File size: 11,954 Bytes
86116e3 569f6f1 86116e3 7f60265 86116e3 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 |
{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
"!pip install git+https://github.com/chiral-carbon/diffusers@advdiff_sdxl -q"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [],
"source": [
"import os, re\n",
"\n",
"from diffusers.utils.hub_utils import load_or_create_model_card, populate_model_card\n",
"\n",
"def save_model_card(\n",
" repo_id: str,\n",
" use_dora: bool,\n",
" images: list = None,\n",
" base_model: str = None,\n",
" train_text_encoder: bool = False,\n",
" train_text_encoder_ti: bool = False,\n",
" token_abstraction_dict=None,\n",
" instance_prompt: str = None,\n",
" validation_prompt:str = None,\n",
" repo_folder=None,\n",
" vae_path=None,\n",
"):\n",
" lora = \"lora\" if not use_dora else \"dora\"\n",
"\n",
" widget_dict = []\n",
" if images is not None:\n",
" for i, image in enumerate(images):\n",
" image.save(os.path.join(repo_folder, f\"image_{i}.png\"))\n",
" widget_dict.append(\n",
" {\"text\": validation_prompt if validation_prompt else \" \", \"output\": {\"url\": f\"image_{i}.png\"}}\n",
" )\n",
" else:\n",
" widget_dict.append(\n",
" {\"text\": instance_prompt}\n",
" )\n",
" embeddings_filename = f\"{repo_folder}_emb\"\n",
" instance_prompt_webui = re.sub(r\"<s\\d+>\", \"\", re.sub(r\"<s\\d+>\", embeddings_filename, instance_prompt, count=1))\n",
" ti_keys = \", \".join(f'\"{match}\"' for match in re.findall(r\"<s\\d+>\", instance_prompt))\n",
" if instance_prompt_webui != embeddings_filename:\n",
" instance_prompt_sentence = f\"For example, `{instance_prompt_webui}`\"\n",
" else:\n",
" instance_prompt_sentence = \"\"\n",
" trigger_str = f\"You should use {instance_prompt} to trigger the image generation.\"\n",
" diffusers_imports_pivotal = \"\"\n",
" diffusers_example_pivotal = \"\"\n",
" webui_example_pivotal = \"\"\n",
" if train_text_encoder_ti:\n",
" trigger_str = (\n",
" \"To trigger image generation of trained concept(or concepts) replace each concept identifier \"\n",
" \"in you prompt with the new inserted tokens:\\n\"\n",
" )\n",
" diffusers_imports_pivotal = \"\"\"from huggingface_hub import hf_hub_download\n",
"from safetensors.torch import load_file\n",
" \"\"\"\n",
" diffusers_example_pivotal = f\"\"\"embedding_path = hf_hub_download(repo_id='{repo_id}', filename='{embeddings_filename}.safetensors', repo_type=\"model\")\n",
"state_dict = load_file(embedding_path)\n",
"pipeline.load_textual_inversion(state_dict[\"clip_l\"], token=[{ti_keys}], text_encoder=pipeline.text_encoder, tokenizer=pipeline.tokenizer)\n",
"pipeline.load_textual_inversion(state_dict[\"clip_g\"], token=[{ti_keys}], text_encoder=pipeline.text_encoder_2, tokenizer=pipeline.tokenizer_2)\n",
" \"\"\"\n",
" webui_example_pivotal = f\"\"\"- *Embeddings*: download **[`{embeddings_filename}.safetensors` here 💾](/{repo_id}/blob/main/{embeddings_filename}.safetensors)**.\n",
" - Place it on it on your `embeddings` folder\n",
" - Use it by adding `{embeddings_filename}` to your prompt. {instance_prompt_sentence}\n",
" (you need both the LoRA and the embeddings as they were trained together for this LoRA)\n",
" \"\"\"\n",
" if token_abstraction_dict:\n",
" for key, value in token_abstraction_dict.items():\n",
" tokens = \"\".join(value)\n",
" trigger_str += f\"\"\"\n",
"to trigger concept `{key}` → use `{tokens}` in your prompt \\n\n",
"\"\"\"\n",
"\n",
" model_description = f\"\"\"\n",
"# SDXL LoRA DreamBooth - {repo_id}\n",
"\n",
"<Gallery />\n",
"\n",
"## Model description\n",
"\n",
"### These are {repo_id} LoRA adaption weights for {base_model}.\n",
"\n",
"## Download model\n",
"\n",
"### Use it with UIs such as AUTOMATIC1111, Comfy UI, SD.Next, Invoke\n",
"\n",
"- **LoRA**: download **[`{repo_folder}.safetensors` here 💾](/{repo_id}/blob/main/{repo_folder}.safetensors)**.\n",
" - Place it on your `models/Lora` folder.\n",
" - On AUTOMATIC1111, load the LoRA by adding `<lora:{repo_folder}:1>` to your prompt. On ComfyUI just [load it as a regular LoRA](https://comfyanonymous.github.io/ComfyUI_examples/lora/).\n",
"{webui_example_pivotal}\n",
"\n",
"## Use it with the [🧨 diffusers library](https://github.com/huggingface/diffusers)\n",
"\n",
"```py\n",
"from diffusers import AutoPipelineForText2Image\n",
"import torch\n",
"{diffusers_imports_pivotal}\n",
"pipeline = AutoPipelineForText2Image.from_pretrained('stabilityai/stable-diffusion-xl-base-1.0', torch_dtype=torch.float16).to('cuda')\n",
"pipeline.load_lora_weights('{repo_id}', weight_name='pytorch_lora_weights.safetensors')\n",
"{diffusers_example_pivotal}\n",
"image = pipeline('{validation_prompt if validation_prompt else instance_prompt}').images[0]\n",
"```\n",
"\n",
"For more details, including weighting, merging and fusing LoRAs, check the [documentation on loading LoRAs in diffusers](https://huggingface.co/docs/diffusers/main/en/using-diffusers/loading_adapters)\n",
"\n",
"## Trigger words\n",
"\n",
"{trigger_str}\n",
"\n",
"## Details\n",
"All [Files & versions](/{repo_id}/tree/main).\n",
"\n",
"The weights were trained using [🧨 diffusers Advanced Dreambooth Training Script](https://github.com/huggingface/diffusers/blob/main/examples/advanced_diffusion_training/train_dreambooth_lora_sdxl_advanced.py).\n",
"\n",
"LoRA for the text encoder was enabled. {train_text_encoder}.\n",
"\n",
"Pivotal tuning was enabled: {train_text_encoder_ti}.\n",
"\n",
"Special VAE used for training: {vae_path}.\n",
"\n",
"\"\"\"\n",
" model_card = load_or_create_model_card(\n",
" repo_id_or_path=repo_id,\n",
" from_training=True,\n",
" license=\"openrail++\",\n",
" base_model=base_model,\n",
" prompt=instance_prompt,\n",
" model_description=model_description,\n",
" widget=widget_dict,\n",
" )\n",
" tags = [\n",
" \"text-to-image\",\n",
" \"stable-diffusion-xl\",\n",
" \"stable-diffusion-xl-diffusers\",\n",
" \"text-to-image\",\n",
" \"diffusers\",\n",
" lora,\n",
" \"template:sd-lora\",\n",
" ]\n",
" model_card = populate_model_card(model_card, tags=tags)\n",
" "
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [],
"source": [
"from diffusers.utils import load_image\n",
"\n",
"images = [\n",
" load_image(\"https://huggingface.co/datasets/diffusers/docs-images/resolve/main/amused/A%20mushroom%20in%20%5BV%5D%20style.png\")\n",
" for _ in range(3)\n",
"]\n",
"\n",
"save_model_card(\n",
" use_dora=False,\n",
" repo_id=\"abby101/test\",\n",
" images=images,\n",
" base_model=\"runwayml/stable-diffusion-v1-5\",\n",
" repo_folder=\".\",\n",
" instance_prompt=\"A mushroom in [V] style\",\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"---\n",
"license: openrail++\n",
"library_name: diffusers\n",
"tags:\n",
"- text-to-image\n",
"- stable-diffusion-xl\n",
"- stable-diffusion-xl-diffusers\n",
"- text-to-image\n",
"- diffusers\n",
"- lora\n",
"- template:sd-lora\n",
"base_model: runwayml/stable-diffusion-v1-5\n",
"instance_prompt: A mushroom in [V] style\n",
"widget:\n",
"- text: ' '\n",
" output:\n",
" url: image_0.png\n",
"- text: ' '\n",
" output:\n",
" url: image_1.png\n",
"- text: ' '\n",
" output:\n",
" url: image_2.png\n",
"---\n",
"\n",
"<!-- This model card has been generated automatically according to the information the training script had access to. You\n",
"should probably proofread and complete it, then remove this comment. -->\n",
"\n",
"\n",
"# SDXL LoRA DreamBooth - abby101/test\n",
"\n",
"<Gallery />\n",
"\n",
"## Model description\n",
"\n",
"### These are abby101/test LoRA adaption weights for runwayml/stable-diffusion-v1-5.\n",
"\n",
"## Download model\n",
"\n",
"### Use it with UIs such as AUTOMATIC1111, Comfy UI, SD.Next, Invoke\n",
"\n",
"- **LoRA**: download **[`..safetensors` here 💾](/abby101/test/blob/main/..safetensors)**.\n",
" - Place it on your `models/Lora` folder.\n",
" - On AUTOMATIC1111, load the LoRA by adding `<lora:.:1>` to your prompt. On ComfyUI just [load it as a regular LoRA](https://comfyanonymous.github.io/ComfyUI_examples/lora/).\n",
"\n",
"\n",
"## Use it with the [🧨 diffusers library](https://github.com/huggingface/diffusers)\n",
"\n",
"```py\n",
"from diffusers import AutoPipelineForText2Image\n",
"import torch\n",
"\n",
"pipeline = AutoPipelineForText2Image.from_pretrained('stabilityai/stable-diffusion-xl-base-1.0', torch_dtype=torch.float16).to('cuda')\n",
"pipeline.load_lora_weights('abby101/test', weight_name='pytorch_lora_weights.safetensors')\n",
"\n",
"image = pipeline('A mushroom in [V] style').images[0]\n",
"```\n",
"\n",
"For more details, including weighting, merging and fusing LoRAs, check the [documentation on loading LoRAs in diffusers](https://huggingface.co/docs/diffusers/main/en/using-diffusers/loading_adapters)\n",
"\n",
"## Trigger words\n",
"\n",
"You should use A mushroom in [V] style to trigger the image generation.\n",
"\n",
"## Details\n",
"All [Files & versions](/abby101/test/tree/main).\n",
"\n",
"The weights were trained using [🧨 diffusers Advanced Dreambooth Training Script](https://github.com/huggingface/diffusers/blob/main/examples/advanced_diffusion_training/train_dreambooth_lora_sdxl_advanced.py).\n",
"\n",
"LoRA for the text encoder was enabled. False.\n",
"\n",
"Pivotal tuning was enabled: False.\n",
"\n",
"Special VAE used for training: None.\n",
"\n",
"\n",
"\n",
"## Intended uses & limitations\n",
"\n",
"#### How to use\n",
"\n",
"```python\n",
"# TODO: add an example code snippet for running this diffusion pipeline\n",
"```\n",
"\n",
"#### Limitations and bias\n",
"\n",
"[TODO: provide examples of latent issues and potential remediations]\n",
"\n",
"## Training details\n",
"\n",
"[TODO: describe the data used to train the model]"
]
}
],
"source": [
"!cat README.md"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "pydl",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.10.9"
}
},
"nbformat": 4,
"nbformat_minor": 2
}
|