|O
z_4q1ER9iG|KP3xf=UMd&NfKdZ{}?X;SPG?9u`~lnwj|m2ooy|B`Q8_?bZ?iSQZPK~
zY(G#NoONBif!>QhT;PSBlx!?|NXMo_Z+7x4s<6@59oZ)`8P_}G3|*>Eu_p^%W3{M`
z?dtNsHx~ZIYlK#6$k1&mW~4tXwY+)#`P09At{^aW*EhbklI2@g`LZ`Yl2nDa%I-`dKAQ?th?NuBWHLl>E86=S)E%)_H3B4E#{
zo9ES!XyrRoZ~r&=R=R(B-09a>ODrycefjpuqtxF&e}2n-zJ7oA`n`;Y9@je;3QJpJOi2F#~BsOzIyz8@BRA
zs;=p8kqS%ZB)#QZN5y>@tvLL)joCsC-rq0`eY1U81r}Sj$t-Pfw-?Mtk0$=GrTSdi
zkqOP0zr8Qe{qjuP?>Y%I=ohreFIgb#XD#$=6Qp7Z{nyX`diOzxD@9?kyouFmOXps%
zly&bK?@ZO1zRK0HulJRu6q5bhfCT2#g(_#CswM;(MGtB9mIkiJVAvSnBUlx6_kDl4
zmQcubF-Js2p**_%N*Xc-`cs39#qwRjbH5Gw_bd^Gx5^
z?ldtJ?}BU^Oxkh_@%7e*$O>!{;~Cr5%7SGRe0a2_aWY43lxN~ih1IwDeWSX~+6uwo
z-L2^@m*8(DLF#m$ZJ>IM4WQxkne|j3`-4w1T?nCjzl@OAbj0f%#lLi_0oZ`8O0`_<
zXQ*8OK%r}V4d?DIdlt`>9Bm}z8YPwFlcnHrfgmvmpfag?)CQ9(0H5NLTf+&2ipcs^
z=;X7IY(#B6TCS}#LzDj2E_LvkAazHfxC-U4t~s_w2bj%2eJEABz2`k~s>{|ceQ6p;
zuJxu^7LQyblC?Bk*+D*c;k*{ly4p5Ch~QmRzS$#nmH0_V^o|<&-KiJwMpVbdfsq*L2qmlN4uyJ&RR5p~5_cz#ruzREj$a=Rt;#$ZjC;`tz5s
zOFfuT0lYxsM?q5K16sCF>v489}DcdMQ|3sEEpC*YO^TP@MDYx38k
zc3o5A%CpS4tf?dkTZ3`wK+_eu1HwU0_}M4qxr#5`>nT?cIr0Phq2X{>Ewt5XrSz7B
zXA~-S7bS5+mr2BJGVpo?^mSuT^|GQv=?)k6Ph=IBby<4GtV_
znJM-SHxtrsIm_v5aqeawQg8#g-CvTPUSKw@U?Da^s&pD8;Jvpfh}!I^
ztu*4uEr>cn}7FmKHnpvx<9!>sY?_%U
z5Rf~QtJaXjb+R|ZW^cY6p?Hsxd`i!o98m2&mr9C|`riZX-7uCS18)3&BRI)Tea{!jsexj`lP3
zhAFZ)Sg{7qM4K0@2N$@7v^u#Qg|`&9Sr@L+tU>s94Y-^*@1D=%auSGSKI?83wq&1;
zHb@TR+&P;W5|s4pZfvGvB~P5DDRtqzn0bAiqM8|`a>PLY>fiqDzmKe3
zVG!BP$O4#Xh~6NBX@2`@BxE$X-Rl`^`xpSgX>t;AU}|;-*<{kcK8Zh@)GX~}m@NN9
zMt~sA5@KF+q&-v2m$YWoaMDC9{K)eYd-7??9Hs|o7@h&gr?5L$q<
zb;TiRhf})^4s$mb$Dc%e-R&)I79g~B3{Dw|Z&s#XyuBB^+r$Q*+Vy1?dhAeS3tn1T
zLlHpzP?7(|Q@U^FX|a_&#YzH1oe}djbiwyI%?hoa%sVgk@`5+-Bx*l?|NU>5#iA!;1JEFEa3G%d%gc3z`df^nR
z{Oio*#Kg>Q`8ad
ztNww6#k-zI^Hm=Fc_~ujdKw9k6j8#!bZfU^6|cl!(MmKBja3J-ra&h4Bdmru3=5%|
z%(qFhXP>tX3&G@x3_Tot3e(qyc~zvOU-if7VLk8@E7_`0--jm)e_;+F3gqP{aw!E(
z#o(b0--0zhf_zG+!Za~rZ4I5-6gRTd1sySrWoG$#M)Fi^x9e$kUH7j&q8X`1FOp{y
zCEop(nr4|iWgIW7t|f|0)TNacf2Xg^9c2jcE|d`xI|Pi*
z@zPE&7-$(aXCq=VcvDuV2u*urp2%FFTsxCIVX}uh#mu_aH_=n4^E$?vL|D_1fxdT+DW
z1@n>%caW6cDVZ^(P&hCCt8zcV)U6Z-HL~oa(VYOWTo!gI8q*L|1ujvL<;=6vqHQ~-
zFA7cGCS%#5>J5E4#XH#(zUtI2I9)#}PnfbOCaD95QqOr|H(KrV|Fh|x7
zGRqY0jenZPf26On%0;EfijSg~PVxDXB!RW1^DG>k*dxxHQ%fXn7g
zHafz;8lhNdRplCWp?}tNa@zkz!`tQ3Y+5hYRaaXeaUYs+Dk+Jw%7-Zq(QHL-;!zLV
zJ2%&`?gJ#!P6V#4Q;`VT#C*%0Uf}XSWJ_+Vrg;Y-KQVUCjG%aO$wl-Fq`MuQ1hzNH
zp&ONoM3bkm4*Mli&P54?
zXMz_ZiGfgZ8{hEV0Bf9^ae1qOaBe;OfqxIT!-_VHB#8F@7pxI8?PjA-2x}j_n5G{j#}Bi7Kprb%HxwhCs0QVSEO{0?n&efBf;uSY2ibTDy>-{jR@#
zu`Km5cA%j<98wAqkHKUCrx?)+uZ!{+t^wxlEz|4WHI`u(p?0)C^6>vxj@;5>ve|Jz
zoyoKFjzTtm#Ek2va+pVz`SZ!V3eTukIaRV)*H6}8u>;uC#X7JGp2{WxZLz1L^H`oq
z5YAKV&a0cTIC+Y9bNg(*0S!J!ynmIWPjf;fRCOQ`aP1er@9%z_6VOp{R(-Pvormal
z^(A?32aJj*hoTsRcW$z`#B?+3lO!Ar7tKvNPh;^;qgK0#To1~p2nfKLl9Rg1phOQWpF>IJUMIE==D({$m&P_F&znlzVj@C()_rc)sTfh5k
z^CEE(pv<#(hRt%K;ecTf&&@)f^M-vZzmS|6un^bBtXkwK+BQJBa?;SPl7mh>#rC~a
zqFF1Ge6IJ`AV>N8cOPM+YJV#9S6E7M@asX{aq0g@h`k8aq+sJK<<~$tTB%oy3Jmt;qN4Pls7KXg{^<
z0$*YtNECsDEgyN+2thQ!-OYV~beGVa<-NPiM8W79rkcE%v7rQk2&Rh(=bIT*_+bF5
zx<))fQI)wAR1WY;67o~JgKGHs5nZs)VW`FU-E1puTGf`KxlmEhjO=Bw*cav6l%qwZ
z6KSpiQby5NQZ)1Ww+|+6pkp3yKd^k-!eoe+TFUUTt5#E~McUVbp(z-Y%#NtOa5uq1
zGiJ`&=!%U!=`;+RWd}R@3z)8*-O>?Xf8=MaEJ>h1l%sZfnO!X)9H$zosDOW}C)lD5
zaWo1|o4~7f8=cxrHj_ScZpbIy*=b`w0{$Ztoyv
zKgS>qSuK$iwh!j>=RewR^`|}TUQ2#2)5@c8?V#84X`D8aQXTz#6@IC0<5!Cp=lc;>
zsBPnjM%DF&CoML4<5`C{@4Wd>-nPEBK@~7hIkBp#!>X3&Mcn`O`Kvu8SoaR$#H%_l
zG8?jKAd!W)q$XTl8bN(t)K1LkQQp%NWzx{IOQyzG77?g(HNI(q1DGETV|G10WTa!Wzr@nT?Pnw1F;GC3?`1I8)q;GP{l+(^4vfzZPWaE
zTU(!ONwK_6zkO07+0Y02-kOm^OVS%h8H#H=+nO6N)!{QgTfK5ujVevT+PXAFIkelJ
z#XDYo{^MUvLP_K`c-d(4&6__yeNvSuAlEE^KmA5+;rMvbIPWdm=2cvHqjB-W)JE#x
z_%mDo?N9I7N2wNP0Aap`Fr3wf`NYCh5<9aKj1hy%Pk;NvGk}JJ`wK+
zOY@SBru&eVY1ffxcVKNaQyoIw%>ipSz3(6^{~{otpw4$ZH_Ox@y|;O3cLAui7R33-
z95z8-z;vd1U*f_$`Jh2t_!bT`30C_^{b4O$g};Q?1lxN;@nIX*>nL}Tnk>GB;bnt-
z<&29+Gm>qy9`V@&HcvI|ewit&E=tT7)HT=e5w@`qkLe`U6nw~&RYH!x{jIoum~>1~
zl}8e%H$06U!YtKD4trp+1m(})K7W-^W!&)GT^9u9o2@n0#$C!a(U`vUN^a%#CTiL|
z+3IsIYcH@~ZaZ(prNt&U*gqRt8YSEB*_@BQmR1m!criU>Yy6l?`V%k%8w|
zFMZ96blDXxG|z0e`>Y5@^ah8zuGMPJZx*Kz%As|xl@5jM!jn=BCSS6dSO56?e{hkx
zXV_J6X!j^nQmuAW*immQrG!_Sl61rZ_F~4OaCrT3`+-W@5T`qV0TDVWkjj~3q4=fm
z2Q+!B6{AMT0k0KN1sX}W+*Ph9&af`aU~T);)d|ZQZdf6WYtvgZ$}39J70x8Hmi
zgQ56?R)z#dP~&;^Q^n=18L)EK(@MNZ>FP3<68niN<3*5&e1@r+mAEk0f!4tjuCSp@
zGAWX!Ds7#*5c#q%uqDP-ch))@;Q80T{pFA`~tLbYE9n&ALWwtKG_dX$&R?xTEE=Fic>PV~I>HEqwt={8g5=K^2BvHPwK_PWF23#IYqh+LBlcRa@|-{`UvU?CXJGl=gI{0)
zUavfbVoeE>V$~;jzM0q83c`*&LV-z%pChx#hEmtEWOWl+8j1UeDb%FXIAX5#e?qgz
zV(J~nS<8+G3@pg)=VngInOPOj2mobW&DagB2k(MGOr2I?vz~q-N+cmOBUkL3dsw8`>%n4w5heGm%kZKynX*xLhl+$
z(G$<16+@~x{VyBivtff-c|H&Mc)5}}b^5XKecyEy)ibMQIl8T~v?B(PhjNM-*)TuE
zY^_n4b2iSXdZ>E6)fA-SMi*7lZcErRGw0n$l8c)yZxkxW#m0$UyVHECA9Lo>>PPSh
znI3$E#nbpH{Cav!>wc=jh$c^-Teqm3klAr@ZF@O-g9<-WA3owzYs+7%vYz0Jjpc%_
z2}h%F!i+o!(U$E^8j1`sK74p@(>lWyJwLvEFE7~#{eCf*xo{B=NyyeS!g)tGW3Z@#
zjx+VdwO)xa0Q19Hb13g&(76L7>t;)p?$!j7_Md(-pYmj$%NKFV17U#e&MG>R;HtcT
z_rb8t?`kbOcM~wd4kT|X0PO&9wS?=lz$jF)p3BBi)1sVBw~`A`J;C`CgLy&ZX^|^g
z3UB<`Fl>g7>EeJ!9WIbJUPowOoNAEALsE+EmH*d2{weM#IyRiE_Sj8d7t`oa9gi-W
z$smAUO(aZ&ZV|SQV^b^?TdvQAVQ4$&vQ*@h#BEV$f}-+pwmqeE5QzmNKG-G;T8w1g
z;+=C}&kpeCbGnc!tc6+zR!-wczOcOd?aLpZ4bfrNmUF;281xOF|
zatJ@X?Im7PK%}w~M8=_^dGck?MSPI^%lZq08Wi18SeD<@@<$AqC&DmyRTMQ9L|C6c
z_n?Wojf4!RUcUJB`45W-QwvJ5Tz@KOF;kL`ge30*_}X?N(1s56Ds@QMyuM)?*_R~H
zki|Xnp@)@fPX7}r3#w=nHQ7ZMb?`i;YfL%s;>f5=K+NSNLs_?7o|f>7>a!U<^~9Va
z^x%TIHcR~i!6mG!)G|E5gF-|lr5>rUS`N=O&obKmpw{(o08nmKs+5O^l8hSczvovT
zZ{?6pgrZenNrt3_X%KHH5ay5ihy`Dmy$Dh)o8nZ|pRBwsn@Kd@`jdvLBJ|+VL5uVy
z97bzP$<$)VG8TEyZ_V^4jXL;+^=OLYB%=T|2^ks$CMhjvKu;saddN~VWpc`&VwAgoca@WQI&AYl^P
ziIgRz2Xwjd%0rKRSXKq%GD@eKNtn0MkQqg3M<#@EGiE(+8T{RJ6zQ3%m^vAlk%+3+PTbmviL2b<&%y$y3fg8**%!
zIAU0}j@eI5fOjz1b~Q-r>)V=LgP+Kw278N8s&G|xp<<;zTIx=z2VFU}q})OZ3=Rs!
ztpF{1X%_HJ+a+NjI$Q6_kWHl<%+q>dq_*eYm5BVhiFsc!#Dzz3CtF=s434~1-&{sC
zW-z2OQp39UU{tKSgFNRis`pejsXn=C@pDN-9XUu?fKLg2m6Fm=1W!FXPo#UP!fGhf
zUb0%BuA|6M|XTUVj0G=$xxiR#
zlq^I!@e*=lgs=Pl@1-V8sgawi`)a(LGVf-`+MYb$V>Lz*oe44B%${;xmqp=*+t6*C@Ws2dt%&tQO$t|P`fFj
zL#>N}(K6pzI`mk(2SR9)&cDNNIPFnW1AIY9Gw>@l=#z2s9s~v^)+HRH3bEWQtnX%7-v`
zwwVp5>8_^_*NqF01{bSaxW5EZNvv+wRmH=m#5E7j-qK|2gQKoi+#TkGS8plx6G6AF
zK7wa?@Tf>vYHLgTsJ&>b5is-$M(e;Jymnn*Kfyr|0C&N3!#i9a3~Yz|D1i-6P8R0HwkD(zeaq
z)R&ATW~ONoE#8a4$?!=1^y$;z{`NOgvQ<}G^vFsPWYXw3#lCVfRCfSzp>7EGzkJdb
z<9f2$1{!AtHC<5FmZxi_1-#A=hyF!}FE4M(7!$e+Ge*c~5;#q^_
z>2u^+#g^J%_GI*Y&0Ty^5;P43>jf1R1`!LM1k6=Dr4HS6oQT}jnx73_yLu3-u4n|t
z1(-dvl^o36%*Oumlo;NTNsJ!EsB}#z92p{*z>o2bqWlpxfUYQJOBRs?;343dcKGD`0noH
zJYL3TCP02%!|O%9f6cGqM}QKPNPm@Ro@SZlBzHZmODd=z33EqK*g@3U>LpDE!+erN0xl%oIY;bhbb2O`%xnAq8$7W3@csWLXpgU6~P1*hR>fip`e*@ICUfo8A
z8&ZeZm};R$v%o5J3SD!d^*4KWB1w-33XKgCAq|u8;E2yEWDIi;s9yc{?eiCil{_~T
zS9Z)8Xd8QE>w91bFa)3OX
zx)yYbxV2v3AYD3biD&zn5aga|i41tD-YCqQ(<`%RY635}$!}SX4o;JMBpCs~oWZN;
z7;TNfw4#vQ`Xp;Hx4?QVJD(`v_r=~)&g>ibS#!GM_KYQRJ?!}uod*E7u1nC&U?zm@
z<>U#i5+u2G^bjUmY*bVY7XY*lXAh@kmL24?C^0RV`s(Qx;Md6goMxAl}0iZgG80H{tUe&
zQO>hzX?&dw6Yw3!!hhj6`>8VyUBER5TU|KGz;qKEF5KN?C~PsaOqCvxcg>%eN#m?X
zbJs5Jv;#jgUvwzIIWC^oT8~DRu2s5m*LpN5LNgg;W572dWgADp$WId&pwH%)DrAJy
zSH#;%D^2-UUh?AohY!fO>Sv#uy)I|cBXSY9v~uic!|0lMAqx3*VVrLBSSwFTXgm9N
zhJ=*>e&&}v;W&CI4!rZ>tk%3hbR+25R~?2(QgxNcv_~qUM~U;Ad!&pQku80*o2Q78
zhiaGJg$~`u|NDRcKRR_1*FXLC&Z>)@sXe^U-%WDeqw%Vs!ExlZoJEPBZMv#09b}nY
zeprjF>yJPF#WEAAo%J&tMkQ;S
z6mh3H%5tIOe9}%iT**%MkB~ymsnUg?8uah3rw%g~FJJuezlHf9AAfJNGGJ&4YHg)!
zqJt*GUzX!;8A1meKt`Fg^Fh{=u3fjgmz^{MKeL3EY!DPb2WN=hmfTh$Jbq`_O(sKy
zvzaJ;;sbsM%1NH9lEV-P4og`U&*6OLE(P6DuUAEPXVm%is8ismBZC;f_xqxR!bzjn
z$P?K|?=Ip{f2pr0uG=v)eWE*li39+N)iEP!`PON6YAWU3E!7QR`ks@3A7nZ*v~KX0
z%y)!pUI~C_1tK(UCzTR|6XzNoeVvD@X~>ta{`mASJN1ifl3mJ}o1j^^69&0hRin_S
z!{2Q2y>_=(`_Y?guJx9>_vEY=B4I$45~Lt$qRzTBfi;H-Hb0q)^SXoFla-W~j!s85&5j6n%BLo)_yt7Uy9W4;8YH%sLZsH#)R!dGkq{ICk3syrc3Ikh|5_=^WoU@IEcsY62yYu8FI@e+yDuqtt
zj0aWgCt?kral>OC8uyPM76Xs&lO7SQE1d;}S=%^D+v=GR1)P$U^U&e|&0gw5-z663m(78w{l3^C%V
zx{Xb_Dq^ZNdnS3;|MfW8nwlCH)xC;pgy-p?ceOAE|BlC~FH~0B>zj*f2QFb!w^@`&
zUV^kd0*~rHY~?Yke&D%A!9ZdfV<|Fktvl4HM#K%7_rH7f@dn@PRxIi<
zWAXZ#!V%vW<(1^)y{W!oNj=4TZ?|bbdzQaBl5}kea$_Xrh$&1Z>QjMf%OXQ1>t9`@
z@AV(t4bgrn6HqN`(;`aF)S<#KV#o$xdF}$Nw!pjA#BgTGBh?TqyG_Mwy|zY)yAIdv
zgeq3tNj;G6xa$_2ln%qOfMwPd-yPv}OgIe)?gYMqxkVI8MtNQXk0$KQagmg#U+^h4mR)wZz@oCJV?ElUis}ke
z;$wSd0dqeOf3;u^tjUWD1^S@tRWEUjJdf-of%27Ac?c?zA>ewz5lBL6Sv1T7M_a-i
zVSwwS75Zt|0->J}I`C90JMc2GJp8eAaz~RPTfl+&k{)D(ga^(WWCDaIc8FzB*WC(X
zS?@W>5^wimg97xnA^
zT_`!PApFC5A;UGyaN>y?{RFJaEd4KPRmUs-m>QMv@WZJh@^XH&BlDA^6Z~oy6V$|>nAW3*-sD#CDdx>kPI$PQ-5v4Fl?nkPo>&H678M<Oo1#fSBbIv(4|N*9nF4Yg&$}iHVF-Rrqs<(Ha!%wHO(}6f6c=39mV%mjHVEG*{J5%r%(UV1oRGA*;8Un2#I>s
zlH>*7{9-J?_Xtu-fE}cY%JF-Gb1U9DBHID%{^`1GTl!KID%)$-w7WCKY>TC(rbB+2_wq`)%mdhe6$M
z6Yz!&>o)Rt((~8kTin10rZ@(hGC~)zO{upmxVKPI3SOp#V#v9<%Bzyl8s&_b!-jek
zw&L>hmk*{BRbvp?_a;;rR6+3jckhkWML+DL4TJhE=bIqWiiAv~E;Ix`3BsYJli>-(
z^^t*->CwNJuom!-$V5IZ0f@<`5Jq&m%OpeVK-m^IU5;F_q7YR6BHZW}Jpaqf
zp7!T(w|H3=0!Dv<97y>-Lg_OXa9!js<>7w`$vfu9&?;8e5_-lei)>R&xb1Dqhp5)$51^M!CY(d
z8M(UvsR&=r1S0bB=;A{=h|1eeQK@W?N|GUNv|?7mwP)@SuA1s60q~?jd~m}ba}&Crk%o034@&UXTX}Ctzp~!+4+N%*C2h-GtOyseGQ^n^GRuvN)_O_z53_h|IX4I7Cm)Z;L_0Yaai(3QF|$TDw9DAB~-piI{@0C#s(kyZ1K*%
z>3Iv33bUQ8NxfdY(u?KRd*OjioWdZ%GeUPU_f0qPHdR(R0xeqYY11Yb+07D*;|-X%
zNsB2&a3nc9$Tayzz5m0k@T$xTD-k7u;o5OD?B~Ohw5<4)bY~?&gYpfMv1V%
z`I_?7+~ow%7yvjR*ZMW#tS_07