Hugging Face's logo

Datasets:
Bluwynd
/
LoraTrain15

Languages:
Portuguese
Tags:
art
lora
trainer
Dataset card Files Community
Bluwynd commited on
Commit
45ae36b
·
verified ·
1 Parent(s): 5c95ca9

Delete autotrain (1).ipynb

Browse files
Files changed (1) hide show
  1. autotrain (1).ipynb +0 -773
autotrain (1).ipynb DELETED
@@ -1,773 +0,0 @@
1
- {
2
- "cells": [
3
- {
4
- "cell_type": "code",
5
- "execution_count": null,
6
- "metadata": {
7
- "colab": {
8
- "base_uri": "https://localhost:8080/"
9
- },
10
- "id": "efSw4FzN89ia",
11
- "outputId": "e8733152-31fd-420f-8a14-0bca6af46641"
12
- },
13
- "outputs": [
14
- {
15
- "output_type": "stream",
16
- "name": "stdout",
17
- "text": [
18
- "Cloning into '/content/Kohya-Colab'...\n",
19
- "remote: Enumerating objects: 2158, done.\u001b[K\n",
20
- "remote: Counting objects: 100% (893/893), done.\u001b[K\n",
21
- "remote: Compressing objects: 100% (231/231), done.\u001b[K\n",
22
- "remote: Total 2158 (delta 738), reused 671 (delta 662), pack-reused 1265 (from 1)\u001b[K\n",
23
- "Receiving objects: 100% (2158/2158), 4.43 MiB | 6.82 MiB/s, done.\n",
24
- "Resolving deltas: 100% (1418/1418), done.\n",
25
- "The following additional packages will be installed:\n",
26
- " libaria2-0 libc-ares2\n",
27
- "The following NEW packages will be installed:\n",
28
- " aria2 libaria2-0 libc-ares2\n",
29
- "0 upgraded, 3 newly installed, 0 to remove and 49 not upgraded.\n",
30
- "Need to get 1,513 kB of archives.\n",
31
- "After this operation, 5,441 kB of additional disk space will be used.\n",
32
- "Selecting previously unselected package libc-ares2:amd64.\n",
33
- "(Reading database ... 123621 files and directories currently installed.)\n",
34
- "Preparing to unpack .../libc-ares2_1.18.1-1ubuntu0.22.04.3_amd64.deb ...\n",
35
- "Unpacking libc-ares2:amd64 (1.18.1-1ubuntu0.22.04.3) ...\n",
36
- "Selecting previously unselected package libaria2-0:amd64.\n",
37
- "Preparing to unpack .../libaria2-0_1.36.0-1_amd64.deb ...\n",
38
- "Unpacking libaria2-0:amd64 (1.36.0-1) ...\n",
39
- "Selecting previously unselected package aria2.\n",
40
- "Preparing to unpack .../aria2_1.36.0-1_amd64.deb ...\n",
41
- "Unpacking aria2 (1.36.0-1) ...\n",
42
- "Setting up libc-ares2:amd64 (1.18.1-1ubuntu0.22.04.3) ...\n",
43
- "Setting up libaria2-0:amd64 (1.36.0-1) ...\n",
44
- "Setting up aria2 (1.36.0-1) ...\n",
45
- "Processing triggers for man-db (2.10.2-1) ...\n",
46
- "Processing triggers for libc-bin (2.35-0ubuntu3.4) ...\n",
47
- "/sbin/ldconfig.real: /usr/local/lib/libur_loader.so.0 is not a symbolic link\n",
48
- "\n",
49
- "/sbin/ldconfig.real: /usr/local/lib/libur_adapter_opencl.so.0 is not a symbolic link\n",
50
- "\n",
51
- "/sbin/ldconfig.real: /usr/local/lib/libtbbbind_2_0.so.3 is not a symbolic link\n",
52
- "\n",
53
- "/sbin/ldconfig.real: /usr/local/lib/libur_adapter_level_zero.so.0 is not a symbolic link\n",
54
- "\n",
55
- "/sbin/ldconfig.real: /usr/local/lib/libtbbbind.so.3 is not a symbolic link\n",
56
- "\n",
57
- "/sbin/ldconfig.real: /usr/local/lib/libtbbmalloc_proxy.so.2 is not a symbolic link\n",
58
- "\n",
59
- "/sbin/ldconfig.real: /usr/local/lib/libtbb.so.12 is not a symbolic link\n",
60
- "\n",
61
- "/sbin/ldconfig.real: /usr/local/lib/libtbbbind_2_5.so.3 is not a symbolic link\n",
62
- "\n",
63
- "/sbin/ldconfig.real: /usr/local/lib/libtbbmalloc.so.2 is not a symbolic link\n",
64
- "\n",
65
- " Preparing metadata (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
66
- "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m100.3/100.3 kB\u001b[0m \u001b[31m3.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
67
- "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m266.3/266.3 kB\u001b[0m \u001b[31m12.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
68
- "\u001b[?25h Installing build dependencies ... \u001b[?25l\u001b[?25hdone\n",
69
- " Getting requirements to build wheel ... \u001b[?25l\u001b[?25hdone\n",
70
- " Installing backend dependencies ... \u001b[?25l\u001b[?25hdone\n",
71
- " Preparing metadata (pyproject.toml) ... \u001b[?25l\u001b[?25hdone\n",
72
- " Installing build dependencies ... \u001b[?25l\u001b[?25hdone\n",
73
- " Getting requirements to build wheel ... \u001b[?25l\u001b[?25hdone\n",
74
- " Preparing metadata (pyproject.toml) ... \u001b[?25l\u001b[?25hdone\n",
75
- " Preparing metadata (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
76
- "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m4.0/4.0 MB\u001b[0m \u001b[31m54.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
77
- "\u001b[?25h Preparing metadata (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
78
- "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m43.9/43.9 kB\u001b[0m \u001b[31m3.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
79
- "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m191.5/191.5 kB\u001b[0m \u001b[31m13.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
80
- "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m6.3/6.3 MB\u001b[0m \u001b[31m82.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
81
- "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m53.1/53.1 kB\u001b[0m \u001b[31m4.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
82
- "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m125.7/125.7 kB\u001b[0m \u001b[31m9.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
83
- "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m61.7/61.7 MB\u001b[0m \u001b[31m14.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
84
- "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m41.6/41.6 kB\u001b[0m \u001b[31m3.2 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
85
- "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m503.1/503.1 kB\u001b[0m \u001b[31m28.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
86
- "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m825.8/825.8 kB\u001b[0m \u001b[31m41.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
87
- "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m92.6/92.6 MB\u001b[0m \u001b[31m8.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
88
- "\u001b[2K \u001b[91m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[91m╸\u001b[0m \u001b[32m475.2/475.2 MB\u001b[0m \u001b[31m130.3 MB/s\u001b[0m eta \u001b[36m0:00:01\u001b[0m"
89
- ]
90
- }
91
- ],
92
- "source": [
93
- "import os\n",
94
- "import zipfile\n",
95
- "import shutil\n",
96
- "from subprocess import getoutput\n",
97
- "from IPython.utils import capture\n",
98
- "import random\n",
99
- "import concurrent.futures\n",
100
- "from tqdm import tqdm\n",
101
- "from PIL import Image\n",
102
- "import time\n",
103
- "import re\n",
104
- "import json\n",
105
- "import glob\n",
106
- "import gdown\n",
107
- "import requests\n",
108
- "import subprocess\n",
109
- "from urllib.parse import urlparse, unquote\n",
110
- "from pathlib import Path\n",
111
- "import toml\n",
112
- "\n",
113
- "#root_dir\n",
114
- "root_dir = \"/content\"\n",
115
- "deps_dir = os.path.join(root_dir,\"deps\")\n",
116
- "repo_dir = os.path.join(root_dir,\"Kohya-Colab\")\n",
117
- "training_dir = os.path.join(root_dir,\"Dreamboot-Config\")\n",
118
- "pretrained_model = os.path.join(root_dir,\"pretrained_model\")\n",
119
- "vae_dir = os.path.join(root_dir,\"vae\")\n",
120
- "config_dir = os.path.join(training_dir,\"config\")\n",
121
- "\n",
122
- "#repo_dir\n",
123
- "accelerate_config = os.path.join(repo_dir, \"accelerate_config/config.yaml\")\n",
124
- "tools_dir = os.path.join(repo_dir,\"tools\")\n",
125
- "finetune_dir = os.path.join(repo_dir,\"finetune\")\n",
126
- "\n",
127
- "for store in [\"root_dir\", \"deps_dir\", \"repo_dir\", \"training_dir\", \"pretrained_model\", \"vae_dir\", \"accelerate_config\", \"tools_dir\", \"finetune_dir\", \"config_dir\"]:\n",
128
- " with capture.capture_output() as cap:\n",
129
- " %store {store}\n",
130
- " del cap\n",
131
- "\n",
132
- "repo_url = \"https://github.com/phamhungd/Kohya-Colab\"\n",
133
- "bitsandytes_main_py = \"/usr/local/lib/python3.10/dist-packages/bitsandbytes/cuda_setup/main.py\"\n",
134
- "branch = \"\"\n",
135
- "verbose = False\n",
136
- "\n",
137
- "def read_file(filename):\n",
138
- " with open(filename, \"r\") as f:\n",
139
- " contents = f.read()\n",
140
- " return contents\n",
141
- "\n",
142
- "\n",
143
- "def write_file(filename, contents):\n",
144
- " with open(filename, \"w\") as f:\n",
145
- " f.write(contents)\n",
146
- "\n",
147
- "\n",
148
- "def clone_repo(url):\n",
149
- " if not os.path.exists(repo_dir):\n",
150
- " os.chdir(root_dir)\n",
151
- " !git clone {url} {repo_dir}\n",
152
- " else:\n",
153
- " os.chdir(repo_dir)\n",
154
- " !git pull origin {branch} if branch else !git pull\n",
155
- "\n",
156
- "\n",
157
- "def install_dependencies():\n",
158
- " s = getoutput('nvidia-smi')\n",
159
- "\n",
160
- " if 'T4' in s:\n",
161
- " !sed -i \"s@cpu@cuda@\" library/model_util.py\n",
162
- "\n",
163
- " !pip install {'-q' if not verbose else ''} --upgrade -r requirements.txt\n",
164
- "\n",
165
- " from accelerate.utils import write_basic_config\n",
166
- "\n",
167
- " if not os.path.exists(accelerate_config):\n",
168
- " write_basic_config(save_location=accelerate_config)\n",
169
- "\n",
170
- "\n",
171
- "def remove_bitsandbytes_message(filename):\n",
172
- " welcome_message = \"\"\"\n",
173
- "def evaluate_cuda_setup():\n",
174
- " print('')\n",
175
- " print('='*35 + 'BUG REPORT' + '='*35)\n",
176
- " print('Welcome to bitsandbytes. For bug reports, please submit your error trace to: https://github.com/TimDettmers/bitsandbytes/issues')\n",
177
- " print('For effortless bug reporting copy-paste your error into this form: https://docs.google.com/forms/d/e/1FAIpQLScPB8emS3Thkp66nvqwmjTEgxp8Y9ufuWTzFyr9kJ5AoI47dQ/viewform?usp=sf_link')\n",
178
- " print('='*80)\"\"\"\n",
179
- "\n",
180
- " new_welcome_message = \"\"\"\n",
181
- "def evaluate_cuda_setup():\n",
182
- " import os\n",
183
- " if 'BITSANDBYTES_NOWELCOME' not in os.environ or str(os.environ['BITSANDBYTES_NOWELCOME']) == '0':\n",
184
- " print('')\n",
185
- " print('=' * 35 + 'BUG REPORT' + '=' * 35)\n",
186
- " print('Welcome to bitsandbytes. For bug reports, please submit your error trace to: https://github.com/TimDettmers/bitsandbytes/issues')\n",
187
- " print('For effortless bug reporting copy-paste your error into this form: https://docs.google.com/forms/d/e/1FAIpQLScPB8emS3Thkp66nvqwmjTEgxp8Y9ufuWTzFyr9kJ5AoI47dQ/viewform?usp=sf_link')\n",
188
- " print('To hide this message, set the BITSANDBYTES_NOWELCOME variable like so: export BITSANDBYTES_NOWELCOME=1')\n",
189
- " print('=' * 80)\"\"\"\n",
190
- "\n",
191
- " contents = read_file(filename)\n",
192
- " new_contents = contents.replace(welcome_message, new_welcome_message)\n",
193
- " write_file(filename, new_contents)\n",
194
- "\n",
195
- "\n",
196
- "def main():\n",
197
- " os.chdir(root_dir)\n",
198
- "\n",
199
- " for dir in [\n",
200
- " deps_dir,\n",
201
- " training_dir,\n",
202
- " config_dir,\n",
203
- " pretrained_model,\n",
204
- " vae_dir\n",
205
- " ]:\n",
206
- " os.makedirs(dir, exist_ok=True)\n",
207
- "\n",
208
- " clone_repo(repo_url)\n",
209
- "\n",
210
- " if branch:\n",
211
- " os.chdir(repo_dir)\n",
212
- " status = os.system(f\"git checkout {branch}\")\n",
213
- " if status != 0:\n",
214
- " raise Exception(\"Failed to checkout branch or commit\")\n",
215
- "\n",
216
- " os.chdir(repo_dir)\n",
217
- "\n",
218
- " !apt install aria2 {'-qq' if not verbose else ''}\n",
219
- "\n",
220
- " install_dependencies()\n",
221
- " time.sleep(3)\n",
222
- "\n",
223
- " remove_bitsandbytes_message(bitsandytes_main_py)\n",
224
- "\n",
225
- " os.environ[\"TF_CPP_MIN_LOG_LEVEL\"] = \"3\"\n",
226
- " os.environ[\"BITSANDBYTES_NOWELCOME\"] = \"1\"\n",
227
- " os.environ[\"SAFETENSORS_FAST_GPU\"] = \"1\"\n",
228
- "\n",
229
- " cuda_path = \"/usr/local/cuda-11.8/targets/x86_64-linux/lib/\"\n",
230
- " ld_library_path = os.environ.get(\"LD_LIBRARY_PATH\", \"\")\n",
231
- " os.environ[\"LD_LIBRARY_PATH\"] = f\"{ld_library_path}:{cuda_path}\"\n",
232
- "\n",
233
- "main()\n",
234
- "\n",
235
- "\n",
236
- "print(f\"Your train data directory : {train_data_dir}\")\n",
237
- "\n",
238
- "os.chdir(finetune_dir)\n",
239
- "\n",
240
- "config = {\n",
241
- " \"_train_data_dir\": train_data_dir, # Manter a referência original\n",
242
- " \"batch_size\": 8, # Valor do segundo código\n",
243
- " \"repo_id\": \"SmilingWolf/wd-v1-4-convnextv2-tagger-v2\", # Valor do segundo código\n",
244
- " \"beam_search\": beam_search, # Manter a referência original\n",
245
- " \"min_length\": min_length, # Manter a referência original\n",
246
- " \"max_length\": max_length, # Manter a referência original\n",
247
- " \"debug\": True, # Do segundo código\n",
248
- " \"caption_extension\": \".txt\", # Valor do segundo código\n",
249
- " \"max_data_loader_n_workers\": 2, # Valor do segundo código\n",
250
- " \"recursive\": True, # Do segundo código\n",
251
- " \"remove_underscore\": True, # Do segundo código\n",
252
- " \"general_threshold\": Threshold, # Do segundo código\n",
253
- " \"character_threshold\": 0.50 # Do segundo código\n",
254
- "}\n",
255
- "\n",
256
- "args = \"\"\n",
257
- "for k, v in config.items():\n",
258
- " if k.startswith(\"_\"):\n",
259
- " args += f'\"{v}\" '\n",
260
- " elif isinstance(v, str):\n",
261
- " args += f'--{k}=\"{v}\" '\n",
262
- " elif isinstance(v, bool) and v:\n",
263
- " args += f\"--{k} \"\n",
264
- " elif isinstance(v, float) and not isinstance(v, bool):\n",
265
- " args += f\"--{k}={v} \"\n",
266
- " elif isinstance(v, int) and not isinstance(v, bool):\n",
267
- " args += f\"--{k}={v} \"\n",
268
- "\n",
269
- "# Verificar qual script executar com base em NoAutoCaption\n",
270
- "if 'NoAutoCaption' not in locals() or not NoAutoCaption:\n",
271
- " final_args = f\"python tag_images_by_wd14_tagger.py {args}\"\n",
272
- "else:\n",
273
- " final_args = f\"python make_captions.py {args}\"\n",
274
- "\n",
275
- "os.chdir(finetune_dir)\n",
276
- "!{final_args}\n",
277
- "\n",
278
- "os.chdir(root_dir)\n",
279
- "\n",
280
- "extension = \".txt\"\n",
281
- "custom_tag = CustomCaption\n",
282
- "\n",
283
- "def read_file(filename):\n",
284
- " with open(filename, \"r\") as f:\n",
285
- " contents = f.read()\n",
286
- " return contents\n",
287
- "\n",
288
- "def write_file(filename, contents):\n",
289
- " with open(filename, \"w\") as f:\n",
290
- " f.write(contents)\n",
291
- "\n",
292
- "def process_tags(filename, custom_tag, append, remove_tag):\n",
293
- " contents = read_file(filename)\n",
294
- " tags = [tag.strip() for tag in contents.split(',')]\n",
295
- " custom_tags = [tag.strip() for tag in custom_tag.split(',')]\n",
296
- "\n",
297
- " for custom_tag in custom_tags:\n",
298
- " custom_tag = custom_tag.replace(\"_\", \" \")\n",
299
- " if remove_tag:\n",
300
- " while custom_tag in tags:\n",
301
- " tags.remove(custom_tag)\n",
302
- " else:\n",
303
- " if custom_tag not in tags:\n",
304
- " if append:\n",
305
- " tags.append(custom_tag)\n",
306
- " else:\n",
307
- " tags.insert(0, custom_tag)\n",
308
- "\n",
309
- " contents = ', '.join(tags)\n",
310
- " write_file(filename, contents)\n",
311
- "\n",
312
- "def process_directory(train_data_dir, tag, append, remove_tag, recursive):\n",
313
- " for filename in os.listdir(train_data_dir):\n",
314
- " file_path = os.path.join(train_data_dir, filename)\n",
315
- " if os.path.isdir(file_path) and recursive:\n",
316
- " process_directory(file_path, tag, append, remove_tag, recursive)\n",
317
- " elif filename.endswith(extension):\n",
318
- " process_tags(file_path, tag, append, remove_tag)\n",
319
- "\n",
320
- "if not any(\n",
321
- " [filename.endswith(extension) for filename in os.listdir(train_data_dir)]\n",
322
- "):\n",
323
- " for filename in os.listdir(train_data_dir):\n",
324
- " if filename.endswith((\".png\", \".jpg\", \".jpeg\", \".webp\", \".bmp\")):\n",
325
- " open(\n",
326
- " os.path.join(train_data_dir, filename.split(\".\")[0] + extension),\n",
327
- " \"w\",\n",
328
- " ).close()\n",
329
- "if not NoAutoCaption :\n",
330
- " process_directory(train_data_dir, custom_tag, False, False, True)\n",
331
- "\n",
332
- "#3.Setting\n",
333
- "\n",
334
- "MODEL_URLS = {\n",
335
- " \"GSMaletoPhotoreal_v4\" : \"https://civitai.com/api/download/models/164715\",\n",
336
- " \"GSMaletoFusion_v1\" : \"https://civitai.com/api/download/models/138518\",\n",
337
- " \"GSMaletoAnime_v1\" : \"https://civitai.com/api/download/models/503605\",\n",
338
- "}\n",
339
- "MODEL_URL = MODEL_URLS.get(Model, Model)\n",
340
- "drive_dir = os.path.join(root_dir, \"drive/MyDrive\")\n",
341
- "def get_supported_extensions():\n",
342
- " return tuple([\".ckpt\", \".safetensors\", \".pt\", \".pth\"])\n",
343
- "\n",
344
- "def get_filename(url, quiet=True):\n",
345
- " extensions = get_supported_extensions()\n",
346
- "\n",
347
- " if url.startswith(drive_dir) or url.endswith(tuple(extensions)):\n",
348
- " filename = os.path.basename(url)\n",
349
- " else:\n",
350
- " response = requests.get(url, stream=True)\n",
351
- " response.raise_for_status()\n",
352
- "\n",
353
- " if 'content-disposition' in response.headers:\n",
354
- " content_disposition = response.headers['content-disposition']\n",
355
- " filename = re.findall('filename=\"?([^\"]+)\"?', content_disposition)[0]\n",
356
- " else:\n",
357
- " url_path = urlparse(url).path\n",
358
- " filename = unquote(os.path.basename(url_path))\n",
359
- "\n",
360
- " if filename.endswith(tuple(get_supported_extensions())):\n",
361
- " return filename\n",
362
- " else:\n",
363
- " return None\n",
364
- "\n",
365
- "def get_most_recent_file(directory):\n",
366
- " files = glob.glob(os.path.join(directory, \"*\"))\n",
367
- " if not files:\n",
368
- " return None\n",
369
- " most_recent_file = max(files, key=os.path.getmtime)\n",
370
- " basename = os.path.basename(most_recent_file)\n",
371
- "\n",
372
- " return most_recent_file\n",
373
- "\n",
374
- "def parse_args(config):\n",
375
- " args = []\n",
376
- "\n",
377
- " for k, v in config.items():\n",
378
- " if k.startswith(\"_\"):\n",
379
- " args.append(f\"{v}\")\n",
380
- " elif isinstance(v, str) and v is not None:\n",
381
- " args.append(f'--{k}={v}')\n",
382
- " elif isinstance(v, bool) and v:\n",
383
- " args.append(f\"--{k}\")\n",
384
- " elif isinstance(v, float) and not isinstance(v, bool):\n",
385
- " args.append(f\"--{k}={v}\")\n",
386
- " elif isinstance(v, int) and not isinstance(v, bool):\n",
387
- " args.append(f\"--{k}={v}\")\n",
388
- "\n",
389
- " return args\n",
390
- "def aria2_download(dir, filename, url):\n",
391
- " aria2_config = {\n",
392
- " \"console-log-level\" : \"error\",\n",
393
- " \"summary-interval\" : 10,\n",
394
- " \"continue\" : True,\n",
395
- " \"max-connection-per-server\" : 16,\n",
396
- " \"min-split-size\" : \"1M\",\n",
397
- " \"split\" : 16,\n",
398
- " \"dir\" : dir,\n",
399
- " \"out\" : filename,\n",
400
- " \"_url\" : url,\n",
401
- " }\n",
402
- " aria2_args = parse_args(aria2_config)\n",
403
- " subprocess.run([\"aria2c\", *aria2_args])\n",
404
- "\n",
405
- "def gdown_download(url, dst, filepath):\n",
406
- " if \"/uc?id/\" in url:\n",
407
- " return gdown.download(url, filepath, quiet=False)\n",
408
- " elif \"/file/d/\" in url:\n",
409
- " return gdown.download(url=url, output=filepath, quiet=False, fuzzy=True)\n",
410
- " elif \"/drive/folders/\" in url:\n",
411
- " os.chdir(dst)\n",
412
- " return gdown.download_folder(url, quiet=True, use_cookies=False)\n",
413
- "\n",
414
- "def download(url, dst):\n",
415
- " print(f\"Starting downloading from {url}\")\n",
416
- " filename = get_filename(url)\n",
417
- " filepath = os.path.join(dst, filename)\n",
418
- "\n",
419
- " if \"drive.google.com\" in url:\n",
420
- " gdown = gdown_download(url, dst, filepath)\n",
421
- " else:\n",
422
- " if \"huggingface.co\" in url and \"/blob/\" in url:\n",
423
- " url = url.replace(\"/blob/\", \"/resolve/\")\n",
424
- " aria2_download(dst, filename, url)\n",
425
- "\n",
426
- " print(f\"Download finished: {filepath}\")\n",
427
- " return filepath\n",
428
- "\n",
429
- "def get_gpu_name():\n",
430
- " try:\n",
431
- " return subprocess.check_output(\"nvidia-smi --query-gpu=name --format=csv,noheader,nounits\", shell=True).decode('ascii').strip()\n",
432
- " except:\n",
433
- " return None\n",
434
- "\n",
435
- "def main():\n",
436
- " global model_path, vae_path\n",
437
- " model_path, vae_path = None, None\n",
438
- " download_targets = {\n",
439
- " \"model\": (MODEL_URL, pretrained_model),\n",
440
- " }\n",
441
- " for target, (url, dst) in download_targets.items():\n",
442
- " if url and not url.startswith(f\"PASTE {target.upper()} URL OR GDRIVE PATH HERE\"):\n",
443
- " filepath = download(url, dst)\n",
444
- " if target == \"model\":\n",
445
- " model_path = filepath\n",
446
- " print()\n",
447
- " if model_path:\n",
448
- " print(f\"Selected model: {model_path}\")\n",
449
- "\n",
450
- "if Model.startswith(\"/content/drive/\"):\n",
451
- " model_path = Model\n",
452
- " print(f\"Diffusers model is loaded : {Model}\")\n",
453
- "else:\n",
454
- " main()\n",
455
- "\n",
456
- "!aria2c --console-log-level=error --summary-interval=10 -c -x 16 -k 1M -s 16 https://huggingface.co/stabilityai/sd-vae-ft-mse-original/resolve/main/vae-ft-mse-840000-ema-pruned.ckpt -d /content/VAE -o VAE84EMA.vae.pt\n",
457
- "vae = \"/content/VAE/VAE84EMA.vae.pt\"\n",
458
- "\n",
459
- "#Dataset Config\n",
460
- "\n",
461
- "activation_word = \"GSGI\"\n",
462
- "caption_extension = \".txt\"\n",
463
- "token_to_captions = False\n",
464
- "dataset_repeats = Repeats\n",
465
- "keep_tokens = 0\n",
466
- "flip_aug = False\n",
467
- "\n",
468
- "if ',' in activation_word or ' ' in activation_word:\n",
469
- " words = activation_word.replace(',', ' ').split()\n",
470
- " class_token = words[-1]\n",
471
- "\n",
472
- "\n",
473
- "def read_file(filename):\n",
474
- " with open(filename, \"r\") as f:\n",
475
- " contents = f.read()\n",
476
- " return contents\n",
477
- "\n",
478
- "\n",
479
- "def write_file(filename, contents):\n",
480
- " with open(filename, \"w\") as f:\n",
481
- " f.write(contents)\n",
482
- "\n",
483
- "\n",
484
- "def get_supported_images(folder):\n",
485
- " supported_extensions = (\".png\", \".jpg\", \".jpeg\", \".webp\", \".bmp\")\n",
486
- " return [file for ext in supported_extensions for file in glob.glob(f\"{folder}/*{ext}\")]\n",
487
- "\n",
488
- "\n",
489
- "def get_subfolders_with_supported_images(folder):\n",
490
- " subfolders = [os.path.join(folder, subfolder) for subfolder in os.listdir(folder) if os.path.isdir(os.path.join(folder, subfolder))]\n",
491
- " return [subfolder for subfolder in subfolders if len(get_supported_images(subfolder)) > 0]\n",
492
- "\n",
493
- "\n",
494
- "def process_tags(filename, custom_tag, remove_tag):\n",
495
- " contents = read_file(filename)\n",
496
- " tags = [tag.strip() for tag in contents.split(',')]\n",
497
- " custom_tags = [tag.strip() for tag in custom_tag.split(',')]\n",
498
- "\n",
499
- " for custom_tag in custom_tags:\n",
500
- " custom_tag = custom_tag.replace(\"_\", \" \")\n",
501
- " # if remove_tag:\n",
502
- " # while custom_tag in tags:\n",
503
- " # tags.remove(custom_tag)\n",
504
- " # else:\n",
505
- " if custom_tag not in tags:\n",
506
- " tags.insert(0, custom_tag)\n",
507
- "\n",
508
- " contents = ', '.join(tags)\n",
509
- " write_file(filename, contents)\n",
510
- "\n",
511
- "\n",
512
- "def process_folder_recursively(folder):\n",
513
- " for root, _, files in os.walk(folder):\n",
514
- " for file in files:\n",
515
- " if file.endswith(caption_extension):\n",
516
- " file_path = os.path.join(root, file)\n",
517
- " extracted_class_token = get_class_token_from_folder_name(root, folder)\n",
518
- " train_supported_images = get_supported_images(train_data_dir)\n",
519
- " tag = extracted_class_token if extracted_class_token else activation_word if train_supported_images else \"\"\n",
520
- " if not tag == \"\":\n",
521
- " process_tags(file_path, tag, remove_tag=(not token_to_captions))\n",
522
- "\n",
523
- "\n",
524
- "def get_num_repeats(folder):\n",
525
- " folder_name = os.path.basename(folder)\n",
526
- " try:\n",
527
- " repeats, _ = folder_name.split('_', 1)\n",
528
- " num_repeats = int(repeats)\n",
529
- " except ValueError:\n",
530
- " num_repeats = dataset_repeats\n",
531
- "\n",
532
- " return num_repeats\n",
533
- "\n",
534
- "\n",
535
- "def get_class_token_from_folder_name(folder, parent_folder):\n",
536
- " if folder == parent_folder:\n",
537
- " return class_token\n",
538
- "\n",
539
- " folder_name = os.path.basename(folder)\n",
540
- " try:\n",
541
- " _, concept = folder_name.split('_', 1)\n",
542
- " return concept\n",
543
- " except ValueError:\n",
544
- " return \"\"\n",
545
- "\n",
546
- "train_supported_images = get_supported_images(train_data_dir)\n",
547
- "train_subfolders = get_subfolders_with_supported_images(train_data_dir)\n",
548
- "\n",
549
- "subsets = []\n",
550
- "config = {\n",
551
- " \"general\": {\n",
552
- " \"enable_bucket\": True,\n",
553
- " \"caption_extension\": caption_extension,\n",
554
- " \"shuffle_caption\": True,\n",
555
- " \"keep_tokens\": keep_tokens,\n",
556
- " \"bucket_reso_steps\": 64,\n",
557
- " \"bucket_no_upscale\": False,\n",
558
- " },\n",
559
- " \"datasets\": [\n",
560
- " {\n",
561
- " \"resolution\": resolution,\n",
562
- " \"min_bucket_reso\": 320 if resolution > 640 else 256,\n",
563
- " \"max_bucket_reso\": 1280 if resolution > 640 else 1024,\n",
564
- " \"caption_dropout_rate\": 0,\n",
565
- " \"caption_tag_dropout_rate\": 0,\n",
566
- " \"caption_dropout_every_n_epochs\": 0,\n",
567
- " \"flip_aug\": flip_aug,\n",
568
- " \"color_aug\": False,\n",
569
- " \"face_crop_aug_range\": None,\n",
570
- " \"subsets\": subsets,\n",
571
- " }\n",
572
- " ],\n",
573
- "}\n",
574
- "\n",
575
- "if token_to_captions and keep_tokens < 2:\n",
576
- " keep_tokens = 1\n",
577
- "\n",
578
- "process_folder_recursively(train_data_dir)\n",
579
- "\n",
580
- "if train_supported_images:\n",
581
- " subsets.append({\n",
582
- " \"image_dir\": train_data_dir,\n",
583
- " \"class_tokens\": activation_word,\n",
584
- " \"num_repeats\": dataset_repeats,\n",
585
- " })\n",
586
- "\n",
587
- "for subfolder in train_subfolders:\n",
588
- " num_repeats = get_num_repeats(subfolder)\n",
589
- " extracted_class_token = get_class_token_from_folder_name(subfolder, train_data_dir)\n",
590
- " subsets.append({\n",
591
- " \"image_dir\": subfolder,\n",
592
- " \"class_tokens\": extracted_class_token if extracted_class_token else None,\n",
593
- " \"num_repeats\": num_repeats,\n",
594
- " })\n",
595
- "\n",
596
- "for subset in subsets:\n",
597
- " if not glob.glob(f\"{subset['image_dir']}/*.txt\"):\n",
598
- " subset[\"class_tokens\"] = activation_word\n",
599
- "\n",
600
- "dataset_config = os.path.join(config_dir, \"dataset_config.toml\")\n",
601
- "\n",
602
- "for key in config:\n",
603
- " if isinstance(config[key], dict):\n",
604
- " for sub_key in config[key]:\n",
605
- " if config[key][sub_key] == \"\":\n",
606
- " config[key][sub_key] = None\n",
607
- " elif config[key] == \"\":\n",
608
- " config[key] = None\n",
609
- "\n",
610
- "config_str = toml.dumps(config)\n",
611
- "\n",
612
- "with open(dataset_config, \"w\") as f:\n",
613
- " f.write(config_str)\n",
614
- "\n",
615
- "print(config_str)\n",
616
- "\n",
617
- "#Config\n",
618
- "optimizer_args = False\n",
619
- "conv_dim = 4\n",
620
- "conv_alpha = 1\n",
621
- "\n",
622
- "network_module = \"networks.lora\"\n",
623
- "network_args = \"\"\n",
624
- "\n",
625
- "config = {\n",
626
- " \"model_arguments\": {\n",
627
- " \"v2\": False,\n",
628
- " \"v_parameterization\": False,\n",
629
- " \"pretrained_model_name_or_path\": model_path,\n",
630
- " \"vae\": vae,\n",
631
- " },\n",
632
- " \"additional_network_arguments\": {\n",
633
- " \"no_metadata\": False,\n",
634
- " \"unet_lr\": float(unet_lr),\n",
635
- " \"text_encoder_lr\": float(text_encoder_lr),\n",
636
- " \"network_module\": network_module,\n",
637
- " \"network_dim\": 64,\n",
638
- " \"network_alpha\": 48,\n",
639
- " \"training_comment\": \"GSGI Trainer\",\n",
640
- " },\n",
641
- " \"optimizer_arguments\": {\n",
642
- " \"optimizer_type\": \"AdamW8bit\",\n",
643
- " \"optimizer_args\": eval(optimizer_args) if optimizer_args else None,\n",
644
- " \"learning_rate\": unet_lr,\n",
645
- " \"max_grad_norm\": 1.0,\n",
646
- " \"lr_scheduler\": \"cosine_with_restarts\",\n",
647
- " \"lr_scheduler_num_cycles\": 4,\n",
648
- " },\n",
649
- " \"dataset_arguments\": {\n",
650
- " \"cache_latents\": True,\n",
651
- " \"debug_dataset\": False,\n",
652
- " \"vae_batch_size\": Batch_size,\n",
653
- " },\n",
654
- " \"training_arguments\": {\n",
655
- " \"output_dir\": output_dir,\n",
656
- " \"output_name\": Loraname,\n",
657
- " \"save_precision\": \"fp16\",\n",
658
- " \"save_every_n_epochs\": save_n_epochs_type_value,\n",
659
- " \"train_batch_size\": Batch_size,\n",
660
- " \"max_token_length\": 225,\n",
661
- " \"mem_eff_attn\": False,\n",
662
- " \"xformers\": True,\n",
663
- " \"max_train_epochs\": num_epochs,\n",
664
- " \"max_data_loader_n_workers\": 8,\n",
665
- " \"persistent_data_loader_workers\": True,\n",
666
- " \"gradient_checkpointing\": False,\n",
667
- " \"gradient_accumulation_steps\": 1,\n",
668
- " \"mixed_precision\": \"fp16\",\n",
669
- " \"clip_skip\": 1,\n",
670
- " \"logging_dir\": \"/content/Dreamboot-Config/logs\",\n",
671
- " \"log_prefix\": Loraname,\n",
672
- " \"lowram\": True,\n",
673
- " \"training_comment\" : \"train by GSGI Trainer\",\n",
674
- " },\n",
675
- " \"sample_prompt_arguments\": {\n",
676
- " \"sample_every_n_steps\": 200,\n",
677
- " \"sample_every_n_epochs\": 1,\n",
678
- " \"sample_sampler\": \"euler\",\n",
679
- " },\n",
680
- " \"dreambooth_arguments\": {\n",
681
- " \"prior_loss_weight\": 1,\n",
682
- " },\n",
683
- " \"saving_arguments\": {\n",
684
- " \"save_model_as\": \"safetensors\",\n",
685
- " },\n",
686
- "}\n",
687
- "SamplePrompt = f\"{Loraname},front view, masterpiece,best quality\"\n",
688
- "sample_str = f\"\"\"\n",
689
- " {SamplePrompt}\\\n",
690
- " --n lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry \\\n",
691
- " --w 512 \\\n",
692
- " --h 768 \\\n",
693
- " --l 7 \\\n",
694
- " --s 30\n",
695
- "\"\"\"\n",
696
- "config_path = os.path.join(config_dir, \"config_file.toml\")\n",
697
- "prompt_path = os.path.join(config_dir, \"sample_prompt.txt\")\n",
698
- "\n",
699
- "for key in config:\n",
700
- " if isinstance(config[key], dict):\n",
701
- " for sub_key in config[key]:\n",
702
- " if config[key][sub_key] == \"\":\n",
703
- " config[key][sub_key] = None\n",
704
- " elif config[key] == \"\":\n",
705
- " config[key] = None\n",
706
- "\n",
707
- "config_str = toml.dumps(config)\n",
708
- "\n",
709
- "def write_file(filename, contents):\n",
710
- " with open(filename, \"w\") as f:\n",
711
- " f.write(contents)\n",
712
- "\n",
713
- "write_file(config_path, config_str)\n",
714
- "write_file(prompt_path, sample_str)\n",
715
- "\n",
716
- "print(config_str)\n",
717
- "\n",
718
- "os.chdir(repo_dir)\n",
719
- "\n",
720
- "\n",
721
- "train_file = \"train_network.py\"\n",
722
- "ConfigFolder = \"/content/Dreamboot-Config/config\"\n",
723
- "sample_prompt = f\"{ConfigFolder}/sample_prompt.txt\"\n",
724
- "config_file = f\"{ConfigFolder}/config_file.toml\"\n",
725
- "dataset_config = f\"{ConfigFolder}/dataset_config.toml\"\n",
726
- "accelerate_conf = {\n",
727
- " \"config_file\" : accelerate_config,\n",
728
- " \"num_cpu_threads_per_process\" : 1,\n",
729
- "}\n",
730
- "\n",
731
- "train_conf = {\n",
732
- " \"sample_prompts\" : sample_prompt,\n",
733
- " \"dataset_config\" : dataset_config,\n",
734
- " \"config_file\" : config_file\n",
735
- "}\n",
736
- "\n",
737
- "def train(config):\n",
738
- " args = \"\"\n",
739
- " for k, v in config.items():\n",
740
- " if k.startswith(\"_\"):\n",
741
- " args += f'\"{v}\" '\n",
742
- " elif isinstance(v, str):\n",
743
- " args += f'--{k}=\"{v}\" '\n",
744
- " elif isinstance(v, bool) and v:\n",
745
- " args += f\"--{k} \"\n",
746
- " elif isinstance(v, float) and not isinstance(v, bool):\n",
747
- " args += f\"--{k}={v} \"\n",
748
- " elif isinstance(v, int) and not isinstance(v, bool):\n",
749
- " args += f\"--{k}={v} \"\n",
750
- "\n",
751
- " return args\n",
752
- "\n",
753
- "accelerate_args = train(accelerate_conf)\n",
754
- "train_args = train(train_conf)\n",
755
- "final_args = f\"accelerate launch {accelerate_args} {train_file} {train_args}\"\n"
756
- ]
757
- }
758
- ],
759
- "metadata": {
760
- "language_info": {
761
- "name": "python"
762
- },
763
- "colab": {
764
- "provenance": []
765
- },
766
- "kernelspec": {
767
- "name": "python3",
768
- "display_name": "Python 3"
769
- }
770
- },
771
- "nbformat": 4,
772
- "nbformat_minor": 0
773
- }