Ezi commited on
Commit
cb79a56
1 Parent(s): e78a8c4

restructuring the MC

Browse files

Hello hello!
This PR has a preliminary model card, based on the format we are using as part of our effort to standardisee model cards at Hugging Face. Please see this [doc of an annotated model card](https://huggingface.co/docs/hub/model-card-annotated)
Any additions, inputs and/or edits are greatly welcomed 🤗

Files changed (1) hide show
  1. README.md +216 -8
README.md CHANGED
@@ -1,12 +1,186 @@
1
  ---
2
  license: openrail
 
 
 
3
  ---
4
 
5
- This is the pretrained weights and some other detector weights of ControlNet.
6
 
7
- See also: https://github.com/lllyasviel/ControlNet
8
 
9
- # Description of Files
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10
 
11
  ControlNet/models/control_sd15_canny.pth
12
 
@@ -68,11 +242,42 @@ ControlNet/annotator/ckpts/upernet_global_small.pth
68
 
69
  - Third-party model: Uniformer semantic segmentation.
70
 
71
- ControlNet/training/fill50k.zip
72
 
73
- - The data for our training tutorial.
74
 
75
- # Related Resources
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
76
 
77
  Special Thank to the great project - [Mikubill' A1111 Webui Plugin](https://github.com/Mikubill/sd-webui-controlnet) !
78
 
@@ -81,8 +286,11 @@ We also thank Hysts for making [Gradio](https://github.com/gradio-app/gradio) de
81
  Thank haofanwang for making [ControlNet-for-Diffusers](https://github.com/haofanwang/ControlNet-for-Diffusers)!
82
 
83
  We also thank all authors for making Controlnet DEMOs, including but not limited to [fffiloni](https://huggingface.co/spaces/fffiloni/ControlNet-Video), [other-model](https://huggingface.co/spaces/hysts/ControlNet-with-other-models), [ThereforeGames](https://github.com/AUTOMATIC1111/stable-diffusion-webui/discussions/7784), [RamAnanth1](https://huggingface.co/spaces/RamAnanth1/ControlNet), etc!
 
84
 
85
- # Misuse, Malicious Use, and Out-of-Scope Use
86
 
87
- The model should not be used to intentionally create or disseminate images that create hostile or alienating environments for people. This includes generating images that people would foreseeably find disturbing, distressing, or offensive; or content that propagates historical or current stereotypes.
 
 
88
 
 
1
  ---
2
  license: openrail
3
+ datasets:
4
+ - HighCWu/fill50k
5
+ pipeline_tag: text-to-image
6
  ---
7
 
 
8
 
9
+ # Model Card for ControlNet
10
 
11
+ <!-- Provide a quick summary of what the model is/does. -->
12
+ ControlNet is a neural network structure to control diffusion models by adding extra conditions.
13
+
14
+ *This is the pretrained weights and some other detector weights of ControlNet.*
15
+
16
+ ## Model Details
17
+
18
+ ### Model Description
19
+
20
+ The [associated paper](https://arxiv.org/pdf/2302.05543.pdf) details:
21
+ > ControlNet, an end-to-end neural network architecture that controls large image diffusion models (like Stable Diffusion) to learn task-specific input conditions.
22
+
23
+
24
+ - **Developed by:** Lvmin Zhang and Maneesh Agrawala
25
+ - **Shared by:** [Lvmin Zhang](https://huggingface.co/lllyasviel)
26
+ - **Model type:** Text to Image
27
+ - **Language(s) (NLP):** [More Information Needed]
28
+ - **License:** OpeRAIL
29
+ - **Finetuned from model [optional]:** [More Information Needed]
30
+
31
+ ### Model Sources]
32
+
33
+ <!-- Provide the basic links for the model. -->
34
+
35
+ - **Repository:** [Github](https://github.com/lllyasviel/ControlNet)
36
+ - **Paper:** [Paper](https://arxiv.org/abs/2302.05543)
37
+ - **Demo:** [Hugging Face Space](https://huggingface.co/spaces/hysts/ControlNet)
38
+
39
+ ## Uses
40
+
41
+ <!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
42
+
43
+ ### Direct Use
44
+
45
+ <!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
46
+ Control pretrained large diffusion models to support additional input conditions.
47
+
48
+ ### Downstream Use
49
+
50
+ <!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
51
+
52
+ See [Description of Files](####Software) section for different models to control Stable Diffusion.
53
+
54
+ ### Out-of-Scope Use
55
+
56
+ <!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
57
+
58
+ The model should not be used to intentionally create or disseminate images that create hostile or alienating environments for people. This includes generating images that people would foreseeably find disturbing, distressing, or offensive; or content that propagates historical or current stereotypes.
59
+
60
+ ## Bias, Risks, and Limitations
61
+
62
+ <!-- This section is meant to convey both technical and sociotechnical limitations. -->
63
+
64
+ [More Information Needed]
65
+
66
+ ### Recommendations
67
+
68
+ <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
69
+
70
+ Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
71
+
72
+ ### How to Get Started with the Model
73
+
74
+ Use the code below to get started with the model.
75
+
76
+ [More Information Needed]
77
+
78
+ ## Training Details
79
+
80
+ ### Training Data
81
+
82
+ <!-- This should link to a Data Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
83
+
84
+ See [fill50k Dataset Card](https://huggingface.co/datasets/HighCWu/fill50k) and the ControlNet/training/fill50k.zip file for the data for our training tutorial.
85
+
86
+ ### Training Procedure
87
+
88
+ <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
89
+
90
+ #### Preprocessing [optional]
91
+
92
+ [More Information Needed]
93
+
94
+
95
+ #### Training Hyperparameters
96
+
97
+ - **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
98
+
99
+ #### Speeds, Sizes, Times [optional]
100
+
101
+ <!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
102
+
103
+ [More Information Needed]
104
+
105
+ ## Evaluation
106
+
107
+ <!-- This section describes the evaluation protocols and provides the results. -->
108
+
109
+ ### Testing Data, Factors & Metrics
110
+
111
+ #### Testing Data
112
+
113
+ <!-- This should link to a Data Card if possible. -->
114
+
115
+ [More Information Needed]
116
+
117
+ #### Factors
118
+
119
+ <!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
120
+
121
+ [More Information Needed]
122
+
123
+ #### Metrics
124
+
125
+ <!-- These are the evaluation metrics being used, ideally with a description of why. -->
126
+
127
+ [More Information Needed]
128
+
129
+ ### Results
130
+
131
+ [More Information Needed]
132
+
133
+ #### Summary
134
+
135
+ [More Information Needed]
136
+
137
+ ## Model Examination [optional]
138
+
139
+ <!-- Relevant interpretability work for the model goes here -->
140
+
141
+ [More Information Needed]
142
+
143
+ ## Environmental Impact
144
+
145
+ <!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
146
+
147
+ Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
148
+
149
+ - **Hardware Type:** Nvidia A100 80G
150
+ - **Hours used:** 600
151
+ - **Cloud Provider:** [More Information Needed]
152
+ - **Compute Region:** [More Information Needed]
153
+ - **Carbon Emitted:** [More Information Needed]
154
+
155
+ The [associated paper](https://arxiv.org/pdf/2302.05543.pdf) detials the compute hardware and GPU hours for:
156
+ > several implementations of ControlNets with different image-based conditions to control large diffusion models in various ways.
157
+ The above hardware and GPU hours corresponds to the **Canny Edge detector**
158
+
159
+ ## Technical Specifications
160
+
161
+ ### Model Architecture and Objective
162
+ As noted in the [associated Github](https://github.com/lllyasviel/ControlNet):
163
+
164
+ >It copys the weights of neural network blocks into a "locked" copy and a "trainable" copy.
165
+ >The "trainable" one learns your condition. The "locked" one preserves your model.
166
+ >Thanks to this, training with small dataset of image pairs will not destroy the production-ready diffusion models.
167
+ >The "zero convolution" is 1×1 convolution with both weight and bias initialized as zeros.
168
+ >Before training, all zero convolutions output zeros, and ControlNet will not cause any distortion.
169
+ >No layer is trained from scratch. You are still fine-tuning. Your original model is safe.
170
+ >This allows training on small-scale or even personal devices.>
171
+
172
+
173
+ ### Compute Infrastructure
174
+
175
+ [More Information Needed]
176
+
177
+ #### Hardware
178
+
179
+ [More Information Needed]
180
+
181
+ #### Software
182
+
183
+ **Description of Files**
184
 
185
  ControlNet/models/control_sd15_canny.pth
186
 
 
242
 
243
  - Third-party model: Uniformer semantic segmentation.
244
 
 
245
 
 
246
 
247
+ ## Citation
248
+
249
+ <!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
250
+
251
+
252
+ **BibTeX:**
253
+ ```
254
+ @misc{https://doi.org/10.48550/arxiv.2302.05543,
255
+ doi = {10.48550/ARXIV.2302.05543},
256
+
257
+ url = {https://arxiv.org/abs/2302.05543},
258
+
259
+ author = {Zhang, Lvmin and Agrawala, Maneesh},
260
+
261
+ keywords = {Computer Vision and Pattern Recognition (cs.CV), Artificial Intelligence (cs.AI), Graphics (cs.GR), Human-Computer Interaction (cs.HC), Multimedia (cs.MM), FOS: Computer and information sciences, FOS: Computer and information sciences},
262
+
263
+ title = {Adding Conditional Control to Text-to-Image Diffusion Models},
264
+
265
+ publisher = {arXiv},
266
+
267
+ year = {2023},
268
+
269
+ copyright = {arXiv.org perpetual, non-exclusive license}
270
+ }
271
+ ```
272
+
273
+
274
+ ## Glossary [optional]
275
+
276
+ <!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
277
+
278
+ [More Information Needed]
279
+
280
+ ## More Information
281
 
282
  Special Thank to the great project - [Mikubill' A1111 Webui Plugin](https://github.com/Mikubill/sd-webui-controlnet) !
283
 
 
286
  Thank haofanwang for making [ControlNet-for-Diffusers](https://github.com/haofanwang/ControlNet-for-Diffusers)!
287
 
288
  We also thank all authors for making Controlnet DEMOs, including but not limited to [fffiloni](https://huggingface.co/spaces/fffiloni/ControlNet-Video), [other-model](https://huggingface.co/spaces/hysts/ControlNet-with-other-models), [ThereforeGames](https://github.com/AUTOMATIC1111/stable-diffusion-webui/discussions/7784), [RamAnanth1](https://huggingface.co/spaces/RamAnanth1/ControlNet), etc!
289
+ ## Model Card Authors
290
 
291
+ [Lvmin Zhang](https://huggingface.co/lllyasviel) in collaboration with [Ezi Ozoani](https://huggingface.co/Ezi) and the Hugging Face team.
292
 
293
+ ## Model Card Contact
294
+
295
+ [More Information Needed]
296